UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The impact of consumer information on brand sales : a field experiment with point-of purchase nutritional… Muller, Thomas Edward 1982

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1982_A1 M95.pdf [ 17.01MB ]
Metadata
JSON: 831-1.0095612.json
JSON-LD: 831-1.0095612-ld.json
RDF/XML (Pretty): 831-1.0095612-rdf.xml
RDF/JSON: 831-1.0095612-rdf.json
Turtle: 831-1.0095612-turtle.txt
N-Triples: 831-1.0095612-rdf-ntriples.txt
Original Record: 831-1.0095612-source.json
Full Text
831-1.0095612-fulltext.txt
Citation
831-1.0095612.ris

Full Text

THE IMPACT OF CONSUMER INFORMATION ON BRAND SALES: A FIELD EXPERIMENT WITH POINT-OF-PURCHASE NUTRITIONAL INFORMATION LOAD by THOMAS EDWARD MULLER M.B.A., Simon Fraser U n i v e r s i t y , 1975 L.N.C.R.T., National College o f Rubber Technology, 1960 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY i n THE FACULTY OF GRADUATE STUDIES Faculty of Commerce and Business A d m i n i s t r a t i o n We accept t h i s t h e s i s as conforming to the re q u i r e d standard THE UNIVERSITY OF BRITISH COLUMBIA A p r i l 1982 © Thomas Edward M u l l e r , 1982 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y o f B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and study. I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e copying o f t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the head of my department or by h i s or her r e p r e s e n t a t i v e s . I t i s understood t h a t copying o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be' allowed without my w r i t t e n p e r m i s s i o n . Department of Commerce and Business A d m i n i s t r a t i o n The U n i v e r s i t y of B r i t i s h Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date A p r i l 21, 1982 DE-6 (3/81) ABSTRACT The issue of r e q u i r i n g marketers to d i s c l o s e o b j e c t i v e product performance information to t h e i r customers has presented a problem to both policymakers and researchers. A major concern i s that the p o t e n t i a l usefulness of such information w i l l be negated i f consumers, t r y i n g to evaluate a l t e r n a t i v e products a t the points of s a l e , are hindered by large amounts of such comparative data. Decision-making experiments i n c o g n i t i v e psychology i n d i c a t e t h a t , because of the ca p a c i t y l i m i t a t i o n s of short-term memory, people provided with high input rates of information can experience "information overload," which reduces the q u a l i t y of t h e i r d e c i s i o n s . However, consumer research performed, to date, i n the la b o r a t o r y has f a i l e d to r e s o l v e whether consumers i n a n a t u r a l i s t i c brand-choicemaking s i t u a t i o n would a l s o experience "information overload," i f confronted with l a r g e amounts of product data on which to base t h e i r choices. A f i e l d experiment was performed to extend the f i n d i n g s of t h i s l a b o r a t o r y research stream and to help r e s o l v e the controversy regarding consumer "information overload." A second o b j e c t i v e of t h i s experiment was to c o n t r i b u t e to p o l i c y - o r i e n t e d research on i n f o r m a t i o n - p r o v i s i o n formats. The study examined the behavioural e f f e c t s of d i s p l a y i n g o b j e c t i v e product performance cues a t the poi n t of purchase, e a s i l y a c c e s s i b l e to consumers and organized i n a format a l l o w i n g d i r e c t comparisons of a l t e r n a t i v e brands. An input-output experimental design used p o i n t - o f - s a l e signs to provide d i f f e r e n t amounts (loads) of n u t r i t i o n a l information on the brands of several food products i n two co-operating supermarkets. The outputs, or information e f f e c t s , were measured by c o l l e c t i n g brand-sales data v i a i i i e l e c t r o n i c checkout f a c i l i t i e s to determine whether the information treatments were having the hypothesized e f f e c t s on the shape of the brand-sales d i s t r i b u t i o n . The f i n d i n g s do not appear to support the "information overload" hypothesis. In f a c t , information load d i d not emerge as an explanatory v a r i a b l e . With c e r t a i n products, there i s evidence that p r o v i d i n g n u t r i t i o n a l i n f o r m a t i o n , i n an organized format a t the poin t of s a l e , w i l l lead to brand choices being made on the basis of such data. A l s o , the o v e r a l l response to t h i s data was s i g n i f i c a n t l y weaker i n the second of two weeks during which they were made a v a i l a b l e to shoppers. i v TABLE OF CONTENTS Page ABSTRACT . i i LIST OF TABLES v i i i LIST OF FIGURES x i i ACKNOWLEDGEMENTS x i i i Chapter I. INTRODUCTION AND PROBLEM DESCRIPTION . 1 Int r o d u c t i o n to the Problem 1 D e f i n i t i o n s o f Information Environment Design Factors 6 Type of Information 7 Information Format 7 Amount of Information 9 General O u t l i n e of the Study 12 General L i m i t a t i o n s o f the Study 14 J u s t i f i c a t i o n f o r the Research and P o t e n t i a l C o n t r i b u t i o n s . 15 C o n t r i b u t i o n to Theory 16 C o n t r i b u t i o n to P u b l i c P o l i c y 18 I I . STATEMENT OF RESEARCH HYPOTHESES 21 Research Questions . 21 D e f i n i t i o n s 23 Research Hypotheses 24 I I I . SUMMARY OF RELEVANT RESEARCH 26 Hypothesis-Related Research 26 Hypothesis H. 26 •Short-Term Memory-and Information Overload . . . . 26 Schroder, D r i v e r and S t r e u f e r t 32 Gerald S t i l e s 35 ' Jacoby,-Speller and Kohn 36 ' Debra:Scammon 41 •Roger Best 44 Goodwin and-Etgar 45 Summary 47 V Table of Contents (Continued) Chapter Page Hypothesis H 2 48 Hypothesis H~ 51 Hypothesis ti* 52 Kendall and Fenwick 54 F r e d r i c a Rudell 55 John Quelch 56 Others 58 The Issue of Information Format 59 J . Edward Russo 59 M. Venkatesan . . . 60 Debra Scammon . 60 Summary 61 Hypothesis H 5 61 Other Related Research . 63 Conclusions 63 IV. RESEARCH METHODOLOGY 64 Overview 64 Research Design 66 Rationa l e For Using N u t r i t i o n a l Cues 69 Test Products 71 N u t r i t i o n a l Cue S e l e c t i o n 72 Obtaining Measures of.Cue Importance 79 Survey Questionnaire 79 Survey Sample 81 Survey Results 81 Independent V a r i a b l e s 87 Load 87 Cue Importance 89 Sign Construction 91 Experimental Controls 92 Experimental Store S e l e c t i o n 94 Experimental Procedure 95 Dependent V a r i a b l e 97 Overview 97 R i d i t A n a l y s i s 98 The Mean R i d i t 103 Weighting of Unit Sales 108 Ranking o f Brands i n Other Products 109 Database Manipulation 117 Pooling of Data 120 v i Table of Contents (Continued) Chapter Page General Methodological Approach . . 124 S t a t i s t i c a l Tests Employed 126 ANOVA and t-Tests 126 Mean R i d i t S i g n i f i c a n c e Test 131 Combining Independent S i g n i f i c a n c e Tests 134 V. ANALYSIS AND RESULTS 139 The Data Entering the Analyses 139 Tests For Treatment-Product and Treatment-Week I n t e r -a c t i o n s 140 Tests For E f f e c t s Due to Number of Brands 141 ANOVA of Treatment E f f e c t s 144 Load and Importance . 144 Conclusion 146 Tests o f Hypotheses 149 Tests o f Hypothesis H. 149 A n a l y t i c a l Approach 150 t-Test Results 150 Test f o r Trends 151 Summary and Conclusions 151 Tests o f Hypothesis H 2 154 A n a l y t i c a l Approacn 155 t-Test Results 156 ANOVA Results 158 Summary and Conclusions 158 Tests of Hypothesis H 3 161 A n a l y t i c a l Approach . . . 162 S i g n i f i c a n c e Test Results 163 Summary and Conclusion 163 Tests of Hypothesis H. 165 A n a l y t i c a l Approach 165 In d i v i d u a l and Combined S i g n i f i c a n c e Results . . . . 169 Canned Soup 175 Mayonnaise 175 Ketchup 176 Macaroni & Cheese Dinner 176 Bran Cereal 177 Results Combined by Experimental Week 177 Week 2 178 Week 3 178 Summary and Conclusions 179 Tests of Hypothesis Hg 180 A n a l y t i c a l Approacn 181 Product Interpurchase I n t e r v a l s 183 v i i Table of Contents (Continued) Chapter Page Canned Soup 186 Mayonnaise 190 Ketchup 190 M&C Dinner 195 Bran Cereal 195 Summary and Conclusions 201 A d d i t i o n a l Analyses 202 A n a l y t i c a l Approach 202 A n a l y s i s of Variance Results 203 E f f e c t s Due to Products 203 E f f e c t s Due to Weeks 207 Summary and Conclusions 208 Chapter V Summary 210 VI. SUMMARY AND IMPLICATIONS 212 Summary of the Research Objectives and Procedures . . 212 Summary of the Findings 216 Hypothesis H, 217 Hypothesis W7 218 Hypothesis WZ : 219 Hypothesis W6. 220 Hypothesis H^ 221 A d d i t i o n a l Analyses 222 Imp l i c a t i o n s of the Findings 223 Research L i m i t a t i o n s 226 Data and A n a l y s i s 226 Research Design 227 D i r e c t i o n s f o r Future Research 228 Conclusion 229 BIBLIOGRAPHY 231 APPENDIX A. Replicas of "Sign C o n t r o l s " and Point-of-Purchase Information Signs Constructed f o r Each Treatment/Product 238 B. N u t r i t i v e Composition Data Obtained From Manufacturers . . 252 C. Survey Questionnaire and Store Patronage Responses . . . . 259 D. D a i l y Coding Sheets f o r Item Movement Data and R e p l i c a of A S h e l f Tag 269 E. "Noise" Monitoring Records on Test Products 273 F. Determination o f Ov e r a l l N u t r i t i v e Performance Ranks i n Treatments With Mayonnaise and Ketchup Brands 280 v i i i LIST OF TABLES Table Page 1. Information Load Treatments Used i n Scammon's (1977) Experiment 42 2. Stated Importance Scores of 19 Cereal A t t r i b u t e s . . . . . 50 3. Aggregate Frequencies of Cue Use Leading to a Brand Choice 57 4. Extent to Which Test Product Brands Were Found to Dominate i n N u t r i t i o n a l Performance 74 5. Percent of Respondents Using Each Type of Information at Least Once i n a Food Evaluation and Choice Task 77 6. Results o f Personal Interview Survey t o Measure R e l a t i v e Importance of and D i r e c t i o n o f U t i l i t y f o r N u t r i e n t s . . . 82 7. Cue-Load Levels Used and Results Reported i n Five Consumer Information Load Studies 88 8. Cues (Numbered by Rank Order o f Cue-Importance) U t i l i z e d to Construct the Eight Treatments f o r the 4 x 2 F a c t o r i a l Design ,. 90 9. F a c s i m i l e of an Information Sign Employed f o r M&C Dinner . 100 10. C a l c u l a t i o n of Brand R i d i t s From Reference D i s t r i b u t i o n . 102 11. C a l c u l a t i o n o f the Mean R i d i t f o r the Reference D i s t r i b u t i o n 103 12. C a l c u l a t i o n of the Mean R i d i t f o r a Treatment D i s t r i b u t i o n 104 13. C a l c u l a t i o n of the Mean R i d i t f o r a D i s t r i b u t i o n Changed by One Purchase 106 14. C a l c u l a t i o n o f the Mean R i d i t f o r a D i s t r i b u t i o n Changed by One or Two Purchases 106 15. C a l c u l a t i o n o f the Mean R i d i t F ollowing a R e l a t i v e l y Large Change i n the Sales D i s t r i b u t i o n 107 16. C a l c u l a t i o n of Brand R i d i t s From Reference D i s t r i b u t i o n of Sales Weighted by Package S i z e 110 i x L i s t of Tables (Continued) Table Page 17. C a l c u l a t i o n of the Mean R i d i t f o r the Data of Table 15, Weighted by Package S i z e 110 18. Point-of-Purchase Information Sign Employed f o r Treatment "8/high" (Bran Cereal) ,. 112 19. Brand Performance Ranks (R.) by N u t r i e n t Ratings D i s c l o s e d i n Table 18 (Bran Cereal) . 113 20. Determination o f Ov e r a l l N u t r i t i v e Performance Ranks (R") of Bran Cereal Brands i n Treatments Containing More Than One Cue, Using Table 19 Data 115 21. Chi Square A n a l y s i s of Brand Sales D i s t r i b u t i o n s Observed Under Two Types of Experimental Controls f o r M&C Dinner i n Week 2, Using Unit Sales Weighted by Package S i z e . . 123 22. Database i n Terms of Total Number o f Containers Sold o f Each Product i n Each Week 125 23. Weighted Frequency D i s t r i b u t i o n s f o r T e s t i n g the S i g n i f i c a n c e o f the D i f f e r e n c e Between Two Mean R i d i t s Obtained From Table 16 and 17 Data 135 24. I l l u s t r a t i o n of the Procedure f o r T e s t i n g the S i g n i f i c a n c e o f Combined Treatment Results Using the Data o f Canned Soup i n Week 3 138 25. Two-Way A n a l y s i s o f Variance to Test f o r Treatment-Product I n t e r a c t i o n 142 26. Two-Way A n a l y s i s o f Variance to Test f o r Treatment-Week I n t e r a c t i o n 143 27. Three-Way A n a l y s i s o f Variance to Test f o r E f f e c t s Due to Number of Brands 145 28. Mean R i d i t C e l l Means and Two-Way A n a l y s i s of Variance f o r Load and Cue Importance 147 29. Mean R i d i t C e l l Means and One-Way A n a l y s i s of Variance of Load E f f e c t s and Tests f o r Trends 152 30. Results of t-Tests f o r D i f f e r e n c e Between Means of Various P a i r s o f "High-Importance"/"Low-Importance" Samples 157 X L i s t of Tables (Continued) Table Page 31. Two-Way A n a l y s i s of Variance of Cue-Importance and Product E f f e c t s (Pooled Across Loads of 1, 2 and 4) . . . . 159 32. Two-Way A n a l y s i s of Variance of Cue-Importance and Product E f f e c t s (Pooled Across a l l Loads) 160 33. Results o f z-Tests on Mean R i d i t D i f f e r e n c e s Between Treatments "8/high" and "8/low" f o r Five Products i n Two Weeks - 164 34. Mean R i d i t S i g n i f i c a n c e Results f o r 15 Treatments With Canned Soup and Results of Ov e r a l l S i g n i f i c a n c e Test of Hypothesis H^ 170 35. Mean R i d i t S i g n i f i c a n c e Results f o r 14 Treatments With Mayonnaise and Results of Ov e r a l l S i g n i f i c a n c e Test o f Hypothesis H^ 171 36. Mean R i d i t S i g n i f i c a n c e Results f o r 16 Treatments With Ketchup . and Results o f O v e r a l l S i g n i f i c a n c e Test o f Hypothesis H^ 172 37. Mean R i d i t S i g n i f i c a n c e Results f o r 16 Treatments With M&C Dinner and Results of Ov e r a l l S i g n i f i c a n c e Test of Hypothesis H^ 173 38. Mean R i d i t S i g n i f i c a n c e Results f o r 16 Treatments With Bran Cereal and Results of Ov e r a l l S i g n i f i c a n c e Test of Hypothesis H^ 174 39. Treatments With a Common Brand Ordering by Over a l l N u t r i t i v e Performance f o r Each of Five Products 184 40. Mean Interpurchase I n t e r v a l , i n Weeks, f o r Five Products Based on Consumer Survey Responses •'. . . . v . 185 41. Mean R i d i t s f o r Canned Soup Brand Sales D i s t r i b u t i o n s i n Each o f Four Weeks Compared to an Experimental Week Control 187 42. Mean R i d i t s f o r Mayonnaise Brand Sales D i s t r i b u t i o n s i n Each of Four Weeks Compared to an Experimental Week Control 191 43. Mean R i d i t s f o r Ketchup Brand Sales D i s t r i b u t i o n s i n Each of Four Weeks Compared to an Experimental Week Control . . 193 44. Mean R i d i t s f o r Macaroni & Cheese Dinner Brand Sales D i s t r i b u t i o n s i n Each of Four Weeks Compared to an E x p e r i -mental Week Control 196 x i L i s t of Tables (Continued) Table Page 45. Mean R i d i t s f o r Bran Cereal Brand Sales D i s t r i b u t i o n s i n Each of Four Weeks Compared to an Experimental Week Control 199 46. Mean R i d i t C e l l Means and A n a l y s i s o f Variance of Product and Week E f f e c t s 205 x i i LIST OF FIGURES Figure Page 1. A Summary o f Two Perspectives on Product Information D i s c l o s u r e E f f o r t s 3 2. A Brand x Cue Information Matrix 10 3. An Example o f "Information Overload" 22 4. A Model o f Memory St r u c t u r e and Operation 28 5. Em p i r i c a l R e l a t i o n s h i p Between Information Complexity and Level o f Information Processing 34 6. Database Manipulation and Plan of A n a l y s i s 118 7. D i s t r i b u t i o n o f 77 Mean R i d i t s Obtained From Treatments on Five Products i n Two Experimental Weeks .. 129 8. Load and Cue-Importance E f f e c t s on the Mean R i d i t . . . . 148 9. Mean R i d i t s a t Four Information-Load Levels 153 10. Mean R i d i t s f o r Load and Cue-Importance Treatments With Five Products i n Week 2 167 11. Mean R i d i t s f o r Load and Cue-Importance Treatments With Five Products i n Week 3 168 12. Mean R i d i t s f o r Canned Soup Brand Sales D i s t r i b u t i o n s i n Each of Five Weeks . . 188 13. Mean R i d i t s f o r Mayonnaise Brand Sales D i s t r i b u t i o n s i n Each of Fi v e Weeks .-• 192 14. Mean R i d i t s f o r Ketchup Brand Sales D i s t r i b u t i o n s i n Each of F i v e Weeks 194 15. Mean R i d i t s f o r Macaroni & Cheese Dinner Brand Sales D i s t r i b u t i o n s i n Each o f Five Weeks 197 16. Mean R i d i t s f o r Bran Cereal Brand Sales D i s t r i b u t i o n s i n Each of Fi v e Weeks 200 17. Mean R i d i t (Averaged Over A l l Treatments) P e r t a i n i n g to Each of Five Products and Two Experimental Weeks 204 18. E f f e c t s o f Various U n i t - P r i c e Formats i n Russo's (1977) 20-Week In-Store Experiment 209 ACKNOWLEDGEMENTS The author g r a t e f u l l y acknowledges the h e l p f u l inputs and suggestions of many i n d i v i d u a l s throughout a l l phases of t h i s research. In p a r t i c u l a r , the i n s i g h t s o f Dr. Bruce C. Fauman on problems o f design and data c o l l e c t i o n were very v a l u a b l e . The knowledgeable programming a s s i s t a n c e provided by Mr. Kevin O'Connor of the Concordia U n i v e r s i t y Computer Centre are s i n c e r e l y a p p r e c i a t e d , and my s p e c i a l thanks go to my w i f e , Edith M u l l e r , who helped me immeasurably i n her m u l t i p l e r o l e s of s e c r e t a r y , research a s s i s t a n t , t y p i s t and c r i t i c . F i n a n c i a l support from the Consumer Research Branch of the Canadian Federal Department of Consumer and Corporate A f f a i r s and the cooperation of Canada Safeway, Li m i t e d are a l s o g r a t e -f u l l y acknowledged. F i n a l l y , the author, who i s a recent Canadian by choice r a t h e r than by a c c i d e n t , expresses h i s deep and s i n c e r e g r a t i t u d e to the people o f Canada: a country of p r i c e l e s s o p p o r t u n i t y , t o l e r a n c e and u n i f i e d d iverseness. I t i s to them t h a t t h i s d i s s e r t a t i o n i s dedicated. 1 CHAPTER I INTRODUCTION AND PROBLEM DESCRIPTION This d i s s e r t a t i o n research represents an attempt to t e s t a hypothesis from human information processing theory v i a a f i e l d experiment. The purpose of t h i s study i s to extend the f i n d i n g s of past l a b o r a t o r y research on consumer information load i n order to gain a be t t e r under-standing of how shoppers respond to d i f f e r e n t amounts o f o b j e c t i v e product performance data provided at the poi n t of s a l e . I n t r o d u c t i o n to the Problem The present and f u t u r e a v a i l a b i l i t y of standardized product information to consumers i s a t o p i c o f p i v o t a l i n t e r e s t and concern to p u b l i c policymakers ( c f . Beales et a l . , 1981; Hutton, McNeill and W i l k i e , 1978; M i l l e r , 1977; Montgomery, 1977), consumer advocates ( c f . Peterson et a l . , 1978; St. Marie, 1978), marketers and academics ( c f . Claxton and Anderson, 1980; R u d e l l , 1979; McEwen, 1978). T r a d i t i o n a l l y , the type of product information provided to consumers has been marketer-dominated, persuasive information which i s s u b j e c t i v e and non-neutral (e.g. a d v e r t i s i n g and sal e s promotion). In recent y e a r s , the evolvement of consumerism has l e d to a spreading demand f o r more o b j e c t i v e f a c t s about products and comparative product performance data (e.g. f u e l consumption r a t i n g s on automobiles; t a r and n i c o t i n e content i n c i g a r e t t e s ; energy consumption-data on a p p l i a n c e s ) . As a r e s u l t , p u b l i c policymakers are under pressure to increase the number of information d i s c l o s u r e programs which r e q u i r e marketers to provide t h i s kind of data to buyers i n the marketplace (Day, 1976). 2 Consider the information d i s c l o s u r e programs which have p r o l i f e r a t e d i n the market f o r food products. Programs that have been studied and reported on include n u t r i e n t and i n g r e d i e n t l a b e l l i n g on foods (Martinsen and McCuTlough, 1977; Daly, 1976:; L i e f e l d , 1976; Gorman, 1975; Lenahan e t a l . , 1973), open dating on food.,packages (Nayak and Rosenberg, 1975; Friedman, 1972) u n i t p r i c i n g i n the store (Russo, 1977; Russo, K r i e s e r and M i y a s h i t a , 1975), and n u t r i t i o n a l information d i s c l o s u r e i n food a d v e r t i s i n g ( R u d e l l , 1979; Venkatesan, 1977; Scammon, 1977). The o b j e c t i v e o f these studies was to gauge the e f f e c t i v e n e s s and/ or usefulness to the end user o f e x i s t i n g or proposed information programs. The f o r c e behind these s t u d i e s appears to be the sharp debate about whether or not these information d i s c l o s u r e programs should be continued and expanded. In ge n e r a l , two major views have emerged from t h i s debate. These have been summarized i n Figure 1. On one s i d e , are the proponents of a "the more i n f o r m a t i o n , the b e t t e r " philosophy — on the grounds that consumers who are armed with complete, o b j e c t i v e l y described market information are b e t t e r able to make sound choices. This t y p i f i e s the consumerist view. The demand f o r greater amounts of such information i s supported by the argument th a t the consumer has the " r i g h t to know" (e.g. Bymers, 1972). I t i s al s o f e l t t h a t complete and f a c t u a l information d i s c l o s u r e to the consuming p u b l i c makes i n d u s t r y more accountable f o r i t s a c t i o n s and st i m u l a t e s competition ( c f . Lenahan et a l . , 1973). A second view about p r o v i d i n g product information to consumers i s consider a b l y more reserved. I t t y p i f i e s the perspective of academic researc h e r s , marketers and p u b l i c policymakers e v a l u a t i n g proposals f o r 3 Figure 1 A Summary o f Two Perspectives On Product Information D i s c l o s u r e E f f o r t s PUBLIC DEBATE ON CONSUMER INFORMATION PROVISION IN THE MARKETPLACE BETTER CONSUMPTION DECISIONS THE "RIGHT TO KNOW" MAKES INDUSTRY MORE ACCOUNTABLE STIMULATES COMPETITION A MORE CAUTIOUS VIEW ASSESSMENT OF CONSUMER... . DESIRE; . UNDERSTANDING, AND . USE LEVEL OF CONSUMER EDUCATION COST OF PROVIDING INFORMATION HUMAN INFORMATION-PROCESSING LIMITATIONS 4 consumer information p r o v i s i o n ( R u d e l l , 1979; Wall S t r e e t J o u r n a l , 1978). A review of the l i t e r a t u r e e s t a b l i s h e s two c l e a r l y defined arguments put forward by t h i s group. F i r s t , the purpose served by these information programs needs to be assessed by determining whether consumers d e s i r e the i n f o r m a t i o n , understand the information provided, and are a c t u a l l y motivated to use i t i n decision-making ( c f . Jacoby, Chestnut and Silberman, 1977). Tied i n with these questions i s the i n t r i c a t e i s s u e of a corresponding need f o r consumer education. Moreover, i n the present era of government budget cuts and a d m i n i s t r a t i v e d e r e g u l a t i o n , p u b l i c p o l i c y programs to inform the consumer are under p a r t i c u l a r pressure to j u s t i f y t h e i r costs (Byron, 1981). Second, the usefulness of e v e r - i n c r e a s i n g product information f o r decision-making w i l l be c u r t a i l e d by the l i m i t e d c a p a c i t y of humans to process the a v a i l a b l e data i n the time h a b i t u a l l y a l l o c a t e d to such decision-making. This l i m i t a t i o n stems from the c a p a c i t y c o n s t r a i n t s of human short-term memory ( c f . Simon, 1974; M i l l e r , 1956). In conceptual terms, short-term memory serves as a temporary storage b u f f e r between a person's sensory receptors of incoming s t i m u l i .(such as product performance data on the package of a supermarket product) and the person's permanent or long-term memory (Bettman, 1979: 140; W i l k i e , 1975: 42). Laboratory research i n the behavioural sciences c o n s i s t e n t l y i n d i c a t e s t h a t the r e l a t i v e l y l i m i t e d c a p a c i t y of human short-term memory acts as a "bottleneck" between sensory receptors and long-term memory. Thus, the f i x e d c a p a c i t y of the short-term memory b u f f e r r e s t r i c t s the rate of information processing. This i m p l i e s t h a t excessive product 5 information inputs during a given time period can lead to confusion and i n f e r i o r decision-making -- a c o n d i t i o n commonly r e f e r r e d to as "information overload". In s h o r t , on the issue of consumer d e s i r e f o r , understanding o f and use of market i n f o r m a t i o n , researchers appear to favour a continued monitoring of the e f f e c t i v e n e s s of e x i s t i n g information d i s c l o s u r e programs, before- f u r t h e r such programs are developed and implemented ( c f . Friedman, 1977; Jacoby, Chestnut and Silberman, 1977; Day, 1976). In terms o f c e r t a i n c a p a c i t y l i m i t a t i o n s i n human information p r o c e s s i n g , the concern i s that a l a r g e volume of product data made a v a i l a b l e to consumers w i l l overtax t h i s c a p a c i t y and negate the p o t e n t i a l usefulness of an information d i s c l o s u r e program. Nonetheless, even among consumer researchers a controversy c u r r e n t l y e x i s t s as to whether a c o n d i t i o n o f excessive product information can be r e a l i s t i c a l l y induced such that consumer decision-making i n , say, a r e t a i l s t o r e w i l l be impaired as a r e s u l t . "Information overload" f i n d i n g s i n the l a b o r a t o r y are i n c o n f l i c t with the r e s u l t s obtained i n more r e a l i s t i c consumer s e t t i n g s ( c f . Jacoby, S p e l l e r and Kohn, 1974 a; b; Scammon, 1977). I f consumers' processing c a p a c i t i e s are l i m i t e d , then l a r g e amounts of product information d i s c l o s e d i n a market w i l l lead to suboptimal c h o i c e s ; but c o n c l u s i v e proof of t h i s notion i s l a c k i n g . Day (1976: 48) sums up the current s i t u a t i o n : . . . However, there i s as yet no c o n c l u s i v e evidence that buyers are e a s i l y overloaded with information which reduces t h e i r decision-making a b i l i t y . The absence of c l e a r - c u t r e s u l t s i s a product o f : (1) methodological problems i n the reported experimental s t u d i e s , (2) the use o f too rigorous a c r i t e r i o n of choice accuracy (choice 6 behavior i s more l i k e l y to r e f l e c t s a t i s f i c i n g than o p t i m i z i n g b e h a v i o r ) , (3) the f a c t that the basic information overload premise derives from t h e o r i e s o f short-term memory, and (4) the r e l i a n c e on l a b o r a t o r y experiments i n constrained time and exposure s e t t i n g s . Thus, the question remains whether the human information processing l i m i t a t i o n s documented i n the behavioural sciences w i l l deter consumers from e f f e c t i v e l y using l a r g e amounts of product information i n a r e a l -l i f e s e t t i n g . Part of the controversy surrounding t h i s question stems from the accumulation of c o n f l i c t i n g answers, and part from the lack of n a t u r a l i s t i c f i e l d s t u d i e s . The present research e f f o r t was undertaken to resolve some of these outstanding i s s u e s . While i t may not f u l l y answer the above q u e s t i o n , i t i s towards t h i s goal t h a t the d i s s e r t a t i o n i s d i r e c t e d . D e f i n i t i o n s of Information Environment^Design Factors Before attempting a f u l l e r d e s c r i p t i o n of the research problem, i t i s e s s e n t i a l to introduce b r i e f d e f i n i t i o n s o f f a c t o r s r e l a t e d to the design of product information environments. Bettman (1975) defines a consumer information environment as the e n t i r e array of pro d u c t - r e l a t e d data a v a i l a b l e to the consumer f o r purchase decision-making. Every information environment can be c h a r a c t e r i z e d by three major design f a c t o r s . The three f a c t o r s are type of information provided, format i n which the information i s s t r u c t u r e d and presented, and amount o f information provided. These design f a c t o r s are considered important i n consumer information processing research because they are thought to a f f e c t the p r o c e s s a b i l i t y and p o t e n t i a l usefulness to consumers of p r o d u c t - r e l a t e d data. 7 Type of Information. W i l k i e (1975:52) points out that the funda-mental purpose of consumer information p r o v i s i o n , from a p u b l i c p o l i c y p e r s p e c t i v e , i s to enable consumers to evaluate the " q u a l i t y " of products i n brand choice d e c i s i o n s . Consequently, p o l i c y - o r i e n t e d information environments c o n s i s t of o b j e c t i v e and/or test e d product information cues (e.g. brand performance r a t i n g s , standardized analyses) as opposed to the type of information o r i g i n a t i n g from commercial or buyer sources (e.g. sales promotional, word-of-mouth, use experience). Recent examples of usage and e f f e c t i v e n e s s research on the o b j e c t i v e type of information are energy consumption r a t i n g s of major appliance brands (Claxton and Anderson, 1980); comparative n u t r i e n t l e v e l s i n food product brands ( R u d e l l , 1979); i n g r e d i e n t information and open dating information on grocery foods (Kendall and Fenwick, 1979); Consumer Reports r a t i n g s on breakfast c e r e a l s ( S t a n l e y , 1977). This d i s s e r t a t i o n examines some unresolved t h e o r e t i c a l and e m p i r i c a l questions about consumer processing of o b j e c t i v e types of market i n f o r m a t i o n . Research r e l e v a n t to the problem which employed t h i s type of information i s reviewed i n Chapter I I I . Information Format. This design f a c t o r r e l a t e s to the manner i n which product information data are presented and p h y s i c a l l y organized w i t h i n an information environment. Day (1976) has noted that unless o b j e c t i v e product information i s 1. E a s i l y a c c e s s i b l e at the point of purchase, 2. Can be understood by the consumer, and 3. Is p h y s i c a l l y d i s p l a y e d i n a form which allows d i r e c t comparisons of a l t e r n a t i v e brands, 8 the information w i l l generally. not enter i n t o the consumer's choice process. Manipulating information format can mean s i m p l i f y i n g or "pre-processing" o b j e c t i v e brand data so as to make them more understandable to consumers or e a s i e r to store i n short-term memory. For example, brand r a t i n g s expressed i n a d j e c t i v a l form (good, f a i r , e x c e l l e n t ) have been found to be more e f f e c t i v e i n communicating brand performance than r a t i n g s comprised of numbers (see Scammon, 1977). Format design a l s o subsumes any arrangement o f such data to brin g them i n t o c l o s e p r o x i m i t y , thereby l e s s e n i n g the burden on a consumer's short-term memory when brand comparisons need to be made. An example i s the l i s t i n g o f brands and container s i z e s i n order o f i n c r e a s i n g p r i c e - p e r - u n i t on a s i n g l e r e t a i l sign ( c f . Russo, 1977). In summary, the demands on short-term memory during an in f o r m a t i o n -gathering and processing task are much greater when brand performance data on several cues ( a t t r i b u t e s ) must f i r s t be l o c a t e d , then i n t e r p r e t e d i n t o common terms, and then mentally organized. Such would be the case i n a grocery shopping s i t u a t i o n where a consumer i s comparing the per-formances o f several brands by searching f o r n u t r i t i o n a l information on the i n d i v i d u a l packages. An appropriate change i n format, e.g. by t a b l i n g the same n u t r i t i o n a l data as a s i n g l e brand-by-cue matrix (see Figure 2) and posting i t on a sign at the point of purchase, would reduce the p o t e n t i a l burden on short-term memory. This would enhance a consumer's a b i l i t y and motivation to process the a v a i l a b l e data. 9 Amount of Inforriiation. I f , as mentioned e a r l i e r , information processing i s constrained by the short-term memory's c a p a c i t y to deal with a f i x e d amount of information per u n i t time, how should one define or measure the amount of product information provided to consumers? The choice of an appropriate d e f i n i t i o n f o r amount of information was c a r e f u l l y considered i n t h i s d i s s e r t a t i o n . Several consumer researchers have used the term "information load" to denote the amount of information provided i n a brand e v a l u a t i o n and choice task (Goodwin and Etgar, 1980; Best, 1978; Scammon, 1977; Jacoby, S p e l l e r and Kohn, 1974 a;b). In.keeping with t h i s p r a c t i c e , the term "information l o a d " or j u s t "load" w i l l be employed henceforth to denote amount of information provided to consumers. The term "information l o a d " i s p a r t i c u l a r l y apt because i t recognizes the f a c t that a load or burden i s being placed on the processing c a p a c i t y o f short-term memory, much l i k e a load of e l e c t r i c current might be placed on a cir c u i t . ' ' ' Since the type of information d e a l t with i n t h i s research i s o b j e c t i v e information i n the form of performance r a t i n g s on a l t e r n a t i v e brands, load has to be defined and o p e r a t i o n a l i z e d i n that context. Suppose that an information environment i s designed i n the format of a brand-by-cue information m a t r i x , as i n Figure 2. Three brands of a product are l i s t e d , along with f o u r d i f f e r e n t information cues. The In c o g n i t i v e psychology, the term "information l o a d " has been o p e r a t i o n a l i z e d as the number of information i n p u t s , per time p e r i o d , i n an information processing task ( c f . Schroder, D r i v e r and S t r e u f e r t , 1967). 10 matrix e n t r i e s (A^, B^, etc.) represent the performance r a t i n g s of each 2 brand on each a t t r i b u t e or cue. Figure 2 A Brand x Cue Information Matrix Information Cue Brand 1 2 •; 3 4 A A i A 2 A 3 A* B: Bi B 2 B 3 C C i C 2 c 3 Ck Since the matrix can be expanded or c o n t r a c t e d , depending on how many brands or cues are i n c l u d e d , i t becomes apparent t h a t there are two sources of load i n such a matrix. More information can come from i n c l u d i n g more brands or from a greater number of cues. I n t u i t i o n would suggest th a t information load be defined and measured as the t o t a l number of r a t i n g s i n the m a t r i x , i . e . , the number of brands, m u l t i p l i e d by the number of cues on which information has been provided. In the case of Figure 2, t h i s would be a load of 3 x 4 = 12 r a t i n g s . This d e f i n i t i o n i s acceptable as long as one of these sources of load i s held constant while the e f f e c t s of varying the other source This method of c o n c e p t u a l i z i n g an information environment i s presented i n w i l k i e and F a r r i s (1976) and Chaffee and McLeod (1973). n are s t u d i e d . However, when both sources o f load vary simultaneously t h i s d e f i n i t i o n i s conceptually problematic; i t t r e a t s the two load sources as e q u i v a l e n t , when i n f a c t they may d i f f e r p s y c h o l o g i c a l l y . A measure of load based on the a r i t h m e t i c a l product of brands and cues makes no d i s t i n c t i o n , f o r example, between a matrix with 3 brands and 4 cues, and one with 4 brands and 3 cues or 6 brands and 2 cues --a l l o f which contain a t o t a l of 12 r a t i n g s . Because consumers have been found to possess a r e p e r t o i r e of processing r u l e s and s i m p l i f y i n g h e u r i s t i c s which they match to the information processing task at hand (see Bettman and Kakkar, 1977; Payne, 1976; Russo and Rosen, 1975; Tversky, 1972), i t i s u n l i k e l y t h a t two matrices c o n s i s t i n g of 3 brands x 4 cues and 6 brands x 2 cues make the same demands on short-term memory ( W i l k i e , 1975:37). In s h o r t , one source of load does not compensate f o r another source because the two load sources i n such a matrix are not p s y c h o l o g i c a l l y e q u i v a l e n t to consumers (Russo, 1974). Information load i n the context of a brand-by-cue matrix should be defined as e i t h e r the number of cues i n the matrix or the number of brands, while holding the other load source constant. Because the present research focuses on the d i s c l o s u r e of product information on a l l a v a i l a b l e brands i n a r e t a i l shopping environment, load i s more a p p r o p r i a t e l y defined as the number of information cues presented i n the m a t r i x , while holding the number of brands constant. In summary, information load i n t h i s d i s s e r t a t i o n i s defined as the number of cues included i n a brand-by-cue m a t r i x ; load i s experiment-a l l y manipulated by varying the number of cues on which brand information 12 i s d i s c l o s e d i n such a matrix. Although the number of brands a l s o v a r i e d i n the matrices because d i f f e r e n t t e s t products were used, t h i s source o f load was c o n t r o l l e d f o r i n the a n a l y s i s o f cue-load e f f e c t s . General O u t l i n e o f the Study The purpose of t h i s research i s to extend the f i n d i n g s of past l a b o r a t o r y research on information load. The i n t e n t i s to gain a b e t t e r understanding o f how consumers react to d i f f e r e n t amounts of a s p e c i f i c type of o b j e c t i v e product information when i t i s posted i n a p a r t i c u l a r format at the point of s a l e . This study represents a f i e l d experiment to t e s t a hypothesis from human information processing theory. Laboratory t e s t s with information load i n d i c a t e d t h a t people can only a s s i m i l a t e a l i m i t e d amount of information f o r decision-making i n a given time p e r i o d . Exposure to too high an information load leads to l e s s e f f e c t i v e information usage. The question i s , can the phenomenon of "information overload", r e p o r t e d l y observed i n past l a b o r a t o r y s t u d i e s , be induced with r e l a t i v e l y high loads of product inform a t i o n cues placed i n a r e a l i s t i c shopping environ-ment? Like many of the l a b o r a t o r y experiments with information l o a d , t h i s f i e l d experiment uses an input-output (stimulus-response) research design. Information inputs or s t i m u l i were v a r i e d by changing the number of o b j e c t i v e product performance cues presented to shoppers i n a brand-by-cue matrix format. The outputs o f , or response t o , each input were measured uno b t r u s i v e l y by t r a c k i n g brand sales i n the s t o r e . This study makes an e f f o r t to b u i l d on previous s t u d i e s of 13 consumer information load and to introduce improvements over the experimental designs of e a r l i e r such research. The present design b e n e f i t t e d from an understanding of the methodological d e f i c i e n c i e s of e a r l y information load experiments, as pointed out by other scholars i n t h i s area. These design improvements are d e t a i l e d i n Chapter IV. An experiment was designed around the d i s c l o s u r e of d i f f e r e n t amounts of n u t r i t i o n a l information on the brands of s i x non-staple food products i n s i d e two co-operating supermarkets. The primary reason f o r choosing i n d i v i d u a l n u t r i e n t s (e.g. p r o t e i n , sodium, vitamin n i a c i n ) as information cues was that such information i s t o p i c a l , a v a i l a b l e from manufacturers, and h i g h l y amenable to experimental manipulation f o r the present design. The n u t r i t i o n a l cues s e l e c t e d f o r each t e s t product were organized i n a brand-by-cue matrix format and d i s p l a y e d on signs w i t h i n the s t o r e s , at the product s h e l v i n g . Shoppers purchasing any of these products could evaluate the brand performance data before making a brand choice. The experimental design focused on information load as the main 3 f a c t o r . Load was manipulated by varying the number of n u t r i t i o n a l cues included i n a product's brand-by-cue information s i g n . S i x products increase the g e n e r a l i z a b i l i t y of the f i n d i n g s and serve as r e p l i c a t i o n s of the chosen load manipulations. Information treatment e f f e c t s were measured by c o l l e c t i n g brand sales data v i a computerized checkout f a c i l i t i e s i n the s t o r e s . The data Cue importance was a secondary design f a c t o r and i s discussed i n d e t a i l i n Chapter IV. 14 were subsequently analyzed with a s t a t i s t i c a l technique known as r i d i t a n a l y s i s which i s uniquely s u i t e d to reveal any d i r e c t i o n a l changes i n the sales d i s t r i b u t i o n s o f the brands of a product. The r e s u l t s from r i d i t a n a l y s i s , used i n conjunction with more conventional s t a t i s t i c a l techniques, gave an i n d i c a t i o n of whether the information signs had the hypothesized e f f e c t s on purchase behaviour. The experiment extended over f i v e consecutive weeks, which included "before" and " a f t e r " brand share measurements, to e s t a b l i s h a causal r e l a t i o n s h i p between the posted information and the expected changes i n brand s a l e s . The r e s u l t s of t h i s study are discussed i n d e t a i l i n Chapter V. P r i o r to the i n - s t o r e experiment, an independent f i e l d survey was conducted among consumers of the s e l e c t e d t e s t products. Measurements were obtained on consumers' sta t e d importance of each i n d i v i d u a l n u t r i t i o n a l cue so th a t t h i s f a c t o r could be incorporated i n the stimulus c o n s t r u c t i o n s (the information s i g n s ) . The d i r e c t i o n of consumers' u t i l i t i e s f o r each i n d i v i d u a l n u t r i e n t was al s o confirmed i n t h i s survey, .since t h i s knowledge would be needed f o r the s t a t i s t i c a l analyses of the i n - s t o r e data. General L i m i t a t i o n s of the Study This study was c o n s c i o u s l y r e s t r i c t e d to examining the short-term e f f e c t s of product information loads on consumers. Because the major t h r u s t of the experiment was to t e s t a hypothesis from human information p r o c e s s i n g , d e a l i n g with short-term memory, i t d e a l t with inputs and t h e i r corresponding responses at a s i n g l e point i n time. A c c o r d i n g l y , no inference can be made about the long-term e f f e c t s of pr o v i d i n g various 15 amounts of information to consumers at the point of sale. To repeat, the emphasis was on a single exposure and the immediate market response resulting from i t . The findings of this study are also limited to a specif ic type of information (nutritional values of foods) and a specif ic type of product (frequently purchased, non-durables). Even so, no conclusions can be drawn about the extended effects of providing this type of information to food shoppers. Only, a single format design was tested in this experiment, thus no comparisons could be made with alternate ways of presenting such information at the point of sale. Since the study focused on behavioural impacts of the information provided, the lack of any assessment of possible intermediate effects (e.g. perceptions, changes in cognitions, attitudes or behavioural i n -tentions) is a further l imitation of this research. By nature of this input-output design, nothing can be said about any intervening cognitive process that took place between stimulus exposure and behavioural response. These interim processes are treated as a "black box." F inal ly , the generalizabi1ity of the findings might be restricted by the use of only two stores of a specif ic supermarket chain in a single c i ty . Just i f icat ion for the Research and Potential Contributions The problem described at the beginning of this chapter indicated that among consumer researchers the question remains as to whether the capacity l imitations of short-term memory could lead to "information overload" in a natural ist ic information environment. 16 C o n t r i b u t i o n to Theory. Given the published studies to date, only f i n d i n g s from l a b o r a t o r y experiments or f i e l d surveys are a v a i l a b l e to i n f e r the behavioural e f f e c t s o f information loads i n the marketplace. Missing i n t h i s research stream are c o n t r i b u t i o n s from experiments con-ducted under c o n d i t i o n s i n v o l v i n g actual brand information d i s c l o s e d to consumers i n the marketplace and actual purchases. Therefore, f i e l d experiments on consumer information load represent the next l o g i c a l step i n the research stream.begun i n the l a b o r a t o r y by Jacoby and h i s colleagues (Jacoby, S p e l l e r and Kohn, 1974 a;b). The Jacoby s t u d i e s , as wel l as one or two l a t e r l a b o r a t o r y i n v e s t i g a t i o n s of consumer information l o a d , seem to confirm what c o g n i t i v e p s y c h o l o g i s t s had found i n l a b o r a t o r y experiments with human information processing t a s k s . Excessive information loads cause the q u a l i t y of consumer d e c i s i o n s to d e t e r i o r a t e because the excess i n f o r m a t i o n cannot be processed and only serves to hinder or c l u t t e r the processing task. What has not been s a t i s f a c t o r i l y r e s o l v e d , to date, i s the problem suggested by various other researchers that t h i s f i n d i n g might be an a r t i f a c t of the la b o r a t o r y environment. I t i s argued that " r e a l world" information en-vironments do not resemble the c o n d i t i o n s created i n the l a b o r a t o r y . Therefore, q u i t e d i f f e r e n t r e s u l t s might be obtained i n natural processing and choice-making s i t u a t i o n s . R e f e r r i n g to one of the l a b o r a t o r y experiments on information load using package information f o r food products (Jacoby, S p e l l e r and Kohn, 1974 a ) , Russo (1974:71,72) emphasizes, the tenuousness of e x t r a -p o l a t i n g from these l a b o r a t o r y f i n d i n g s to the re a l s i t u a t i o n : 17 The experimental s i t u a t i o n of Jacoby et a l . i s c r i t i c a l l y d i f f e r e n t from most shopping s i t u a t i o n s . For how many supermarket purchases with 12 a v a i l a b l e brands would a t y p i c a l consumer spend 5 minutes making a choice? . . . In s h o r t , there was no detrimental e f f e c t o f information overload because the subjects took enough time to process the presented i n f o r m a t i o n . I t remains an open e m p i r i c a l question whether a performance decrement would occur i n a t y p i c a l shopping s i t u a t i o n . S i m i l a r research i n the area o f buyer/consumer information proces-sing reveals the very d i f f e r e n t r e s u l t s and conclusions t h a t can emanate from l a b o r a t o r y studies and f i e l d s t u d i e s of the same phenomenon. S t i l e s (1974:126) i n v e s t i g a t e d the information processing a c t i v i t i e s o f s i x t y i n d u s t r i a l buyers i n a f i e l d study of the "information overload" hypo-t h e s i s and concluded: Therefore, using data obtained from c o n d i t i o n s that are more r e a l i s t i c than those simulated i n past l a b o r a t o r y s t u d i e s of information-processing behavior, the present study f u r n i s h e s evidence denying the notion t h a t people often work under c o n d i t i o n s of information overload --c o n d i t i o n s under which they are unable to f i n d ways of managing the load or escape from i t . The study does not say, however, that one cannot create c o n d i t i o n s of com-p l e x i t y overload or that one cannot induce people to work under such c o n d i t i o n s . I t does say t h a t such s i t u a t i o n s do not tend to occur i n the r e a l world. Day (1976:51) reviewed the problems and defects of past studies designed to assess the e f f e c t s of consumer information d i s c l o s u r e environments and suggested some o p p o r t u n i t i e s and d i r e c t i o n s f o r f u t u r e research i n t h i s area: 1. L i t t l e i s known about information-processing behavior: how i t changes over time, how much information can be absorbed (both i n q u a n t i t y and q u a l i t y ) , and how such behavior d i f f e r s among segments. 2. There i s a pressing need f o r f i e l d s t udies rather than l a b o r a t o r y s t u d i e s , . . . (which) focus s p e c i f i c a l l y on actual purchase behavior. . . 18 With s p e c i f i c regard to f u t u r e s t u d i e s of the "information over-load" concept, Scammon (1977:154) recommends greater a t t e n t i o n to the task environment: . . . Further p u r s u i t of t h i s l i n e of research --i n v e s t i g a t i n g the impact of task p r o p e r t i e s on consumers' information processing a b i l i t y -- would suggest studies varying the a v a i l a b l e processing time and/or the number of dimensions of information presented .... Undoubtedly, a strong concern i s c u r r e n t l y being voiced that not enough research has been done on information load i n a r e a l i s t i c consumer environment. A f i e l d experiment designed along the l i n e s of previous l a b o r a t o r y work with information l o a d , but made as unobtrusive as p o s s i b l e , would c o n t r i b u t e to consumer information processing theory. Moving t h i s research from the l a b o r a t o r y i n t o a f i e l d s e t t i n g would serve to develop e m p i r i c a l g e n e r a l i z a t i o n s about consumer responses at various information loads. The present study i s a step towards that g o a l . The i n - s t o r e experiment reported here i s seen as an extension o f e a r l i e r work to t e s t a t h e o r e t i c a l model of consumer information processing. As Doyle and Gidengil (1977:59) have noted, too oft e n i n the past marketing experiments have been i s o l a t e d e f f o r t s which do not lead to e m p i r i c a l g e n e r a l i z a t i o n s : At the t h e o r e t i c a l l e v e l , progress has been hindered by an i n a b i l i t y to b u i l d on previous r e s u l t s and to r e p l i c a t e and g e n e r a l i z e f i n d i n g s across d i v e r s e areas and products. Without such an approach, e m p i r i c a l g e n e r a l i z a t i o n s and theory cannot advance. C o n t r i b u t i o n to P u b l i c P o l i c y . In a minor sense, the r e s u l t s of t h i s research could have some u t i l i t y to those with a p u b l i c p o l i c y o r i e n t a t i o n towards the p r o v i s i o n of n u t r i t i o n a l information on grocery food products. 19 At the present time, Canadian f e d e r a l r e g u l a t i o n s s t i p u l a t e that n u t r i e n t information (e.g. "Iron 5.5 mg per 100g") be provided to consumers on the packages of c e r t a i n non-staple foods (e.g. breakfast c e r e a l s , alimentary pastes) which have been f o r t i f i e d with vitamins and minerals (Health and Welfare Canada, 1977:20). There does not appear to be any published research on the immediate e f f e c t s of posting t h i s same package information on p o i n t - o f - s a l e signs to al l o w d i r e c t n u t r i t i o n a l comparisons among a l l a v a i l a b l e brands o f the product. Given t h a t the n u t r i t i o n a l information used i n t h i s experiment i s of the type described above, the e f f e c t o f presenting i t i n a brand-by-cue format w i l l be measured i n terms of actual brand s a l e s . Jacoby, Chestnut and Silberman (1977:126) urge that past surveys r e v e a l i n g the extent of consumers' d e s i r e f o r , understanding.and claimed use of n u t r i t i o n a l information be supplemented with behavioural measures of the impact of such information on the p u b l i c : Of course, knowledge and usage are but two of many d i f f e r e n t c r i t e r i a which could be used to assess the e f f e c t i v e n e s s of n u t r i t i o n information p r o v i s i o n . . . A d d i t i o n a l research i n v o l v i n g these and other r e l a t e d c r i t e r i a needs to be conducted before any d e f i n i t i v e statement regarding the impact of n u t r i t i o n information on consumer d e c i s i o n making and well-being can be made. In terms of an information format f o r the d i s c l o s u r e of comparative n u t r i t i o n a l data on branded food products i n the supermarket, the r e s u l t s of t h i s study would c o n t r i b u t e to the small but growing body of e x p e r i -mental f i n d i n g s on n u t r i t i o n a l formats i n the s t o r e ( c f . Russo, undated). Day (1976:48) sums up the format question i n terms of p u b l i c p o l i c y e f f o r t s to gauge the usefulness o f information d i s c l o s u r e programs: . . . U n t i l the best p o s s i b l e form of presentation i s used, i t i s premature to conclude t h a t a d i s c l o s u r e requirement i s i n e f f e c t i v e . Echoing t h i s sentiment i s Friedman's (1977:81) conclusion a f t e r reviewing a f i v e - y e a r span of research on consumer use of product information i n the supermarket: To sum up, while there i s l i t t l e research evidence that l a r g e numbers of consumers are making r e g u l a r use of i n f o r m a t i o n a l aids i n r e t a i l food markets, i t would seem imprudent to jump from these f i n d i n g s to a p o l i c y d e c i s i o n of d i s c o n t i n u i n g aids i n supermarkets. Less c l e a r how-ever are the i m p l i c a t i o n s i n the other d i r e c t i o n . In p a r t i c u l a r , i s there a need f o r more in f o r m a t i o n a l aids i n supermarkets to help consumers get more value f o r t h e i r money? 21 CHAPTER I I STATEMENT OF RESEARCH HYPOTHESES The previous chapter described the current gap i n research on consumer information load and h i g h l i g h t e d the problem area addressed by t h i s d i s s e r t a t i o n . This b r i e f chapter i s a d i s t i l l a t i o n of s p e c i f i c research questions which were chosen f o r i n v e s t i g a t i o n and which are re s t a t e d as t e s t a b l e hypotheses. Chapter I I I reviews the l i t e r a t u r e found r e l e v a n t to each of the research hypotheses. Research Questions The problem i n t r o d u c t i o n i n Chapter I discussed the groundwork already l a i d i n the research t r a d i t i o n on consumer information load. The overview d i r e c t l y suggested some research questions which would represent the next l o g i c a l e m p i r i c a l step i n the study of consumer information load. Therefore, the present author has chosen to i n v e s t i g a t e research questions which are intended to c l a r i f y the r e l a t i o n s h i p between consumer information loads and consumer behaviour. I t i s hoped that the f i n d i n g s on these research questions w i l l make a d i r e c t c o n t r i b u t i o n to that body o f knowledge. The research questions are enumerated below. 1. Does market response to information input loads reach a maximum at lower loads and d i m i n i s h at the highest l o a d , as a r e s u l t o f "information overload?" The questioned r e l a t i o n s h i p between information load and market response i s i l l u s t r a t e d i n Figure 3. In other words, can an in v e r t e d U-shaped r e l a t i o n s h i p between load and market response be found, as predicted by a theory from human information processing? 22 Figure 3 An Example of "Information Overload" 4 Max LxJ 00 I— ^ U J o ^ O -CtL to <C U J 21 Qi 2. I f an information input load i s constructed from n u t r i t i o n a l cues which are r e l a t i v e l y important (based on stated importance measures obtained from consumers), w i l l the market response be greater than the response to a s i m i l a r load comprised of l e s s important n u t r i t i o n a l cues? Is there a response i n t e r a c t i o n between input load and n u t r i t i o n a l cue importance? 3. W i l l a r e v e r s a l i n the l e f t - t o - r i g h t arrangement of cues l i s t e d i n a brand-by-cue information input a f f e c t consumer response? 4. W i l l consumers base t h e i r brand choices on the n u t r i t i o n a l information now p r i n t e d on some food packages, i f that information i s posted i n a format which f a c i l i t a t e s d i r e c t brand comparisons? 5. Does removal of the point-of-purchase information from the stores i n the post-experimental weeks c o i n c i d e with the r e t u r n of a product's brand s a l e s d i s t r i b u t i o n to the pre-experimental shape? Can a c a u s a l i t y between information and market response be e s t a b l i s h e d ? 23 D e f i n i t i o n s Since these research questions w i l l be r e s t a t e d as t e s t a b l e hypotheses i n t h e i r o p e r a t i o n a l i z e d form, several terms which are used i n the wording of the hypotheses are defined below. 1. Information load: The number of n u t r i t i o n a l cues on which brand performances are d i s c l o s e d i n a brand-by-cue information m a t r i x ; s p e c i f i c a l l y , the number of columns of brand performance r a t i n g s i n such a matrix. 2. Mean r i d i t : A s t a t i s t i c from r i d i t a n a l y s i s i n d i c a t i n g the changes which have occurred i n the brand s a l e s d i s t r i b u t i o n f o l l o w i n g an experimental treatment when compared to the basel i n e ( c o n t r o l ) brand s a l e s d i s t r i b u t i o n . The mean r i d i t s t a t i s t i c serves as the dependent v a r i a b l e and summarizes the d i r e c t i o n a l changes i n the brand sales d i s t r i b u t i o n , as compared to the basel i n e d i s t r i b u t i o n . 1 B r i e f l y , brands of a product are f i r s t rank ordered by t h e i r o v e r a l l n u t r i t i o n a l r a t i n g s , given the cues d i s c l o s e d i n a p a r t i c u l a r treatment. The sales d i s t r i b u t i o n s of these brands f o l l o w i n g c o n t r o l and f o l l o w i n g the treatment are matched and a mean r i d i t f o r the treatment d i s t r i b u t i o n i s computed (the mean r i d i t f o r the b a s e l i n e d i s t r i b u t i o n i s always 0.5, by d e f i n i t i o n ) . A mean r i d i t s t a t i s t i c greater than 0.5 i n d i c a t e s t h a t the treatment d i s t r i b u t i o n i s more skewed towards brands with b e t t e r n u t r i t i o n a l r a t i n g s . The higher the mean r i d i t above 0.5, the greater i s the skewness towards "more As opposed to chi square a n a l y s i s , which i s i n s e n s i t i v e to any d i r e c t i o n a l d i f f e r e n c e s between two frequency d i s t r i b u t i o n s whenever df > 1. 24 n u t r i t i o u s brands" when compared to the bas e l i n e d i s t r i b u t i o n . Thus, the mean r i d i t provides a measure of the extent to which purchases are s h i f t i n g towards those brands whose n u t r i t i o n a l performances are more favourable to consumers. The greater the s h i f t , the l a r g e r w i l l be the corresponding ( p o s i t i v e ) d e v i a t i o n from 0.5 i n the mean r i d i t s t a t i s t i c . The mean r i d i t and i t s d e r i v a t i o n are discussed i n greater d e t a i l i n Chapter IV. 3. "High-importance"/"low-importance" cues: For each product, e i g h t n u t r i t i o n a l cues were s e l e c t e d f o r the experimental manipulations of load. A s i n g l e aggregate measure of the importance of each cue was derived from the responses obtained i n a consumer survey. Any one of the four most important n u t r i t i o n a l cues was designated as a "high-importance" cue. Any one of the remaining four cues served as a "low-importance" cue. 4. Brand-by-cue matrix format: A t a b u l a r l i s t i n g of each a v a i l a b l e brand of a product with i t s corresponding performance r a t i n g on each of the n u t r i t i o n a l cues included i n the t a b l e (see Figure 2). Research Hypotheses The research questions enumerated e a r l i e r can now be sta t e d as t e s t a b l e hypotheses. The hypotheses have been o p e r a t i o n a l i z e d i n a form which makes them d i r e c t l y amenable to s t a t i s t i c a l a n a l y s i s . The hypothesized d e v i a t i o n i s p o s i t i v e since i t i s expected that information w i l l cause a s h i f t towards more n u t r i t i o u s brands, not l e s s n u t r i t i o u s brands. 25 H.: The mean r i d i t reaches a maximum at some lower information l o a d , and diminishes a t the highest information load. hL. The mean r i d i t at an information load c o n s i s t i n g o f "high-importance" n u t r i t i o n a l cues i s grea t e r than the mean r i d i t at the same load using "low-importance" n u t r i t i o n a l cues. Ho-* A point-of-purchase s i g n l i s t i n g e i g h t cues i n decreasing order of importance, from l e f t to r i g h t , y i e l d s a greater mean r i d i t than a sign l i s t i n g the same eigh t cues i n reverse order. H.: The mean r i d i t f o r a product's brand sales d i s t r i b u t i o n i s greater than 0.5 when n u t r i t i o n a l information i s placed i n a brand-by-cue matrix format at the point o f purchase. Hr' Following the removal of the point-of-purchase information from the s t o r e s , the mean r i d i t f o r a product's weekly brand s a l e s d i s t r i b u t i o n remains at the experimental b a s e l i n e l e v e l o f 0.5. 26 CHAPTER I I I SUMMARY OF RELEVANT RESEARCH Chapter I l a i d out the framework of t h i s research and zeroed i n on the problem to be i n v e s t i g a t e d . Chapter II enumerated the research questions which were then r e s t a t e d as t e s t a b l e hypotheses. This chapter summarizes the research uncovered i n the l i t e r a t u r e d e a l i n g s p e c i f i c a l l y with each o f the f i v e hypotheses. What f o l l o w s i s a review of past e m p i r i c a l f i n d i n g s p e r t i n e n t to each hypothesis and attempting to reveal the present s t a t e o f knowledge with respect to the hypothesis. Hypothesis-Related Research H 1: The mean r i d i t reaches a maximum at some lower information l o a d , and diminishes at the highest information load. This hypothesis postulates t h a t a c u r v i l i n e a r (inverted-U) r e l a t i o n -ship w i l l be found i n t h i s f i e l d experiment between the information load v a r i a b l e and the behavioural response v a r i a b l e (mean r i d i t ) , as a r e s u l t of "information overload." P r i o r to reviewing e m p i r i c a l research which t e s t e d a model o f "information overload," an overview of short-term memory i s presented to reveal the conceptual underpinnings of that model. Short-Term Memory and Information Overload. A b r i e f d i s c u s s i o n of the components of human memory i s necessary to gain an understanding of why consumers have a l i m i t e d c a p a c i t y to process information i n a given time period and why an i n v e r t e d U-shaped response to i n c r e a s i n g load inputs i s hypothesized. Bettman (1979) r e c e n t l y i n t e g r a t e d the r e l e v a n t concepts 27 and f i n d i n g s from psychology and synthesized them i n t o a theory o f consumer information processing and choice. His overview on the d i f f e r e n t f unctions of memory are r e l i e d upon i n the f o l l o w i n g d i s c u s s i o n . Figure 4 t y p i f i e s a g e n e r a l l y accepted model of memory s t r u c t u r e and o p e r a t i o n . I t recognizes t h a t although there may a c t u a l l y be only a s i n g l e memory s t o r e , t o t a l memory cap a c i t y i s a l l o c a t e d to d i f f e r e n t storage systems, each with s p e c i f i c f u n c t i o n s and c h a r a c t e r i s t i c s . When processing takes place s t i m u l i received by the sense organs are r e g i s t e r e d i n the r e s p e c t i v e sensory s t o r e . This information decays w i t h i n f r a c t i o n s of a second unless a t t e n t i o n i s paid to the stimulus and processed by being t r a n s f e r r e d to short-term memory. In short-term memory the i n -coming s t i m u l i are i n t e r p r e t e d and given meaning by meshing with information stored i n long-term memory. Thus, short-term memory i s the focus o f current processing a c t i v i t y and serves to brin g together the s t i m u l i received from the senses with the information already permanently r e s i d i n g i n long-term memory (Bettman, 1979:140). I f a s u f f i c i e n t amount of processing i s maintained i n short-term memory through rehearsal of the i n t e r p r e t e d i n p u t , thereby not a l l o w i n g i t to fade away, t h i s a c t i v i t y w i l l u s u a l l y r e s u l t i n some form of output: a d e c i s i o n made, an a t t i t u d e changed or some of the new information t r a n s -f e r r e d to long-term memory ( W i l k i e and F a r r i s , 1976:2; Bettman, 1979:144). The c a p a c i t y p o t e n t i a l o f the sensory receptors at one end of t h i s memory system i s bel i e v e d . t o be immense, as i s the storage c a p a c i t y of long-term memory at the other end ( W i l k i e , 1975:42). However there are d e f i n i t e c a p a c i t y and temporal l i m i t a t i o n s to short-term memory. This i s because only a l i m i t e d p o r t i o n o f the e n t i r e memory i s a l l o c a t e d to the Figure 4 A Model o f Memory St r u c t u r e and Operation Stimulus Input Sensory Receptors + Sensory Stores V i s u a l Auditory Haptic Short- Long-Term <— Term Memory Memory (Temporary — > (Permanent Working Memory Memory) Store) Response Output (Adapted from Bettman, 1979:140, F i g . 6.1) 29 a c t i v e processing of incoming s t i m u l i - a t any- given time. That l i m i t e d p o r t i o n o f memory which i s a c t i v a t e d f o r t h i s purpose i s c a l l e d s h o r t -term memory. In the quarter century since George A. M i l l e r (1956) introduced his magical number of seven to define the ca p a c i t y l i m i t s o f short-term memory, some f a i r l y b asic parameters have been e s t a b l i s h e d to c h a r a c t e r i z e short-term memory. Herbert A. Simon (1974:487) was able to e x t r a c t some of these parameters by examining and combining the r e s u l t s of several experiments on c o g n i t i v e processes reported i n the psycho l o g i c a l 1 i t e r a t u r e : The work of i d e n t i f y i n g and measuring the basic parameters of the human information processing system has j u s t begun, but already important information has been gained. The psychological r e a l i t y of the chunk has been f a i r l y well demonstrated, and the chunk c a p a c i t y of short-term memory has been shown to be i n the range of f i v e to seven. F i x a t i o n of information i n long-term memory has been shown to take about 5 or 10 seconds per chunk. The term "chunk" was coined by M i l l e r (1956:93) to denote u n i t s of compound s t r u c t u r e s ( b u i l t up of information input components) which served as basic human memory u n i t s i n order to measure the ca p a c i t y of short-term memory: The c o n t r a s t of the terms b i t and chunk also serves to h i g h l i g h t the f a c t t h a t we are not very d e f i n i t e about what c o n s t i t u t e s a chunk o f in f o r m a t i o n . For example, the memory span of f i v e words that Hayes obtained . . . might j u s t as a p p r o p r i a t e l y have been c a l l e d a memory span of 15 phonemes, since each word had about three phonemes i n i t . I n t u i t i v e l y , i t i s c l e a r t h a t the subjects were r e c a l l i n g f i v e words, not 15 phonemes, but the l o g i c a l d i s t i n c t i o n i s not immediately apparent. We are d e a l i n g here with a process o f o r g a n i z i n g o r grouping the input i n t o f a m i l i a r u n i t s o r chunks, and a great deal of l e a r n i n g has gone i n t o the formation of these f a m i l i a r u n i t s . 30 Bettman (1979:147) e x p l a i n s the chunking concept i n the consumer context: . . . A chunk was defined to be a c o n f i g u r a t i o n that was f a m i l i a r to an i n d i v i d u a l and could be manipulated as a u n i t , i n essence an organized c o g n i t i v e s t r u c t u r e that could grow as information i s i n t e g r a t e d i n t o i t . For example, a brand name can summarize a good deal o f more d e t a i l e d information f o r a consumer f a m i l i a r with t h a t brand, and hence the brand name and a l l i t stands f o r can be thought o f as a chunk. The ac t u a l amount o f under-l y i n g m a t e r i a l t h a t can be processed simultaneously can thus be expanded by formation of l a r g e r chunks . . . The accumulating research evidence i n d i c a t e s t h a t the c a p a c i t y of short-term memory i s a constant number o f chunks -- independent o f the s t i m u l i from which those chunks are b u i l t , be i t " . . . f i v e chunks worth of words, f i v e chunks o f d i g i t s , f i v e chunks of c o l o r s , f i v e chunks of shapes, f i v e chunks of poetry or prose . . . " (Simon, 1974:483). Thus, only about f i v e to seven chunks of information can be processed by humans at any one time. This i s so because the processing c a p a c i t y necessary to rehearse these chunks i s l i m i t e d ; and rehearsal (keeping the stimulus a c t i v a t e d i n short-term memory while i t i s analyzed f u r t h e r so t h a t some output can r e s u l t ) i s c r u c i a l because i f the information i s not rehearsed i t w i l l vanish from short-term memory i n about 30 seconds or l e s s (Bettman, 1979:148). More i m p o r t a n t l y , as Bettman (1979:147-148) points out, these f a i r l y well e s t a b l i s h e d c a p a c i t i e s of short-term memory (STM) are reduced even f u r t h e r whenever some of t h i s f i x e d memory c a p a c i t y must be a l l o c a t e d f o r other simultaneous t a s k s : The c a p a c i t y of STM i s lowered i f other processing de-mands are made. This f o l l o w s immediately from the notion of the l i m i t s on STM as processing c a p a c i t y l i m i t s . I f part o f t o t a l c a p a c i t y must be used f o r another t a s k , that leaves l e s s f o r processing chunks of inform a t i o n . The normal c a p a c i t y of seven chunks or so may be reduced 31 to a c a p a c i t y of two or three chunks i f other tasks are undertaken simultaneously, such as search processes or counting tasks . . . I t i s apparent t h a t a lowering of cap a c i t y a v a i l a b l e f o r processing information due to competing demands made by the information environment (such as searching f o r or monitoring incoming information) i n conjunction with a high information input r a t e imposed by a task can lead to a d e t e r i o r a t i o n i n processing performance. As the rate of information inputs i n c r e a s e s , a greater r a t e of  i n t e r n a l processing may be required to analyze and t r a n s f e r the information i n t o memory. The heavier the task of monitoring and processing the incoming data i n a l i m i t e d time p e r i o d , the sooner the a v a i l a b l e c a p a c i t y of memory i s used up. Once t h i s c a p a c i t y i s completely used up, the i n -d i v i d u a l can no longer process a d d i t i o n a l i n p u t s . Moreover, the excess inputs may a c t u a l l y c l u t t e r the i n t e r n a l processing task which i s already operating at c a p a c i t y . The c h a r a c t e r i s t i c r e s u l t i s an o v e r a l l d e t e r i o r -a t i o n i n processing performance o r a c o n d i t i o n r e f e r r e d to as "information overload." In summary, the greater the processing demands imposed by a set of input data i n terms of using up short-term memory ca p a c i t y per f i x e d time p e r i o d , the greater i s the p o t e n t i a l f o r an information overload (Bettman, 1979:127). Having presented an o u t l i n e of human memory s t r u c t u r e to e x p l a i n why the p o t e n t i a l f o r information overload e x i s t s , the review now con-centrates on a model o f human information processing performance developed and t e s t e d i n numerous e m p i r i c a l s t u d i e s . 32 Schroder, D r i v e r and S t r e u f e r t . The programmatic research on information load and human information processing performed i n the lab o r a t o r y by the "Princeton" group of Harold Schroder, Michael D r i v e r and S i e g f r i e d S t r e u f e r t (1967) i s probably the seminal work i n t h i s area. To c l a r i f y the term "information processing" f o r the ensuing d i s c u s s i o n , the d e f i n i t i o n given by Haines (1974 :17-18) i s useful here: . . . [Schroder, Driver,,and S t r e u f e r t ] present as an example humans making judgments about d i f f e r e n t kinds of l i g h t . People appear to be able . t o d i f f e r e n t i a t e a t l e a s t three d i f f e r e n t aspects or dimensions of l i g h t : b r i g h t n e s s , s a t u r a t i o n , and hue. These u n i t s are oft e n combined by people i n t o a s i n g l e e n t i t y ; t h i s e n t i t y may be c a l l e d a symbol. When data are processed and encoded i n t o a symbol o r symbols, the raw data have been converted i n t o i n f o r m a t i o n . Thus an information-processing system i s conceived as a s e r i e s of i n t e r r e l a t e d programs t h a t convert data i n t o information . . . . . . When information processing occurs, i t i s the process of a c t u a l l y using an information-processing system to encode data i n t o symbols. This process may or may not r e s u l t i n some output. U s u a l l y there i s some output. Using an input-output paradigm to model human information p r o c e s s i n g , Schroder, D r i v e r and S t r e u f e r t (1967) manipulated the complexity o f the information environment presented to subjects i n t h e i r experiments. The Sch r o d e r - D r i v e r - S t r e u f e r t model defines the complexity of an information environment as the number of div e r s e dimensional u n i t s o f information provided to a subje c t i n a decision-making task. Complexity, t h e r e f o r e , i s d i r e c t l y a f f e c t e d by three primary f a c t o r s : 1. Information load (e.g. from 2 to 25 pieces of information r e -ceived by subjects per h a l f - h o u r ) , 33 2. Information d i v e r s i t y , and 3. Rate of information change (Suedfeld and Hagen, 1966; Schroder, 1 D r i v e r and S t r e u f e r t , 1967). The output (response) measure i n the S c h r o d e r - D r i v e r - S t r e u f e r t model i s the subject's l e v e l of information processing. This was o p e r a t i o n -a l l ' zed as the amount of data t h a t had been i n t e g r a t e d by the subject t o -wards s o l v i n g a problem and making a d e c i s i o n . Thus, the number of i n t e g r a t i o n s gave an i n d i c a t i o n of how much of the data provided to the subject had been converted i n t o information and employed i n the problem-s o l v i n g task. The S c h r o d e r - D r i v e r - S t r e u f e r t model postulates a c u r v i l i n e a r (inverted-U) r e l a t i o n s h i p between information complexity and the l e v e l o f information processing. F a i r l y c o n s i s t e n t e m p i r i c a l r e s u l t s i n the l a b o r a t o r y appeared to bear t h i s r e l a t i o n s h i p out (Schroder, 1971; S t r e u f e r t , 1970; S t r e u f e r t , Suedfeld and D r i v e r , 1965; Schroder, D r i v e r and S t r e u f e r t , 1967:60,151). A t y p i c a l f i n d i n g i s depicted i n Figure 5. Following the pattern of r e s u l t s i n Figure 5, i t i s apparent t h a t the subject group's l e v e l o f information processing was a monotonically i n c r e a s i n g f u n c t i o n of information l o a d , up to a maximum p o i n t . As load continued to increase beyond t h i s point the group's l e v e l o f information processing d e c l i n e d sharply i n a monotonic f a s h i o n . Excess information load had become "information overload" and the q u a l i t y of decision-making d e t e r i o r a t e d . •^According to the model, complexity i s a l s o i n d i r e c t l y i n f l u e n c e d by the c o s t - b e n e f i t aspects of the information environment as well as the subject's l e v e l of involvement i n the task. Figure 5 Empirical R e l a t i o n s h i p Between Information Complexity and Level of Information Processing Note, however, from Figure 5 that although some subjects (the high conceptual l e v e l group) reached a higher l e v e l o f processing at t h e i r optimal performance, the sharp downturn occurred at almost e x a c t l y the same point i n load as subjects whose processing performance peaked at a much lower performance l e v e l (the low conceptual l e v e l group). 35 The c o l l e c t i v e research evidence i n these.studies supports the notion of a f i x e d c a p a c i t y o f chunks i n human short-term memory with apparently no i n d i v i d u a l d i f f e r e n c e s . The e f f e c t s of "information overload" begin to appear at e s s e n t i a l l y the same point i n l o a d , f o r a l l i n d i v i d u a l s . Looking at the curves i n Figure 5, one might summarize these r e s u l t s i n the f o l l o w i n g terms.: e v o l u t i o n has given humans-about f i v e chunks of short-term memory cap a c i t y . Some humans b u i l d l a r g e r chunks than others and thus i n t e g r a t e information inputs at a more s o p h i s t i c a t e d l e v e l , but f i v e chunks i s a l l they get. Gerald S t i l e s . To t e s t the postulates of the Schroder-Driver-S t r e u f e r t model outside the l a b o r a t o r y , S t i l e s (1974) undertook a f i e l d study of the information processing a c t i v i t i e s o f s i x t y i n d u s t r i a l buyers. He found the l e v e l of information processing among these buyers to be a p o s i t i v e and l i n e a r f u n c t i o n of environmental complexity. Thus, the r e s u l t s of h i s f i e l d study d i d not support the S c h r o d e r - D r i v e r - S t r e u f e r t model. S t i l e s (1974) argued t h a t decision-makers i n natural working con-d i t i o n s w i l l take various a c t i o n s to avoid s i t u a t i o n s which might lead to e x c e s s i v e l y complex information environments. He reasoned that the i n d u s t r i a l buyers adjusted the r a t e of information input i n several ways so as not to become overwhelmed by a buying process which became i n c r e a s i n g l y complex. Coping s t r a t e g i e s could be used, and as a purchase became more complex: 1. More time could be a l l o c a t e d to i t , 2. Needed information could be obtained piecemeal from s u p p l i e r s and u s e r s , 36 3. Spread sheets could be used to c o l l e c t and g r a d u a l l y i n t e g r a t e the data on purchase a l t e r n a t i v e s , and, 4. Other p a r t i c i p a n t s could be involved i n the a l t e r n a t i v e e v a l u a t i o n process so t h a t the buyer r e s t r i c t s his decision-making to areas where he i s most competent ( I b i d . ,126). This r e a l - w o r l d information environment contra s t s sharply with the information complexity experiments i n the l a b o r a t o r y . Schroder, D r i v e r and S t r e u f e r t ' s (1967) subjects had no c o n t r o l over the information input rate and d i v e r s i t y . S t i l e s (1974:126) w r i t e s t h a t h i s data r e f u t e the notion that people often work i n such environments: i n d i v i d u a l s do not u s u a l l y place themselves i n . a p o s i t i o n where they are unable to c o n t r o l the information input load. Commenting on the f i n d i n g made by S t i l e s (1974) i n h i s f i e l d a p p l i c a t i o n of the S c h r o d e r - D r i v e r - S t r e u f e r t model, Wilson (1974:140) proposes: . . . Since the research was done i n the f i e l d , i t l i k e l y only measured problems t h a t have been reduced to t h e i r s i m p l e s t forms. I t i s not l i k e l y t hat a c u r v i l i n e a r r e l a t i o n s h i p between l e v e l s of information processing and task complexity w i l l be found i n a f i e l d study. Complex tasks are l i k e l y decomposed i n t o a set simple task as the o r g a n i z a t i o n works toward reducing u n c e r t a i n t y . Jacoby, S p e l l e r and Kohn. Two landmark e f f o r t s to t e s t the hypo-t h e s i s of "information overload" i n a consumer context were the l a b o r a t o r y studies of Jacoby, S p e l l e r and Kohn (1974 a;b). In the f i r s t experiment, Jacoby, S p e l l e r and Kohn (1974 b) used student subjects and various cues of package information (e.g. bleach content, p r i c e , q u a n t i t y required per wash load) on h y p o t h e t i c a l brands 37 of laundry detergent. Subjects assigned to one of nine load treatments received e i t h e r 2, 4 or 6 information cues on e i t h e r 4, 8 or 12 brands, studied the information at t h e i r own pace, and then made a brand choice. Following the d e c i s i o n , subjects completed a questionnaire assessing the p s y c h o l o g i c a l impact of information load. The dependent measure used by Jacoby e t a l . was "performance accuracy": the number of subjects c o r r e c t l y choosing "the brand which most c l o s e l y approximated t h e i r p r e v i o u s l y determined i d e a l brand ( I b i d . , 6 5 ) . " A n a l y s i s of t h e i r r e s u l t s i n d i c a t e d t h a t " c o r r e c t l y s e l e c t i n g one's 'best' brand" . . . i s i n v e r s e l y r e l a t e d to number of brands and p o s i t i v e l y r e l a t e d to number.of items o f information per brand. Conceivably, t h i s l a t t e r f i n d i n g might be reversed given more items of information (n_ > 6) per brand ( I b i d . , 6 5 ) . In a d d i t i o n , Jacoby e t a l . m u l t i p l i e d the number of brands times the number of information cues i n t o a measure of " t o t a l i n f o r m a t i o n " load f o r each experimental treatment. Upon p l o t t i n g t h i s measure of information load against "performance accuracy", Jacoby et a l . r e p o r t e d l y found an.inverted 'U-shaped r e l a t i o n s h i p between load and number of c o r r e c t choices. Jacoby, S p e l l e r and Kohn (1974:67) concluded: Within the confines of the subjects and procedures used i n t h i s study, i t would appear that i n c r e a s i n g package information load tends to produce: (1) d y s f u n c t i o n a l consequences i n terms of the consumer's a b i l i t y to s e l e c t t h a t brand which was best f o r him, and (2) b e n e f i c i a l e f f e c t s upon the consumer's degree of s a t i s f a c t i o n , c e r t a i n t y , and confusion regarding h i s s e l e c t i o n . In other words, our subjects f e l t b e t t e r with more information but a c t u a l l y made poorer purchase d e c i s i o n s . Obviously, before g e n e r a l i z i n g from such data, comparable studies must be conducted using d i f f e r e n t consumer products and sampling from d i f f e r e n t consumer populations (e.g., housewives). 38 In t h e i r second l a b o r a t o r y study, Jacoby, S p e l l e r and Kohn (1974 a) employed housewives as subjects and various package information cues (e.g. p r i c e , c o n t a i n e r s i z e , n u t r i t i o n a l components) on bogus brands of two d i f f e r e n t food products. Subjects assigned to one of 16 load treatments received e i t h e r . 4, 8, 12 or 16 information cues on e i t h e r 4, 8, 12 or 16 brands, the information on each brand having been placed on a separate 4" x 6" index card. Subjects examined a l l the information cards they r e c e i v e d , made a s i n g l e brand choice and ranked the remaining brands i n order of decreasing preference. As i n the e a r l i e r study, subjects a l s o completed a questionnaire to assess the e f f e c t s o f load on a number o f psychological dimensions. In Jacoby, S p e l l e r and Kohn's (1974 a) second study, two dependent measures were used to assess the "performance accuracy" of brand choices: " C o r r e c t l y s e l e c t i n g the brand c l o s e s t to one's i d e a l " , and a c o r r e l a t i o n measure between "each subject's actual preference ranking and the ranking predicted from her d e s c r i p t i o n of her i d e a l brand" ( I b i d . , 36). Information load was a l s o o p e r a t i o n a l i z e d as the combined number of brands x the number o f information cues which a subject received and which Jacoby et a l . c a l l e d "total-amount-of-information" ( I b i d . , 38). The r e s u l t s reported were s i m i l a r to the f i n d i n g s i n t h e i r f i r s t study. Combining the evidence of the four "performance accuracy" curves (two f o r each product) i n the second study with t h a t of the f i r s t study, Jacoby, S p e l l e r and Kohn (1974 a:40-41) concluded: When p l o t t e d . a g a i n s t total-amount-of-information, three of the four major performance accuracy curves . . . d i s p l a y the pr e d i c t e d decrease i n accuracy at the higher end of the continuum and the fo u r t h curve . . . shows a 39 : taperi n g o f f which could be i n t e r p r e t e d as a p o s s i b l e prelude to a d e c l i n e . . . In sum, the f i v e accuracy curves from both studies suggest that p r o v i d i n g s u b s t a n t i a l amounts of package information can r e s u l t i n poorer purchase d e c i s i o n s . In c o n c l u s i o n , the research conducted thus f a r suggests that there are f i n i t e l i m i t s to the consumer's a b i l i t y to accommodate s u b s t a n t i a l amounts of package information w i t h i n a l i m i t e d time span. However, numerous questions remain to be answered and a d d i t i o n a l research i s required before any d e f i n i t i v e conclusions or p u b l i c p o l i c y d e c i s i o n s can or should be made based on these data . . . Jacoby, S p e l l e r and Kohn's (1974 a; b) conclusions t h a t t h e i r subjects made poorer purchase d e c i s i o n s with more information received a l i v e l y r e b u t t a l from several sources. A reexamination o f the experimental design, a n a l y t i c a l methods, and operati o n a l d e f i n i t i o n of load used by Jacoby et a l . brought three c r i t i c s to the opposite conclusion about the e f f e c t s of information l o a d : the subjects of Jacoby et a l . made b e t t e r purchase d e c i s i o n s with more information ( W i l k i e , 1974; Summers, 1974; Russo, 1974). F i r s t , a design problem was detected i n the Jacoby et a l . method of co n s t r u c t i n g each load treatment by randomly s e l e c t i n g information cues from the t o t a l array of cues a v a i l a b l e . As i n d i c a t e d by Summers (1974:467), t h i s procedure does not c o n t r o l f o r v a r i a t i o n across d i f f e r e n t load t r e a t -ments due to "the s p e c i f i c , types of information presented." W i l k i e (1975:36) e x p l a i n s t h i s troublesome is s u e as f o l l o w s : . . . Subjects i n the 2 - a t t r i b u t e , 4-brand c o n d i t i o n , f o r example, not only received l e s s information than those i n the 4 - a t t r i b u t e , 4-brand c o n d i t i o n , but al s o could have received d i f f e r e n t a t t r i b u t e s than those i n 4, 4, which may al s o have d i f f e r e d from those used i n the 2 - a t t r i b u t e , 8-brands c e l l , and so on. I f , as we would suspect, some a t t r i b u t e s are more s a l i e n t to consumers than are o t h e r s , such d i f f e r e n t i a l p r o v i s i o n of information could lead to d i f f e r e n t i a l e f f e c t s apart from those of qu a n t i t y of information per se. 40 Second, the a n a l y t i c a l method used by Jacoby et a l . f a i l e d to make allowance f o r the d i f f e r e n t c o n d i t i o n a l p r o b a b i l i t i e s o f making a " c o r r e c t " brand choice i n d i f f e r e n t number-of-brand c e l l s ( W i l k i e , 1974; Summers, 1974). That i s , Jacoby et a l . overlooked the f a c t t h a t fewer " c o r r e c t " brand choi ces should be expected i n the 12 or 16-brand treatments than i n the 4 or 8-brand treatments, when comparing subjects' choice "performance acc u r a c i e s " . When the Jacoby et a l . data are adjusted f o r these unequal base r a t e s , actual "performance accuracy" improves monotonically with the number of brands, or the number of information cues, i n a load treatment ( W i l k i e , 1974:464). F i n a l l y , the operat i o n a l d e f i n i t i o n of load by Jacoby et a l . as " t o t a l i n f o r m a t i o n " , i . e . the number of brands m u l t i p l i e d by the number of information cues presented to a s u b j e c t , s u f f e r s from conceptual drawbacks ( W i l k i e , 1974; Russo,' 1974). This point was explained i n Chapter I , when an appropriate d e f i n i t i o n of information load was chosen f o r t h i s d i s s e r t a t i o n . The Jacoby et a l . o p e r a t i o n a l i z a t i o n of load i n c o r r e c t l y assumes, f o r example, that e v a l u a t i n g 8 brands on 8 cues and 16 brands on 4 cues y i e l d s p s y c h o l o g i c a l l y equivalent amounts of information to s u b j e c t s , since both treatments contain a " t o t a l i n -formation" load of 64 r a t i n g s . As Russo (1974:70) emphasizes, f o r consumers "there i s no t r a d e - o f f between[^number of brands and number o f cues) : th a t i s , one does not compensate f o r the other." In a d d i t i o n , t h i s " t o t a l i n f o r m a t i o n " measure confounds any response d i f f e r e n c e s due e i t h e r to increases i n the number of cues or increases i n the number of brands ( W i l k i e , 1975:37). 41 The p e r t i n e n t measure o f information load i n the Jacoby et a l . studies i s the number of information cues present i n the treatment, thus c o n f i n i n g comparisons to d i f f e r e n t cue-load treatments with the same number of brands. When t h i s i s done, the Jacoby e t a l . r e s u l t s show tha t s u b j e c t s ' brand choice "performance a c c u r a c i e s " improved with the number o f information cues i n a load treatment (Russo, 1974; W i l k i e , 1974; Summers, 1974). In summary, these r e v a l u a t i o n s o f the r e s u l t s from Jacoby, S p e l l e r and Kohn's (1974 a; b) two studies lead to a re v e r s a l o f the claim by Jacoby et a l . t h a t subjects r e c e i v i n g high information loads experience "information overload." Russo's (1974:71) r e a n a l y s i s sums up the t r i o o f r e b u t t a l s reviewed here: The preceding r e a n a l y s i s of the data o f Jacoby e t a l . (1974 a) r e s t s on two c l a i m s , t h a t [number of cu e s ] , not [number of brands x number of cu e s ] , i s the p s y c h o l o g i c a l l y r e l e v a n t information measure and that choice accuracy cannot be compared over d i f f e r e n t values o f [number o f brands]. Both of these claims r e c e i v e considerable support, from the choice accuracy data and als o from the verbal r a t i n g s and d e c i s i o n times. The importance of these a s s e r t i o n s i s to r e s t r i c t the t e s t of the information overload hypothesis to co n d i t i o n s varying [the number of cues] and holding [the number o f brands] constant. When t h i s i s done, the [data] c l e a r l y show th a t more information improves choice accuracy. Debra Scammon. C a r e f u l l y noting the c r i t i c i s m s l e v e l e d at the Jacoby et a l . s t u d i e s i n terms o f design, a n a l y s i s and measurement i s s u e s , Scammon (1977) conducted a la b o r a t o r y study of consumer information load which sought to c o r r e c t most o f these methodological d e f i c i e n c i e s . The study was designed around a proposed p u b l i c p o l i c y r e g u l a t i o n f o r the d i s c l o s u r e o f n u t r i t i o n a l information i n t e l e v i s i o n commercials f o r food. Two brands of peanut bu t t e r and various n u t r i t i o n a l information cues 42 served as the t e s t s t i m u l i . In a d d i t i o n to a con t r o l c o n d i t i o n (no n u t r i t i o n a l information i n the t e s t commercials), two l e v e l s of information load were employed: c a l o r i e s plus 4 other cues, c a l o r i e s plus 8 other cues. The brand r a t i n g s on the n u t r i t i o n a l cues employed i n the design were manipulated i n such a way as to make one of the brands appear n u t r i t i o n a l l y s u p e r i o r i n both load treatments. Scammon (1977) used two formats f o r presenting the load treatments: brand r a t i n g s were d i s c l o s e d as a percentage of Recommended D a i l y Allowance or as a d j e c t i v a l des-c r i p t i o n s of n u t r i t i o n a l performance. Table 1, taken from Scammon (1977:150), d i s p l a y s the load treatments and the r a t i n g s d i s c l o s e d on the two brands (Koogle and Skippy). Table 1 Information Load Treatments Used i n Scammon's (1977) Experiment Experimental group Ingredient 1" 2 3 4 5 6 a Koogle Calories — 180 180 180 180 — Protein — 35 35 Excellent Excellent — Niacin — 30 30 Good Good — Thiamine — 20 20 Good Good — Iron — 20 20 Good Good — Calcium — — 10: — fair — Riboflavin — — 10 — fair — Vitamin A — — 0 • — none — Vitamin C — — 0 — none — Skippy Calories — 190 190 190 190 — Protein — 15 15 fair fair — Niacin — 15 15 fair fair — Thiamine — 10 10 fair fair — Iron — 10 10 fair fair — Calcium — — 0 — none — Riboflavin — — 0 — none — Vitamin A — — 0 — none — Vitamin C — — 0 — none — " Experimental groups 1 and 8 resulted from the random split of the no information control group into two equal subgroups to be used as controls for the two experi-mental manipulations. (From Scammon, 1977: 150, Table 1) 43 Note t h a t Scammon increased information load from f i v e to nine cues by r e t a i n i n g the f i v e cues of the lower load and adding on four new cues, thereby s k i r t i n g one design problem of the Jacoby et a l . s t u d i e s . The load treatments were presented to subjects ( C a l i f o r n i a n s from various demographic groups) during the l a s t s i x seconds o f 30-second t e s t commercials. Two of Scammon's (1977) information load e f f e c t s measures are of i n t e r e s t here: 1. A b i l i t y to c o r r e c t l y i d e n t i f y the "more n u t r i t i o u s " brand; 2. Brand preference/intention-to-buy. With regard to the f i r s t c r i t e r i o n , Scammon (1977; 151) found no s i g -n i f i c a n t d i f f e r e n c e s between the groups exposed to d i f f e r e n t information loads: . . . Over a l l i t appeared t h a t i d e n t i f i c a t i o n of the "more n u t r i t i o u s " brand was independent of the amount of n u t r i t i o n information received by respondents . . . However, p a i r - w i s e group comparisons d i d reveal a s i g n i f i c a n t d i f f e r e n c e between the "no in f o r m a t i o n " c o n t r o l group and each of the "information" groups, suggesting t h a t some information i s b e t t e r than no informa t i o n . The data f u r t h e r suggest t h a t as long as some information i s a v a i l a b l e , the amount a v a i l a b l e does not a f f e c t consumers' a b i l i t y to i d e n t i f y the "more n u t r i t i o u s " brand. The data were a l s o analyzed f o r load treatment d i f f e r e n c e s i n the number o f subjects choosing one brand as t h e i r most l i k e l y next purchase. Again, Scammon (1977:152) found no s i g n i f i c a n t d i f f e r e n c e s i n the proportions of brand choices between the three groups (no i n f o r m a t i o n , c a l o r i e s + 4 cues, c a l o r i e s + 8 cues): 44 Chi-square a n a l y s i s and a n a l y s i s of variance of the d i f f e r e n c e s between the expected and the observed c e l l frequences of respondents p r e f e r r i n g each brand revealed no s i g n i f i c a n t d i f f e r e n c e s between groups r e c e i v i n g d i f f e r e n t amounts of information . . . Neither the amount nor the format of the information presented to the subjects a f f e c t e d t h e i r brand p r e f e r -ence/intention-to-buy. Apparently, i n the context of two (6-second) exposures on t e l e -v i s i o n , d i f f e r e n c e s i n information load n e i t h e r a f f e c t e d behavioural i n t e n t i o n s nor accounted f o r changes i n a t t i t u d e . Roger Best. Best (1978) performed an information load experiment with student subjects i n which he manipulated other information design f a c t o r s i n conjunction with v a r i a t i o n s i n load. His f i n d i n g s reported here are l i m i t e d to those on l o a d , i . e . number of information cues. Best (1978) constructed brand-by-cue information matrices on three products (10-speed b i c y c l e s , h a i r c o n d i t i o n e r s , r e n t a l apartments) using e i t h e r 2, 4, 6 or 8 cues on e i t h e r 3 or 6 bogus brands. Cue importance measures were obtained a month e a r l i e r from each subject and respondents were given an information matrix c o n t a i n i n g t h e i r n most important cues, l i s t e d i n decreasing order of importance. The brand r a t i n g s i n t h i s information matrix were expressed i n terms of f i v e a d j e c t i v a l d e s c r i p t i o n s of performance (very good, good, f a i r , poor, very poor). A f t e r processing the information they r e c e i v e d , sub-j e c t s noted t h e i r most p r e f e r r e d brand. The dependent measure used by Best (1978:12) was "choice accuracy:" Using a subject's s t a t e d a t t r i b u t e importance weights, an a d d i t i v e model was used to compute summated brand scores f o r each a l t e r n a t i v e evaluated. When the highest summated brand score corresponded with the respondent's f i r s t choice the choice was scored as accurate or c o r r e c t . . . 45 Then, f o r each experimental c o n d i t i o n choice accuracy was computed as the proportion of c o r r e c t or c o n s i s t e n t brand choices. A n a l y s i s of the data revealed t h a t very l i t t l e of the variance i n the dependent measure was a t t r i b u t a b l e to the information load f a c t o r . In any case, Best (1978:14) found no basis of support f o r the "information overload" hypothesis: Information load which was v a r i e d across 2, 4, 6, and 8 l e v e l s o f a t t r i b u t e load was only found to have a s i g -n i f i c a n t e f f e c t on choice accuracy i n the s e l e c t i o n of 10-speed b i c y c l e s . This v a r i a t i o n i n information load had no e f f e c t on choice accuracy i n choosing among a l t e r n a t i v e h a i r c o n d i t i o n e r s and r e n t a l apartments . . . Though the p l o t o f choice accuracy as a f u n c t i o n o f information load . . . f o r the 10-speed b i c y c l e s suggests an information o v e r l o a d , an a n a l y s i s o f c o n s t r u c t s associated with t h i s r e l a t i o n s h i p reveal t h a t choice accuracy does not d i f f e r f o r 6 a t t r i b u t e and 8 a t t r i b u t e information loads (p = .10). Thus, f o r brand s e l e c t i o n among 10-speed b i c y c l e s , choice accuracy increased with increases i n information load . . . Goodwin and Etgar. In a l a b o r a t o r y i n v e s t i g a t i o n o f comparative a d v e r t i s i n g appeals, Goodwin and Etgar (1980) manipulated the amount of product information contained i n mock p r i n t advertisements f o r a f i c t i t i o u s brand of beer and a f i c t i t i o u s brand of cold/headache remedies. This experimental design d i f f e r e d from the information load designs described so f a r , i n t h a t brand information was incorporated i n t o the (somewhat puffy) ads r a t h e r than being presented i n a brand-by-cue format. Thus, Goodwin and Etgar's (1980) measure of information load was the number of d i f f e r e n t a t t r i b u t e s (information cues) which were discussed somewhere i n the p r i n t ad. A p r e t e s t , requesting subjects to l i s t the most important a t t r i -butes that they use to compare brands of the two t e s t products, i d e n t i f i e d 46 the seven most f r e q u e n t l y mentioned cues f o r use i n the experimental manipulation of load (e.g. smooth t a s t e , alcohol content, c a l o r i e content, f o r beer; quickness of r e l i e f , recommended.by.doctors, p r i c e , f o r c o l d / headache remedies). Advertisements were prepared i n which e i t h e r the 2, 5 or 7 most important cues were mentioned i n the ad. Manipulation checks with a p i l o t sample revealed that the ads with d i f f e r e n t l e v e l s of information load were perceived as p r o v i d i n g monotonically i n c r e a s i n g amounts of i n f o r m a t i o n . Student subjects were given one ad and.after studying i t responded to a b a t t e r y of 15 dependent measures designed to tap c o g n i t i v e and a f f e c t i v e responses to the product and the advertisement. Although other experimental f a c t o r s were manipulated simultaneously, i n t e r e s t here focuses on the main e f f e c t s of l o a d . A n a l y s i s of variance revealed only a s i n g l e s i g n i f i c a n t main e f f e c t of load on the 15 dependent measures: q u a l i t y assessment of the a d v e r t i s e d brand was highest at the 5-cue l e v e l of information load. Goodwin and Etgar (1980:196) c i t e t h i s f i n d i n g as some weak evidence of "information overload:" . . . Second, there i s some (though weak) evidence of information overload. In the one s i g n i f i c a n t main e f f e c t of the a t t r i b u t e information v a r i a b l e , the intermediate number of a t t r i b u t e s i s s u p e r i o r to other treatments with low or high l e v e l s of a t t r i b u t e s . U n f o r t u n a t e l y , t h i s evidence i s f u r t h e r weakened by the f a c t t h a t Goodwin and Etgar (1980) d i d not make a post hoc s t a t i s t i c a l comparison to check i f the 7-cue load treatment produced a s i g n i f i c a n t l y lower response than the 5-cue load treatment. This weakness, along with other methodological i s s u e s , was pointed out by T r a y l o r (1981). 47 Summary. The several consumer information load studies reviewed above cap be viewed as e f f o r t s to r e p l i c a t e the Schroder, D r i v e r and S t r e u f e r t (1967) experiments i n which a c u r v i l i n e a r (inverted-U shaped) r e l a t i o n s h i p between load and subject responses was c o n s i s t e n t l y ob-served. A l l of the consumer studies reviewed were l a b o r a t o r y experiments, none measured a c t u a l purchase behaviour, and the o v e r a l l p i c t u r e t h a t emerges i s one of weak evidence f o r the S c h r o d e r - D r i v e r - S t r e u f e r t hypo-t h e s i s of "information overload" i n the consumer context. I t should be remembered, however, th a t the information load mani-pu l a t i o n s i n the S c h r o d e r - D r i v e r - S t r e u f e r t stream of experiments incorporated time c o n s t r a i n t s and, t h e r e f o r e , subjects had no c o n t r o l over the information input r a t e . Subjects could not a l t e r the time span a l l o t t e d to the d e c i s i o n task as the information became more voluminous and complex. Contrast t h i s s i t u a t i o n with the l a b o r a t o r y experiments performed by Jacoby, S p e l l e r and Kohn (1974 a;b), Best (1978) and Goodwin and Etgar (1980). In each study, subjects were under no p a r t i c u l a r time c o n s t r a i n t to make a brand choice d e c i s i o n or other response, from the information load they received. The information input rate was c o n t r o l l e d by each i n d i v i d u a l . Perhaps the f i n d i n g s i n these studies may have been m i t i g a t e d by the f a c t that the load manipu-l a t i o n s d i d not c o n t r o l the d e c i s i o n times a v a i l a b l e to s u b j e c t s . The c r u c i a l question remains whether the n a t u r a l i s t i c and r e l a t i v e l y l i m i t e d time period that a shopper a l l o c a t e s to the purchase of a s i n g l e product i n a supermarket w i l l c o n s t r a i n the amount of product information he or she can process at the point of purchase. I f there i s a c o n s t r a i n t (due to time and the l i m i t a t i o n s of human short-term memory) 48 and i f the highest information load presented a t the point of purchase exceeds t h i s l i m i t , then shoppers are expected to avoid the complex d i s p l a y and to make a choice which does not u t i l i z e the information provided. W i l k i e (1975:41) proposes a s i m i l a r scenario f o r such a s i t u a t i o n : Within the natural consumer context the most l i k e l y r e s u l t would be a suspension of [consumer information processing] a c t i v i t i e s on the information i t s e l f , f o llowed by e i t h e r a s h i f t of a t t e n t i o n to other s t i m u l i i f a purchase i s not imminent, or a choice being made on bases other than the information provided . . . HL: The mean r i d i t at an information load c o n s i s t i n g of "high-importance" n u t r i t i o n a l cues i s greater than the mean r i d i t at the same load using "low-importance" n u t r i t i o n a l cues. This hypothesis t e s t s the notion that observed d i f f e r e n c e s i n consumer response to various n u t r i t i o n a l cues are al s o a t t r i b u t a b l e to d i f f e r e n c e s i n the importance of those cues. The preceding reviews of consumer information load experiments i n d i c a t e d that some researchers took precautions to include t h e i r s u b j e c t s ' most important information cues i n t h e i r load treatments. The e a r l y work on load by Jacoby, S p e l l e r and Kohn (1974 a; b) f a i l e d to co n t r o l f o r d i f f e r e n c e s between any two load treatments a r i s i n g from d i f f e r e n c e s i n the r e l a t i v e importance of.cues randomly chosen to make up those treatments. Thus, t h e i r r e s u l t s may have r e f l e c t e d responses to d i f f e r e n t l o a ds, confounded by responses to cues which may have been important or unimportant to the sub j e c t . Note, however, that Jacoby et a l . employed widely d i v e r s e kinds of information cues, e.g., enzyme content, p r i c e , q u a n t i t y required 49 per wash load ( f o r d e t e r g e n t s ) ; p r i c e , storage i n s t r u c t i o n , type of c o n t a i n e r , c a l o r i e s - p e r - s e r v i n g ( f o r food products). The information cues employed i n the present research are a l l of the n u t r i t i o n a l kind (e.g., f a t , p r o t e i n , v i t a m i n , n i a c i n , sodium). Moreover, research conducted with housewives reveals that the r e l a t i v e d i f f e r e n c e s i n importance between these i n d i v i d u a l n u t r i t i o n a l cues, from the perspective of grocery shopping, are q u i t e small ( R u d e l l , 1979:46; Quelch, 1978). To i l l u s t r a t e , the study by Quelch (1978:21) had 250 housewives r a t e the importance of 19 cereal a t t r i b u t e s on a s i x - p o i n t L i k e r t .scale anchored at "very important" (+ 3) and "not at a l l important" (- 3) i n the purchase of a cereal brand. Aggregated across s u b j e c t s , the 19 cues are rank ordered by t h e i r mean normalized importance scores i n Table 2. Of i n t e r e s t i s t h a t the f i v e ( s t a r r e d ) n u t r i t i o n a l cues i n Table 2 (vitamins and m i n e r a l s , p r o t e i n , sugar, f i b e r , c a l o r i e s ) are a l l rated very c l o s e l y i n r e l a t i v e importance and near the midpoint of the s c a l e . Thus, d i f f e r e n c e s between i n d i v i d u a l n u t r i e n t s i n stated importance to consumers are co n s i d e r a b l y smaller than d i f f e r e n c e s between, say, i n g r e d i e n t s , p r o t e i n content and p r i c e of the c e r e a l . Given p o s s i b l y small i n t e r - n u t r i e n t importance d i f f e r e n c e s , i t remains to be e m p i r i c a l l y determined whether load treatments constructed from a subset of more important n u t r i t i o n a l cues w i l l r e s u l t i n a greater number of purchase changes than loads c o n s i s t i n g of l e s s important n u t r i t i o n a l cues. As mentioned in-Chapter II and discussed more f u l l y i n the next chapter, measures of importance were obtained from a consumer survey before the i n - s t o r e experiment was begun. Table 2 Stated Importance Scores of 19 Cereal A t t r i b u t e s Mean A t t r i b u t e Normalized A t t r i D u t e Importance Score Taste 1.441 N u t r i t i o n a l content 1.245 Ingredients 1.217 Vitamin and mineral content 0.929 P r o t e i n content 0.697 A d d i t i v e s and p r e s e r v a t i v e s content 0.431 Sugar content 0.425 Type of cereal 0.285 Fiber content 0.257 C a l o r i e content 0.189 P r i c e 0.089 Crunchiness -0.063 Brand name -0.099 Appearance of the cereal -0.507 Aroma of cereal pieces -0.539 Sweetness -0.607 Color -1.378 Siz e of cereal pieces -1.970 Shape of cereal pieces -2.098 (From Quelch, 1978:21, Table 4) 51 H~: A point-of-purchase s i g n l i s t i n g e i g h t cues i n decreasing order of importance, from l e f t to r i g h t , y i e l d s a greater mean r i d i t than a sign l i s t i n g the same e i g h t cues i n reverse order. A t e s t of t h i s hypothesis was made p o s s i b l e by a p a r t i c u l a r aspect of the research design. Two of the information load treatments i n the experiment c o n s i s t of the i d e n t i c a l set of e i g h t cues. The only d i f f e r e n c e between these treatments i s the l e f t - t o - r i g h t arrangement of the e i g h t cues on the point-of-purchase s i g n s . On one sign the cues were p r i n t e d , from l e f t to r i g h t , i n decreasing order of importance, a f t e r s t a r t i n g with the most important cue. On another sign the same cues were p r i n t e d , from l e f t to r i g h t , i n i n c r e a s i n g order of importance, a f t e r s t a r t i n g with the l e a s t important cue (see Appendix A). Best (1978:4-5) hypothesizes t h a t information environment designs can be made more amenable to processing by consumers i f the cues s e l e c t e d are not only important to the consumer, but a l s o presented i n order of importance. He c a l l s t h i s design f a c t o r or " c o n f i g u r a l property" of the information environment i t s " h i e r a r c h i c a l order:" . . . That i s , the order i n which consumers acquire information i n a brand choice d e c i s i o n . There i s e m p i r i c a l evidence that consumers acquire information i n a meaningful and systematic manner . . . Therefore, presentations of information t h a t deviate from the consumers' a c q u i s i t i o n pattern create a s i t u a t i o n where a consumer could make a brand choice d e c i s i o n t h a t i s l e s s than optimal given the a v a i l a b l e choice set . . . t h i s property Ccan] be c o n t r o l l e d f o r by p r o v i d i n g d e c i s i o n makers with information i n order [ o f ] t h e i r preference f o r t h a t i n f o r m a t i o n . In h i s own experiment on information l o a d , Best (1978) c o n t r o l l e d f o r " h i e r a r c h i c a l order" by always presenting subjects with information matrices c o n t a i n i n g t h e i r most important cues, arrayed i n order of importance. 52 Thus, one of the 8-cue-load signs i n the present research r e -presents a meaningful " h i e r a r c h i c a l order" from most to l e a s t important cue, s i n c e people c h a r a c t e r i s t i c a l l y read a t a b l e from l e f t to r i g h t ( c f . Quelch, 1978:6). The second 8-cue-load sign presents the same set of cues arrayed i n reverse " h i e r a r c h i c a l order," from l e a s t important cue on the l e f t , to most important cue on the r i g h t . According to Best's (1878) p o s t u l a t e , the l a t t e r d i s p l a y i s l e s s e f f e c t i v e f o r consumers processing the matrix. The r e s u l t i n g response i n terms of the mean r i d i t i s expected to be smaller than the mean r i d i t f o r the former d i s p l a y . H.: . The mean r i d i t f o r a product's brand sales d i s t r i b u t i o n i s greater than 0.5 when n u t r i t i o n a l information i s placed i n a brand-by-eue matrix format at the point of purchase. This hypothesis s t a t e s t h a t , i n ge n e r a l , i f n u t r i t i o n a l information i s obtained f o r each brand, formatted i n a brand-by-cue matrix and d i s -played a t the point o f purchase, consumers w i l l base t h e i r brand choices on the brand r a t i n g s d i s c l o s e d . Obviously, any f i e l d experiment designed to assess the e f f e c t s of d i f f e r e n t information loads on consumer behaviour must i m p l i c i t l y assume that consumers w i l l use the type of information provided, i n the f i r s t place. Probably the l a r g e s t element of r i s k shouldered i n the present research endeavour was the p o s s i b i l i t y that the type of information s e l e c t e d f o r the experimental design, i . e . , n u t r i t i o n a l information posted at the poi n t of purchase, would not r e s u l t i n any behavioural changes. Somewhat m i t i g a t i n g t h i s experimental r i s k were two c o n s i d e r a t i o n s . 53 F i r s t , n u t r i t i o n a l information had been a v a i l a b l e on the packages of some non-staple foods since 1975 (Health and Welfare Canada, 1977:6). Thus, a c e r t a i n l e v e l of awareness of t h i s information could be ex-pected among consumers with perhaps a l i m i t e d degree of use i n personal d i e t d e c i s i o n s . Second, the format f o r presenting t h i s information to shoppers i n the experiment was deemed v a s t l y s u p e r i o r to the e x i s t i n g array of n u t r i t i o n a l data on i n d i v i d u a l brand packages. The arrangement of the data on brand-by-cue signs f a c i l i t a t e d i n t e r - b r a n d comparisons f o r a given product category; the signs themselves were new to the store environment and deployed i n such a way as to a t t r a c t shopper a t t e n t i o n . I t was f e l t t hat p u b l i c concern about the intake of c e r t a i n "negative" n u t r i e n t s ( f a t , sodium, c a l o r i e s ) and " p o s i t i v e " n u t r i e n t s ( p r o t e i n , f i b r e , v i t a m i n s ) , i n tandem with a su p e r i o r format f o r d i s -c l o s i n g the l e v e l s of these n u t r i e n t s i n each brand on the supermarket s h e l f , j u s t i f i e d the use of n u t r i t i o n a l cues f o r t h i s experiment on information l o a d . Support f o r t h i s researcher's optimism about the use of n u t r i t i o n a l information by consumers, given an appropriate present-a t i o n format, came from some of the stu d i e s which are reviewed below. While there i s considerable published evidence t h a t consumers want n u t r i t i o n a l information to be a v a i l a b l e to them, but show a l i m i t e d t e c h n i c a l understanding of i t ( R u d e l l , 1979; Schrayer, 1978; Jacoby, Chestnut and Silberman, 1977; Murray, 1977; Daly, 1976; Lenahan et a l . , 1973), fewer studies have assessed whether they make use of t h i s i n f o r m a t i o n . With one or two exceptions, the s t u d i e s focusing s p e c i f i c a l l y on n u t r i t i o n a l information use are e i t h e r surveys or l a b o r a t o r y research. 54 Kendall arid Fenwick. A f i e l d study to determine the extent o f la b e l reading by shoppers was c a r r i e d out i n two H a l i f a x , Canada super-markets by Kendall and Fenwick (1979). Researchers, dressed as super-market personnel, were s t a t i o n e d i n an a i s l e and un o b t r u s i v e l y recorded the time that every purchaser of the t e s t products spent examining the l a b e l s of the brands before moving on. The foods were r i c e , pasta products, canned m e a t / f i s h , and powdered (dehydrated) soup. Observations were made over a fi v e - d a y period and covered 662 purchasers of those products. Kendall and Fenwick (1979:155) dichotomized the shoppers thus observed i n t o '"grabbers' who simply snatched the product from the s h e l f with v i r t u a l l y no d i s c e r n i b l e product viewing time, and 'lookers' who observed the products f o r at l e a s t one second." They found that only 25% of the 662 buyers were "grabbers", with the other 75% r e v e a l i n g a t l e a s t some i n t e r e s t i n gathering product i n f o r m a t i o n . Of p a r t i c u l a r i n t e r e s t i s the extent of time which the average "looker" devoted to a product choice: . . . Furthermore, of those showing some i n t e r e s t , over h a l f spent more than 8 seconds with the products, g i v i n g a mean product viewing time of 38 seconds. Although no s p e c i f i c i nformation was gathered on the p a r t i c u l a r product/package/label a t t r i b u t e s t h a t warranted t h i s a t t e n t i o n time, the extent of exposure suggests much more information p r o c e s s i n g , or product comparisons, than i s normally assumed . . . ( I b i d . , 156). Kendall and Fenwick (1979) note t h a t the extent of l a b e l reading uncovered i n t h e i r study c o n f l i c t s with a number of o l d e r s e l f - r e p o r t surveys which found a l e s s e r degree (30-50%) of claimed product information usage. 55 In a d d i t i o n to the time recording procedure, a random sample of the shoppers observed was interviewed by other researchers i n a d i f f e r e n t part o f the s t o r e . As part of the i n t e r v i e w , subjects were shown nine mock l a b e l s f o r a new food product and asked to rank these according to h e l p f u l n e s s i n making a buying d e c i s i o n f o r the new product. Each l a b e l contained a v a r i a t i o n on four types o f inform a t i o n : brand name, n u t r i t i o n a l i n f o r m a t i o n , i n g r e d i e n t i n f o r m a t i o n , and open-dating i n f o r m a t i o n . Using c o n j o i n t a n a l y s i s to analyze the preference rankings f o r the l a b e l s , Kendall and Fenwick (1979:158) were able to determine the importance of each type of information to "grabbers" and "looke r s " : Of p a r t i c u l a r i n t e r e s t i s the d i f f e r e n c e i n the order of importance o f information items between the two groups. For the average grabber brand name i s the most important information source, followed by i n -gredient i n f o r m a t i o n . The average l o o k e r , however, i s most concerned with n u t r i t i o n i n f o r m a t i o n , although i n g r e d i e n t s are a c l o s e second. The importance of n u t r i t i o n information when buying a new food product i s most i n t e r e s t i n g as Canada, u n l i k e the U.S.A., has no mandatory l e g i s l a t i o n on n u t r i t i o n i n f o r m a t i o n . Ottawa, and Consumer and Corporate A f f a i r s , has been r e t i c e n t about such l e g i s l a t i o n and l i t t l e e m p i r i c a l research has been reported . . . Kendall and Fenwick (1979:159) concluded t h a t , f o r the m a j o r i t y o f shop-pers interviewed i n the s t o r e , n u t r i t i o n a l and i n g r e d i e n t information dominate t h e i r l a b e l preferences, at l e a s t with respect to choosing a new food product. F r e d r i c a R u d e l l . A l a b o r a t o r y study of the c o g n i t i v e and behavioural e f f e c t s of presenting n u t r i t i o n a l information to 183 housewives was c a r r i e d out by Rudell (1979). Respondents were i n t i t i a l l y asked to make one choice from each of three product groups: e i t h e r whole or skim m i l k , 56 e i t h e r white or whole wheat bread, e i t h e r r e a l or i m i t a t i o n bacon. Following t h i s , they were i n v i t e d to examine as many or as few i n -formation cards as they wished c o n t a i n i n g n u t r i t i o n a l information on each a l t e r n a t i v e (e.g. "A one cup se r v i n g of whole milk contains 8 grams of p r o t e i n , which i s 20% of the recommended d a i l y allowance"). The information p r i n t e d on the cards im p l i e d that one a l t e r n a t i v e from each product group was more n u t r i t i o u s than the other. Having acquired t h i s i n f o r m a t i o n , subjects were again asked to choose one a l t e r n a t i v e from the three product groups. Rudell (1979:87) found that between 11 and 14 percent of respondents changed t h e i r choices a f t e r exposure to the n u t r i t i o n a l data, and t h a t the major s h i f t s i n choice were to the more n u t r i t i o u s a l t e r n a t i v e s . John Quelch. Quelch (1978) used an information d i s p l a y board technique to determine the importance of n u t r i t i o n a l cues i n the formation of brand choice. This d i s p l a y board i s a p h y s i c a l r e p l i c a of a brand-by-cue information matrix where the i n d i v i d u a l e n t r i e s of the matrix c o n s i s t o f cards on which a brand r a t i n g has been p r i n t e d . Subjects are i n s t r u c t e d to gather as much or as l i t t l e i nformation as they need, i n order to make a brand c h o i c e , by c o n s u l t i n g the brand r a t i n g s cards i n the d i s p l a y board. R e c r u i t i n g 250 housewives at a shopping mall Quelch (1978) had h i s subjects acquire information cards on e i t h e r four or f i v e cues f o r breakfast cereal before making a brand choice. Although the n u t r i t i o n a l information entered on the cards had been taken from the packages of the s i x brands used i n the e x e r c i s e , the actual brand i d e n t i t i e s were d i s g u i s e d . One treatment group had access to a p h y s i c a l sample i n order 57 to t a s t e a cereal brand i f t h i s cue was e l i c i t e d ; t h i s cue was not a v a i l a b l e to a second group. Aggregating across s u b j e c t s , the frequency with which each of the information cues was used by members of the two treatment groups, be-fore making a brand c h o i c e , i s shown i n Table 3. Table 3 Aggregate Frequencies o f Cue Use Leading to a Brand Choice Treatment Groups No Physical Physical Sample (NPS) Sample (. n=124 n=126 Ingredient Information 91 66 Nutrition Information 91 70 Physical Appearance 50 20 Price 47 44 Physical Sample N/A 86 Total Elicitations 279 286 (From Quelch, 1978:15, Table 1) With no p h y s i c a l sample of the cereal a v a i l a b l e f o r t r i a l , n u t r i t i o n a l and i n g r e d i e n t information were the two most f r e q u e n t l y used cues i n forming a brand preference. With access to a ph y s i c a l sample, n u t r i t i o n a l information was the second most f r e q u e n t l y used cue. 58 Others. Friedman (1972) reported two e a r l y s t u d i e s i n the f i e l d which gauged the i n i t i a l e f f e c t s of n u t r i e n t - l a b e l e d food on purchase patterns. One study found small s h i f t s i n purchases i n the d i r e c t i o n of n u t r i e n t - l a b e l e d food f o l l o w i n g i t s market i n t r o d u c t i o n . A second study, by the Consumer Research I n s t i t u t e , examined the e f f e c t s of i n c l u d i n g n u t r i e n t i n f o r m a t i o n , i n the c a t a l o g d e s c r i p t i o n s of various food products, on households who shopped by c a t a l o g . There was some evidence t h a t purchase patterns s h i f t e d toward brands with n u t r i t i o n a l advantages. Asam and Buck!in (1973), presented grocery shoppers with d i f f e r e n t l a b e l s on bogus brands of canned peas. To a l i m i t e d e x tent, n u t r i t i o n a l cues made shoppers' a t t i t u d e s and preferences s h i f t toward brands rated n u t r i t i o n a l l y higher. A s i m i l a r f i n d i n g was made by Stanley (1977) a f t e r subjects r e -ceived Consumer Reports r a t i n g s on the n u t r i t i o n a l content of breakfast c e r e a l s . However, Peterson's (1977) experiment with bread samples of d i f f e r e n t darknesses revealed that people's preconceptions about bread c o l o r were unaffected by n u t r i t i o n a l information cues. Manipulating n u t r i t i o n a l cues on bread l a b e l s d i d not a l t e r s u b j e c t s ' r e l i a n c e on the c o l o r cue as an i n d i c a n t o f the n u t r i t i o n a l q u a l i t y o f the bread. Scammon (1977) found that b r a n d - s p e c i f i c n u t r i t i o n a l r a t i n g s d i s -closed to subjects i n t e l e v i s i o n commercial format s i g n i f i c a n t l y a f f e c t e d t h e i r a b i l i t y to c o r r e c t l y i d e n t i f y the more n u t r i t i o u s o f two brands of peanut b u t t e r . However, the same exposure to n u t r i t i o n a l information did not a f f e c t s u b j e c t s ' preferences f o r one or the other brand. 59 The Issue of Information Format. A second aspect of hypothesis i s that consumer use of n u t r i t i o n a l information i s contingent upon d i s p l a y i n g the n u t r i t i o n a l cues i n a brand-by-cue matrix format, at the point o f purchase. Although presentation format was not discussed i n the preceding review o f the extent to which consumers use n u t r i t i o n a l i n f o r m a t i o n , the two issues are ob v i o u s l y i n t e r r e l a t e d . As was i n d i c a t e d i n Chapter I , the p resentation format of any o b j e c t i v e product information environment w i l l determine the demands made on a consumer's short-term memory, which i s the c o n s t r i c t i n g element i n information processing a c t i v i t i e s . Information formats which make product data e a s i l y a c c e s s i b l e , under-standable to the consumer, and which f a c i l i t a t e comparisons among brands w i l l be more e f f e c t i v e i n changing brand choice behaviour than information formats which do not (Day, 1976). The three s t u d i e s reviewed below underscore the importance o f choosing an appropriate format to make a product information environment u s e r - e f f e c t i v e . J . Edward Russo. In two f i e l d experiments, Russo (1977) and Russo, K r i e s e r and M i y a s h i t a (1975) showed that consumer usage o f i n - s t o r e information w i l l increase when data p r e v i o u s l y a v a i l a b l e i n an i n f e r i o r  format are reorganized i n t o a format which f a c i l i t a t e s processing. For example, Russo (1977) presented u n i t p r i c e s i n l i s t format f o r several products i n s i d e a supermarket. I n i t i a l l y , the u n i t p r i c e s had been a v a i l a b l e only i n the t r a d i t i o n a l format, i . e . on separate s h e l f tags. The s h e l f - t a g format was then supplemented with p r i n t e d signs on which 60 brands and p r i c e s were l i s t e d i n order of i n c r e a s i n g p r i c e - p e r - u n i t . These signs were posted beside the product s h e l v i n g and g r e a t l y f a c i l i t -ated the d i r e c t comparison of p r i c e s f o r brands with m u l t i p l e container s i z e s . The a d d i t i o n of t h i s l i s t format to the store's previous information environment r e s u l t e d i n an average saving to consumers of about 2tt per purchase, compared to the average cost o f a purchase under the t r a d i t i o n a l u n i t p r i c e format. M. Venkatesan. Venkatesan (1977) describes three presentation formats f o r n u t r i t i o n a l information on various food products which were te s t e d on housewives i n H a l i f a x , Canada, using the t e l e v i s i o n and magazine media. The e f f e c t i v e n e s s c r i t e r i a were several measures of information r e c a l l and the a b i l i t y to determine which products were more n u t r i t i o u s . Two of the formats c o n s i s t e d of d i f f e r e n t composite i n d i c e s to summarize a product's n u t r i t i o n a l value i n a s i n g l e i n t e g e r . In g e n e r a l , both composite-index formats were more e f f e c t i v e than a t h i r d multidimensional format presenting U.S. RDA l i s t i n g s of nine n u t r i t i o n a l cues. Debra Scammon. Scammon (1977) experimented with two formats of presenting n u t r i t i o n a l information i n t e l e v i s i o n commercials f o r a branded food product. The n u t r i e n t information was presented e i t h e r as a percentage of U.S. RDA (Recommended D a i l y Allowance) or as an a d j e c t i v a l d e s c r i p t i o n of content ( E x c e l l e n t , Good, F a i r ) . See Table 1. A b i l i t y to i d e n t i f y the "more n u t r i t i o u s " brand was greater among people exposed to the simpler a d j e c t i v a l format than among those exposed to the percentage format. 61 Summary. Regarding consumer use of n u t r i t i o n a l information i n gen e r a l , the studies reviewed here show an encouraging tendency f o r n u t r i t i o n a l cues to i n f l u e n c e brand choice when n u t r i t i o n a l d i f f e r e n c e s are found between choice a l t e r n a t i v e s . However, the evidence i s by no means i r r e f u t a b l e and c o n s i s t e n t ; f i n d i n g s from some of the stu d i e s are c o n t r a d i c t o r y . Published research t r a c i n g changes i n actual purchase behaviour to the a v a i l a b i l i t y of n u t r i t i o n a l cues i s c e r t a i n l y r a r e . Thus, the evidence c i t e d on n u t r i t i o n a l information use w i l l be tempered by the f a c t t h a t most of i t was obtained from subjects who were f u l l y aware th a t they were p a r t i c i p a t i n g i n a study. Concerning information presentation format, the research by Russo c l e a r l y i n d i c a t e s t h a t consumers use of e x i s t i n g information i n a r e t a i l s t o r e w i l l increase i f the same information i s made e a s i l y a c c e s s i b l e and reorganized to f a c i l i t a t e comparisons among competing items. The s i t u a t i o n i n the present f i e l d experiment i s much l i k e t h a t i n the experiments by'.Russo (1977) and Russo, K r i e s e r and Miyashita (1975). N u t r i t i o n a l information on c e r t a i n products i n the t e s t stores i s normally a v a i l a b l e . o n the i n d i v i d u a l brand packages, i f at a l l . With the same type of information organized i n t o a complete brand-by-cue matrix and p r i n t e d on a s i n g l e point-of-purchase s i g n , the proportion of brand choices based on t h i s information i s expected to increase. Hr-: Following the removal o f the point-of-purchase information from the s t o r e s , the mean r i d i t f o r a product's weekly brand sales d i s t r i b u t i o n remains at the experimental b a s e l i n e l e v e l o f 0.5. 62 In t h i s hypothesis, a product's weekly brand sales d i s t r i b u t i o n , observed during the post-experimental weeks (signs removed), i s compared with the brand s a l e s d i s t r i b u t i o n observed i n experimental c o n t r o l periods during the experimental weeks. This hypothesis p r e d i c t s t h a t the brand s a l e s d i s t r i b u t i o n s f o r post-experimental and experimental c o n t r o l w i l l not d i f f e r , hence the mean r i d i t w i l l remain at 0.5. Worded d i f f e r e n t l y , the hypothesis s t a t e s t h a t the observed mean r i d i t d i f f e r e n c e s between load treatments and experimental c o n t r o l are not due to sampling e r r o r because experimental c o n t r o l and post-experimental brand s a l e s d i s t r i b u t i o n s w i l l be the same. Removal of the point-of-purchase signs represents a r e t u r n - t o -bas e l i n e sequence i n the 5-week experiment and acceptance o f t h i s hypo-t h e s i s would give credence to the causal e f f e c t of the information on 2 purchase behaviour. In Russo's (1977) i n - s t o r e u n i t p r i c i n g format experiment, a f t e r the l i s t s of u n i t p r i c e s had been i n place f o r several weeks they were again replaced by the t r a d i t i o n a l format of separate u n i t p r i c e s h e l f tags. Continued monitoring of the t e s t product sales revealed that the average per-purchase savings brought on by the s u p e r i o r l i s t format per-s i s t e d f o r about two weeks. T h e r e a f t e r , the average p r i c e per u n i t paid by shoppers showed a complete r e t u r n to the l e v e l i n i t i a l l y observed under the t r a d i t i o n a l format. This hypothesis i s also p r o d u c t - s p e c i f i c since " i t requires comparisons The a u t h o r i s g r a t e f u l to Professor Gerald Albaum, U n i v e r s i t y o f Oregon, f o r suggesting t h i s hypothesis as a t e s t of c a u s a l i t y . 63 to be made across several weeks of sales data. Since week-to-week price/promotional changes were expected to a f f e c t at l e a s t some of the t e s t product s a l e s d a t a , t h i s hypothesis was t e s t e d on each product, s e p a r a t e l y , whenever the weekly data permitted v a l i d comparisons across d i f f e r e n t weeks. Other Related Research Subsequent to the i n i t i a t i o n of t h i s d i s s e r t a t i o n , a study of n u t r i t i o n a l information usage i n the grocery s t o r e was a l s o i n progress i n the United S t a t e s . Professor J . Edward Russo.(1981a) of the U n i v e r s i t y of Chicago conducted a 30-week f i e l d experiment i n several supermarkets to assess the l o n g i t u d i n a l e f f e c t s o f posting a l t e r n a t i v e formats of n u t r i t i o n a l information on awareness, a t t i t u d e s and purchase behaviour of shoppers. The r e s u l t s o f t h i s study are not yet a v a i l a b l e . Conclusions The l i t e r a t u r e synthesized and summarized f o r each of the f i v e hypotheses i n t h i s chapter aimed to present evidence i n support of the o p e r a t i o n a l i z e d research questions of i n t e r e s t i n t h i s study. Never-t h e l e s s , the l i t e r a t u r e reviewed was by no means unanimously i n support of each hypothesis. This f a c t more or l e s s r e f l e c t s the s t a t e of the a r t with respect to information processing questions addressed i n past consumer research and remaining only p a r t l y r e s o l v e d . There i s c e r t a i n l y a need f o r f u r t h e r t e s t s of the present hypotheses r e l a t i n g to consumer responses to product i n f o r m a t i o n , i n g e n e r a l , and product information l o a d , i n p a r t i c u l a r . 64 CHAPTER IV RESEARCH METHODOLOGY This chapter presents a step-by-step approach to the research design so tha t the appropriate data could be c o l l e c t e d and analyzed i n order to t e s t the research hypotheses. Following a b r i e f overview of the design o b j e c t i v e s , the bas i c experimental design i s described. This i s followed by several s e c t i o n s d e a l i n g with the i n t r i c a t e de-velopment o f the research design and covering t e s t product s e l e c t i o n , information cue s e l e c t i o n , a consumer survey to obta i n measures of importance and u t i l i t y on the cues s e l e c t e d , stimulus c o n s t r u c t i o n , experimental s t o r e s e l e c t i o n , and experimental procedure. A d i s c u s s i o n of the dependent v a r i a b l e precedes the secti o n s on database manipulation and the methodological approach used. The chapter ends with a des-c r i p t i o n o f the s t a t i s t i c a l t e s t s employed. Overview In Chapter I , the general o u t l i n e of t h i s d i s s e r t a t i o n i n d i c a t e d t h a t an input-output research design was se l e c t e d to assess the e f f e c t s of information load i n a r e a l i s t i c f i e l d s e t t i n g . The focus o f the present research was not on i n t e r v e n i n g c o g n i t i v e processes r e s u l t i n g from exposure to information but r a t h e r on the "end-s t a t e s " of consumers i n terms of actual , u n o b t r u s i v e l y measured behaviour. This c o n s i d e r a t i o n e l i m i n a t e d several a l t e r n a t i v e information processing methodologies, e.g., eye f i x a t i o n s , information d i s p l a y boards, verbal p r o t o c o l s . These techniques are a p t l y s u i t e d to t r a c i n g intermediate 65 c o g n i t i v e a c t i v i t i e s between stimulus input and behavioural response, but are u s u a l l y r e s t r i c t e d to f a i r l y o b t r u s i v e l a b o r a t o r y - l i k e s e t t i n g s . 1 Since the o v e r r i d i n g goal of t h i s research was to extend the f i n d i n g s from past l a b o r a t o r y s t u d i e s , the choice of an input-output design r e f l e c t e d the d e s i r e to o b t a i n e m p i r i c a l data of high v a l i d i t y by l o c a t i n g the experiment i n a natural consumer s e t t i n g . Moreover, the output generated by the experiment was recorded i n the most un-o b t r u s i v e manner p o s s i b l e , by t r a c k i n g the purchases of st o r e customers e n t i r e l y unaware o f t h e i r p a r t i c i p a t i o n i n an experiment. The experimental variables" were determined by the hypotheses stated i n the previous chapter. The input v a r i a b l e s c o n s i s t e d of information load and the importance of the i n d i v i d u a l information cues. These were manipulated w i t h i n the context of a f i x e d p r e s e n t a t i o n format. The output v a r i a b l e c o n s i s t e d of store s a l e s data which were s t a t i s t i c a l l y analyzed to reveal whether s t o r e customers were i n c o r p o r a t i n g the inputs i n t h e i r brand choice d e c i s i o n s . This a n a l y s i s , i n t u r n , was designed to t e s t a model from information processing theory and to c o n t r i b u t e to p o l i c y - o r i e n t e d research on the e f f e c t i v e n e s s of n u t r i t i o n a l information d i s p l a y e d at the point of s a l e . Throughout the development of the research design, p a r t i c u l a r a t t e n t i o n was paid to avoid repeating the design and a n a l y t i c a l e r r o r s committed i n the e a r l i e s t consumer information load experiments of Russo (1978) provides a comprehensive comparison among several methodologies f o r consumer information processing research on the basis of data q u a l i t y , a p p l i c a b i l i t y and c o s t . 66 Jacoby, S p e l l e r and Kohn (1974 a; b). Therefore, at appropriate stages i n the d e s c r i p t i o n of the methodology, the s p e c i f i c procedures taken to remedy these d e f i c i e n c i e s w i l l be pointed out. Research Design The basic experiment manipulated two f a c t o r s and employed a 4 (load l e v e l s ) x 2 (cue importance l e v e l s ) between-treatments a n a l y s i s of variance design. Four information load l e v e l s were constructed from e i t h e r 1, 2, 4 or 8 cues; two cue importance l e v e l s were created by employing a subset of e i t h e r r e l a t i v e l y important cues or r e l a t i v e l y unimportant cues i n the load treatments. The basic design, t h e r e f o r e , c o n s i s t e d of e i g h t experimental t r e a t -ments and was executed w i t h i n one e n t i r e week of a store's' business hours. With the basic design e s t a b l i s h e d , numerous r e p l i c a t i o n s of the same design served to generate m u l t i p l e within-treatment observations f o r t e s t i n g hypotheses to H^. Towards t h i s end, the same complete 4 x 2 design was r e p l i c a t e d : 1. With each of s i x products, 2. In each o f two s t o r e s , 3. In each o f two successive shopping weeks. Subsequently, the sales data f o r each treatment with each product were pooled across the two s t o r e s , so as to reduce treatment sampling e r r o r because some t e s t product brands had r e l a t i v e l y low s a l e s frequencies i n i n d i v i d u a l s t o r e treatment periods. The combined store data on s i x products and two weekly measurements could provide up to 12 67 observations of the dependent measure f o r each of the e i g h t ( 4 x 2 ) experimental treatments, depending on which of several s t a t i s t i c a l analyses was subsequently used. Because the m u l t i p l e observations w i t h i n experimental treatments v a r i e d over products and weeks, each datum e n t e r i n g the between-treatments a n a l y s i s was a c t u a l l y expressed as a change from b a s e l i n e . To permit t h i s , a unique bas e l i n e or. c o n t r o l was incorporated i n the design f o r each i n d i v i d u a l product i n each o f the two experimental weeks. These baselines were obtained from s t o r e s a l e s data c o l l e c t e d during periods when the treatment signs were e i t h e r removed or replaced with c o n t r o l signs comprised of a l i s t i n g o f the t e s t product's brands, but g i v i n g no in f o r m a t i o n . Furthermore, because the m u l t i p l e within-treatment observations were based on data from products with d i f f e r e n t numbers of brands, i n those s t a t i s t i c a l analyses where main e f f e c t s were aggregated across products, the number of brands i n a product was s t a t i s t i c a l l y c o n t r o l l e d f o r . The e i g h t treatments and c o n t r o l s f o r each product were randomly assigned to designated time s l o t s created from a store's t o t a l weekly business hours. The complete f a c t o r i a l experiment was executed and the r e s p e c t i v e sales measurements obtained simultaneously i n both s t o r e s . Since within-week experimental c o n t r o l s were b u i l t i n t o the design, i t was unnecessary to use one store as a c o n t r o l while the experiment proceeded i n the second s t o r e . The simultaneous execution of the design i n both stores o f f e r e d the f l e x i b i l i t y of pooling the brand sales data across 68 stores i n order to increase sample s i z e s , i f low treatment sales -volumes warranted t h i s . A l t e r n a t i v e l y , the s t o r e s ' data could be taken as separate treatment observations since the randomized block design used 2 permitted the b l o c k i n g of observations on s t o r e s . In a d d i t i o n to the basic f a c t o r i a l experiment performed i n two successive weeks, t h i s study incorporated a l o n g i t u d i n a l b e f o r e - a f t e r design to i n v e s t i g a t e the existence of a causal r e l a t i o n s h i p between the posting of information and changes i n brand s a l e s . Therefore, f i e l d measurements extended over f i v e consecutive weeks and permitted a t e s t of hypothesis Hg. The execution of the b e f o r e - a f t e r design can be represented i n i t s temporal sequence with the f o l l o w i n g n o t a t i o n , read from l e f t to r i g h t : Store 1: 01 X 0 2 X 0 3 0h 0 5 Store 2: Ol X 0 2 X O3 0^ O5 where the 0^  and 0"C represent the sales measurements taken simultaneously i n both stores i n Week i , and X represents the execution of the 4 x 2 f a c t o r i a l experiment described above. As can be seen, Weeks 4 and 5 of the experiment allowed f o r r e t u r n - t o - b a s e l i n e measurements, f o l l o w i n g the permanent removal of the point-of-purchase signs from the s t o r e . This b e f o r e - a f t e r design was applied to a l l s i x products, i n both s t o r e s . The h e l p f u l suggestions of Professor Gerald Albaum, U n i v e r s i t y of Oregon, are acknowledged i n t h i s regard. 69 Rationale For Using N u t r i t i o n a l Cues The r a t i o n a l e f o r choosing information p e r t a i n i n g to n u t r i t i o n to serve as the information s t i m u l i needs to be thoroughly understood before d e t a i l s of the experimental design are presented. The d e c i s i o n to u t i l i z e n u t r i t i o n a l cues was an important p r e l i m i n a r y step i n the e a r l i e s t stages of design development. A fundamental design o b j e c t i v e was to f i n d an appropriate type of consumer information which would s a t i s f y the f o l l o w i n g three c r i t e r i a : 1. The information should c o n s i s t of o b j e c t i v e , performance-o r i e n t e d product data; 2. The information should lend i t s e l f to p a r t i t i o n i n g i n t o d i f f e r e n t subsets of numerous i n d i v i d u a l cues so that information load could be p r e c i s e l y defined and c o n t r o l l e d ; 3. The information should be a v a i l a b l e from manufacturers and be s u i t a b l e f o r exposure i n the marketplace so t h a t the immediate impact of exposure could be measured i n terms of actual consumer purchases. A studious e v a l u a t i o n of various types of information to meet the above design c r i t e r i a ended i n the s e l e c t i o n of data on the n u t r i t i v e composition of branded food p r o d u c t s . 3 In p a r t i c u l a r , i n d i v i d u a l n u t r i e n t s served as cues, with n u t r i t i v e values d i s c l o s e d on every brand. In terms of the three c r i t e r i a , n u t r i t i o n a l information can be o b j e c t i v e l y described on the basis of t e s t e d performance (e.g. Brand A contains Some a l t e r n a t i v e types of information which were considered: t e x t i l e composition l a b e l l i n g ; c l o t h i n g care l a b e l l i n g ; product package information (country of o r i g i n , open date, p r i c e , net weight); household detergent performance r a t i n g s . 70 623 mg of Sodium per lOOg of packaged product); n u t r i t i o n a l information can be p a r t i t i o n e d i n t o numerous i n d i v i d u a l cues (e.g. sodium, calcium, f a t , p r o t e i n , vitamin B^); the n u t r i t i v e values of each brand were a v a i l a b l e from the manufacturers and could be posted at the p o i n t of purchase. Another p o t e n t i a l design problem may have been extenuated through the e x c l u s i v e use of n u t r i t i o n a l i n f o r m a t i o n . In e a r l i e r l a b o r a t o r y studies of consumer information l o a d , Jacoby, S p e l l e r and Kohn (1974 a; b) had employed very d i v e r s e types of information cues to c o n s t r u c t d i f f e r e n t information loads (e.g. p r i c e , storage i n s t r u c t i o n , c o n t a iner s i z e , c a l o r i e s - p e r - s e r v i n g f o r food products; enzyme content, p r i c e , q u a n t i t y required per wash load f o r detergents.) Day (1976:48) has pointed out the p o s s i b i l i t y t h a t such widely divergent cues, i n themselves, present widely d i f f e r e n t processing tasks to consumers and are, t h e r e f o r e , not e q u i v a l e n t f o r the purposes of d e f i n i n g information load as the sheer number of cues present: The relevance of most research or information processing i s f u r t h e r compromised by the d i f f i c u l t y of adequately d e f i n i n g information l o a d . The emphasis has been on the q u a n t i t y of i n f o r m a t i o n , f o r example, number of cues per a l t e r n a t i v e , number of a l t e r n a t i v e s , and the t o t a l set of cues competing f o r the i n d i v i d u a l ' s a t t e n t i o n . But not a l l cues are e q u i v a l e n t . A f a m i l i a r cue may be more e a s i l y used than a t e c h n i c a l cue, which i t s e l f may be a composite of several f a c t o r s . Thus, i t i s f a l l a c i o u s to equate a f a m i l i a r q u a n t i t y cue, such as ounces, with the r e c e n t l y proposed measure of cost per e f f e c t i v e wash of detergent as having the same demands on consumer i n f o r m a t i o n -processing c a p a b i l i t y . I f cDay's (1976) contention i s t r u e , then the e x c l u s i v e use of n u t r i ^ t i o n a l cues should minimize the methodological problem of mixing q u a l i t a t i v e l y d i s s i m i l a r cues while equating t h e i r i n d i v i d u a l c o n t r i b u t i o n 71 toward t o t a l i nformation load. Test Products In the i n i t i a l stages of the research design, a d e t a i l e d l i s t o f 18 p o t e n t i a l packaged food products was drawn up. This l i s t i n g c o n s i s t e d of data on a l l i n v e n t o r i e d brands and package s i z e s of each product and t h e i r a v a i l a b i l i t y i n the two stores s e l e c t e d f o r the experiment. Further-more, the weekly s a l e s volumes of each p o t e n t i a l t e s t product brand i n the two s t o r e s was obtained from the cooperating supermarket chain. Co n s u l t a t i o n s with the management of the cooperating c h a i n , over a period of four months, l e d to a short l i s t of 11 p o t e n t i a l products, on the basis of a l o g i s t i c a l l y f e a s i b l e number of brands, a c c e p t a b i l i t y to the management, and high u n i t sales volume. Working from t h i s l i s t , l e t t e r s were sent by the chain to manufacturers of a l l brands i n v o l v e d , e x p l a i n i n g the purpose of the research and requesting t h e i r cooperation i n supplying the n u t r i t i v e composition of each brand. A l l but one manufacturer cooperated and, w i t h i n three months from the i n i t i a l request, the n u t r i t i o n a l data had been received and t a b u l a t e d . A f i n a l s e l e c t i o n o f s i x t e s t products was made on the basis of the f o l l o w i n g c r i t e r i a : 1. The product had a r e l a t i v e l y high weekly s a l e s volume; 2. The product c o n s i s t e d of a l o g i s t i c a l l y and a n a l y t i c a l l y f e a s i b l e number of brands and package s i z e s ( i t e m s ) ; 3. The products avoided redundancy i n food c a t e g o r i e s ; 4. A l l packages of the product were l a b e l l e d with the Universal Product Code (U.P.C.), since item movements i n the 72 stores were recorded on e l e c t r o n i c checkout f a c i l i t i e s ; 5. At l e a s t e i g h t d i f f e r e n t n u t r i t i o n a l information cues had been provided by the manufacturer o f each a v a i l a b l e brand of the product. A t e s t product set s i z e of s i x was judged to be l a r g e enough to permit the d e l e t i o n of a product i f u n c o n t r o l l a b l e f a c t o r s during the experiment c r i t i c a l l y a f f e c t e d i t s sales data, yet small enough f o r the time required to manually key-in 42 i n d i v i d u a l U.P.C. coded items and o b t a i n s a l e s measurements at the automated checkout. F i n a l l y , the need f o r m u l t i p l e products to serve as design r e p l i c a t e s had to be balanced against the cost and labour of c o n s t r u c t i n g 18 p o i n t - o f -purchase information signs f o r each product, f o r a t o t a l of 108 two-sided s t o r e s i g n s . The s i x products used i n the experiment were 1. Canned cream-of-mushroom soup (2 brands/3 items) 2. Tomato ketchup (2 brands/7 items) 3. Macaroni & cheese dinner (3 brands/3 items) 4. Mayonnaise (3 brands/6 items) 5. Bran-type breakfast c e r e a l (7 brands/10 items) 6. Peanut b u t t e r (9 brands/13 items) N u t r i t i o n a l Cue S e l e c t i o n A l i s t i n g of the n u t r i t i v e composition data obtained on each product i s contained i n Appendix B. Unless otherwise noted i n the appendix, only the usable cues have been l i s t e d , i . e . , cues f o r which data were a v a i l a b l e on every brand of a product. The t o t a l number of usable cues 73 f o r each product exceeded the e i g h t required to implement the experimental design and v a r i e d between 9 and 13 (see Table 4). The exception was mayonnaise, f o r which e x a c t l y e i g h t n u t r i t i o n a l cues became a v a i l a b l e , of which only seven cues gave data on two or a l l three brands.^ Eight cues were judgmentally s e l e c t e d f o r a product i n a stepwise f a s h i o n . I n i t i a l l y , brand values on a l l usable cues o f a product were inspected to see whether or not a s i n g l e brand ( o r , perhaps, a small subset of brands) rated the highest i n n u t r i t i v e performance on at l e a s t e i g h t cues. A single-brand dominance across nine cues was discovered i n the brand r a t i n g s of canned soup and macaroni & cheese dinner: i n each of these products, one brand outperformed the remaining brand(s) on nine cues. In the case of bran c e r e a l , three of the seven brands were found to c o l l e c t i v e l y dominate on e i g h t cues i n the sense that one of these three brands rated e i t h e r f i r s t , second or t h i r d highest i n n u t r i t i v e performance on e i g h t of the a v a i l a b l e cues. Of the two tomato ketchup brands, one was found to have the higher r a t i n g on f i v e cues. For peanut b u t t e r , no brand rated the highest i n n u t r i t i v e performance on more than f o u r cues, and no small subset of brands was found which could improve the number of s u p e r i o r r a t i n g s . HThe product mayonnaise-type salad dressing had i n i t i a l l y been s e l e c t e d as one o f the s i x products to be used. J u s t p r i o r to the experiment, the stores d e l i s t e d one of three brands of t h i s product. A last-minute d e c i s i o n was made to s u b s t i t u t e r e a l mayonnaise as the s i x t h product f o r which seven n u t r i t i o n a l cues were a v a i l a b l e , but with two missing values on one o f the three brands (see Appendix B, Table B-4). Table 4 Extent to Which Test Product Brands Were Found To Dominate i n N u t r i t i o n a l Performance Product No. of Brands Total No. of Usable Cues A v a i l able Max. No. of Cues on Which Any One Brand Found to Rate Highest Canned Soup 2 13 9 M&C Dinner 3 10 9 Bran Cereal 7 16 a 8 C Mayonnaise 3 7 b 4 Ketchup 2 9 5 Peanut Bu t t e r 9 13 4 Includes three cues with missing values on one brand ^Includes two cues with missing values on one brand cThree of the brands c o l l e c t i v e l y dominated the r e s t on 8 cues 75 The extent to which brands were found to dominate i n n u t r i t i o n a l performance w i t h i n each product-is summarized i n Table 4. Among the two products with a brand which rated highest on nine cues, one cue was deleted from each, more or l e s s a r b i t r a r i l y . In canned soup, thiamine was e l i m i n a t e d because i t s l e v e l s i n the two brands were judged to be the most n e g l i g i b l e among the nine cues. In m&c di n n e r , c a l o r i e s was e l i m i n a t e d because i t s l e v e l s i n two of the three brands were almost i d e n t i c a l . Thus, apart from the e i g h t cues already predetermined f o r mayonnaise, the f i r s t d e c i s i o n a l step i d e n t i f i e d the e i g h t cues to be used i n the experiment f o r 1. Canned soup, 2. M&c di n n e r , and 3. Bran c e r e a l . This primary cue s e l e c t i o n procedure was n e c e s s i t a t e d by the a n a l y t i c a l design o f the dependent v a r i a b l e which i s described l a t e r i n t h i s chapter. B r i e f l y , d i r e c t i o n a l changes i n purchase frequencies from l e s s n u t r i t i o u s to more n u t r i t i o u s brands can be traced with g r e a t e r s e n s i t i -v i t y i f one or a small subset of brands i s dominant i n n u t r i t i v e per-formance, regardless of the information load employed. For the two remaining products, the cue s e l e c t i o n procedure described above d i d not y i e l d at l e a s t e i g h t cues. Therefore, of the nine usable cues a v a i l a b l e f o r tomato ketchup, one was a r b i t r a r i l y e l i m i n a t e d . For peanut b u t t e r , the extent of between-brand d i f f e r e n c e s was 76 r e l a t i v e l y low across a l l 13 usable cues. Therefore, the ei g h t cues f o r t h i s product were a r b i t r a r i l y s e l e c t e d to in c l u d e c a l o r i e s , p r o t e i n and f a t , two d i f f e r e n t v i t a m i n s , and three d i f f e r e n t minerals. The set of e i g h t cues s e l e c t e d f o r each product i s d i s c l o s e d i n the sign r e p l i c a s given i n Appendix A. Perhaps the cue s e l e c t i o n procedure described here provokes the su s p i c i o n that n u t r i e n t s which are r e l a t i v e l y important to consumers were e l i m i n a t e d from some of the products. This p o s s i b i l i t y was i n v e s t i g a t e d and the cases c i t e d below tend to confirm t h a t the cues s e l e c t e d f o r a product are r e p r e s e n t a t i v e o f the major n u t r i e n t s and tha t c r u c i a l omissions are u n l i k e l y . In the United S t a t e s , the Federal Trade Commission has proposed a requirement f o r the d i s c l o s u r e o f c e r t a i n n u t r i t i o n a l information whenever n u t r i t i o n a l claims are made i n a d v e r t i s i n g ( c f . Scammon, 1977). The p a r t i c u l a r n u t r i e n t s s i n g l e d out f o r d i s c l o s u r e are c a l o r i e s , p r o t e i n and seven vitamins and minerals. Thus, the p u b l i c s e c t o r has defined c e r t a i n important n u t r i t i o n a l cues by v i r t u e o f t h i s requirement. The study by Quelch (1978), reported i n Chapter I I I , provides f u r t h e r i n s i g h t i n t o the perceived importance of these n u t r i e n t s among housewives. Quelch (1978) had h i s subjects r a t e the importance of 19 breakfast cereal a t t r i b u t e s i n the process of choosing a f a v o u r i t e brand. L i s t e d among these a t t r i b u t e s were f i v e n u t r i e n t s : vitamins and m i n e r a l s , p r o t e i n , sugar, f i b r e , and c a l o r i e s . The cues vitamins and minerals and p r o t e i n were given greater importance than sugar, f i b r e and c a l o r i e s . Nevertheless, a l l f i v e n u t r i e n t s were rated very c l o s e l y 77 in relative importance, and near the midpoint of the bipolar importance cale used in the study (see Table 2, p.50). Rudell (1979) conducted a nutritional information usage experiment among 187 grocery shoppers. She found the percentage of respondents using each type of information at least once in a simulated purchase task to be as in Table 5. Table 5 Percent of Respondents Using Each Type of Information At Least Once In a Food Evaluation and Choice Task Type Of Information Popularity Percent of Respondents Ingredients 1 82.4 Vitamins 2 69.0 Protein 3 67.9 Consumer Reports 4 65.8 Fat content 5 63.6 Calories 6 59.9 Advertising claims 7 41.2 Government booklet 8 39.0 Friends' comments 9 33.2 (From Rudell, 1979:46, Table 5.4) As in Quelch's (1978) study, vitamins and minerals and protein turn 5 out to be more important than fat content and calor ies. Although only "vitamins" is entered in Rudell's (1979) Table 5.4, subjects were actually given information on vitamins and minerals (see Rudell, 1979:34, 111). 78 When Rudell's (1979) subjects were asked to r a t e the value of each of these cues i n terms of p r e d i c t i n g the n u t r i t i o n a l q u a l i t y of food, e x a c t l y the same rank order by value was obtained from the aggregated r e s u l t s . Once again, d i f f e r e n c e s i n r e l a t i v e importance between these four n u t r i t i o n a l cues appear to be q u i t e s m a l l , as i n d i c a t e d by the percentages i n Table 5. In summary, vitamins and minerals and p r o t e i n are probably the two n u t r i e n t s of highest importance r e l a t i v e to other n u t r i t i o n a l cues such as f a t or c a l o r i e s . An examination of the n u t r i t i o n a l cues s e l e c t e d f o r t h i s experiment (see Appendix A) reveals t h a t p r o t e i n has been included i n a l l of the products. Moreover, i n the case of canned soup, m&c d i n n e r , bran cereal and peanut b u t t e r , at l e a s t one vitamin and several minerals are among the cues chosen. For the product tomato ketchup, there are no vitamins to speak of (only l a b o r a t o r y - d e t e c t a b l e t r a c e s ) , so the remainder of the cues c o n s i s t of m i n e r a l s , c a l o r i e s and f a t . For the product mayonnaise, the s e l e c t i o n o f n u t r i t i o n a l cues a v a i l a b l e p a r a l l e l s the conclusions of a consumer t e s t o f mayonnaise brands published i n Consumer Reports (1977). The report points out that there are n e g l i g i b l e amounts of vitamins and minerals i n mayonnaise and that brands are d i f f e r e n t i a t e d e s s e n t i a l l y by: 1. Poly-unsaturated f a t content, 2. Saturated f a t content, and 3. Sodium content. These three n u t r i t i o n a l cues are among those employed f o r the product mayonnaise. 79 In summary, the cue s e l e c t i o n procedure probably d i d not r e s u l t i n any c r i t i c a l omissions o f important n u t r i t i o n a l cues. In a d d i t i o n , the evidence a v a i l a b l e suggests t h a t consumers reveal q u i t e small d i f f e r e n c e s i n the r e l a t i v e importance they attach to various n u t r i t i o n a l cues. Obtaining Measures o f Cue Importance Since cue importance was one of the experimental f a c t o r s i n t h i s study, the next step i n preparation f o r the f i e l d experiment was to determine the r e l a t i v e importance o f each s e l e c t e d cue i n the formation of brand preferences w i t h i n each t e s t product. For t h i s purpose, a personal i n t e r v i e w survey was conducted one month p r i o r to the commencement of the i n - s t o r e experiment, to obt a i n measures o f cue importance among frequent purchasers of each product. S p e c i f i c a l l y , the consumers sampled i n t h i s survey were randomly s e l e c t e d from a geographic area with a high p r o b a b i l i t y of i n c l u d i n g patrons of the two stores used i n the research. In a d d i t i o n to cue importance, the d i r e c t i o n o f a consumer's u t i l i t y f o r each n u t r i e n t i n the que s t i o n n a i r e was measured i n the survey. This information was needed to confirm the ranking of brands by n u t r i t i o n a l performance i n order to o p e r a t i o n a l i z e the dependent v a r i a b l e . Survey Questionnaire. A copy o f the complete q u e s t i o n n a i r e i s given i n Appendix C. Respondents were f i r s t screened to a s c e r t a i n t h a t they g e n e r a l l y d i d the grocery shopping f o r t h e i r household. Following t h i s , respondents were screened f o r usage of each product at the beginning of the i n t e r v i e w . I f a product was purchased l e s s o f t e n than once i n 80 4 months, i t was deleted from the l i s t o f products on which f u r t h e r questions were asked. Next, f o r each product on which the respondent q u a l i f i e d to answer, a l i s t of nine n u t r i e n t s f o r that product was given to the respondent. Although only e i g h t n u t r i t i o n a l cues were needed f o r the experimental design, a "reserve" cue was added to each l i s t o f n u t r i e n t s i n a n t i c i p a t i o n o f i n s u f f i c i e n t data on (or other unforseen problems with) a p a r t i c u l a r cue.^ Following the completion o f the survey, the same e x t r a cues which had been added to these l i s t s were e l i m i n a t e d , since no problems were encountered on the e i g h t o r i g i n a l l y s e l e c t e d cues. Respondents were asked to assign a score of 10 to the s i n g l e most important cue on the l i s t , i n terms of i t s help f u l n e s s f o r choosing which brand of the product to buy. Then, using the.score of 10 as an anchor, respondents assigned a number between 0 and 10 to denote the importance of each remaining cue. Following t h i s , respondents were asked whether they would g e n e r a l l y seek " l o t s o f " or " l i t t l e o f " each of the n u t r i e n t s i n choosing which brand of a product to buy. Responses other than these two were a l s o noted and l a t e r coded as "other." A f t e r n u t r i e n t importance and u t i l i t y responses had been c o l l e c t e d on a l l the e l i g i b l e products, respondents were questioned about which grocery s t o r e , or s t o r e s , they u s u a l l y - p a t r o n i z e d . This question served as a check on whether most of the respondents surveyed were i n f a c t The l i s t f o r mayonnaise contained only e i g h t n u t r i e n t s since no more than e i g h t were a v a i l a b l e . The l i s t f o r m&c dinner contained ten n u t r i e n t s , of which two were l a t e r d e l e t e d . 81 patrons of the two experimental s t o r e s . F i n a l l y , a set of demographic questions were included at the end of the q u e s t i o n n a i r e . Two forms of the same questionnaire were used, so t h a t about h a l f o f the survey sample responded to the l i s t o f n u t r i e n t s f o r each product i n the reverse order given to the other h a l f . A d d i t i o n a l l y , the order i n which the s i x products appeared was rotated from qu e s t i o n n a i r e to q u e s t i o n n a i r e . Both precautions were taken to o f f s e t order bias and l e a r n i n g and f a t i g u e e f f e c t s . Survey Sample. The sampling plan f o r the personal i n t e r v i e w survey was developed as f o l l o w s . Centered on each of the two cooperating stores,, a one-mile radius was used to d e l i n e a t e a store's patronage area. Within t h i s area, 10 r e s i d e n t i a l c i t y blocks w i t h a common perimeter were randomly s e l e c t e d . Each block was v i s i t e d by t h i s researcher and was given a predesignating s t a r t i n g address. One t r a i n e d female i n t e r v i e w e r conducted a l l the i n t e r v i e w s during morning, afternoon and evening hours, over a period bf 10 days. A given block was covered by i n t e r v i e w i n g at the s t a r t i n g address and at a l l adjacent houses, working clockwise from the s t a r t i n g address, u n t i l four completed i n t e r v i e w s were obtained i n each block. No c a l l - b a c k s were attempted. T h i r t y - f i v e i n t e r v i e w s were completed i n each store's patronage ar e a , f o r a t o t a l o f 70 i n t e r v i e w s . Survey R e s u l t s . Table 6 presents the r e s u l t s o f the cue importance and n u t r i e n t u t i l i t y measurements from the survey. Cue importance was determined w i t h i n the context of a t e s t product. In other words, the r e l a t i v e importance of each cue l i s t e d under a product was c a l c u l a t e d 82 Table 6 Results of Personal Interview Survey to Measure R e l a t i v e Importance of and D i r e c t i o n o f U t i l i t y f o r Nu t r i e n t s Cue i Mean R e l a t i v e Importance (%) Number of Importance Scores (N) Response Frequencies For U t i l i t y Measure "Lots of" " L i t t l e " "Other" Bran C e r e a l : Food F i b r e .189 34 32 0 1 P r o t e i n .162 34 30 0 4 Vitamin N i a c i n .147 31 25 3 4 Calcium .130 34 26 2 6 Magnesium .095 33 16 11 5 Potassium .094 32 14 12 7 Phosphorus .082 34 13 15 6 * C a l o r i e s .069 34 2 27 3 Sodium .055 33 1 30 2 Mushroom Soup: Calcium .204 41 36 3 2 Pr o t e i n .199 41 37 0 4 ' Vitamin B 2 .190 39 34 1 4 Potassium .100 38 13 ,17 7 Phosphorus .089 37 9 21 6 *Carbohydrate .075 40 1 27 12 C a l o r i e s .064 41 1 36 2 Fat .057 40 0 32 9 Sodium .051 40 1 35 2 Macaroni & Cfeeese Dinner: Iron .161 30 29 0 1 Pr o t e i n .151 30 29 0 1 Vitamin B~ .136 27 25 0 2 Vitamin B, • .1321 27 24 0 3 Calcium .1320 30 24 4 2 Vitamin N i a c i n .125 27 22 1 4 Phosphorus .069 26 8 14 6 *Carbohydrate .055 28 3 13 12 * C a l o r i e s .050 29 2 23 5 Fat .045 29 2 23 5 83 Table 6 (Continued) Number of Mean R e l a t i v e Importance Response Frequencies For Importance Scores U t i l i t y Measure Cue i (N) "Lots o f " " L i t t l e " 1 'Other" Tomato Ketchup Iron .242 47 41 4 1 P r o t e i n .161 47 30 13 2 C a l o r i e s .146 47 2 41 2 Potassium .114 42 20 15 4 Phosphorus .108 40 15 17 7 Sodium .071 45 1 37 5 Carbohydrate .065 44 3 28 9 *Sugar .055 47 2 42 1 Fat .053 46 3 39 2 Peanut Butter: P r o t e i n .208 51 49 1 1 Iron .179 51 47 1 3 Vitamin B ? .165 48 42 1 6 Vitamin N i a c i n .140 46 34 6 6 Potassium .092 43 17 19 8 Phosphorus .091 44 12 22 10 Fat .067 50 2 34 13 C a l o r i e s .059 51 3 36 7 *Sodium .054 49 2 40 5 Mayonnaise: Poly-Unsaturated Fats.247 53 24 17 11 P r o t e i n .197 54 33 17 4 Total Fats .121 54 2 42 9 Carbohydrate .106 54 5 33 15 -Saturated Fats .102 53 3 47 2 C a l o r i e s .102 54 4 45 4 Sodium .071 54 1 48 3 Sugars .061 54 1 51 1 84 from responses given only by purchasers of th a t product. The r e l a t i v e importance o f a cue p a r t i c u l a r to a product was c a l c u l a t e d as f o l l o w s : X' i r i r I X. i = l 1 1 (4.1) where X'. = the normalized importance score f o r cue i f o r respondent r , and 0 <_ X'. <_ 1; X. = raw importance score, from 0 to 10, given by respondent r to cue i ; n = number of n u t r i t i o n a l cues responded to i n a product category. Next, aggregating the above scores across a l l respondents f o r that product, the mean r e l a t i v e importance o f cue i , X\, was obtained as f o l l o w s : N y x1. X = I I I > (4.2) 1 N where < X.. < 1 — i — = t o t a l number of respondents p r o v i d i n g an importance score on cue i . 85 The N's vary w i t h i n products i n Table 6 because some respondents claimed not to recognize c e r t a i n n u t r i e n t s w i t h i n a l i s t , or gave no r e p l y . Response frequencies f o r the u t i l i t y measure of each n u t r i e n t are a lso l i s t e d i n Table 6. In g e n e r a l , consumers' u t i l i t i e s f o r the various n u t r i e n t s were reasonably c l e a r - c u t and i n the d i r e c t i o n s expected. Larger amounts of p r o t e i n , v i t a m i n s , i r o n , c a l c i u m , magnesium, food f i b r e , and poly-unsaturated f a t s were p r e f e r r e d to smaller amounts. On the other hand, smaller q u a n t i t i e s of c a l o r i e s , sodium, f a t , saturated f a t s , carbohydrate and sugars were pref e r r e d to l a r g e r q u a n t i t i e s of these. Mixed r e s u l t s occurred i n determining the d i r e c t i o n o f u t i l i t y f o r the n u t r i e n t s potassium and phosphorus. Among products which l i s t e d these two n u t r i e n t s , s u b s t a n t i a l numbers of responses were given i n both the " l o t s o f " and " l i t t l e " c a t e g o r i e s . These mixed r e s u l t s p a r a l l e l the intermediate l e v e l s i n mean r e l a t i v e importance of both potassium and phosphorus found by the survey. Where potassium or phosphorus was more important to the respondent, more was p r e f e r r e d to l e s s . Where potassium or phosphorus was l e s s important to the respondent, the tendency was to p r e f e r " l i t t l e " r a t h e r than " l o t s o f " . On t h i s b a s i s , the survey r e s u l t s were i n t e r p r e t e d to mean t h a t , to consumers f o r whom e i t h e r potassium or phosphorus has meaning and some importance, l a r g e r amounts of the n u t r i e n t are p r e f e r r e d to smaller amounts. The cues preceded by a s t a r i n Table 6 were the "reserve" ones to be deleted i n order to a r r i v e at the f i n a l set of e i g h t cues f o r each product. 86 The r e s u l t s of the question on store patronage are presented i n Appendix C. Although the responses i n d i c a t e t h a t the surveyed i n d i v i d u a l s patronized many other grocery s t o r e s , as expected the m a j o r i t y i n each store patronage area d i d i n f a c t u s u a l l y shop f o r g r o c e r i e s i n the re-specti v e experimental s t o r e . F i n a l l y , i t i s conceded here that the measures of r e l a t i v e importance obtained from the survey may a c t u a l l y represent measures of r e l a t i v e s a l i e n c e . Myers and A l p e r t (1977) pointed out the semantic and conceptual d i s t i n c t i o n between the "importance" and " s a l i e n c e " o f . i n f o r m a t i o n cues. S p e c i f i c a l l y , cue "importance" denotes the degree to which a cue i s of s i g n i f i c a n c e to the consumer i n ev a l u a t i o n and choicemaking. The " s a l i e n c e " of a cue denotes the extent to which i t i s "top-of-mind" to the consumer, i n s o f a r as i t i s commonly encountered by the consumer and, t h e r e f o r e , f r e q u e n t l y mentioned or discussed. However, a s a l i e n t cue i s not n e c e s s a r i l y an important cue, i n the sense def i n e d . Since the survey r e s u l t s are based on the stated importance of a cue they may not n e c e s s a r i l y r e f l e c t how consumers w i l l behave with respect to informat i o n on th a t cue. In t h i s r espect, the measures may represent r e l a t i v e cue s a l i e n c e but not n e c e s s a r i l y r e l a t i v e importance. To sum up, the d i r e c t i o n of consumers' u t i l i t i e s f o r the n u t r i e n t s represented by i n d i v i d u a l cues was e s t a b l i s h e d from the survey. In a d d i t i o n , the survey data revealed the mean stated r e l a t i v e importance, to purchasers, o f each o f the cues designated f o r the experiment. Having s e l e c t e d the f i n a l set of e i g h t n u t r i t i o n a l information cues f o r the s i x products, the procedures taken to o p e r a t i o n a l i z e the independent v a r i a b l e s are discussed next. 87 Independent V a r i a b l e s Load. As pointed out i n Chapters I and I I , load was o p e r a t i o n a l l y defined i n t h i s research as the number of cues included i n a brand-by-cue information matrix. Therefore, s t i m u l i representing d i f f e r e n t load l e v e l s were constructed by varying the number of n u t r i t i o n a l cues pre-sented w i t h i n such a matrix. Although e a r l i e r chapters i n d i c a t e d the r a t i o n a l e f o r using a brand-by-cue matrix on a point-of-purchase information sign as a v e h i c l e f o r manipulating load., these reasons are b r i e f l y h i g h l i g h t e d below. F i r s t , the consumer has d i r e c t access to the information at the place of decision-making -- an important determinant of the e f f e c t i v e -ness of product information (Day, 1976). Second, the data are presented to consumers i n a h i g h l y processable format which f a c i l i t a t e s d i r e c t comparisons of choice a l t e r n a t i v e s along the a t t r i b u t e s d i s c l o s e d (Russo, 1977; Day, 1976; Bettman, 1975). F i n a l l y , as information load i s v a r i e d the information format can be held constant. The load f a c t o r was v a r i e d over four l e v e l s . A product's brand-by-cue matrix contained e i t h e r 1, 2, 4 or 8 cues. These p a r t i c u l a r l e v e l s were chosen a f t e r c o n s u l t i n g the cue-loads s e l e c t e d i n some of the l a b o r a t o r y studies of consumer information load reviewed i n Chapter '.' I I I . The r e l e v a n t data on a l l f i v e s t u d i e s reviewed are presented i n Table 7. I n t e r e s t i n g l y , none of the s t u d i e s reported i n Table 7 employed a load of one cue, thereby perhaps s a c r i f i c i n g a useful lower reference point against which the e f f e c t s of a l l higher loads could be compared. Table 7 Cue-Load Levels Used and Results Reported In Five Consumer Information Load Studies "Information Control No. of Cues Selected Overload" Study Used For Load Manipulation Reported Jacoby, S p e l l e r & Kohn (1974b) No 2 4 6 Yes a Jacoby, S p e l l e r & Kohn (1974a) No 4 8 12 16 Yes a Scammon (1977) Yes 5 9 No Best (1978) No 2 4 6 8 No Goodwin & Etgar (1980) No 2 5 7 Yes b In these s t u d i e s the number of brands was also manipulated as a load f a c t o r , thus confounding the e f f e c t s o f cue-load. b 0 n l y weak evidence of overload i s c i t e d by the authors. 89 Note, a l s o , t h a t i n a l l studies the maximum load represented an increase of the minimum load by a f a c t o r of f o u r , or l e s s . The loads used i n t h i s study r e f l e c t a number of judgmental c r i t e r i a . A p o t e n t i a l l y i n t e r e s t i n g s t a r t i n g point of one cue was included i n the load range. M u l t i p l e s of load were chosen so t h a t each higher load represented a doubling of the previous l o a d . The highest load used provided e i g h t times as much information as the lowest load. The l i m i t was placed at e i g h t cues r a t h e r than at the next m u l t i p l e because of the l i m i t e d amount of data received from manufacturers (see Table 4). The next task was to devise a procedure of cumulating s p e c i f i c cues from the set of e i g h t a v a i l a b l e , i n order to c o n s t r u c t the four load by two cue importance treatments. Successive increases i n load were achieved by s t a r t i n g with a s i n g l e cue and adding on "unused" cues to reach the required load l e v e l s . In other words, once an information cue was included i n a treatment, i t was r e t a i n e d i n a l l s u c c e s s i v e l y higher load treatments. The net e f f e c t o f t h i s procedure was t h a t consumers exposed to the next higher load treatment would see every one of the cues employed i n al1 lower load treatments, plus the required number of cues which had not appeared before. Cue Importance. In a d d i t i o n , a s p e c i f i c hierarchy f o r cue s e l e c t i o n was used so t h a t , as the load treatments were co n s t r u c t e d , the two cue-importance treatments were created simultaneously. Having de-termined the mean r e l a t i v e importance of each cue from the survey r e s u l t s (see Table 6 ) , the e i g h t cues f o r a product could be rank ordered by importance, from most to l e a s t important. 90 Therefore, one load s e r i e s was constructed by s t a r t i n g with the s i n g l e most important cue and cumulating cues i n order of decreasing importance to o b t a i n each d e s i r e d load l e v e l . A second load s e r i e s was constructed by s t a r t i n g with the s i n g l e l e a s t important cue and cumulating i n order of i n c r e a s i n g importance to obtain the same four load l e v e l s ( 1 , 2, 4 and 8 ) . This procedure produced the e i g h t t r e a t -ments f o r the 4 x 2 f a c t o r i a l design. Table 8 summarizes the way i n which s p e c i f i c cues (numbered by t h e i r rank order of importance) were u t i l i z e d to c o n s t r u c t the treatments. Table 8 Cues (Numbered by Rank Order of Cue-Importance) U t i l i z e d To Construct the Eight Treatments f o r the 4 x 2 F a c t o r i a l Design Load Level Cue-Importance Level 1 2 4 8 "High" 1 1,2 1,2,3,4 1,2,3,4,5,6,7,8 "Low" 8 8,7 8,7,6,5 8,7,6,5,4,3,2,1 On the point-of-purchase signs representing these experimental s t i m u l i , the cues were p r i n t e d i n the same l e f t - t o - r i g h t order as l i s t e d i n Table 8. For example, the sign representing a load of 8 "high-importance" cues (henceforth, treatments w i l l be denoted i n the f o l l o w i n g manner: "8/high") l i s t e d the cues, from l e f t to r i g h t , i n decreasing order of importance, s t a r t i n g with the most important cue. 91 Note that the treatment c o n s t r u c t i o n procedure designated the same e i g h t cues f o r the "8/high" and "8/low" s t i m u l i . As shown i n Table 8, the only d i f f e r e n c e between these two s t i m u l i i s the l e f t -t o - r i g h t arrangement of the e i g h t cues on the s i g n s . This s i t u a t i o n permitted the t e s t i n g of hypothesis H^ -Sign C o n s t r u c t i o n . A l l signs were i n i t i a l l y prepared i n 8h" x 11" t y p e s c r i p t form, then photographed and enlarged to a p r i n t s i z e of 20" x 27". Two copies of each photo-enlargement were bonded to cardboard of the same s i z e to create a double-sided s i g n . Signs were hung from a h o r i z o n t a l wooden bar to f a c i l i t a t e t h e i r placement and removal at s p e c i f i e d times during the experimental period. The bar, i t s e l f , was suspended from.the store c e i l i n g , about 8 f e e t above the a i s l e f l o o r , throughout the d u r a t i o n of the experiment. Signs were p o s i t i o n e d halfway along the s h e l v i n g width a l l o c a t e d to a t e s t product. A l l signs and t h e i r hanging bars were suspended i n such a way as to be perpendicular to the a i s l e s and product s h e l v i n g s , thus making the signs f u l l y . v i s i b l e from both ends of t h e i r r e s p e c t i v e a i s l e s . ' ' Each product required nine d i f f e r e n t signs ( e i g h t treatment signs plus one c o n t r o l s i g n , described below) f o r a t o t a l of 54 signs per se t . Two sets of these signs were made up because the experiment was run simultaneously i n the two s t o r e s , making i t necessary to c o n s t r u c t The author i s indebted to Professor J . Edward Russo, U n i v e r s i t y of Chicago, f o r the idea of how to p o s i t i o n these signs so as to make them more v i s i b l e to stor e customers. 92 a t o t a l o f 108 two-sided s i g n s . Replicas of the 54 signs are provided i n Appendix A. Experimental C o n t r o l s . Table 7 shows t h a t , of the f i v e information load s t u d i e s , only Scammon's (1977) experiment included a c o n t r o l against which the e f f e c t s of two d i f f e r e n t loads could be compared. In Jacoby, S p e l l e r and Kohn's (1974a; b) s t u d i e s , no baselines were incorporated i n the designs, so t h a t responses to various loads could not be adjusted f o r v a r i a t i o n s i n task d i f f i c u l t y a r i s i n g from treatment c a l l s with d i f f e r e n t numbers of brands. As discussed i n the review of Chapter I I I , t h i s l e d to i n c o r r e c t conclusions about the true e f f e c t s of load ( W i l k i e , 1974; Russo, 1974). In the present research, two types of experimental c o n t r o l s were designed i n order to assess treatment e f f e c t s . The s t o r e s a l e s data comprising one of these c o n t r o l s were c o l l e c t e d during periods when no sign was attached to the hanging bar at a t e s t product's s h e l v i n g (hence-f o r t h denoted as "no-sign c o n t r o l " ) . The i n c l u s i o n o f t h i s experimental c o n t r o l was e s s e n t i a l f o r four reasons: 1. Changes i n purchase behaviour had to be measured as changes i n the brand s a l e s d i s t r i b u t i o n of a t e s t product. R i d i t a n a l y s i s , the basic s a l e s a n a l y t i c a l t o o l employed f o r the f i e l d data, c a l l s f o r an i d e n t i f i e d b a s e l i n e or "reference" d i s t r i b u t i o n , against which d i s -t r i b u t i o n changes can be measured. 2. The t e s t products d i f f e r by nature and by number of brands. These d i f f e r e n c e s are already incorporated i n each product's unique brand s a l e s d i s t r i b u t i o n under experimental c o n t r o l , against which 93 treatment e f f e c t s are measured. 3. Brand p r i c e s and promotions, which are t y p i c a l l y held constant w i t h i n a s i n g l e week, may vary from week to week. Since experimental treatments extended across two successive weeks, c o n t r o l s unique to each week were used f o r each product. 4. Controls i n the two experimental weeks can be compared with "before" and " a f t e r " b a s e l i n e measures during the remaining weeks i n the f i e l d , i n order to t e s t hypothesis H,. and answer the question "Is the c o n t r o l d i s t r i b u t i o n observed i n each of the two experimental weeks r e p r e s e n t a t i v e of the 'before' or ' a f t e r ' weekly s a l e s d i s t r i b u t i o n i n the s t o r e s ? " A second type of c o n t r o l was created with signs i d e n t i c a l i n appearance and s i z e to an information treatment s i g n . However, these signs included only the name of the t e s t product as a heading, below which were l i s t e d the brands i n the same top-to-bottom order as they appeared on a l l information signs (see Appendix A). No brand information was d i s c l o s e d on these s i g n s , henceforth denoted as "sign c o n t r o l . " This s i g n c o n t r o l served two purposes: 1. I t provided a c o n t r o l f o r any behavioural changes among t e s t product customers due s o l e l y to the p h y s i c a l presence of a sign at the product s h e l v i n g , where none had e x i s t e d before (see Henion, 1972, f o r an i n - s t o r e experiment t a k i n g s i m i l a r p r e c a u t i o n s ) . I f there i s any order bias i n the l i s t i n g of a product's brands i t i s kept constant across a l l treatments. 94 2. I t c o n t r o l l e d f o r any behavioural changes a r i s i n g from the l i s t i n g of a l l o f a t e s t product's brands a v a i l a b l e i n the s t o r e . Should no s i g n i f i c a n t d i f f e r e n c e s be detected i n the brand s a l e s d i s t r i b u t i o n s obtained under no-sign c o n t r o l and sign c o n t r o l , the two sets of data could be pooled to reduce sampling e r r o r i n the c o n t r o l measurements. Experimental Store S e l e c t i o n In the i n i t i a l design planning stages, i n t e r e s t focused on o b t a i n i n g the cooperation of a supermarket chain which had i n s t a l l e d automated checkouts i n t h e i r s t o r e s . Since the experiment c a l l e d f o r a l a r g e number of treatment manipulations w i t h i n the span of a s i n g l e shopping week, the o b j e c t i v e was to perform the experiment i n a s t o r e where a l a r g e amount of s a l e s data could be c o l l e c t e d i n a b r i e f p e r i o d . The recent i n t r o d u c t i o n of computerized checkout f a c i l i t i e s i n t o a few supermarkets by the two leading chains i n M e t r o p o l i t a n Vancouver, Canada (population 1.2 m i l l i o n ) o f f e r e d an opportunity to secure the cooperation needed. The l e a d i n g supermarket chain was the f i r s t one contacted and granted the necessary permission. Two of i t s stores were equipped with automated checkouts and i t was decided to proceed with the experiment i n these s t o r e s . The stores were w i t h i n the same metropolitan area but i n d i f f e r e n t m u n i c i p a l i t i e s . Both stores had annual s a l e s volumes w i t h i n 5% of each o t h e r , almost p r e c i s e l y the same weekly customer t r a f f i c ( c 1 3 , 0 0 0 ) , s i m i l a r s h e l f f a c i n g s a l l o c a t e d to the t e s t products, 95 and i n the case of each t e s t product, the same brands and assortment of container s i z e s . Experimental Procedure Store business hours extended from 9 A.M. to 6 P.M., s i x days a week (Monday through Saturday), with the exception of two days (Thursday and F r i d a y ) when s t o r e c l o s i n g time was 9 P.M. Since the bas i c f a c t o r i a l experiment consisted of e i g h t treatments and two types o f c o n t r o l s to be executed w i t h i n one week, a store's t o t a l hours were p a r t i t i o n e d i n t o two treatment periods per day, y i e l d i n g 12 treatment periods per week. The durations of the two d a i l y treatment periods were s e l e c t e d so that the expected volume of t o t a l s t o r e customer t r a f f i c would be approximately the same i n each treatment p e r i o d . Actual customer t r a f f i c counts were obtained during p r e t e s t s i n the two stores several months before the experiment commenced. The r e s u l t i n g treatment period durations were f i x e d at 9:00 A.M. to 2:15 P.M. and 3:00 P.M. to 6:00 P.M. (or 9:00 P.M.) each day, i n both s t o r e s . The 45-minute i n t e r v a l separating the two d a i l y periods ensured that the same t e s t product customer was not exposed to two d i f f e r e n t treatments during the same shopping t r i p . A separate p r e t e s t was c a r r i e d out several months before the experiment to measure the shopping t r i p d urations of these s t o r e customers. This researcher p o s i t i o n e d himself i n one of the two cooperating stores so as to u n o b t r u s i v e l y monitor the time taken by shoppers between entering the store and entering the check-out queue. Observations were made on 111 shoppers during three d i f f e r e n t weeks and s i x d i f f e r e n t periods o f the day and days of the week. Mean 96 shopping time was 22.08 minutes, with 95% of the sample taki n g 45 minutes or l e s s . With 12 treatment periods a v a i l a b l e i n the s t o r e , the e i g h t t r e a t -ments were a l l o c a t e d to one period each while the two c o n t r o l s were a l l o c a t e d to two periods each. The 12 a l l o c a t i o n s were made randomly f o r each product i n each s t o r e , i n order to negate time-of-day and day-of-week e f f e c t s . The same procedure was followed i n the second e x p e r i -mental week, with the 12 a l l o c a t i o n s per product, per s t o r e being randomly redetermined. At the end of each treatment period s a l e s measurements were obtained instantaneously at an automated checkout and the data f o r each item (a running count of the number of u n i t s s o l d since the s t a r t o f the week) t r a n s c r i b e d onto d a i l y coding sheets (see Appendix D). Signs were then changed according to treatment schedule. Two researchers were needed i n order to c a r r y out the experimental procedures i n both stores simultaneously. In order to draw f u r t h e r a t t e n t i o n to an information s i g n , 5" x 8" m u l t i c o l o u r e d s h e l f tags were prepared f o r each t e s t product and attached to the product s h e l v i n g i n the manner t y p i c a l of in-.store p r i c e promotions on a brand. The message on these tags stated t h a t n u t r i t i o n a l i nformation on the r e s p e c t i v e product was a v a i l a b l e on a sign above the a i s l e (Appendix D has a"tag r e p l i c a ) . These tags were always removed during periods a l l o c a t e d to no-sign c o n t r o l or sign c o n t r o l f o r the product. The t e s t product shelves were monitored several times a day to 97 check f o r stock outages of any items and to note any changes i n p o i n t -of-purchase promotions or to replace damaged s h e l f tags. A l l stock outages and i n - s t o r e promotions were recorded on a "noise monitoring" report throughout the five-week period i n the f i e l d (see Appendix E). In a d d i t i o n to i n - s t o r e checks, a weekly record was maintained of any a d v e r t i z e d and unadvertized promotions r e l a t i n g to the t e s t products c a r r i e d out by competing stores i n the v i c i n i t y . This was done as a precaution i n the event t h a t some experimental r e s u l t s turned out to be d i f f i c u l t to i n t e r p r e t . For t h i s purpose, a l l major competing super-markets' w i t h i n a radius of about 3-4 miles from the experimental stores were v i s i t e d once a week, since promotions are t y p i c a l l y held constant during a week. The f i e l d experiment reported here was performed during A p r i l and May, 1979. Although no data were c o l l e c t e d on the extent to which shoppers were seen examining the point-of-purchase s i g n s , i t was obvious to both i n - s t o r e researchers that shoppers were n o t i c i n g the signs and o c c a s i o n a l l y appearing to study them. Dependent V a r i a b l e Overview. The considerable amount of e f f o r t and time t h a t was expended on developing a s u i t a b l e dependent v a r i a b l e f o r t h i s f i e l d experiment undoubtedly marks i t as the s i n g l e most d i f f i c u l t aspect of the d i s s e r t a t i o n . Some s a t i s f a c t o r y measure was needed to reveal whether the information signs were having an e f f e c t on purchasers of the t e s t products, w.ithout having to ask themi The p r i c e paid f o r t h i s unobtrusive measurement i n the f i e l d was to s a c r i f i c e i n d i v i d u a l 98 responses to information i n favour of gaining an aggregate p i c t u r e o f actual market behaviour. Nonetheless, i t i s necessary to e x t r a p o l a t e from expected i n d i v i d u a l -l e v e l behaviour i n order to assess responses i n the aggregate. The expectation was t h a t consumers who were not already purchasing the brand with the highest n u t r i t i o n a l performance would tend to s h i f t t h e i r pur-chases from t h e i r c u r r e n t l y p r e f e r r e d brand to one with a n u t r i t i o n a l advantage. Therefore, given the n u t r i t i o n a l cues d i s c l o s e d on a p a r t i c u l a r s i g n , consumers were expected to assess the data a v a i l a b l e and to purchase a b e t t e r performing brand i n s p i t e o f the c u r r e n t l y p r e v a i l i n g p r i c e s and e s t a b l i s h e d brand images. Consequently, an aggregate measure was sought which would i n d i c a t e or summarize the extent to which t e s t product purchases had s h i f t e d t o -wards brands with n u t r i t i o n a l advantages. This required the r e s o l u t i o n of two i s s u e s : how to o p e r a t i o n a l i z e n u t r i t i o n a l advantages and how to measure such d i r e c t i o n a l s h i f t s . Since the sales data representing brand purchases were i n the form o f frequency counts, an i n i t i a l candidate f o r t r a c k i n g d i s t r i b u t i o n a l changes i n these frequencies was two-sample chi square a n a l y s i s . U n f o r t u n a t e l y , i n cases where df > 1, chi square i s i n s e n s i t i v e to the d i r e c t i o n of brand sales d i s t r i b u t i o n changes ( S i e g e l , 1956:110). With the exception of the two-brand products, two-sample chi square a n a l y s i s would r e s u l t i n the l o s s of information on d i r e c t i o n a l s h i f t s . R i d i t A n a l y s i s . An a n a l y t i c technique known as r i d i t a n a l y s i s was e v e n t u a l l y found and i t provided a s u i t a b l e dependent v a r i a b l e with which to measure treatment e f f e c t s . R i d i t a n a l y s i s has a twofold s t r u c t u r e : 99 1. The basel i n e (reference) frequency d i s t r i b u t i o n , against which d i s t r i b u t i o n a l changes due to a treatment can be measured and summarized i n a s i n g l e s t a t i s t i c . The d i s t r i b u t i o n to serve as a reference i s se l e c t e d by the researcher. 2. The ordering of the frequency d i s t r i b u t i o n categories (brands) on some q u a l i t a t i v e s c a l e or c r i t e r i o n such as n u t r i t i o n a l performance. This ordering i s a l s o determined by the researcher. R i d i t a n a l y s i s handles the p a r t i c u l a r a n a l y t i c a l problem encountered i n t h i s research because i t 1. Is a p p l i c a b l e to the aggregate brand s a l e s frequencies recorded i n each treatment p e r i o d , 2. Takes advantage of an a p r i o r i ranking of the brands from lowest to highest n u t r i t i v e performance, and 3. Y i e l d s a s i n g l e s t a t i s t i c , the mean r i d i t ( r ) , which i n d i c a t e s the net change i n the brand sales d i s t r i b u t i o n towards more n u t r i t i v e brands, as compared to b a s e l i n e . The mean r i d i t s t a t i s t i c serves as the dependent measure. As w i l l be discussed i n the remaining parts of t h i s chapter, the mean r i d i t has the f o l l o w i n g advantages: 1. I t i s a r a t i o - s c a l e d dependent v a r i a b l e s a t i s f y i n g the measure-ment assumption of•parametric s t a t i s t i c a l t e s t s . Each mean r i d i t derived from the s a l e s data i s s t a t i s t i c a l l y independent of every other mean r i d i t . The mean r i d i t i s nearly normally d i s t r i b u t e d , although f a l l i n g i n the closed i n t e r v a l 0 < r < 1 (Bross, 1958:36). 2. I t i s independent of the number of brands underlying the sal e s data from which i t was d e r i v e d , thus making the mean r i d i t comparable Table 9 F a c s i m i l e of an Information Sign Employed f o r M&C Dinner MACARONI & CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE* Brand Code BRANDS IRON PROTEIN VITAMIN B. VITAMIN B. CALCIUM VITAMIN PHOSPHORUS NIACIN mg mg mg mg mg mg FAT 9 A B TOWN HOUSE 0,4 4.8 0.4 0.2 21 2.6 54 10.8 KRAFT DELUXE 3.6 6.9 1.5 0.4 114 2.9 208 4.5 KRAFT DINNER 1.8 5.3 0.7 0.4 72 2.7 111 9.2 * c a l c u l a t e d f r o m , o r o b t a i n e d f r o m m a n u f a c t u r e r s ' d a t a o o 101 across products with d i f f e r e n t numbers of brands. 3. Each mean r i d i t value i s a l s o amenable to a d i r e c t s t a t i s t i c a l t e s t to determine whether i t i s s i g n i f i c a n t l y d i f f e r e n t from i t s ex-pected value. The f o l l o w i n g d i s c u s s i o n e x p l a i n s r i d i t a n a l y s i s and the d e r i v a t i o n of the mean r i d i t , with one of the t e s t products serving as an example. R i d i t a n a l y s i s was developed by Bross (1958:19), who based the name on the c a l c u l a t i o n o f r i d i t s from an i d e n t i f i e d e m p i r i c a l d i s t r i b u t i o n whose d i s c r e t e categories could be q u a l i t a t i v e l y ordered: The name ' r i d i t ' was chosen because of the analogy with ' p r o b i t s ' and ' l e g i t s ' . L i k e other members of the ' i t ' f a m i l y r i d i t s represent a type of transformation. But whereas p r o b i t s are r e l a t i v e to a t h e o r e t i c a l d i s -t r i b u t i o n (the normal d i s t r i b u t i o n ) , r i d i t s are re-l a t i v e to an e m p i r i c a l d i s t r i b u t i o n . The f i r s t three l e t t e r s stand f o r R e l a t i v e to an I d e n t i f i e d D i s t r i b u t i o n . The ordering o f the categories which r i d i t a n a l y s i s takes advantage of i s , i n t h i s case, the ranking of brands by n u t r i t i v e performance. Only one major assumption i s made i n . r i d i t a n a l y s i s , namely, t h a t the cate-gories represent i n t e r v a l s of an underlying but unobservable continuum on which they have been ordered ( F l e i s s , 1981:151). Table 9 i s a f a c s i m i l e of an information sign f o r one of the three-brand products and d i s c l o s e s the brand r a t i n g s on the e n t i r e set of eight n u t r i t i o n a l cues. From these r a t i n g s i t i s c l e a r t h a t , given any subset o f these cues, Brand A i s lowest i n n u t r i t i v e performance, Brand C ranks second and Brand B ranks highest. Therefore, i n the case of t h i s p a r t i c u l a r product, whichever treatment consumers were exposed t o , Brands B, C and A would c o n s i s t e n t l y rank 1 s t , 2nd and 3r d , r e s p e c t i v e l y , on n u t r i t i v e performance. 1 0 2 Upon ranking the brands on t h i s c r i t e r i o n , r i d i t - a n a l y s i s i s per-formed on a "reference d i s t r i b u t i o n " ( b a s e l i n e d i s t r i b u t i o n ) of the brand sales frequencies. The reference d i s t r i b u t i o n used i s the data obtained under experimental c o n t r o l . Table 10 i l l u s t r a t e s the mechanics of r i d i t a n a l y s i s i n order to obtain.the r i d i t s a ssociated with each brand. Table 10 C a l c u l a t i o n of Brand R i d i t s From Reference D i s t r i b u t i o n Brand (1) (2) (3): (4) (5) = r i d i t A 84 42 0 42 .1094 C 296 148 84 232 .6042 B 4 2 380 382 .9948 Total (N) 384 The order i n which the brands have been t a b l e d i s from lowest to  highest rank on n u t r i t i v e performance. Column 1 contains the u n i t sales frequencies ( i n d i v i d u a l packages sold) obtained during c o n t r o l periods. The data used here are s i m i l a r to actual r e s u l t s , but the frequencies are even numbers to s i m p l i f y the e x p o s i t i o n o f the technique. The column 2 e n t r i e s are h a l f the corresponding e n t r i e s i n column 1. The e n t r i e s i n column 3 are the cumulated frequencies of column 1, but d i s p l a c e d one brand downwards. Column 4 sums the corresponding e n t r i e s of columns 2 and 3. The r i d i t s entered i n column 5 are obtained by d i v i d i n g the 103 corresponding e n t r i e s i n column 4 by the t o t a l u n i t s a l e s , i . e . 384 packages. The r i d i t a s s o c iated with a brand represents the combined market share ( i n u n i t s sold) of a l l brands which rank lower i n n u t r i t i v e per-formance, plus h a l f the market share of that p a r t i c u l a r brand, as observed i n the reference d i s t r i b u t i o n . For example, the r i d i t f o r Brand C, 0.6042, i s the market share of the lower ranking brand (Brand A ) , plus h a l f the market share of Brand C. The Mean R i d i t . The summary s t a t i s t i c of r i d i t a n a l y s i s i s the mean r i d i t , r , and serves as the dependent v a r i a b l e . The mean r i d i t f o r the reference d i s t r i b u t i o n i s always, by d e f i n i t i o n , e x a c t l y 0.5. This i s i l l u s t r a t e d i n Table 11. Each r i d i t i s m u l t i p l i e d by i t s corresponding o r i g i n a l frequency and the products are summed. D i v i d i n g t h i s sum (192.0120) by the frequency t o t a l o f 384, one obtains an average r i d i t of e x a c t l y 0.5 f o r the reference d i s t r i b u t i o n ( a f t e r a l l o w i n g f o r round-o f f e r r o r s ) . Table 11 C a l c u l a t i o n o f the Mean R i d i t For the Reference D i s t r i b u t i o n Brand Unit Sales R i d i t (r ) Product A 84 .1094 9.1896 C 296 .6042 178.8432 B 4 .9948 3.9792 Total 384 - _ 192.0120 _ 0 5 r 3 8 4 U.D 192.0120 104 One can now use the mean r i d i t of 0.5 as a standard against which to gauge net changes i n the brand s a l e s d i s t r i b u t i o n o f t h i s product, as a r e s u l t of treatment e f f e c t s . To i l l u s t r a t e , consider the data i n Table 12. A hy p o t h e t i c a l brand s a l e s d i s t r i b u t i o n f o l l o w i n g some treatment has been t a b l e d , such t h a t the u n i t sales of each brand are e x a c t l y h of those given i n Table 10 f o r the reference d i s t r i b u t i o n ( r e c a l l t h a t experimental c o n t r o l s were given four times as many i n - s t o r e measurement periods as any s i n g l e treatment). Table 12 C a l c u l a t i o n of the Mean R i d i t For a Tr e a t m e n t ' D i s t r i b u t i o n Unit R i d i t -::v. Brand Sales ( r ) Product 2.2974 44.7108 .9948 48.0030 A 21 .1094 C 74 .6042 B 1 .9948 Total 96 - = 4^0030 96 Each brand's r i d i t (obtained from Table 10) has been m u l t i p l i e d by i t s r e s p e c t i v e s a l e s frequency, the r e s u l t i n g products summed, and t h i s sum d i v i d e d by the t o t a l u n i t s a l e s f o r t h i s treatment, i . e . 96. The mean r i d i t obtained i s (save f o r rounding e r r o r ) e x a c t l y 0.5, i n -d i c a t i n g that the r e l a t i v e brand d i s t r i b u t i o n remains unchanged from t h a t 105 of the reference d i s t r i b u t i o n . One concludes t h a t there has been no s h i f t i n the brand d i s t r i b u t i o n f o l l o w i n g t h i s treatment. The next three examples i l l u s t r a t e what happens to the mean r i d i t i f there i s any s h i f t , at a l l , i n the brand sales d i s t r i b u t i o n towards n u t r i t i v e l y b e t t e r performing brands, as a r e s u l t of exposure to information treatments. For the sake o f c l a r i t y , the t o t a l u n i t sales i n these examples w i l l be held at 96. The data i n Table 13 shows the r e s u l t of changing the brand s a l e s d i s t r i b u t i o n i n Table 12 by a s i n g l e purchase from the lowest ranked brand to the next higher ranked brand. The increase i n the mean r i d i t from 0.500 to 0.505 i s small but i n s t r u c t i v e . The u n i t s a l e s data i n Table 14 can be i n t e r p r e t e d as the r e s u l t of two s h i f t s i n the d i s t r i b u t i o n o f purchases given i n Table 12: one u n i t o f Brand A (the lowest ranked on n u t r i t i v e performance) to Brand C, and one u n i t of Brand C to Brand B. A l t e r n a t e l y , the d i s t r i b u t i o n can represent a s i n g l e purchase s h i f t from Brand A to Brand B. The r e s u l t o f e i t h e r event i s a s l i g h t l y l a r g e r increase i n the mean r i d i t , from 0.500 to 0.509. The s a l e s data i n Table 15 represent a s u b s t a n t i a l change i n the brand sales d i s t r i b u t i o n as a r e s u l t o f treatment e f f e c t s . The mean r i d i t has increased to 0.573, which i s a h i g h l y s i g n i f i c a n t increase (p < .001, . q on e - t a i l e d ) over the mean r i d i t of 0.5 f o r the reference d i s t r i b u t i o n . D i s c u s s i o n of the s i g n i f i c a n c e t e s t f o r mean r i d i t s i s deferred to a l a t e r s e c t i o n of t h i s chapter. 106 Table 13 C a l c u l a t i o n o f the Mean R i d i t For a D i s t r i b u t i o n Changed by One Purchase Unit R i d i t Brand Sales ( r ) Product A 20 .1094 2.1880 C 75 .6042 45.3150 B 1 .9948 .9948 Total 96 48.4978 - = 48^978 = ( L 5 0 5 Table 14 C a l c u l a t i o n of the Mean R i d i t For a D i s t r i b u t i o n Changed by One or Two Purchases Unit R i d i t Brand Sales (r ) Product A 20 .1094 2.1880 C 74 .6042 44.7108 B 2 .9948 1.9896 Total 96 48.8884 ^ 4 = 0.509 107 Table 15 C a l c u l a t i o n o f the Mean R i d i t Following a R e l a t i v e l y Large Change i n the Sales D i s t r i b u t i o n Unit R i d i t Brand Sales ( r ) Product A 10 .1094 1.0940 C 81 .6042 48.9402 B 5 .9948 4.9740 Total 96 55.0082 - 55.0082 n C 7 Q r = - g g = 0.573 Note t h a t the hypothesized change i n the mean r i d i t i s u n i d i r e c t i o n a l . The mean r i d i t i s expected to increase from the basel i n e l e v e l o f 0.5 as a r e s u l t of treatment e f f e c t s , because consumers are expected to s h i f t to brands which are n u t r i t i v e l y s u p e r i o r . Were the mean r i d i t to decrease from the basel i n e o f 0.5, t h i s would s i g n i f y that the brand s a l e s d i s -t r i b u t i o n had s h i f t e d away from higher performing brands and toward brands with a poorer n u t r i t i v e performance. The mean r i d i t a c t u a l l y can be i n t e r p r e t e d as a p r o b a b i l i t y ( F l e i s s , 1981:152). The value o f the mean r i d i t gives the p r o b a b i l i t y that a purchase chosen randomly from the treatment d i s t r i b u t i o n w i l l turn out to be a more n u t r i t i o u s brand than a purchase chosen randomly from the reference d i s t r i b u t i o n . A mean r i d i t o f , say, 0.57 f o r a treatment d i s t r i b u t i o n denotes t h a t 57% of the time, a purchase randomly s e l e c t e d 108 from among a l l purchases made during the treatment p e r i o d , w i l l be a brand which i s n u t r i t i v e l y s u p e r i o r to a brand randomly s e l e c t e d from among a l l purchases made during the c o n t r o l periods. Since i t conveniently summarizes the information contained i n a d i s t r i b u t i o n a l change due to a treatment, the mean r i d i t w i l l represent the dependent v a r i a b l e . This i m p l i e s t h a t i t must be comparable across products with d i f f e r e n t numbers o f brands and across brand d i s t r i b u t i o n s which may vary widely i n shape, not only from product to product, but also from treatment to treatment. In f a c t , no assumptions about the shape of the brand s a l e s d i s t r i b u t i o n being analyzed u n d e r l i e the use of r i d i t a n a l y s i s ( F l e i s s , 1981:151). Bross (1958:26) points out the advantage o f f e r e d by r i d i t a n a l y s i s when one i s faced with data where the frequencies over ordered cate g o r i e s are d i s t r i b u t e d i n many d i f f e r e n t ways, o c c a s i o n a l -l y with zero observations i n some c a t e g o r i e s : From the above d i s c u s s i o n i t i s apparent t h a t f o r sub-j e c t i v e s c a l e s the a n a l y t i c t o o l should p r e f e r a b l y be d i s t r i b u t i o n - f r e e (or non-parametric) i n so f a r as p o s s i b l e . R i d i t a n a l y s i s does have t h i s d e s i r a b l e f e a t u r e . Note t h a t the p r o b a b i l i t y statement associated with the average r i d i t i s a d i s t r i b u t i o n - f r e e statement. Weighting o f Unit Sales. The examples thus f a r have used i n d i v i d u a l packages as the u n i t o f a n a l y s i s , whereas many (but not a l l ) of the t e s t product brands were a v a i l a b l e i n more than one package s i z e . In order to e x t r a c t more information from the u n i t sales data, a l l o f the actual r i d i t analyses were based on brand s a l e s by weight (or volume, f o r some products). Consequently, purchases of l a r g e r packages are weighted more h e a v i l y than purchases of smaller ones i n the c a l c u l a t i o n o f the brand r i d i t s . 109 Table 16 i l l u s t r a t e s the c a l c u l a t i o n of brand r i d i t s based on t o t a l s a l e s i n grammes f o r the reference d i s t r i b u t i o n u n i t s a l e s data i n Table 10. Each brand of t h i s product was a v a i l a b l e i n one package s i z e but the Brand B package was 397g, compared to 206g f o r the other two brands. Thus, the data i n column 1 are the u n i t s a l e s m u l t i p l i e d by each brand's package weight. The r e s u l t i n g brand r i d i t s are s l i g h t l y d i f f e r e n t than those obtained i n Table 10, but the mean r i d i t w i l l s t i l l be 0.5. The r i d i t a s s o c iated with a brand now represents the t o t a l market share (by weight) o f a l l n u t r i t i v e l y lower-ranked brands, plus h a l f the market share (by weight) o f that brand, as observed during b a s e l i n e . Table 17 contains the s a l e s data o f Table 15, weighted by package s i z e . The c a l c u l a t e d mean r i d i t f o r the weighted data i s s l i g h t l y higher than t h a t f o r the unweighted data (0.587 vs. 0.573). I f one can imagine the t o t a l amount o f product purchased during a co n t r o l or treatment period being broken down i n t o one-gram u n i t s , i d e n t i f i a b l e by brand, then a mean r i d i t obtained from weighted data can be i n t e r p r e t e d as: the p r o b a b i l i t y t h a t a randomly picked gram from the t r e a t -ment purchases i s n u t r i t i v e l y s u p e r i o r to a randomly picked gram from the purchases made during c o n t r o l . Ranking of Brands i n Other Products. Only i n the case of two products (m&c d i n n e r , canned soup) d i d the rank order of brands by n u t r i t i v e per-formance remain c o n s i s t e n t across a l l e i g h t n u t r i t i o n a l cues and, there-f o r e , across a l l treatments. With other products, a brand's o v e r a l l n u t r i t i v e performance rank d i f f e r e d across treatments. T h i s s e c t i o n e x p l a i n s the steps taken to determine the various brand orderings f o r n o Table 16 C a l c u l a t i o n o f Brand R i d i t s From Reference D i s t r i b u t i o n of Sales Weighted by Package S i z e Brand (1) Sales ( i n g) (2) (3) (4) (5) R i d i t A 17304 8652 0 8652 .1083 C 60976 30488 17304 47792 .5984 B 1588 794 78280 79074 .9901 Total 79868 Table 17 C a l c u l a t i o n of the Mean R i d i t For the Data of Table 15, Weighted by Package S i z e Total Sales R i d i t Brand ( i n g) ( r ) Product A 2060 .1083 223.0980 C 16686 .5984 9984.9024 B 1985 .9901 1965.3485 Total 20731 12173.3489 - _ 12173.3489  r 20731 = 0.587 I l l the remaining products, so t h a t each treatment's unique brand r i d i t s could be derived and i t s mean r i d i t computed. Table 18 i s a f a c s i m i l e of the sign f o r Treatment "8/high", d i s -c l o s i n g brand r a t i n g s on a l l e i g h t a v a i l a b l e cues f o r bran c e r e a l . ^ Given these r a t i n g s , the seven brands are rank ordered by t h e i r n u t r i t i v e performance on each i n d i v i d u a l cue i n Table 19. The cues are l i s t e d , from l e f t to r i g h t , i n order of decreasing importance. The ranking o f brands i n single-cue treatments i s immediately a v a i l a b l e from the simple ranks given i n Table 19 f o r the cue f i b r e (used i n Treatment "1/high") and the cue sodium (used i n Treatment "1/low"). To determine a brand's o v e r a l l n u t r i t i v e performance rank i n t r e a t -ments where more than one cue was present, the i n d i v i d u a l ranks entered i n Table 19 were summed across the cues included i n a treatment. The r e s u l t i n g rank sums revealed a brand's o v e r a l l rank i n th a t treatment. For example, i n the 2-cue treatment (Treatment "2/high") g i v i n g brand r a t i n g s on f i b r e and p r o t e i n ( f i r s t two columns o f Table 19), Brand A's rank of 1 on f i b r e and 1.5 on p r o t e i n gives a rank sum of 2.5. Since t h i s turns out to be the lowest rank sum among the seven brands, Brand A ranks f i r s t , o v e r a l l , i n t h i s treatment. Any t i e s i n the rank sums were broken by r e f e r r i n g to the re s p e c t i v e brands' ranks on the most The brand code l e t t e r s l i s t e d at the extreme l e f t of Table 18 did not appear on the actual sign but are used here to f a c i l i t a t e the d i s c u s s i o n . Table 18 Point-of-Purchase In.formation Sign Employed For Treatment "8/high" (Bran Cereal) BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* | BRANDS FOOD PROTEIN VITAMIN B CALCIUM MAGNESIUM POTASSIUM PHOSPHORUS SODIUM BRAND CODE I alphabetical FIBRE NIACIN ' order) g g mg mg mg mg mg mg •I • I I A KELLOGG'S 2 8 u 2 1 g n 3 9 5 m g 3 0 6 2 3 ALL-BRAN B KELLOGG' S 2 3 1 2 n g g 2 8 2 m w 2 g 8 BRAN BUDS c nn»L°r,G'L. 10 10 21 47 212 352 450 697 BRAN FLAKES D KELLOGG'S 8 i 8 8 8 2 1 53 163 358 476 1000 RAISIN BRAN E CRUNCH I E S ^ 9 ' 3 1 0 , 3 2 1 ^ 1 6 6 4 6 1 m F G N A B I S C 0 25 11.3 23 87 431 1165 1120 620 100% BRAN POST BRAN N / A « 1 0 t, 21 43 221 543 421 618 FLAKES * data obtained from manufacturer ** data not available Table 19 Brand Performance Ranks (R^) by N u t r i e n t Ratings D i s c l o s e d i n Table 18 (Bran Cereal) Nutrient ( L i s t e d , From L e f t to Right, In order of Decreasing Importance) Brand F i b r e P r o t e i n Niacin Ca Mg K P Na Code ( R 1 } . (R 2) (R 3) (R 4 ) . (R5> (R 6) (R7) (R 8) A l a 1.5 4.5 3 2 3 2 4 B 3 1.5 4.5 1 3 2 3 1 C 4 6 4.5 6 5 7 6 6 D 6 7 4.5 4 7 6 4 7 E 5 5 4.5 5 6 5 5 5 F 2 3 1 2 1 1 1 3 G 7 4 4.5 7 4 4 7 2 aRead: Brand A ranked 1st on f i b r e (cue 1 ) , 1.5 on p r o t e i n (cue 2 ) , 4.5 on n i a c i n (cue 3 ) , 3rd on calcium (cue 4 ) , e t c . 114 important cue contained i n that t r e a t m e n t . x ± Table 20 summarizes the r e s u l t s of t h i s procedure f o r bran cereal n treatments c o n t a i n i n g more than one cue. The £ are the sums, over n cues, o f the ranks on i n d i v i d u a l cues (R^) given i n Table 1 9 . R" i s a brand's o v e r a l l n u t r i t i v e performance rank i n a given treatment. I t i s c l e a r from Table 20 t h a t the rank o r d e r i n g o f brands by o v e r a l l n u t r i t i v e performance changes from treatment to treatment. R e c a l l t h a t r i d i t a n a l y s i s r e q u i r e s the brands of the reference d i s t r i b u t i o n to be ordered by t h e i r n u t r i t i v e performance ranks before a treatment's brand r i d i t s can be derived and i t s mean r i d i t c a l c u l a t e d . Therefore, the reference d i s t r i b u t i o n o f brand s a l e s data was ordered i n each treatment's a n a l y s i s according to the n u t r i t i v e performance ranks e s t a b l i s h e d i n Table 20 o r , i n the case o f the 1-cue treatments, according to the ranks given i n the f i r s t and l a s t columns of Table 19 ( f i b r e and sodium, r e s p e c t i v e l y ) . The procedure described f o r bran cereal was used to determine the A much more complex method o f determining each brand's o v e r a l l rank would be to weight each cue by the mean importance score derived from the survey (see Table 6) and to m u l t i p l y the i n d i v i d u a l ranks by t h e i r r e s p e c t i v e cue importance scores i n order to o b t a i n weighted rank sums. This method was t r i e d and the o v e r a l l n u t r i t i v e performance ranks ob-tained were compared with the r e s u l t s from the simpler procedure described above. The two procedures gave i d e n t i c a l o v e r a l l ranks i n a l l treatments but three. Therefore, mean r i d i t s were computed from these treatment data a f t e r ranking the brands i n the two d i f f e r e n t ways. The two methods y i e l d e d mean r i d i t s w i t h i n .025 of each o t h e r , hence the choice of one ranking method over the other does not c r i t i c a l l y a l t e r the r e s u l t s f o r the product (see Table 38). For the remaining products whose o v e r a l l brand ranking d i f f e r e d across treatments, both the weighted rank sum and the simple rank sum methods were t r i e d on each treatment. Without e x c e p t i o n , both methods gave the same o v e r a l l performance ranks. Table 20 Determination of Overall N u t r i t i v e Performance Ranks (R) of Bran Cereal Brands In Treatments Containing More Than One Cue, Using Table 19 Data Treatment '-'2/high" "4/high" "8/high" "2/low" "4/low" "8/1 ow" Brand 2 4 8 8 8 8 Code £ R t IVR. R I R, R . I R- R I Ri R I R1 R 1 1 1 i=7 1 i=5 1 i = l 1 A 2.5 1 10.0 2 a 21.0 3 6 3 11 3 21.0 3 B 4.5 2 10.0 3 a 19.0 2 4 2 a 9 2 19.0 2 C 10.0 4 a 20.5 5 44.5 6 12 7 24 6 a 44.5 6 D 13.0 7 21.5 6 45.5 7 11 6 24 7 a 45.5 7 E 10.0 5 a 19.5 4 40.5 5 10 5 21 5 40.5 5 F 5.0 3 8.0 1 14.0 r 4 l a 6 1 14.0 1 G 11.0 6 22.5 7 39.5 4 9 4 17 4 39.5 4 The t i e i n rank sums of these two brands was broken by r e f e r r i n g to each brand's rank on the most important cue contained i n the treatment. 116 ranking of brands by n u t r i t i v e performance f o r each treatment with the remaining products. Tables s i m i l a r to 19 and 20 summarize the r e s u l t s o f the ranking procedures on these products and are given i n Appendix F. The researcher's a p r i o r i o r d ering o f brands on some q u a l i t a t i v e s c a l e i s c e n t r a l to the technique o f r i d i t a n a l y s i s . Even so, the mean r i d i t s t a t i s t i c , summarizing d i r e c t i o n a l changes i n the d i s t r i b u t i o n o f brand s a l e s , i s apparently q u i t e robust to problems commonly encountered i n a p p l i c a t i o n s of s u b j e c t i v e s c a l e s , two of which are r e l e v a n t here: 1. V a r i a t i o n s i n the number of categ o r i e s (brands) used, these representing i n t e r v a l s on a n u t r i t i v e performance continuum; 2. Minor v a r i a t i o n s i n the judgmental ordering of ca t e g o r i e s (brands) as might occur, f o r example, i f a d i f f e r e n t procedure of ranking the brands by n u t r i t i v e performance than the one described above had r e s u l t e d i n a s l i g h t l y d i f f e r e n t ordering o f the brands. Bross (1958:34) c l a r i f i e s the p r o p e r t i e s o f the mean r i d i t s t a t i s t i c and underscores i t s robustness with respect to these two problems: . . . One person might use a 5-point s c a l e , another a 15-point s c a l e . The same i n v e s t i g a t o r may, from time to time, wish to c o n s o l i d a t e or elaborate the s u b d i v i s i o n s o f h i s s u b j e c t i v e s c a l e . Now l e t us suppose that two i n v e s t i g a t o r s (using s c a l e s w i t h d i f f e r e n t numbers of s u b d i v i s i o n s but otherwise c o n s i s t e n t with each other) rate the same two sets of subjects ( i . e . a 'reference' s e t and an 'other' s e t ) . Both apply r i d i t s to t h e i r data and then the two i n v e s t i g a t o r s compare r e s u l t s . They w i l l f i n d very n e a r l y the same average r i d i t i n the 'other' s e t ! Or consider another p e c u l i a r i t y o f s u b j e c t i v e s c a l e s , one which can occur even when two i n v e s t i g a t o r s use the same number of s u b d i v i s i o n s and the same name f o r the s u b d i v i s i o n s . I t i s c h a r a c t e r i s t i c of s u b j e c t i v e s c a l e s t h a t a kind of 'slippage' w i l l take place -- the b o r d e r l i n e between two categ o r i e s w i l l 117 not c o i n c i d e f o r the two i n v e s t i g a t o r s . A c l a s s o f cases w i l l e x i s t which one s c i e n t i s t would r a t e as 'minor' but which the other rates as 'moderate!' Yet, i f the two i n v e s t i g a t o r s r a t e the same two s e r i e s of subjects and then use r i d i t s t h e i r r e s u l t s w i l l be i n good numerical agree-ment! These and s i m i l a r p r o p e r t i e s o f r i d i t s are not very sur-p r i s i n g when the meaning o f the average r i d i t i s r e c a l l e d . Both i n v e s t i g a t o r s are estimating the same t h i n g : the chance t h a t an i n d i v i d u a l i n the 'other' s e r i e s i s worse o f f than an i n d i v i d u a l i n the reference s e t . The e f f e c t s of c o n s o l i d a t i n g o r e l a b o r a t i n g c a t e g o r i e s or slippage are s i m i l a r to those i n the est i m a t i o n o f the mean of a d i s -t r i b u t i o n from grouped data. Database Manipulation The data c o l l e c t e d on peanut b u t t e r were deleted a l t o g e t h e r from the database at the very outset. The five-week experiment coincided with an unexpected changeover from avoirdupois to metric container s i z e s by several of the peanut b u t t e r s u p p l i e r s . As a r e s u l t , i n t e r m i t t e n t stock outages o f numerous brands i n both stores marred the treatment and c o n t r o l observations to the point where t h i s product was considered u n r e l i a b l e f o r the data a n a l y s i s . The s t o r e s a l e s data recorded on d a i l y coding forms (Appendix D) were keypunched on computer cards and l a t e r t r a n s c r i b e d onto magnetic disk f i l e s f o r a n a l y s i s by computer. Figure 6 o u t l i n e s the steps taken to process the raw data, beginning w i t h t h e i r conversion to the re l e v a n t v a r i a b l e s and ending with t h e i r a n a l y s i s v i a various s t a t i s t i c a l programs. I n i t i a l l y , the d a t a f i l e manipulations were performed separately f o r each s t o r e . The item movement counts, representing b e f o r e - a f t e r measurements i n the two d a i l y treatment p e r i o d s , were submitted to the SPSS ( S t a t i s t i c a l Package f o r the S o c i a l Sciences) computer program 118 Figure 6 Database Manipulation and Plan o f A n a l y s i s P r i n t and Review f o r E r rors Store 1 Data F i l e I To Disk F i l e i SPSS Program I Convert U.P.C. Scanner!'Data to U n i t Sales by Itern/Period Item Sales Count i n A l l Periods Delete Product Data With C r i t i c a l Stock Outages Add Conversion Program Store 2 Data F i l e To Disk F i l e 1 SPSS Program Convert U.P.C. Scanner Data to Unit Sales by Item/Period Item Sales Count i n A l l Periods P r i nt and Review f o r E r rors Delete Product Data With C r i t i c a l Stock Outages Add Conversion Program Convert Item Sales Into Raw/Weighted Brand Sales by Week/Product/Treatment Add Merging Program 1 Pool Store Data i Convert Item Sales Into Raw/Weighted Brand Sales by Week/Product/Treatment Add Merging Program I ure 6 (Continued) To Output F i l e Pool Control Data Output With Pooled Control Data Run Chi Square Program • Test For "Sign C o n t r o l " / "No-Sign Control' D i f f e r e n c e s l Run B e l l Canada R i d i t A n a l y s i s Program Mean R i d i t Output F i l e Run SPSS Subprograms ANOVA/T-TEST. Test Hypotheses I Run Modified R i d i t A n a l y s i s Program Test Hypotheses H 2 120 (Nie e t a l . , 1975). The counts were converted i n t o t o t a l u n i t s a l e s of each item i n each treatment period. These were then reviewed f o r key-punching e r r o r s . Treatment and c o n t r o l data a f f e c t e d by stock outages i n the stores were removed at t h i s p o i n t . The d e c i s i o n c r i t e r i o n was to d e l e t e such data whenever a l l s i z e s of a brand were out of stock during any part of a treatment p e r i o d . Next, a s p e c i a l l y w r i t t e n program converted the item s a l e s data i n t o unweighted (raw) brand s a l e s and brand sales weighted by package s i z e , by week, product and treatment (or c o n t r o l ) . A merging program was then added to pool t h i s output o f each s t o r e i n t o one output f i l e . T his pooled f i l e provided the basic data inputs to the s t a t i s t i c a l analyses which f o l l o w e d . Pooling o f Data. As was expected, t o t a l purchases of a t e s t product v a r i e d across treatment periods as a r e s u l t o f u n c o n t r o l l a b l e f a c t o r s . O c c a s i o n a l l y , a product's treatment sample s i z e was low enough to con-t a i n zero sales of the lowest volume brand. Therefore, i n order to reduce sampling e r r o r as much as p o s s i b l e , i t was decided to pool the corresponding treatment and c o n t r o l data across both s t o r e s , as was i n d i c a t e d above. This p o o l i n g procedure r e s u l t e d i n combined-store baselines re-presenting each product's and week's reference d i s t r i b u t i o n f o r r i d i t a n a l y s i s . I t i s against these b a s e l i n e s that the r e s p e c t i v e combined- store treatment d i s t r i b u t i o n s are t e s t e d f o r treatment e f f e c t s . R e c a l l t h a t r i d i t a n a l y s i s i s a d i s t r i b u t i o n - f r e e method r e s u l t i n g i n a 121 d i s t r i b u t i o n - f r e e p r o b a b i l i t y statement about the e f f e c t s o f an ex-perimental treatment. Regardless of how the brand s a l e s frequencies are d i s t r i b u t e d i n each separate s t o r e to begin w i t h , treatment e f f e c t s are assessed by i d e n t i f y i n g any d i r e c t i o n a l s h i f t s i n the pooled treatment d i s t r i b u t i o n , as compared to the pooled reference d i s t r i b u t i o n . I f there are no d i r e c t i o n a l s h i f t s then the treatment and reference brand s a l e s d i s t r i b u t i o n s represent samples drawn from the same population of pooled-store purchases. With the data now pooled across s t o r e s , the next step was to determine whether the two types o f experimental c o n t r o l s i n the design ( i . e . , "sign c o n t r o l " and "no-sign c o n t r o l " ) had any d i f f e r e n t i a l e f f e c t s on brand s a l e s . B a r r i n g s i g n i f i c a n t d i f f e r e n c e s , the sa l e s data from both c o n t r o l s could be pooled i n t o one o v e r a l l c o n t r o l to increase the sample s i z e o f a product's reference d i s t r i b u t i o n . P r e f e r a b l y , the s i z e of the reference d i s t r i b u t i o n s e l e c t e d f o r r i d i t a n a l y s i s should be l a r g e enough to insure t h a t the derived r i d i t s w i l l be s t a b l e (Bross, 1958:22) and r e p r e s e n t a t i v e of the pooled-store population o f brand purchases when no point-of-purchase n u t r i t i o n a l information signs are a v a i l a b l e to shoppers. With no a p r i o r i notions of which way a "sign c o n t r o l " might a f f e c t brand s a l e s , the appropriate a n a l y t i c a l method was to t e s t the two types o f c o n t r o l s f o r any kind of d i s t r i b u t i o n a l d i f f e r e n c e at a l l , using two-sample chi square a n a l y s i s . Given the nominally-scaled brand s a l e s data, the chi square t e s t i s a p p l i c a b l e and i s s e n s i t i v e to any kind of d i f f e r e n c e i n the two c o n t r o l d i s t r i b u t i o n s ( S i e g e l , 1956:157). 122 The t e s t s were performed on the brand s a l e s frequencies o f the two c o n t r o l s f o r each product i n the two experimental weeks. Therefore, 10 such comparisons were made (5 products x 2 weeks). The frequencies were weighted by a brand's package s i z e (s) and the sal e s entered i n the c e l l s o f the chi square t a b l e were i n standard u n i t s , so t h a t the t o t a l number of u n i t s sold (N) remained the same a f t e r weighting. Table 21 demonstrates the procedure followed and shows a two-sample chi square t e s t on the two co n t r o l d i s t r i b u t i o n s f o r macaroni & cheese dinner i n Week 2. The c e l l e n t r i e s are brand s a l e s i n standard u n i t s , while the e n t r i e s i n parentheses are the brand s a l e s i n grams. To convert the sal e s data i n t o standard u n i t s the t o t a l weight s o l d (80,074g) was d i v i d e d by the t o t a l u n i t s a l e s (N = 385 packages), to obtai n a weight o f 207.98g f o r one standard u n i t i n t h i s p a r t i c u l a r t a b l e . D i v i d i n g each brand's sales i n grams by t h i s f i g u r e gives the weighted brand s a l e s i n standard u n i t s . The d i s t r i b u t i o n under "sign c o n t r o l " was not s i g n i f i c a n t l y d i f f e r e n t from that under "no-sign c o n t r o l " ( x 2 = 2.02; df = 2; p < .40). With a s i n g l e e xception, these t e s t s i n d i c a t e d no d i f f e r e n c e s between 12 "sign c o n t r o l " and "no-sign c o n t r o l " a t the 5% l e v e l o f s i g n i f i c a n c e . The f i n d i n g on the macaroni & cheese dinner data i n the second experimental For two of these t e s t s , c e l l s with very low sal e s frequencies were combined with an adjacent c e l l so as not to i n v a l i d a t e the chi square t e s t with expected frequencies below those recommended by Sig e l (1956:110). A l s o , i n a l l t e s t s where d f ;= 1, x 2 was computed with Yates' c o r r e c t i o n f o r c o n t i n u i t y (see S i e g e l , 1956:107). 123 Table 21 Chi Square A n a l y s i s o f Brand Sales D i s t r i b u t i o n s Observed Under Two Types o f Experimental Controls f o r M&C Dinner In Week 2, Using Unit Sales Weighted by Package S i z e "Sign "No-Sign Brand C o n t r o l " C o n t r o l " Total B 3.82 3.82 7.64 (794) (794) (1588) C 94.09 198.09 292.19 (19570) (41200) (60770) A 32.69 52.49 85.18 (6798) (10918) (17716) To t a l s 130.60 254.40 385.00 = N (27162) (52912) (80074) Note: X 2 = 2.02; df = 2; n.s. week (Week 3) c o n s t i t u t e d the one exception/Xx 2 =13.28; df = 2; 13 p < .005). Thus, the two co n t r o l d i s t r i b u t i o n s of t h i s product were d i f f e r e n t i n Week 3 but not i n Week 2. On the basis of these r e s u l t s , the d e c i s i o n was made to pool the two types of c o n t r o l data i n t o one l a r g e r c o n t r o l , thereby e s t a b l i s h i n g each product's reference d i s t r i b u t i o n f o r a given experimental week. An examination of the i n d i v i d u a l s t o r e data revealed that t h i s d i f f e r e n c e between c o n t r o l s was s i g n i f i c a n t i n one s t o r e (x 2 = 8.44; df = 1; p < .005) but not i n the other (x 2 = 0.12; df = 1; n.s.). 124 Therefore, i n the output f i l e f o r a l l s t a t i s t i c a l analyses which f o l l o w e d , the reference d i s t r i b u t i o n s c o n s i s t e d o f s a l e s data pooled across the two types of c o n t r o l . With t h i s step i n the data manipulation, the f i n a l output f i l e l eading to the s t a t i s t i c a l a n a l y s i s stage c o n s i s t e d of weighted brand sales data, pooled across the two stores and based on the s a l e s o f 21,833 packages of f i v e products, over f i v e weeks. Table 22 i n d i c a t e s the t o t a l number of containers purchased, by product and week. General Methodological Approach The a n a l y t i c a l methods chosen were judged to be appropriate to the basic f a c t o r i a l design o f t h i s research and to the t e s t i n g o f hypothesized r e l a t i o n s h i p s between the research v a r i a b l e s . The primary focus i n t h i s d i s s e r t a t i o n i s on the separate e f f e c t s of load and cue importance, as we l l as t h e i r j o i n t e f f e c t s , a f t e r accounting f o r any e f f e c t s due to non-experimental v a r i a b l e s (e.g. products, weeks). This i n d i c a t e d the use of n-way analyses of variance. A subsequent s h i f t i n a t t e n t i o n to s p e c i f i c l e v e l s o f load i n order to t e s t p o r tions of the information load curve and to study d i f f e r e n c e s between various "high" and "low" importance treatments suggested the use of two-sample t - t e s t s . Given t h i s choice of methodology, use was made o f the ANOVA, ONEWAY and T-TEST subprograms a v a i l a b l e i n the SPSS -- Version 8.0 computer a n a l y s i s program (Nie et a l . , 1975). P r i o r to the a p p l i c a t i o n of these techniques, the brand s a l e s data (weighted by package s i z e ) had to, be transformed i n t o mean r i d i t s , which 125 Table 22 Database i n Terms o f Total Number of Containers Sold of Each Product i n Each Week Product Week 1 Week 2 a Week 3 a Week 4 Week 5 Total Canned Soup 5107 888 1063 1179 938 9175 Mayonnaise 445 354 331 426 373 1929 Ketchup 675 815 576 835 541 3442 M&C Dinner 1185 1086 747 902 719 4639 Bran Cereal 600 548 474 505 521 2648 Total 21833 aThese two weeks are the experimental weeks; the other weeks served as b a s e l i n e s . 126 which served as the dependent v a r i a b l e . For t h i s purpose, the output f i l e data were submitted to a r i d i t a n a l y s i s program.1'* The mean r i d i t s generated from t h i s r o u t i n e provided the inputs to the SPSS r o u t i n e s , as shown i n Figure 6. Turning to the p r o d u c t - s p e c i f i c research questions, the drawing of inferences about general e f f e c t s o f n u t r i t i o n a l information on brand choices c a l l e d f o r a treatment-by-treatment e v a l u a t i o n of the dependent v a r i a b l e . A l s o , each product's experimental c o n t r o l data had to be com-pared against the product's b a s e l i n e data i n other weeks. These i n d i v i d u a l e valuations s i g n a l l e d the use of separate z - t e s t s on each mean r i d i t to assess whether i t was s i g n i f i c a n t l y d i f f e r e n t from 0.5. For t h i s purpose, a s p e c i f i c a l l y modified r i d i t a n a l y s i s program was w r i t t e n so that the i n d i v i d u a l mean r i d i t s , which were based on weighted brand s a l e s , could be subjected to t e s t s of s i g n i f i c a n c e . The next and f i n a l s e c t i o n o f t h i s chapter presents a concise des-c r i p t i o n of the s p e c i f i c s t a t i s t i c a l t e s t s chosen t o analyze the research r e s u l t s . S t a t i s t i c a l Tests Employed ANOVA and t - T e s t s . Parametric s t a t i s t i c a l models such as a n a l y s i s of variance and the two-sample t - t e s t are most powerful i n the sense o f r e j e c t i n g HQ when HQ i s f a l s e , provided the dependent v a r i a b l e being analyzed meets c e r t a i n assumptions underlying each model. Siegel (1956:19) The author would l i k e to thank Dr. Gideon Vigderhous, Survey Research Group, B e l l Canada, Montreal, f o r p r o v i d i n g the Fortran program on r i d i t a n a l y s i s . 127 has summarized these basic assumptions as f o l l o w s : 1. The observations must be independent. That i s , the s e l e c t i o n o f any one case from the population f o r i n c l u s i o n i n the sample must not bias the chances of any other case f o r i n c l u s i o n , and the score which i s assigned t o any case must not bias the score which i s assigned to any other case. 2. The observations must be drawn from normally d i s t r i b u t e d populations. 3. These populations must have the same variance ( o r , i n s p e c i a l cases, they must have a known r a t i o o f v a r i a n c e s ) . 4. The v a r i a b l e s involved must have been measured i n at l e a s t an i n t e r v a l s c a l e , so th a t i t i s p o s s i b l e to use the operations o f a r i t h m e t i c (adding, d i v i d i n g , f i n d i n g means, etc.) on the scores. In the case of the a n a l y s i s o f variance (the £ t e s t ) , another c o n d i t i o n i s added to those already given: 5. The means of these normal and homoscedastic populations must be l i n e a r combinations o f e f f e c t s due to columns and/or rows. That i s , the e f f e c t s must be a d d i t i v e . The b r i e f d i s c u s s i o n which f o l l o w s i s intended to j u s t i f y the use of these parametric t e s t s by showing t h a t the mean r i d i t observations reason-ably f u l f i l l the underlying assumptions. F i r s t , each mean r i d i t derived from the brand s a l e s data representing a treatment or c o n t r o l i s s t a t i s t i c a l l y independent of every other s i m i l a r l y derived mean r i d i t . Second, the mean r i d i t i s a r a t i o - s c a l e d v a r i a b l e (0 < r < 1). T h i r d , by v i r t u e of the c e n t r a l l i m i t theorem, the mean r i d i t s , r e p r e s e n t i n g means of samples drawn from a population o f r i d i t s , are ne a r l y normally d i s t r i b u t e d . From the s a l e s data on 80 separate treatments (8 signs x 5 products x 2 weeks) 77 mean r i d i t s were c a l c u l a t e d and t h e i r frequency d i s t r i b u t i o n 128 i s shown i n * F i g u r e 7.J"a The mean r i d i t s range from .408 to .664 and the mean of t h i s d i s t r i b u t i o n i s .518. With respect to the assumption of populations with equal variances ( i . e . , homogeneity o f variances or homoscedasticity), the SPSS subprogram T-TEST computes two t s t a t i s t i c s f o r every p a i r w i s e comparison of means. One t i s based on a "pooled variance" estimate of a 2 , representing the weighted average of the sample variances S j 2 and s 2 2 . 1 6 T n e second t i s based on a "separate variance" e s t i m a t e . 1 ^ In the process of computing these two t s t a t i s t i c s , the program i n -corporates an F - t e s t f o r homogeneity of variances among the two samples of mean r i d i t s . As noted by Nie et a l . (1975:270), t h i s gives the T-TEST routi n e user the option of deciding which t s t a t i s t i c w i l l be appropriate Three mean r i d i t s were discarded from the database. One was c r i t i c a l l y a f f e c t e d by a stockout i n canned soup. The other two, r e -presenting treatment "1/low" with mayonnaise were judged to be meaning-l e s s s i n c e the information s i g n representing t h i s treatment provided no n u t r i t i o n a l data on two of these three brands. 16 s 2 ( n r - Dsx2 + (n2 - l ) s 2 2 K - 1) + (n2 - 1) , where X. i s the mean of t n x + s 2 / n 2 a sample of mean r i d i t s f o r Treatment i . 17 t / S i 2/ni + s 2 2 / n 2 Figure 7 D i s t r i b u t i o n of 77 Mean R i d i t s Obtained From Treatments On Five Products i n Two Experimental Weeks 16 -14 -12 -10 -8 -6 - * 1 ^ 1 4 -2 -.40 .42 .44 .46 .48 .50 .52 .54 .56 .58 .60 .62 .64 .66 .68 Mean R i d i t 130 f o r each pairwise comparison: I f i t i s not known whether the two populations have the same v a r i a n c e , an F t e s t of sample variances may be performed: The n u l l hypothesis ti0:ai2 = a22 w i t h a l t e r n a t i v e H\:oiz f a22 and a s i g n i f i c a n c e l e v e l a1* i s chosen ( a ' does not n e c e s s a r i l y have the same value as a used f o r the t - t e s t ) . From the sample v a r i a n c e s , F i s computed. F = l a r g e r s 2 smaller s 2 I f the p r o b a b i l i t y f o r F i s g r e a t e r than, a", H 0 i s accepted; t based on the pooled-variance estimate f o r Op 2 should be issued. I f the p r o b a b i l i t y f o r F i s l e s s than or equal to a', H 0 i s r e j e c t e d ; t based on the separate variance estimate f o r 0-Q2 should be used. In any case, both the t t e s t and the a n a l y s i s of variance have been found to be q u i t e robust with respect to minor d e v i a t i o n s from the assumptions o f normality and equal v a r i a n c e s , even with small sample s i z e s . Ferguson (1976:166) summarizes e m p i r i c a l i n v e s t i g a t i o n s of t h i s property of the t t e s t by noting: . . . The t t e s t should be used only when there i s reason to b e l i e v e that the population d i s t r i b u t i o n s do not depart too g r o s s l y from the normal form and the population variances do not d i f f e r markedly from e q u a l i t y . Tests of normality and homogeneity of variance may be a p p l i e d , but these t e s t s are not very s e n s i t i v e f o r small samples. With respect t o the robustness o f the a n a l y s i s o f v a r i a n c e , Ferguson (1976:235) i n d i c a t e s : With most sets of r e a l data the assumptions underlying the a n a l y s i s of variance are, at best, only roughly s a t i s f i e d . The raw data of experiments f r e q u e n t l y do not e x h i b i t the c h a r a c t e r i s t i c s which the mathematical models r e q u i r e . One advantage of the a n a l y s i s of variance i s t h a t reason-able departures from the assumptions of normality and homogeneity may occur without s e r i o u s l y a f f e c t i n g the v a l i d i t y o f the inferences drawn from the data. 131 Mean R i d i t S i g n i f i c a n c e Test. The s i g n i f i c a n c e o f the d i f f e r e n c e between any obtained mean r i d i t and i t s expected value o f 0.5 can be determined by applying a z - t e s t , as i s commonly done with c e r t a i n non-parametric s t a t i s t i c s . The mean r i d i t , r , i s nearly normally d i s t r i b u t e d and, t h e r e f o r e , i f the standard e r r o r of a mean r i d i t i s found, one can c a l c u l a t e the corresponding standard normal d e v i a t e , z', to determine the p r o b a b i l i t y , under H Q, that z > z 1 . Bross (1958:36) exp l a i n s t h a t i n the reference d i s t r i b u t i o n the frequency d i s t r i b u t i o n o f i n d i v i d u a l r i d i t s w i l l approximate t h a t o f the rec t a n g u l a r d i s t r i b u t i o n : . . . No matter what the nature o f the o r i g i n a l observations may be the d i s t r i b u t i o n of the r i d i t s i s going to be c l o s e l y approximated by a t h e o r e t i c a l curve (the 'rectangular d i s t r i b u t i o n ' ) . The sol e exception occurs when nearly a l l of the observations f a l l i n t o one or two categ o r i e s ( i n which case a c o r r e c t i o n i s needed to reduce the variance and the approximation i s poorer). Since the variance of the rect a n g u l a r d i s t r i b u t i o n i s 1/12, the standard d e v i a t i o n i n a frequency d i s t r i b u t i o n of r i d i t s w i l l be c l o s e to 1//L2, or 1/2/f. A c c o r d i n g l y , the standard e r r o r of an average r i d i t ( r ) w i l l be approximated by s.e.(r) = (4.3) where n i s the number of observations going i n t o the mean. However, the standard e r r o r s of mean r i d i t s w i l l vary not only with sample s i z e but whenever the r i d i t d i s t r i b u t i o n s depart from the 132 rectangular. F l e i s s (1981:154) gives the general formula f o r c a l c u l a t i n g the standard e r r o r of a mean r i d i t , based on the sample s i z e s o f both the reference d i s t r i b u t i o n and the treatment d i s t r i b u t i o n from which the mean r i d i t was c a l c u l a t e d : S' e' ( f ) = ^ r f A / 1 + ^ + N ( N * " ' 1 1 " + » - 1) r <4-4> where N. = the number of packages (weighted as standard u n i t s of Brand j i n the reference d i s t r i b u t i o n ; j = l,2,...,k brands; k N = 'j?N. the t o t a l number of weighted packages i n the reference d i s t r i b u t i o n ; n. = the number of weighted packages o f Brand j i n the t r e a t -ment d i s t r i b u t i o n ; k n = £n . the t o t a l number of weighted packages i n the treatment d i s t r i b u t i o n . Note t h a t expression (4.3) i s now adjusted by the square.root o f three f a c t o r s i n (4.4). The component n + 1/N weights the o r i g i n a l standard e r r o r (4.3) by the r a t i o o f the treatment sample s i z e (n) to the reference sample s i z e (N). This adjusts the standard e r r o r to the pooled sample, N + n, r e s u l t i n g i n one standard e r r o r estimate from the two sample d i s -t r i b u t i o n s being compared. The second component, 1/N(N + n - 1 ) , makes an adjustment f o r the 133 o v e r a l l sample s i z e , N + n, such t h a t small o v e r a l l sample s i z e s w i l l increase the standard e r r o r i n (4.3) more than w i l l l a r g e samples. The t h i r d component, - £(N. + n.) 3/N(N + n)(N + n - 1 ) , provides a c o r r e c t i o n f o r r i d i t d i s t r i b u t i o n s which depart from the rect a n g u l a r . Departures from the rect a n g u l a r have the e f f e c t of reducing the variance o f the k r i d i t s . As a d i s t r i b u t i o n becomes skewed, the numerator, £(N. + n . ) 3 , w i l l increase i n proportion to the denominator. Consequently, the r a t i o 18 - ( : ) w i l l tend towards -1 and reduce the standard e r r o r given by (4.3). Using (4.4) to obtain the standard e r r o r of a mean r i d i t , the s i g -n i f i c a n c e o f the d i f f e r e n c e between r and 0.5 can be teste d by r e f e r r i n g the value of I f , i n f a c t , the d i s t r i b u t i o n o f r i d i t s i s rect a n g u l a r and N i s very l a r g e r e l a t i v e to n, then as N i n c r e a s e s , k " I i * " i ) 3 k 1 im N -> < CO N(N + n)(N + n - 1) where k i s the number of brands, and 0 < The upper bound of j corresponds to the maximum variance of a rect a n g u l a r d i s t r i b u t i o n . 134 z = .5 s.e.(r) (4.5) to a t a b l e of the standard normal d i s t r i b u t i o n . Table 23 shows the r e s u l t s o f such a z - t e s t on the hy p o t h e t i c a l data presented e a r l i e r i n Tables 16 and 17. The s a l e s , i n grams, of three brands f o r the reference and treatment d i s t r i b u t i o n s are given i n the l e f t - m o s t columns, along with the computed r i d i t s . The data here represent combined sales of 480 packages f o r the two d i s t r i b u t i o n s . The l a s t three columns i n d i c a t e the same weighted brand s a l e s , t h i s time i n standard u n i t s , and provide the inputs f o r the c a l c u l a t i o n of the 19 standard e r r o r , using formula (4.4). The obtained value o f i n d i c a t e s t h a t the treatment mean r i d i t o f .587 i s s i g n i f i c a n t l y g reater than the reference mean r i d i t of .5 (p < .0005, o n e - t a i l e d ) . Combining Independent S i g n i f i c a n c e Tests. An a p p l i c a t i o n of the mean r i d i t s i g n i f i c a n c e t e s t to each of the 77 mean r i d i t s observed i n the experiment y i e l d s a set of 77 d i f f e r e n t p value ( p r o b a b i l i t i e s o f a 19 The use of i n d i v i d u a l package s a l e s frequencies as independent samples i n s t a t i s t i c a l a n a l y s i s i s not uncommon f o r i n - s t o r e experiments i n marketing. In Russo's (1977) experiment, i n d i v i d u a l container s a l e s represented independent observations f o r t e s t products l i k e apple j u i c e , ketchup and peanut b u t t e r . I 135 Table .23 Weighted Frequency D i s t r i b u t i o n s f o r T e s t i n g the S i g n i f i c a n c e of the D i f f e r e n c e Between Two Mean R i d i t s Obtained From Table 16 and 17 Data Sales i n g Sales i n i Standard Units Brand Reference Treatment R i d i t (•r) Reference (N.) Treatment (nj) Total (N.+n.) J y A 17304 2060 .1083 82.56 9.83 92.39 C 60976 16686 .5984 290.94 79.62 370.56 B 1588 1985 .9901 7.58 9.47 17.05 Tot a l s 79868 20731 381.08(=N) 98.92(=n) 480.00(=N+n) Mean R i d i t Reference Data = .500 Mean R i d i t Treatment Data = .587 Standard E r r o r = .0238 z Value = 3.66 p < .0005 136 Type I e r r o r ) . For mean r i d i t s g r e a t e r than 0.5 ( i . e . , i n the hypo-; th e s i z e d d i r e c t i o n ) some p values were l e s s than and some greater than a = .05. The same a p p l i e d to p values associated with mean r i d i t s be-low 0.5, although these c o n s t i t u t e d r e v e r s a l s i n the hypothesized d i r e c t i o n . In the sense t h a t a l l o f the experimental treatments provided some amount of n u t r i t i o n a l i n f o r m a t i o n t o shoppers, one o b j e c t i v e was to t e s t whether these d i v e r s e r e s u l t s from numerous i n d i v i d u a l t r e a t -ments support the general hypothesis t h a t n u t r i t i o n a l information leads to an increase i n the mean r i d i t (H^). S p e c i f i c a l l y , i s there an o v e r a l l  tendency f o r the mean r i d i t to increase whenever some n u t r i t i o n a l i n -formation was posted at the point of purchase? Given the p values from several s t a t i s t i c a l l y independent s i g -n i f i c a n c e t e s t s o f the same phenomenon, Scott and Wertheimer (1962:369; see a l s o Jones and F i s k e , 1953; Donnelly and E t z e l , 1973) describe a procedure f o r t e s t i n g the o v e r a l l s i g n i f i c a n c e of the combined p values. The product of k independent p values may be transformed i n t o a composite chi-square value by the formula k x2 = -2 l o g p n p. , (4.7) e i = l 1 which i s d i s t r i b u t e d as chi square with 2k degrees of freedom. An e q u i v a l e n t and more convenient formula f o r o b t a i n i n g the composite X 2 i s given by Jones and Fiske (1953:376): 137 ,2 = k i=l i p g e Pi (4.8) The p r o b a b i l i t y associated with the obtained value o f x 2 provides a t e s t of the n u l l hypothesis that the several s i g n i f i c a n c e r e s u l t s being com-bined could have occurred by chance. Table 24 i l l u s t r a t e s t h i s procedure using a sample of mean r i d i t s obtained from e i g h t treatments g i v i n g n u t r i t i o n a l information on canned soup i n Week 3. S i x of the e i g h t mean r i d i t s are greater than 0 .5 , thereby denoting changes i n the hypothesized d i r e c t i o n . Mean r i d i t s i g n i f i c a n c e t e s t s revealed t h a t only three o f those changes were s i g -n i f i c a n t at a = .05. Given t h i s set of e i g h t independent t e s t r e s u l t s , i s the apparent tendency f o r the mean r i d i t to increase whenever some information on soup was posted i n Week 3 a dependable one? The o n e - t a i l e d p values from the e i g h t separate s i g n i f i c a n c e t e s t s are entered alongside t h e i r r e s p e c t i v e mean r i d i t s and z values. Since two of the mean r i d i t s are l e s s than 0.5 and represent r e v e r s a l s o f the o n e - d i r e c t i o n a l hypothesis on treatment e f f e c t s , t h e i r corresponding p r o b a b i l i t i e s are entered as 1 - p. The f i n a l column i n Table 24 gives the natural logarithm o f each p value and the sum o f e i g h t l o g s . Applying formula (4.8) to t h i s sum gives a x 2 value o f 38.316, which with 2k = 16 degrees of freedom i s h i g h l y s i g n i f i c a n t (p < .005). Therefore, the p r o b a b i l i t y of o b t a i n i n g t h i s p a r t i c u l a r s e t of r e s u l t s i n ei g h t independent t e s t s on the mean r i d i t , when i n f a c t there i s no e f f e c t from the n u t r i t i o n a l i n f o r m a t i o n , i s l e s s than f i v e chances i n a thousand. 138 Table 24 I l l u s t r a t i o n o f the Procedure For Test i n g the S i g n i f i c a n c e o f Combined Treatment Results Using the Data of Canned Soup i n Week 3 Mean One-Treatment R i d i t z Value T a i l e d p log p "1/high" .555 2.224 .013 -4.337 "2/high" .533 1.060 .145 -1.934 "4/high" .485 - .409 .659 - .417 "8/high" .498 - .074 .529 - .636 " l / l o w " .523 .794 .214 -1.544 "2/low" .549 1.782 .037 -3.287 "4/low" .555 2.564 .005 -5.264 "8/low" .521 .932 .176 -1.739 8 I l o g Pi i = l e 1 -19.158 X 2 = "2 I log p. i = l e 1 38.316; df = 16; p < .005 139 CHAPTER V ANALYSIS AND RESULTS Chapter IV presented the development o f the research design to generate the experimental data and discussed the a n a l y t i c a l methodology chosen to analyze t h a t data. This chapter focuses on the r e s u l t s of the analyses to t e s t each hypothesis. F i r s t , a number of p r e l i m i n a r y analyses are presented which provide the j u s t i f i c a t i o n f o r subsequent a n a l y t i c a l procedures. Each hypothesis i s then t e s t e d , i n t u r n , with the appropriate a n a l y t i c a l t o o l s and the corresponding r e s u l t s summarized. The chapter concludes with some a d d i t i o n a l analyses which were suggested by the f i n d i n g s from the e a r l i e r analyses. The Data Entering the Analyses Before presenting the r e s u l t s of parametric s t a t i s t i c a l t e s t s i n the f i r s t part o f t h i s chapter, a comment about the nature o f the data e n t e r i n g i n t o these analyses i s i n order. Since the input data (the mean r i d i t s ) are p r o p o r t i o n s , the p o s s i -b i l i t y o f transforming these data was considered. Some s t a t i s t i c a l t e x t -books suggest an a r c s i n e or angular transformation ( a r c s i n /p..) of pr o p o r t i o n a l data, i n order to brin g e r r o r variances c l o s e r to e q u a l i t y f o r the ANOVA and t - t e s t procedures ( c f . Winer, 1971:397-402; Snedecor and Cochran, 1967:327; Ferguson, 1976:235). Snedecor and Cochran (1967:328) note, however, t h a t the a r c s i n transformation i s only r e a l l y useful when the proportions are c l o s e to 140 0 or 1: Angles may al s o be used with proportions t h a t are subject to other sources of v a r i a t i o n i n a d d i t i o n to the b i n o m i a l , i f i t i s thought t h a t the variance o f p.. i s some m u l t i p l e of P-- (1 - P-j-j)- S i n c e , however, t h i s product v a r i e s l i t t l e f o r p.. l y i n g between 30% and 70%, the angular transformation i s s c a r c e l y needed i f nearly a l l the observed p.. l i e i n t h i s range. In f a c t , t h i s transformation i s u n l i k e l y to produce a n o t i c e a b l e change i n the conclusions unless the p.. range from near zero to 30% and beyond (or from below 70% to 100%). The mean r i d i t s i n the database range only from .408 to .664 and are, there-f o r e , a l l w i t h i n the .30 to .70 range s p e c i f i e d . b y Snedecor and Cochran (1967). Keppel (1973:559) o f f e r s the f o l l o w i n g advice on the c o n t r o v e r s i a l t o p i c of data transformations: I t should be ovbious t h a t the d e c i s i o n to use a transformation of the data i s a complicated business. A researcher should know e x a c t l y why he wants to transform h i s data. I f i t i s to s a t i s f y assumptions of normality and homogeneity of w i t h i n -c e l l v a r i a n c e s , then he might as wel l save himself the t r o u b l e , s i n c e the £ t e s t i s robust with regard to these v i o l a t i o n s . I f i t i s to s i m p l i f y the s t a t i s t i c a l model, he should seek com-petent s t a t i s t i c a l advice. The j u s t i f i c a t i o n of a data t r a n s -formation i s i n the hands of the user. Readers o f research reports are e n t i t l e d to know whether or not the outcome o f the experiment was m a t e r i a l l y a f f e c t e d by the transformation. Whether we l i k e i t or not, our a n a l y s i s and our conclusions are based on the data a c t u a l l y e n t e r i n g i n t o the a n a l y s i s . I f the scores are not transformed, then our conclusions i n v o l v e the means of the o r i g i n a l scores. I f the scores are t r a n s -formed, then our conclusions are r e s t r i c t e d to the transformed  means.' In c o n s i d e r a t i o n of the above p o i n t s , the d e c i s i o n was made not to transform the mean r i d i t data by the a r c s i n . A l l analyses and conclusions are based on the values of the o r i g i n a l mean r i d i t s . Tests For Treatment-Product and Treatment-Week I n t e r a c t i o n s The 10 r e p l i c a t e s w i t h i n each c e l l of the 4 x 2 f a c t o r i a l design 141 v a r i e d across f i v e products and two weeks. Therefore, p r i o r to the analyses of variance on the experimental f a c t o r s , i t was necessary to a s c e r t a i n t h a t treatment e f f e c t s d i d not vary from product to product, or from week to week. For t h i s purpose, analyses of variance were i n i t i a l l y performed to detect p o s s i b l e i n t e r a c t i o n e f f e c t s between treatments and products or treatments and weeks. Tables 25 and 26 summarize these 2-way analyses of variance.''' Neither the treatment-product nor the treatment-week i n t e r a c t i o n e f f e c t s 2 were s i g n i f i c a n t . Table 26 does, however, i n d i c a t e a s i g n i f i c a n t e f f e c t 3 on the mean r i d i t s due to d i f f e r e n t weeks. The l a t t e r part of t h i s chapter examines the s p e c i f i c e f f e c t s due to d i f f e r e n t weeks and d i f f e r e n t products. Tests For E f f e c t s Due to Number of Brands As noted i n Chapter IV, the number of brands i n a treatment sign d i f f e r e d across t e s t products. By pooling within-treatment observations iThe ANOVA program used i n the a n a l y s i s placed a l i m i t a t i o n on the number of c e l l s f o r a given sample s i z e . S p e c i f i c a l l y , the number of c e l l s must be l e s s than the sample s i z e . Therefore, a 3-way a n a l y s i s o f 8 treatments x 5 products x 2 weeks = 80 c e l l s would exceed t h a t l i m i t a t i o n with a sample s i z e of 77. To stay w i t h i n the l i m i t a t i o n , the 77 mean r i d i t s were subjected to separate 2-way analyses of i n t e r a c t i o n e f f e c t s . 2 In a l l a n a l y s i s of variance r e s u l t s reported i n t h i s chapter, i n t e r a c t i o n e f f e c t s are assessed a f t e r a d j u s t i n g f o r main e f f e c t s . 3 Unless s p e c i f i c a l l y noted i n t h i s chapter, a l l r e s u l t s are evaluated against a s i g n i f i c a n c e l e v e l of a = .05. Table 25 Two-Way A n a l y s i s of Variance to Test For Treatment-Product I n t e r a c t i o n Source o f V a r i a t i o n s s a d.f. MS a F P< Main e f f e c t s 3.277 11 .298 1.016 n.s. Treatments .847 7 .121 .413 n.s. Products 2.331 4 .583 1.988 .15 I n t e r a c t i o n s Treatments x Products 4.093 27 .152 .517 n.s. Residual 11.138 38 .293 Total 18.508 76 .244 xlOO ( i . e . decimal point moved two places to the r i g h t ) Table 26 Two-Way A n a l y s i s o f Variance to Test For Treatment-Week I n t e r a c t i o n Source of V a r i a t i o n SS a d.f. MSa F p< Main E f f e c t s 2.281 8 .285 1.214 n. Treatments .907 7 .130 .552 n. Weeks 1.334 1 1.334 5.682 .0 I n t e r a c t i o n s Treatments x Weeks 1. ,901 7 .272 Residual 14, .327 61 .235 Total 18, .508 76 .244 a xlOO 1 across products, one i s also pooling across s t i m u l i w i t h d i f f e r e n t numbers of brands. Although no i n t e r a c t i o n between treatment and product e f f e c t s was detected i n Table 25, a t e s t f o r treatment e f f e c t d i f f e r e n c e s due to v a r i a t i o n s i n the number of brands was deemed appropriate. With two 2-brand products, two 3-brand products and one 7-brand product, the number-of-brands v a r i a b l e took on three values. A 3-way a n a l y s i s of variance was performed on the two experimental f a c t o r s with number-of-brands included as a t h i r d f a c t o r . Table 27 i n d i c a t e s none of the main or i n t e r a c t i o n e f f e c t s to be s i g n i f i c a n t . In p a r t i c u l a r , no s t a t i s t i c a l l y s i g n i f i c a n t e f f e c t s are apparent as a r e s u l t of d i f f e r e n c e s i n the number of brands contained i n an information s i g n . Therefore, the same type of a n a l y s i s could a l s o be performed on the experimental f a c t o r s without regard to the number of brands i n a treatment s i g n . This a n a l y s i s i s presented next. ANOVA of Treatment E f f e c t s Load and Importance. The foregoing analyses d i d not reveal any i n t e r a c t i o n s between treatment e f f e c t s and e f f e c t s due to c e r t a i n e x t r a -neous f a c t o r s , across which the within-treatment r e p l i c a t e s could be pooled ( i . e . , products, weeks, no. of brands i n a s i g n ) . Consequently, i n order to focus on the e f f e c t s of the experimental v a r i a b l e s , the with treatment observations were pooled across products and weeks. This step provided 10 observations i n each c e l l o f the 4 x 2 f a c t o r i a l design, except f o r two c e l l s where the n were 8 and 9, f o r a t o t a l sample of 145 Table 27 3-Way A n a l y s i s of Variance to Test f o r E f f e c t s Due to Number o f Brands Source o f V a r i a t i o n SS a d.f. MS a F P Main E f f e c t s .980 6 .163 .597 n.s. Number of Brands .224 2 .112 .410 n.s. Load .609 3 .203 .741 n.s. Importance .242 1 .242 .885 n.s. 2-Way I n t e r a c t i o n s .955 11 .087 .317 n.s. No. o f Brands x Load .265 6 .044 .161 n.s. No. of Brands x Importance .508 2 .254 .928 n.s. Load x Importance .181 3 .060 .221 n.s. 3-Way I n t e r a c t i o n s No. of Brands x Load x Importance 2.070 6 .345 1.261 n.s. Residual 14.503 53 .274 Total 18.508 76 .244 a xlOO 1 4 6 77 mean r i d i t s . ' * A two-way a n a l y s i s o f variance was performed on the mean r i d i t s f o r four l e v e l s o f load and two l e v e l s of cue importance. Table 28 gives the mean r i d i t means f o r each c e l l o f the design and the summary r e s u l t s from ANOVA. Neither load e f f e c t s nor cue-importance e f f e c t s were s i g -n i f i c a n t . A l s o , no i n t e r a c t i o n between the two f a c t o r s i s apparent. These f i n d i n g s on load and cue importance are i l l u s t r a t e d i n Figure 8. Note that the mean r i d i t means f o r a l l e i g h t experimental c e l l s are greater than 0.5, i n d i c a t i n g a general tendency f o r the information signs to a f f e c t the brand sa l e s d i s t r i b u t i o n s i n the hypothesized d i r e c t i o n , r e -gardless of load or cue importance. Under HQ, one would expect h a l f of these means to be l e s s than 0.5 and h a l f greater than 0.5. A d i r e c t t e s t of the a l t e r n a t i v e hypothesis (H^) i s deferred to a l a t e r s e c t i o n . Figure 8 als o shows a c l e a r r e v e r s a l of the expectation that "high-importance" cues would g e n e r a l l y r e s u l t i n higher mean r i d i t s than "low-importance" cues, although d i f f e r e n c e s between the two l e v e l s of t h i s f a c t o r were not s t a t i s t i c a l l y s i g n i f i c a n t . This r e s u l t i s examined more c l o s e l y i n the t e s t s of hypothesis ri^. Conclusion. A n a l y s i s of variance o f load and cue-importance e f f e c t s i n the 4 x 2 experiment revealed no systematic between-treatment d i f f e r e n c e s i n the mean r i d i t . In f a c t , only 4% of the v a r i a t i o n i n the mean r i d i t i s accounted f o r by information load and cue importance ( m u l t i p l e R 2 = .041). Subprogram ANOVA i n the SPSS system assesses main and i n t e r a c t i o n e f f e c t s i n f a c t o r i a l designs with unequal c e l l frequencies a f t e r the sum o f squares due to a d d i t i v e e f f e c t s of the f a c t o r s , the sum of squares due to the i n t e r a c t i o n e f f e c t , and the sum of squares due to e r r o r are a l l made orthogonal to one another (see Nie et a l . , 1975:405-408). Table 28 Mean R i d i t C e l l Means and Two-Way A n a l y s i s of Variance f o r Load and Cue Importance C e l l Means Load Cue Importance 1 2 4 8 "High" .531 .502 .509 .508 "Low" .532 .526 .526 .508 A n a l y s i s o f Variance Source of V a r i a t i o n SS a d.f. MS a F P Main E f f e c t s .756 4 .189 .743 n.s Load .568 3 .189 .743 n.s Cue Importance .221 1 .221 .866 n.s I n t e r a c t i o n s Load x Importance .190 3 .063 .249 n.s Residual 17.562 69 .255 Total 18.508 76 .244 3 xlOO Figure 8 Load and Cue-Importance E f f e c t s on the Mean R i d i t 149 The means i n every c e l l were above 0.5 and, t h e r e f o r e , i n the hypothesized d i r e c t i o n . However, the a n a l y s i s thus f a r i n d i c a t e s that n e i t h e r i n -formation load nor the r e l a t i v e importance of the n u t r i t i o n a l cues present i n a sign e x p l a i n s these r e s u l t s . Tests of Hypotheses The s e c t i o n s which f o l l o w describe the a n a l y s i s and r e s u l t s per-t a i n i n g to each hypothesis, i n t u r n . The database of 77 mean r i d i t s whose frequency d i s t r i b u t i o n was shown i n Figure 7 i s examined i n terms of s p e c i -f i c r e l a t i o n s h i p s between the mean r i d i t and the independent v a r i a b l e s . The procedure followed i n the separate a n a l y s i s of each hypothesis i s as f o l l o w s . F i r s t , the hypothesis i s re s t a t e d and the meaning o f r e l a t i o n s h i p s to be i n v e s t i g a t e d i s b r i e f l y described. Next, the appropriate a n a l y t i c a l approach and r a t i o n a l e are discussed. Then, the actual t e s t r e s u l t s are presented, followed by a summary of the f i n d i n g s . Tests of Hypothesis H^ Hypothesis Hj s t a t e s The mean r i d i t reaches a maximum at some lower information l o a d , and diminishes at the highest information load. The r e l a t i o n s h i p between information load ( i . e . , the number of cues present i n an information sign) and the mean r i d i t expressed i n t h i s hypothesis i s a c u r v i l i n e a r one. That i s , market response to the n u t r i -t i o n a l information on a s i g n i s postulated to reach a maximum at some lower load and then become smaller at the highest load o f eigh t cues because consumers f i n d the maximum load to be detrimental to brand choice decision-making. 150 The t e s t a b l e i m p l i c a t i o n of t h i s hypothesis i s that once a global maximum i s lo c a t e d on the information-load response curve, the mean r i d i t observed at the highest load of 8 cues should be lower than the value of the mean r i d i t at global maximum. A n a l y t i c a l Approach. The approach taken here was to p l o t the information load response curve and i d e n t i f y the load at which a global maximum i n the dependent measure was reached. Once t h i s was determined, the corresponding sample of mean r i d i t s at t h i s load l e v e l was compared with the sample o f mean r i d i t s at the highest load of 8 cues. This i n d i c a t e d a t - t e s t f o r the s i g n i f i c a n c e o f the d i f f e r e n c e between means of two independent samples to determine whether the means at the two load l e v e l s d i f f e r e d , i n p a r t i c u l a r , whether the mean at a load o f 8 cues was s i g n i f i c a n t l y lower than the maximum mean. A second approach was to perform a one-way a n a l y s i s o f variance on the four l e v e l s of load and to apply a trend t e s t to determine whether the quadratic component of the load response curve was s i g n i f i c a n t . This would i n d i c a t e whether the response curve a c t u a l l y was c u r v i l i n e a r (inverted-U-shaped) . Since the a n a l y s i s of variance i n the preceding s e c t i o n revealed no e f f e c t s due to cue importance and no i n t e r a c t i o n e f f e c t between load and cue importance, the within-treatment observations at each load l e v e l were pooled across the cue-importance f a c t o r . t-Test R e s u l t s . The upper part of Table 29 gives the means of the mean r i d i t s a t each l e v e l of load . These are p l o t t e d i n Figure 9 which reveals t h a t the mean r i d i t reached a maximum at a load of 1 cue. There-f o r e , the appropriate comparison was 151 H0: ^ i = ^8 ; tiy vi > y 8 . A o n e - t a i l e d t - t e s t revealed that the d i f f e r e n c e between the two means (.531 vs. .508) i s not s t a t i s t i c a l l y s i g n i f i c a n t ( t = 1.52; df = 36; p < .10, o n e - t a i l e d ) . Thus, no adequate grounds e x i s t f o r r e j e c t i n g the n u l l hypothesis. Test f o r Trends. The second approach to t e s t i n g was to perform a one-way a n a l y s i s of variance between load treatments and to apply a t e s t f o r polynomial t r e n d . The set of hypotheses being t e s t e d here can be represented as f o l l o w s : H Q: ]i\ = V2 = Pit = Us '•> Table 29 presents the summary r e s u l t s from t h i s a n a l y s i s . Treatment means were not s i g n i f i c a n t l y d i f f e r e n t from one another. The t e s t s f o r polynomial trends appear w i t h i n the a n a l y s i s of variance t a b l e . Neither the l i n e a r component nor the quadratic component o f the between-treatments sum of squares i s s i g n i f i c a n t . Thus, no grounds e x i s t f o r b e l i e v i n g t h a t the mean r i d i t i s an inverted-U f u n c t i o n of informa t i o n load i n t h i s experiment. Summary and Conclusions. The i m p l i c a t i o n of hypothesis t h a t the mean r i d i t would peak at some lower load and then d e c l i n e at the maximum experimental load of 8 cues, i s not borne out by the s t a t i s t i c a l r e s u l t s . 152 Table 29 Mean R i d i t C e l l Means and One-way A n a l y s i s of Variance o f Load E f f e c t s and Tests f o r Trends C e l l Mean f o r Load o f 1 2 4 8 .531 .515 .517 .508 A n a l y s i s o f Variance Source o f V a r i a t i o n SS a d.f. MSd F P Between Treatments .536 3 .179 .725 n.s. L i n e a r Term .351 1 .351 1.424 n.s. Quadratic Term .036 1 .036 .148 n.s. Within Treatments 17.973 73 .246 Total 18.508 76 a xlOO Figure 9 Mean R i d i t s at Four Information-Load Levels .55 -CD .45 -i 1 T 7 ' IT 1 2 4 8 Load 154 Moreover, one cannot draw the inference from the a n a l y s i s of variance r e s u l t s that the load v a r i a b l e was e x e r t i n g a d i f f e r e n t i a l e f f e c t on the mean r i d i t . The i m p l i c a t i o n of these f i n d i n g s , which are summarized i n Figure 9, i s t hat although n u t r i t i o n a l information a f f e c t e d the brand sales d i s -t r i b u t i o n s of the t e s t products i n the hypothesized d i r e c t i o n (a rigorous t e s t o f t h i s statement i s made under H^), the trend e x h i b i t e d i n the response curve i s not r e l i a b l e . A l s o , because the mean r i d i t at the highest load i s not s i g n i f i c a n t -l y lower than the glob a l maximum, there i s no evidence t h a t shopper response to an information load of eigh t n u t r i t i o n a l cues i s any smaller than the response to lower information loads. Hypothesis H^ i s r e j e c t e d on the basis of these r e s u l t s . There i s no evidence of "information overload" i n t h i s experiment. Tests of Hypothesis Hp Hypothesis H^ s t a t e s The mean r i d i t at an information load c o n s i s t i n g of "high-importance" n u t r i t i o n a l cues i s greater than the mean r i d i t at the same load using "low-importance" n u t r i t i o n a l cues. This hypothesis i m p l i e s that behavioural responses to d i f f e r e n t loads of n u t r i t i o n a l information on a product w i l l a l s o depend upon the importances assigned to the set of n u t r i e n t s used i n a treatment. R e l a t i v e importances among a set of eigh t n u t r i e n t s were i n f e r r e d i n t h i s research on the basis of responses given by consumers i n a survey. Given these aggregate derived measures of r e l a t i v e importance among a set of e i g h t n u t r i e n t s , any of the four most important n u t r i e n t s was 155 designated as a "high-importance" cue. The t e s t a b l e i m p l i c a t i o n o f t h i s hypothesis i s t h a t , with the load f a c t o r held constant, the mean r i d i t f o r a treatment c o n t a i n i n g "high-importance" cues should be s i g n i f i c a n t l y higher than the mean r i d i t f o r a treatment with "low-importance" cues. A n a l y t i c a l Approach. The p r e l i m i n a r y a n a l y s i s o f the f a c t o r i a l design to detect between-treatment d i f f e r e n c e s already i n d i c a t e d t h a t cue-importance main e f f e c t s , adjusted f o r load e f f e c t s , were n o n s i g n i f i c a n t (Table 28). Hypothesis H£ 5 however, i s l i m i t e d to comparisons between t r e a t -ments c o n t a i n i n g e i t h e r "high-importance" cues o r "low-importance" cues, but not both, as i s the case i n Treatments "8/high" and "8/low." There-f o r e , a d d i t i o n a l analyses were undertaken to t e s t the s i g n i f i c a n c e of the d i f f e r e n c e between means of treatments matched on the load f a c t o r and cont a i n i n g e i t h e r "high-" or "low-importance" cues ( i . e . , treatments with 1, 2 or 4 cues). In a d d i t i o n to t e s t s on means at s p e c i f i c l o a d s , the means of aggregated samples representing "high-" and "low" cue-importances were teste d f o r s i g n i f i c a n t d i f f e r e n c e s . S p e c i f i c a l l y , importance treatments were pooled across s u c c e s s i v e l y l a r g e r load combinations and d i f f e r e n c e s between "high" and "low" cue-importance means were t e s t e d f o r s i g n i f i c a n c e . The purpose o f t h i s approach was to make comparisons between importance treatments with p r o g r e s s i v e l y l a r g e r sample s i z e s . A l l comparisons employed the t - t e s t f o r the s i g n i f i c a n c e o f the d i f f e r e n c e between means of two independent samples. A f i n a l approach to the t e s t o f hypothesis H ? examined cue-156: importance e f f e c t s j o i n t l y with product e f f e c t s . The purpose was to a s c e r t a i n that any cue-importance e f f e c t s were not confounded by d i f f e r e n c e s across products. For t h i s , a two-way a n a l y s i s of variance was performed. t-Test R e s u l t s . Each t e s t on the d i f f e r e n c e between means of a p a i r of samples was an e v a l u a t i o n of the hypothesis set 0 : u h i g h " v l o w ' 2' u h i g h > y l o w ' The r e s u l t s of t - t e s t s on the means of various p a i r s of samples are presented i n Table 30. The d i f f e r e n c e between means of treatments using "high-importance" cues and treatments using "low-importance" cues ( i . e . , at the load l e v e l s 1, 2 and 4) was i n no case s t a t i s t i c a l l y s i g n i f i c a n t , i n d i c a t i n g that should be r e j e c t e d . Tests on the means of aggregated samples a l s o revealed no s i g -n i f i c a n t d i f f e r e n c e s between treatments composed of "high-" and "low-importance" cues. For example, the means d i d not d i f f e r f o r the pooled data from a l l treatments c o n t a i n i n g e i t h e r one or two "high-importance" cues (1,2/high) and a l l treatments c o n t a i n i n g e i t h e r one or two "low-importance" cues (1,2/low). F i n a l l y , f o r a pooled comparison at maximum sample s i z e , the d i f f e r e n c e between means of "high-importance" and "low-importance" t r e a t -ments, pooled across a l l four loads ( 1 , 2, 4 and 8), was t e s t e d f o r s i g n i f i c a n c e . As shown i n t a b l e 30, the d i f f e r e n c e was not r e l i a b l e . 157 Table 30 Results of t-Tests f o r D i f f e r e n c e Between Means of Various P a i r s o f "High-Importance"/"Low-Importance" Samples 1-Tail Comparison P a i r s Means t df p 1/high .531 _ Q 5 . 1 6 1/1ow .532 n.s. 2/high .502 . , ? 2/low .526 ~ 1 - 3 2 1 7 < A S 4/high .509 c n 1 Q 4/low .526 ~ - 6 0 1 8 n' s-Pooled 1,2/high .517 R 1 1,2/low .529 ' " a i J b n - s > 1,2,4/high .514 . m 1,2,4/low .528 i , U i b b n' s* 1,2,4,8/high .513 P f t 7J-1,2,4,8/low .523 ' * a B / b n' s-158 ANOVA R e s u l t s . To check t h a t the above f i n d i n g s on cue importance were not confounded by any e f f e c t s due to d i f f e r e n t products i n the data-base, a two-way a n a l y s i s o f variance was performed on the mean r i d i t . The two f a c t o r s analyzed were 5 products x 2 cue-importance l e v e l s , pooled across loads o f 1, 2 and 4, only. Table 31 summarizes t h i s a n a l y s i s of variance and reveals t h a t cue-importance e f f e c t s are not s i g n i f i c a n t . No i n t e r a c t i o n between cue importance and products i s apparent. The same type of a n a l y s i s was repeated with the "cue-importance" treatments pooled across a l l loads, i n order to increase the c e l l sample s i z e s . The r e s u l t s of t h i s a n a l y s i s o f variance are summarized i n Table 32. As before, n e i t h e r cue-importance e f f e c t s nor i n t e r a c t i o n e f f e c t s are s t a t i s t i c a l l y s i g n i f i c a n t . However, a s i g n i f i c a n t e f f e c t due to d i f f e r e n t products i s i n d i c a t e d i n t h i s t a b l e . Summary and Conclusions. The i m p l i c a t i o n of these f i n d i n g s from t e s t s of hypothesis H2 i s th a t v a r i a t i o n s i n the mean r i d i t are not explained by the cue-importance f a c t o r . In g e n e r a l , the mean r i d i t f o r an information stimulus constructed with subsets of the four most important n u t r i e n t s , among the eig h t s e l e c t e d f o r a product, was no grea t e r than t h a t observed f o r a stimulus created w i t h subsets of the four l e a s t important n u t r i e n t s . Even i n the most extreme case, where the mean f o r treatments using the s i n g l e most important n u t r i e n t i s compared against the mean f o r treatments composed of the s i n g l e l e a s t important n u t r i e n t , the means are v i r t u a l l y i d e n t i c a l (I "1/high" = .531; X "1/low" = .532). I f anything, the magnitudes of the means f o r every comparison p a i r Table 31 Two-Way A n a l y s i s of Variance o f Cue-Importance and Product E f f e c t s (Pooled Across Loads of 1, 2 and 4) Source of V a r i a t i o n s s a d.f. MSa F P< Main E f f e c t s 2.039 5 .408 1.717 .15 Cue Importance .203 1 .203 .853 n.s. Products 1.790 4 .448 1.884 .15 I n t e r a c t i o n s Importance x Products .451 4 .113 .474 n.s. Residual 11.167 47 .238 Total 13.657 56 .244 a xlOO Table 32 Two-Way A n a l y s i s o f Variance o f Cue-Importance and Product E f f e c t s (Pooled Across a l l Loads) Source of V a r i a t i o n SS a d.f. MS a F P< Main E f f e c t s 2.584 5 .517 2 .313 .10 Cue Importance .153 1 .153 .686 n.s Products 2.395 4 .599 2 .680 .05 I n t e r a c t i o n s Importance x Products .953 4 .238 1 .066 n.s Residual 14.971 67 .223 Total 18.508 76 .244 a xlOO 161 i n Table 30 are i n reverse of those hypothesized, although none of the d i f f e r e n c e s i s s t a t i s t i c a l l y s i g n i f i c a n t . This suggests t h a t the v a r i a t i o n s i n the mean r i d i t are due to some f a c t o r ( s ) other than the r e l a t i v e importance of i n d i v i d u a l n u t r i e n t s included i n a s i g n , as measured i n t h i s research. I t was p r e v i o u s l y pointed out i n Chapters I I I and IV th a t e a r l i e r research on food product cue importance had found f a i r l y small d i f f e r e n c e s i n r e l a t i v e importance to consumers among i n d i v i d u a l n u t r i e n t s . As noted i n the d i s c u s s i o n on research methodology, the measures of r e l a t i v e importance derived from the consumer survey may a c t u a l l y represent measures of s a l i e n c e . I f t h a t i s the case, then d i f f e r e n c e s i n r e l a t i v e importance to consumers of i n d i v i d u a l n u t r i e n t s are probably exaggerated by a measure which does not take i n t o account the r e l a t i v e importance o f other cues l i k e p r i c e , brand name and t a s t e , i n a d d i t i o n to n u t r i t i o n a l cues. Hypothesis H 2 i s re j e c t e d on the basis o f these r e s u l t s . Purchase responses to n u t r i t i o n a l information on a product are no greater f o r signs l i s t i n g "high-importance" n u t r i e n t s than they are f o r signs l i s t i n g "low-importance" n u t r i e n t s . Tests o f Hypothesis H., Hypothesis Hg s t a t e s A point-of-purchase sig n l i s t i n g e i g h t cues i n decreasing order of importance, from l e f t to r i g h t , y i e l d s a greater mean r i d i t than a sign l i s t i n g the same e i g h t cues i n reverse order. This hypothesis postulates t h a t a brand-by-cue information matrix i s more e f f e c t i v e i f the product information cues are arranged, from l e f t to r i g h t , i n " h i e r a r c h i c a l order" (see Best, 1978:4-5), i . e . , from most 162 to l e a s t important, s i n c e people c h a r a c t e r i s t i c a l l y read from l e f t to r i g h t . A matrix whose cues are arrayed i n reverse " h i e r a r c h i c a l order," with the l e a s t important cue l i s t e d f i r s t and the most important cue l i s t e d l a s t , w i l l e l i c i t a smaller aggregate response from shoppers. Two of the information treatments i n the experimental design were comprised of the i d e n t i c a l set of e i g h t cues and d i f f e r e d i n only one respect. In Treatment "8/high" the cues were arrayed from l e f t to r i g h t i n descending order of importance, while i n Treatment "8/low" the same cues were arrayed from l e f t to r i g h t i n ascending order of importance. The t e s t a b l e i m p l i c a t i o n of t h i s hypothesis i s th a t the mean r i d i t f o r Treatment "8/high" should be s i g n i f i c a n t l y higher than the mean r i d i t f o r Treatment "8/low." A n a l y t i c a l Approach. In order to t e s t t h i s hypothesis, product-s p e c i f i c comparisons must be made between Treatment "8/high" and "8/low." This i s because the hypothesis deals with d i f f e r e n c e s i n the mean r i d i t a r i s i n g from the l e f t - t o - r i g h t arrangement of the same set of e i g h t cues. Since the same set of cues ( l e t alone the same brand r a t i n g s ) was not used across a l l products, the comparison of Treatment "8/high" and "8/low" e f f e c t s i s meaningful only w i t h i n products. A c c o r d i n g l y , the a n a l y t i c a l procedure was to t e s t the s i g n i f i c a n c e of the d i f f e r e n c e between the mean r i d i t s f o r two brand s a l e s d i s t r i b u t i o n s . One treatment's data served as the reference d i s t r i b u t i o n (whose mean r i d i t i s a u t o m a t i c a l l y 0.5), against which the d i s t r i b u t i o n of the second treatment could be compared. Treatment "8/low" was s e l e c t e d as the reference and the brand r i d i t s were computed from t h i s treatment's data 163 f o r each product i n each experimental week. The mean r i d i t f o r the corresponding Treatment "8/high" was then d e r i v e d . A z - t e s t , using formulas (4.4) and (4.5), was a p p l i e d to determine whether the mean r i d i t f o r the Treatment "8/high" was s i g n i f i c a n t l y greater than that f o r the corresponding "8/low" mean r i d i t of 0.5. These procedures were t e s t s o f the f o l l o w i n g hypothesis s e t : H 0 : y 8 / h i g h = y8/low ;  H 3 : u 8 / h i g h > y8/low ' S i g n i f i c a n c e Test R e s u l t s . Table 33 presents the r e s u l t s of the z- t e s t s on the d i f f e r e n c e between each p a i r of mean r i d i t s . Ten comparisons between Treatment "8/high" and "8/low" e f f e c t s were made f o r the f i v e products i n two d i f f e r e n t weeks. In no case i s the d i f f e r e n c e between Treatment "8/high" and "8/low" mean r i d i t s s t a t i s t i c a l l y s i g n i f i c a n t . Thus no basis e x i s t s f o r r e j e c t i n g the n u l l hypothesis. Summary and Conclusion. The r e s u l t of t h i s t e s t i n d i c a t e s that the n u l l hypothesis cannot be r e j e c t e d . Whether the same set of eigh t n u t r i t i o n a l cues on a product i s arrayed i n " h i e r a r c h i c a l order" or i n reverse " h i e r a r c h i c a l order" the mean r i d i t remains the same. Given t h i s r e s u l t , hypothesis l-L i s r e j e c t e d . A brand-by-cue In t h i s i n s t a n c e , i t i s of no consequence which treatment serves as the "reference" f o r the other treatment. The s e l e c t i o n o f "8/low" as the reference i s c o n s i s t e n t with the expectation t h a t the mean r i d i t f o r "8/high" would be greater than t h a t f o r "8/low" which becomes, by d e f i n i t i o n , 0.5. 164 Table 33 Results of z-Tests on Mean R i d i t D i f f e r e n c e s Between Treatments "8/high" and "8/low" f o r Five Products i n Two Weeks Mean z One-Product R i d i t Value T a i l e d p Canned Soup Week 2 8/low .500 8/high .492 Week 3 Mayonnaise Week 2 8/low .500 8/high .565 Week 3 8/low .500 8/high . .624 ,160 n.s, 8/low .500 _ 7 8 5 8/high .477 ' 1.134 <.15 1.494 <.10 ,027 n.s. Ketchup Week 2 8/low .500 ._ 4 1 5 8/high .482 3 , S' Week 3 8/low .500 8/high .498 M&C Dinner Week 2 8/low .500 8/high .464 Week 3 8/low .500 8/high .508 Bran Cereal Week,2 8/low .500 8/high .467 Week 3 8/low .500 8/high .431 •1.023 n.s. .191 n.s, .605 n.s. ,957 n.s, 165 information matrix does not r e s u l t i n a greater s h i f t towards n u t r i t i v e l y higher performance brands when the product cues are arranged i n a " h i e r a r c h i c a l order," as opposed to a reverse " h i e r a r c h i c a l order." Tests of Hypothesis Hypothesis H^ st a t e s The mean r i d i t f o r a product's brand s a l e s d i s t r i b u t i o n i s greater than 0 . 5 when n u t r i t i o n a l information i s placed i n a brand-by-cue matrix format at the point o f purchase. The premise i s t h a t customers w i l l make t h e i r brand choices according to n u t r i t i o n a l information on a product i f t h a t information i s posted i n a brand-by-cue matrix. This i s because such a presentation format f a c i l i t a t e s d i r e c t brand comparisons, thereby encouraging shoppers t o process the information made a v a i l a b l e . This hypothesis p r e d i c t s the general e f f e c t of n u t r i t i o n a l information on a product's brand s a l e s . The r e l a t i o n s h i p examined here i s one which spans a l l treatments, regardless of load or cue-importance. The t e s t a b l e i m p l i c a t i o n o f t h i s hypothesis i s that the mean r i d i t f o r any treatment w i t h a p a r t i c u l a r t e s t product i s s i g n i f i c a n t l y greater than 0 . 5 , s i n c e a l l treatments d i s c l o s e d n u t r i t i o n a l information on a product i n the same format. A n a l y t i c a l Approach. The a n a l y s i s begins by examining the s t a t i s t i c a l s i g n i f i c a n c e of each o f the 7 7 mean r i d i t s i n the database. Using a s p e c i a l l y w r i t t e n program to compute the standard e r r o r of each mean r i d i t (equation 4 . 4 ) , the z - t e s t was a p p l i e d to determine whether the mean r i d i t was s i g n i f i c a n t l y greater than i t s expected value, under HQ, of 0 . 5 . The r e s u l t s of these t e s t s are g r a p h i c a l l y summarized i n Figures 1 0 166 and 11. The 77 mean r i d i t s have been p l o t t e d according to t h e i r corresponding load and cue-importance treatments, by product and by week. Mean r i d i t s which are s i g n i f i c a n t l y greater than 0.5, at the 5% l e v e l , are c i r c l e d i n the f i g u r e s . The f i g u r e s a l s o i n d i c a t e the mean r i d i t s which were found to be s i g n i f i c a n t l y l e s s than 0.5 (at a = .05). Although a o n e - d i r e c t i o n a l hypothesis i s being t e s t e d , such t h a t these mean r i d i t s represent r e v e r s a l s , they are included to show that only three o f 77 mean r i d i t s were found to be s i g n i f i c a n t l y smaller than 0.5. In comparison, 16 mean r i d i t s were s i g n i f i c a n t i n the hypothesized d i r e c t i o n . I t i s obvious from Figures 10 and 11 th a t not every mean r i d i t i s s i g n i f i c a n t l y g r e a t e r than 0.5, as pr e d i c t e d by hypothesis H^. However, i t would-be premature to r e j e c t H^ on the basis of these f i n d i n g s , alone. The question can be reformulated as f o l l o w s . Is the tendency f o r the mean r i d i t to be greater than 0.5, whenever some n u t r i t i o n a l information was provided, a r e l i a b l e one? By approaching the a n a l y s i s i n t h i s general way, one can t e s t whether these mixed mean r i d i t s i g n i f i c a n c e r e s u l t s d i s p l a y a r e l i a b l e tendency i n the d i r e c t i o n hypothesized by H^. The approach taken here was to apply an overal1 s i g n i f i c a n c e t e s t on a combination o f p values obtained from the i n d i v i d u a l mean r i d i t s i g n i f i c a n c e t e s t s . The a n a l y t i c a l t o o l was the chi-square model f o r t e s t i n g the s i g n i f i c a n c e of combined r e s u l t s . Because preceding analyses revealed t h a t there were s i g n i f i c a n t e f f e c t s on the mean r i d i t due s p e c i f i c a l l y to d i f f e r e n t products and d i f f e r e n t experimental weeks, t h i s hypothesis was t e s t e d s e p a r a t e l y on Figure 10 Mean R i d i t s f o r Load and Cue-Importance Treatments With Five Products i n Week 2 Canned Soup Mayonnaise Ketchup M&C Dinner Bran Cereal T 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r 1 2 4 8 1 2 4 8 1 2 4 8 1 2 4 8 1 2 4 8 Load Treatment ©Denotes a mean r i d i t s i g n i f i c a n t l y > 0.5 (or < 0.5) at a = .05 ©Denotes two overlapping mean r i d i t s , both s i g n i f i c a n t l y > 0.5 at a = .05 Figure 11 Mean R i d i t s f o r Load and Cue-Importance Treatments With Five Products i n Week 3 "High" Cue-Importance Treatments 9 "Low" Cue-Importance Treatments 9 Canned Soup Mayonnaise \ Ketchup ® M f I \ I \ I \ I \ I \ I » I \ M&C Dinner Bran Cereal i i i i i I i i i i 1 i i i i i i i i r 1 2 4 8 1 2 4 8 1 2 4 8 1 2 4 8 1 2 4 8 Load Treatment ® Denotes a mean r i d i t s i g n i f i c a n t l y > 0.5 (or < 0.5) at a = .05 169 each product and again on each of the two weeks. A c c o r d i n g l y , the overal 1 s i g n i f i c a n c e t e s t s were performed on i n d i v i d u a l p val u e s , combined by product and, l a t e r , by week. Results from these t e s t s formed the basis f o r accepting o r r e j e c t i n g with respect to a product or week. I n d i v i d u a l and Combined S i g n i f i c a n c e R e s u l t s . Tables 34 to 38 present the product-by-product r e s u l t s of i n d i v i d u a l z - t e s t s to determine whether a mean r i d i t i s s i g n i f i c a n t l y g reater than 0.5. For each i n d i v i d u a l treatment, the hypothesis set being evaluated i s The computed z value i s given opposite the r e s p e c t i v e mean r i d i t along with i t s o n e - t a i l e d p r o b a b i l i t y . Note that f o r mean r i d i t s l e s s than 0.5, the corresponding p r o b a b i l i t y of the obtained -z i s entered as 1-p, since the t e s t s of are o n e - d i r e c t i o n a l . Tables 34 to 38 al s o summarize the product-by-product t e s t s of H^. These t e s t s use the chi-square model f o r t e s t i n g the s i g n i f i c a n c e o f the combined treatment p values w i t h i n a t a b l e . In t h i s case, one i s t e s t i n g the hypothesis that the composite p value, c a l c u l a t e d from the several treatment p values f o r a product, could have occurred by chance (Jones and F i s k e , 1953:376). This procedure gives the p r o b a b i l i t y , under H n, th a t H y = 0.5 H vi > 0.5 k X 2 = -2 I l o g e p. > chi square (5.1) Table 34 Mean R i d i t S i g n i f i c a n c e Results f o r 15 Treatments With Canned Soup And Results o f Over a l l S i g n i f i c a n c e Test of Hypothesis H. Mean z One- k Treatment R i d i t Value T a i l e d p l o g g p I log -p^ k Week 2 1/high .599 2.275 .011 -4.470 2/high - - - -4/high .562 1.599 .055 -2.902 8/high .551 1.637 .051 -2.980 1/low .527 .604 .273 -1.299 2/low .510 .442 .329 -1.111 4/1 ow .527 1.158 .123 -2.092 8/low .559 1.458 .072 -2.625 Jeek 3 1/high .555 2.224 .013 -4.337 2/high .533 1.060 .145 -1.934 4/high .485 - .409 .659 - .417 8/high .498 - .074 .529 - .636 1/low .523 .794 .214 -1.544 2/low .549 1.782 .037 -3.287 4/low .555 2.564 .005 -5.264 8/low .521 .932 .176 -1.739 -17.478 7 -19.158 8 -36.636 15 15 X 2 = -2 I log p. = 73.272; df = 30; p < .00005 i = l e 1 Table 35 Mean R i d i t S i g n i f i c a n c e Results f o r 14 Treatments With Mayonnaise And Results of Over a l l S i g n i f i c a n c e Test of Hypothesis H„ Mean z One- k Treatment R i d i t Value T a i l e d p l o g g p L l o 9 e Pi k Week 2 1/high .523 .423 .336 -1.090 2/high .451 - .927 .823 - .195 4/high .575 1.343 .090 -2.412 8/high .552 1.148 .125 -2.076 1/low - - - -2/low .511 .201 .420 - .867 4/1 ow .536 .663 .254 -1.372 8/low .488 - .235 .593 - .523 Jeek 3 1/high .527 .450 .326 -1.120 2/high .562 1.333 .091 -2.394 4/high .412 -1.307 .904 - .100 8/high .544 1.068 .143 -1.947 1/low - - - -2/low .505 .083 .467 - .762 4/low .504 .097 .461 - .774 8/low .424 - .959 .831 - .185 - 8.534 7 - 7.281 7 -15.815 14 14 X2 = -2 I l o g o p. = 31.63; df = 28; p < .30 i = l e 1 Table 3,6 Mean R i d i t S i g n i f i c a n c e Results f o r 16 Treatments With Ketchup And Results of Over a l l S i g n i f i c a n c e Test of Hypothesis H. Mean z One- k Treatment R i d i t Value T a i l e d p l o g e p l/}°9e Pi ^ Week 2 1/high .445 -1.809 .965 - .036 2/high .505 .142 .444 - .813 4/high .523 .555 .289 -1.240 8/high .446 -1.739 .959 - .042 1/low .472 - .709 .761 - .273 2/low .519 .642 .260 -1.345 4/low .518 .530 .298 -1.210 8/low .464 - .964 .832 - .183 Jeek 3 1/high .505 .127 .449 - .800 2/high .470 - .958 .831 - .185 4/high .489 - .386 .650 - .430 8/high .515 .327 .372 - .989 1/low .536 .865 .194 -1.642 2/low .505 .146 .442 - .817 4/low .418 -1.662 .952 - .049 8/low .517 .342 .366 -1.005 - 5.143 8 - 5.918 8 -11.061 16 16 X 2 =-2 7 log p. = 22.122; df = 32; p < .95 ,• _ i e l 173 Table 37 Mean R i d i t S i g n i f i c a n c e Results f o r 16 Treatments With M&C Dinner And Results of Ov e r a l l S i g n i f i c a n c e Test o f Hypothesis H^ Mean z One- k Treatment R i d i t Value T a i l e d p l o g e p I l o g e p.. Week 2 1/high 2/high 4/high 8/high 1/low 2/low 4/low 8/low -42.498 Week 3 .569 2.434 .007 - 4.897 .545 1.753 .040 - 3.224 .543 1.746 .040 - 3.209 .563 1.872 .031 - 3.487 .591 4.091 .000 -10.748 .545 1.686 .046 - 3.081 .527 1.182 .119 - 2.132 .596 4.311 .000 -11.720 1/high .454 -1.349 .911 - .093 2/high .523 .590 .278 -1.282 4/high .531 1.145 .126 -2.071 8/high .483 - .378 .647 - .435 1/low .567 1.978 .024 -3.731 2/low .480 - .461 .678 - .389 4/low .588 2.488 .006 -5.048 8/low .476 - .812 .792 - .234 16 X 2 = -2 I l o g p. = 111.56; df = 32; p < i = l e 1 •13.282 8 -55.780 16 •,-10 174 Table 3 8 Mean R i d i t S i g n i f i c a n c e Results f o r 16 Treatments With Bran Cereal And Results of Ov e r a l l S i g n i f i c a n c e Test of Hypothesis H„ Treatment Mean R i d i t z Value One-T a i l e d p l o g e P k I l o g e Pn- k Week 2 1/high .592 1.860 .031 -3.460 2/high .452 - .870 .808 - .213 4/high .511 .211 .416 - .876 8/high .517 .447 .327 -1.116 1/low .532 .578 .282 -1.267 2/low .519 .461 .322 -1.132 4/low .664 3.582 .000 -8.677 8/low .554 1.152 .125 -2.082 -18.823 8 Week 3 1/high .540 .883 .189 -1.668 2/high .479 - .574 .717 - .333 4/high .457 - .963 .832 - .184 8/high .408 -1.500 .933 -.069 1/low .507 .154 .439 - .824 2/low .617 2.493 .006 -5.062 4/low .473 - .657 .744 - .295 8/low .487 - .316 .624 - .472 - 8.906 8 -27.729 16 Week 2 a 4/low .690 4.133 .000 8/high .498 - .058 .523 8/low .548 1.032 .151 Week 3 a 4/low .454 -1.149 .875 8/high .430 -1.149 .875 8/low .474 - .647 .741 16 X 2 = -2 V l o g p p. = 55.458; df = 32; p < .01 i = l e 1 aMean r i d i t s based on weighted rank sum method of determining each brand's o v e r a l l n u t r i t i v e performance rank (see footnote 11, Chapter I V ). 175 with 2k degrees o f freedom, where the p. are the o n e - t a i l e d z - t e s t p r o b a b i l i t i e s being combined across k treatments with a product. The re s p e c t i v e l o g g p f o r each of k treatments are entered i n these t a b l e s . D e t a i l s o f these product-by-product analyses are presented next. Canned Soup. Table 34 shows the mean r i d i t s i g n i f i c a n c e r e s u l t s f o r every one of the 15 treatments with canned soup (the data on Treatment "2/high" i n Week 2 were deleted because of a c r i t i c a l stock outage during t h i s treatment period i n both s t o r e s ) . Four of the 15 mean r i d i t s are s i g n i f i c a n t l y g reater than 0.5 at a = .05. Only two of the 15 mean r i d i t s are l e s s than 0.5 and represent r e v e r s a l s i n the r e l a t i o n s h i p p r e d i c t e d under H^. The question i s , "Is the observed general trend f o r the mean r i d i t to be greater than 0.5 a dependable one?" Applying the t e s t f o r the s i g n i f i c a n c e of combined r e s u l t s a h i g h l y s i g n i f i c a n t f i n d i n g i s obtained ( x 2 = 73.272; df = 30; p < .00005). The j o i n t p r o b a b i l i t y o f o b t a i n i n g t h i s set of independent p r o b a b i l i t i e s from i n d i v i d u a l s i g n i f i c a n c e t e s t s on 15 treatments, when i n f a c t the n u t r i t i o n a l information signs are having no e f f e c t on the mean r i d i t , i s very s m a l l . Hypothesis H^ cannot be r e j e c t e d f o r t h i s product. The tendency f o r the mean r i d i t to be above 0.5 whenever n u t r i t i o n a l information was provided on canned soup i s r e l i a b l e . Mayonnaise. Table 35 gives the r e s u l t s of s i g n i f i c a n c e t e s t s on the mean r i d i t s f o r 14 n u t r i t i o n a l information treatments with the product mayonnaise. None of the mean r i d i t s i s s i g n i f i c a n t l y greater than The data on Treatment "1/low" are el i m i n a t e d i n both weeks because missing r a t i n g s on two of the three brands of t h i s product rendered t h i s p a r t i c u l a r information treatment meaningless (see the sign r e p l i c a i n Appendix A). 176 (or l e s s than) 0.5 at a = .05. The t e s t f o r the s i g n i f i c a n c e of combined r e s u l t s was performed to determine whether there was a r e l i a b l e tendency f o r the mean r i d i t to be great e r than 0.5. The obtained x2> i n t h i s case, i s not s t a t i s t i c a l l y s i g n i f i c a n t (x2 = 31.63; df = 28; p < .30). Hypothesis H^ i s r e j e c t e d f o r t h i s product. There i s no r e l i a b l e tendency f o r the mean r i d i t to be above 0.5 whenever some n u t r i t i o n a l information on mayonnaise was placed at the point of purchase. Ketchup. I n d i v i d u a l s i g n i f i c a n c e t e s t r e s u l t s on the 16 treatments with the product ketchup are summarized i n Table 36. Seven of the 16 mean r i d i t s are l e s s than 0.5, about what might be expected as a r e s u l t of random v a r i a t i o n around the value o f 0.5. Three o f these mean r i d i t s turned out to be s i g n i f i c a n t l y l e s s than 0.5 and represent r e v e r s a l s i n the expected r e l a t i o n s h i p under H^. Summarizing the 16 mean r i d i t f i n d i n g s with t h i s product, a t e s t f o r the s i g n i f i c a n c e of combined r e s u l t s revealed no tendency f o r the mean r i d i t to be l a r g e r than 0.5 (x2 = 22.122; df = 32; n.s.) Hypothesis H^ i s r e j e c t e d f o r t h i s product. There does not appear to be any impact of n u t r i t i o n a l information on the brand s a l e s d i s t r i b u t i o n of ketchup. Macaroni & Cheese Dinner. Table 37 summarizes the s i g n i f i c a n c e t e s t r e s u l t s on the 16 mean r i d i t s obtained f o r n u t r i t i o n a l information treatments w i t h m&c dinner. Nine of the 16 mean r i d i t s are s i g n i f i c a n t l y greater than 0.5 at a = .05. Four o f 16 mean r i d i t s are below 0.5. The t e s t f o r the s i g n i f i c a n c e o f the combined p values gives a h i g h l y s i g n i f i c a n t r e s u l t 177 ( x 2 = 111.56;•df = 32; p < 5 x l 0 " 1 0 ) . Therefore, the tendency f o r the mean r i d i t to be greater than 0.5 whenever n u t r i t i o n a l i nformation was a v a i l a b l e on t h i s product i s dependable. Hypothesis H^ cannot be r e j e c t e d f o r t h i s product. The mean r i d i t f o r the brand s a l e s d i s t r i b u t i o n of m&c dinner tended to be greater than 0.5 whenever n u t r i t i o n a l i nformation on t h i s product was provided a t the point of purchase. Bran C e r e a l. The r e s u l t s o f i n d i v i d u a l s i g n i f i c a n c e t e s t s on the mean r i d i t s f o r the 16 treatments with bran cereal are presented i n Table 38. Three o f the 16 mean r i d i t s are s i g n i f i c a n t l y g r e a t e r than 0.5. To determine whether there i s a tendency f o r the mean r i d i t to be greater than 0.5, the t e s t f o r the s i g n i f i c a n c e of combined r e s u l t s was ap p l i e d to the i n d i v i d u a l mean r i d i t s i g n i f i c a n c e f i n d i n g s . The tendency i s r e l i a b l e ( x 2 = 55.458; df = 32; p < .01). Hypothesis H^ cannot be r e j e c t e d f o r t h i s product. The mean r i d i t f o r the brand s a l e s d i s t r i b u t i o n of bran cereal tended to be greater than 0.5 when n u t r i t i o n a l information was provided on t h i s product. Results Combined by Experimental Week. Tables 34 to 38 separate the mean r i d i t s i g n i f i c a n c e r e s u l t s i n t o Week 2 and Week 3 combinations k to show the £ l o g g p^ t o t a l s over the k treatments f o r a product w i t h i n a given week. These t o t a l s were then summed across products i n a given week and the t e s t f o r the s i g n i f i c a n c e of combined r e s u l t s was performed f o r each week. The o b j e c t i v e was to t e s t whether the mean r i d i t d i s p l a y e d a r e l i a b l e tendency to be greater than 0.5 f o r a given experimental week, as a whole. 178 Week 2. The 38 o n e - t a i l e d p values from the mean r i d i t s i g n i f i c a n c e t e s t s on a l l 38 treatments i n Week 2 were transformed i n t o natural logarithms to obt a i n 38 I l o g e p1 = -92.476. The chi square t e s t on the combined r e s u l t s revealed a h i g h l y s i g n i f i c a n t e f f e c t f o r Week 2: 38 X 2 = -2 I l o g e p. = 184.952 ; df = 76 ; p - 0. Therefore, the tendency f o r the mean r i d i t to be greater than 0.5 during periods i n Week 2 when n u t r i t i o n a l information was provided to shoppers i s dependable. Hypothesis H^ i s accepted f o r Week 2 of the f i e l d experiment. Week 3. The 39 o n e - t a i l e d p values from the mean r i d i t s i g n i f i c a n c e t e s t s on the 39 n u t r i t i o n a l information treatments deployed i n Week 3 were transformed to obtain 39 I l o g e p. = -54.544. The chi square t e s t on the combined.results revealed a s i g n i f i c a n t e f f e c t f o r Week 3: 39 X 2 = -2 I l o g e p. = 109.088 ; df = 78 ; p < .05. 179 This f i n d i n g shows t h a t , despite d i f f e r e n c e s among i n d i v i d u a l products, there was a r e l i a b l e o v e r a l l tendency f o r the mean r i d i t to be grea t e r than 0.5 when n u t r i t i o n a l information was placed on signs at the point of purchase during Week 3. Hypothesis H^ i s accepted f o r Week 3 of the f i e l d experiment. Summary and Conclusions. The mean r i d i t was not s i g n i f i c a n t l y greater than 0.5 f o r every information stimulus used i n the f i e l d experiment. Approached d i f f e r e n t l y , however, the analyses i n d i c a t e t h a t f o r c e r t a i n t e s t products, there was a r e l i a b l e o v e r a l l tendency f o r the mean r i d i t to be greater than i t s expected value, under HQ, of 0.5. For the. products canned soup, m&c dinner and bran c e r e a l , the com-bined r e s u l t s o f i n d i v i d u a l treatment s i g n i f i c a n c e t e s t s support hypothesis H^. There i s a h i g h l y s i g n i f i c a n t tendency f o r the mean r i d i t to be greater than 0.5 when n u t r i t i o n a l information on these three products was provided to. s t o r e customers i n a brand-by-cue matrix format. The s i g n i f i c a n c e t e s t s on the combined r e s u l t s o f treatments w i t h mayonnaise and ketchup i n d i c a t e that H^ should be r e j e c t e d f o r these products. Perhaps consumers are not s e n s i t i v e to n u t r i t i o n a l information on ketchup and mayonnaise. Both products serve as a condiment or dres s i n g and are a c t u a l l y eaten i n r e l a t i v e l y small p o r t i o n s . I t i s a l s o p o s s i b l e that between-brand d i f f e r e n c e s i n n u t r i t i o n a l performance f o r these two products, as d i s c l o s e d by the n u t r i t i o n a l cues used on the s i g n s , were inconsequential to consumers/ (see Appendix A). One p o i n t i s i n order here. Brands of the products m&c dinner and bran cereal (but not canned soup) provided consumers with some n u t r i t i o n a l information on the package. I t i s p o s s i b l e that t h i s p r a c t i c e by the 180 manufacturers s e n s i t i z e s consumers to any a d d i t i o n a l n u t r i t i o n a l i n -formation on m&c dinner and breakfast c e r e a l , as provided by the p o i n t -of-purchase s i g n s . The s i g n i f i c a n c e t e s t s f o r combined treatment r e s u l t s reveal that the mean r i d i t tended to be greater than 0.5 whenever point-of-purchase signs were i n place during Week 2, i n ge n e r a l , and a l s o Week 3, i n general. The pooled r e s u l t s across an e n t i r e week i n d i c a t e t h a t hypothesis should be accepted f o r Week 2 and Week 3, the two experimental weeks. Tests of Hypothesis Hypothesis H^ s t a t e s Following the removal of the point-of-purchase information from the s t o r e s , the mean r i d i t f o r a product's weekly brand s a l e s d i s t r i b u t i o n remains a t the experimental baseline l e v e l of 0.5. This hypothesis p r e d i c t s t h a t the mean r i d i t f o r a product's post-experimental, weekly brand sales d i s t r i b u t i o n w i l l have the same value as that f o r the reference ( c o n t r o l ) d i s t r i b u t i o n during the experiment, namely, 0.5. Tests of hypothesis H^ showed t h a t the mean r i d i t tended to be above 0.5 during Weeks 2 a n d 3, when point-of-purchase signs were i n place. At the end o f Week 3 the store signs were permanently removed. Therefore, a " r e t u r n " to 0.5 i n the mean r i d i t ' s value f o r a product's brand s a l e s d i s t r i b u t i o n i n post-experimental weeks would support the c a u s a l i t y between information signs and the mean r i d i t . A d d i t i o n a l l y , the acceptance of t h i s hypothesis f o r a product would give greater v a l i d i t y to the experimental f i n d i n g s by demonstrating t h a t the c r u c i a l reference d i s t r i b u t i o n s i n experimental weeks (against which 181 experimental e f f e c t s were measured) represented samples drawn from the same population of brand s a l e s when no experiment was t a k i n g place i n the s t o r e s . The t e s t a b l e i m p l i c a t i o n of t h i s hypothesis i s t h a t the mean r i d i t f o r a product's t o t a l brand sales d i s t r i b u t i o n i n a given post-experimental week i s not s i g n i f i c a n t l y d i f f e r e n t from the mean r i d i t f o r Week 2 con t r o l or Week 3 co n t r o l ( t w o - t a i l e d t e s t ) . A n a l y t i c a l Approach. A product-by-product a n a l y t i c approach was taken to t e s t t h i s hypothesis because the reference d i s t r i b u t i o n s on which the e a r l i e r r i d i t analyses of treatment e f f e c t s were based are n e c e s s a r i l y p r o d u c t - s p e c i f i c . Moreover, d i f f e r e n t products were a f f e c t e d by d i f f e r e n t exogenous f a c t o r s from week to week and, t h e r e f o r e , required separate e v a l u a t i o n . The t e s t s of hypothesis Hg assume that there was no carryover of behavioral e f f e c t s from the experimental weeks i n t o the post-experimental p e r i o d . To s u b s t a n t i a t e t h i s assumption, data are provided on the average interpurchase i n t e r v a l f o r each product (taken from the consumer survey made p r i o r to the experiment) before the t e s t r e s u l t s are presented. The post-experimental weeks were weeks 4 and 5 of the f i e l d data c o l l e c t i o n and were included s p e c i f i c a l l y to t e s t t h i s hypothesis. In a d d i t i o n to the two r e t u r n - t o - b a s e l i n e weeks, brand s a l e s data were c o l l e c t e d one week p r i o r to the s t a r t of the experiment, i . e . , i n Week 1 of the f i e l d study. This l o n g i t u d i n a l b e f o r e - a f t e r design, which generated s a l e s data f o r f i v e consecutive weeks i n the s t o r e s , permitted a comparison of the reference d i s t r i b u t i o n s i n the experimental weeks ( i . e . , Week 2 co n t r o l and Week 3 c o n t r o l ) against the brand s a l e s d i s -182 t r i b u t i o n s of Week 1, 4 and 5. Week 2 c o n t r o l and Week 3 c o n t r o l d i s t r i b u t i o n s were a l s o compared against each other. The data underlying Weeks 1, 4 and 5 are the t o t a l s ales of each brand f o r t h a t e n t i r e week. Adjustments were made to these weekly t o t a l s wherever a c r i t i c a l stock outage occurred. For example, i f a l l s i z e s o f one brand o f a t e s t product were out of stock i n one of the predesignated measurement periods i n one s t o r e , the e n t i r e data f o r that p a r t i c u l a r s t o r e measurement were deleted before t o t a l l i n g the weekly s a l e s . A l l weekly t o t a l s represent the combined s a l e s of both stores (see Table 22). The brand sales data e n t e r i n g these analyses were weighted by package s i z e . R i d i t a n a l y s i s served as the a n a l y t i c a l t o o l f o r t e s t s o f t h i s hypothesis. For the weekly comparisons against an experimental c o n t r o l , the c o n t r o l data served as the reference d i s t r i b u t i o n which a u t o m a t i c a l l y gave i t a mean r i d i t of 0.5. Mean<ridits were then computed f o r the d i s t r i b u t i o n s i n other weeks. 7 In these analyses, the c a l c u l a t i o n of brand r i d i t s from the reference d i s t r i b u t i o n was made a f t e r rank ordering the brands by t h e i r o v e r a l l n u t r i t i v e performance i n c e r t a i n treatments, depending on the product. For two products (canned soup, m&c d i n n e r ) , the choice of treatment on which to base the brand ordering was immaterial because the rank order For these pairwise t e s t s , the d e c i s i o n as to which of the two d i s -t r i b u t i o n s serves as the "reference" i s not c r i t i c a l . The mean r i d i t s computed both ways would be symmetrical about 0.5 and have the same standard e r r o r . I n t u i t i v e l y , however, i t makes sense to "anchor" Week 2 c o n t r o l ' s mean r i d i t at 0.5 and to i l l u s t r a t e the l e v e l s of the mean r i d i t before and a f t e r the experiment was conducted i n the s t o r e s . This procedure was repeated with Week 3 con t r o l serving as the reference d i s t r i b u t i o n . 183 of brands by n u t r i t i v e performance was i d e n t i c a l across a l l treatments constructed f o r these products i n the experimental weeks. With the three remaining products, not a l l treatments r e s u l t e d i n the same n u t r i t i v e performance o r d e r i n g of brands. Consequently, the brand ordering which was common to the m a j o r i t y of treatments was chosen. Table 39 l i s t s the treatments which met t h i s c r i t e r i o n f o r each product. The r i d i t analyses to t e s t hypothesis are based on the common ordering of brands i n these treatments. For each product, the hypothesis set being tes t e d i s H Q: y. f 0.5 ; H 5: y. = 0.5 , where ]i- i s the mean r i d i t f o r post-experimental week i . Product Interpurchase I n t e r v a l s . One of the questions i n the survey of consumers preceding the f i e l d experiment provided data on the frequency of t e s t product purchases. The responses to t h i s question are summarized i n Table 40 i n terms of the average number of weeks between purchase o f each t e s t product. Responses are l i m i t e d to those consumers who pur-chased the product at l e a s t once every 16 weeks. On the average, the t e s t products are purchased l e s s o f t e n than once every three weeks. Thus, i t i s not very l i k e l y t hat the t e s t product purchase data i n post-experimental Weeks 4 and 5 contain some carryover from the e f f e c t s of Weeks 2 and 3, when the n u t r i t i o n a l signs were i n place. In a d d i t i o n , there i s some em p i r i c a l evidence that s t o r e patrons' Table 39 Treatments With a Common Brand Ordering by O v e r a l l N u t r i t i v e Performance f o r Each of Five Products Canned M&C Bran Soup Mayonnaise Ketchup Dinner Cereal 1/high 1/high 1/high 2/high 2/high 2/high 2/high 4/high 4/high 4/high 4/high 8/high 8/high 8/high 8/high 8/high 1/low 1/low 2/1 ow 2/1 ow 2/1 ow 4/low 4/1 ow 4/low 4/1 ow 8/low 8/low 8/low 8/low 8/low 185 Table 40 Mean Interpurchase I n t e r v a l , i n Weeks, f o r F i v e Products Based on Consumer Survey Responses (Standard Deviations i n Parentheses) Canned Soup Mayonnaise Ketchup 3.78 6.26 (2.93) (3.89) M&C Bran Dinner Cereal 6.12 3.76 6.71 (4.76) (2.85) (4.56) i 186 behavioural responses to point-of-purchase information aids on grocery products w i l l not ca r r y over beyond two weeks a f t e r t h i s information i s removed. Russo (1977) found that consumer behavioural changes due to point-of-purchase u n i t - p r i c e l i s t s which had been posted i n a super-market f o r e i g h t consecutive weeks disappeared completely w i t h i n two weeks of the removal of these l i s t s . The products i n hi s f i e l d experiment were apple j u i c e , c o f f e e , d i s h detergent, laundry detergent, peanut bu t t e r and syrup. Canned Soup. In Table 41 are the mean r i d i t s obtained f o r the canned soup brand s a l e s d i s t r i b u t i o n s of each week, when compared against the Week 2 co n t r o l d i s t r i b u t i o n . The d i s t r i b u t i o n s o f Week 1, Week 3 c o n t r o l and Week 4 are a l l s i g n i f i c a n t l y d i f f e r e n t from Week 2 con t r o l a t a = .05. Inspection of the "noise" monitoring record f o r t h i s product (Appendix E) reveals t h a t , during Week 1, a chain-wide a d v e r t i z e d p r i c e i n c e n t i v e was i n e f f e c t f o r one of the two brands (the one designated as n u t r i t i v e l y s u p e r i o r i n t h i s experiment, ergo an r > 0 . 5 ) . Furthermore, although t h i s p r i c e i n c e n t i v e was no longer i n e f f e c t during Week 2, a higher p r i c e f o r the same brand was introduced i n Week 3 and maintained through Week 5. This could e x p l a i n why the Week 3 con t r o l and Week 4 d i s t r i b u t i o n s are s i g n i f i c a n t l y d i f f e r e n t from Week 2 c o n t r o l . The mean r i d i t f o r Week 5 i s not s i g n i f i c a n t l y d i f f e r e n t from t h a t f o r Week 2 c o n t r o l , there-by supporting hypothesis H5-The r e s u l t s of these comparisons against Week 2 co n t r o l are summarized i n Figure 12. In s h o r t , the d i s t r i b u t i o n employed as a co n t r o l Table 41 Mean R i d i t s f o r Canned Soup Brand Sales D i s t r i b u t i o n s In Each of Four Weeks Compared to an Experimental Week Control Comparison Mean R i d i t z Value Two-T a i l e d p Week 2 (Control) vs, Week 1 Week 3 (Control) Week 4 Week 5 ,764 ,452 ,473 ,494 25.771 - 2.644 - 1.994 - .404 .000 <.01 <.05 n.s. Week 3 (Control) vs, Week 1 Week 2 (Control) Week 4 Week 5 .812 .548 .521 .542 28.129 2.644 1.389 2.693 .000 <.01 n.s. <.01 Figure 12 189 f o r Week 2 of the experiment i s no d i f f e r e n t from the t o t a l brand s a l e s d i s t r i b u t i o n observed i n Week 5. I t s shape does d i f f e r , however, from the shape of the d i s t r i b u t i o n i n the post-experimental Week 4. Table 41 a l s o gives the r e s u l t s of s i g n i f i c a n c e t e s t s on the mean r i d i t s f o r four weeks when compared against the Week 3 c o n t r o l d i s -t r i b u t i o n . As i n the foregoing analyses, Week 1 and Week 2 c o n t r o l d i f f e r s i g n i f i c a n t l y from Week 3 c o n t r o l . The Week 4 d i s t r i b u t i o n i s not s i g -n i f i c a n t l y d i f f e r e n t from Week 3 c o n t r o l , w h i le the Week 5 d i s t r i b u t i o n i s . 8 To summarize the above t e s t s with canned soup, hypothesis Hg cannot be r e j e c t e d when the Week 2 c o n t r o l d i s t r i b u t i o n i s compared with the d i s t r i b u t i o n i n post-experimental Week 5, thus p r o v i d i n g some v a l i d i t y to the experimental r e s u l t s of Week 2, i f the price-change explanations o f f e r e d e a r l i e r can be accepted. A l s o , hypothesis Hg can only be accepted f o r the comparison of the Week 3 co n t r o l d i s t r i b u t i o n against t h a t of p o s t - e x p e r i -mental Week 4. These f i n d i n g s i n d i c a t e only that the c o n t r o l data used as reference d i s t r i b u t i o n s f o r r i d i t analyses of Week 2 and Week 3 experimental e f f e c t s represent samples drawn from the population of brand sales when no point-of-purchase information e x i s t e d i n the stores during For the product-by-product analyses i n these s e c t i o n s only the comparisons against Week 2 c o n t r o l w i l l be graphed to avoid redundancy. In e f f e c t , the graph of mean r i d i t s i n comparison to Week 3 c o n t r o l would be v i r t u a l l y i d e n t i c a l to the one f o r Week 2 c o n t r o l i f the points are r i g i d l y t r a n s l a t e d , such t h a t the mean r i d i t f o r Week 3 c o n t r o l i s f i x e d at 0.5. Very small departures from i d e n t i t y are due to d i f f e r e n c e s i n the computation o f a weighted package from d i f f e r e n t p a i r s of d i s -t r i b u t i o n s . Save f o r t h i s small d i f f e r e n c e , the mean r i d i t f o r a week compared against Week 3 c o n t r o l can be obtained from i t s mean r i d i t com-pared against Week 2 c o n t r o l . For example, i n Table 41, r f o r Week 1 compared to Week 3 c o n t r o l i s the a l g e b r a i c d i f f e r e n c e between Week 2 and Week 3, plus r f o r Week 1 compared to Week 2 c o n t r o l : (.5-.452) + .764 = .812. 190 Week 5 and Week 4, r e s p e c t i v e l y . Mayonnaise. Table 42 gives the r e s u l t s o f weekly brand s a l e s d i s t r i b u t i o n comparisons against Week 2 co n t r o l and Week 3 c o n t r o l . The mean r i d i t f o r Week 1 i s s i g n i f i c a n t l y d i f f e r e n t from the c o n t r o l s i n both experimental weeks. Since the ord e r i n g o f brands on which these r i d i t analyses are based i s such that the K r a f t brand was ranked lowest i n n u t r i t i v e per-formance, t h i s f i n d i n g could be explained by the i n - s t o r e p r i c e i n c e n t i v e i n e f f e c t during Weeks 1 and 2 on one s i z e of K r a f t (although Week 2 c o n t r o l i s not s i g n i f i c a n t l y d i f f e r e n t from Week 3 c o n t r o l ) . Figure 13 i l l u s t r a t e s the mean r i d i t s f o r other weeks i n comparison to the Week 2 co n t r o l mean r i d i t of 0.5. In n e i t h e r of the post-experimental weeks was the mean r i d i t s i g -n i f i c a n t l y d i f f e r e n t from an experimental c o n t r o l , as shown i n Table 42. Hypothesis Hg cannot be r e j e c t e d f o r t h i s product. The mean r i d i t s f o r the mayonnaise brand sales d i s t r i b u t i o n s i n post-experimental weeks remained at the ba s e l i n e l e v e l of 0.5. Ketchup. The r e s u l t s of weekly comparisons against Week 2 and Week 3 c o n t r o l s f o r ketchup are given i n Table 43. In Figure 14, Week 3 cont r o l i s "anchored" at a mean r i d i t of 0.5 because Weeks 2 and 4 coincided with a chain-wide a d v e r t i s e d p r i c e i n c e n t i v e on one of the two brands (the one designated as n u t r i t i v e l y s u p e r i o r , which exp l a i n s why r > 0.5). The mean r i d i t s f o r these two weeks, t h e r e f o r e , are s i g n i f i c a n t l y d i f f e r e n t from the mean r i d i t f o r Week 3 c o n t r o l , as shown i n Table 43. Comparisons o f other weeks against Week 2 c o n t r o l reveal a s i g n i f i c a n t d i f f e r e n c e i n mean r i d i t s i n every case. This f i n d i n g i s l i k e l y explained 191 Table 42 Mean R i d i t s f o r Mayonnaise Brand Sales D i s t r i b u t i o n s In Each of Four Weeks Compared to an Experimental Week Control Comparison Mean R i d i t z Value Two-T a i l e d p Week 2 (Control) vs, Week 1 Week 3 (Control) Week 4 Week 5 .450 .512 .532 .544 •2.020 .377 1.299 1.697 <.05 n.s. n.s. n.s. Week 3 (Control) vs, Week 1 Week 2 (Control) Week 4 Week 5 .439 ,488 .520 .532 •2.316 • .377 .738 1.158 <.05 n.s. n.s. n.s. Figure 13 Mean R i d i t s f o r Mayonnaise Brand Sales D i s t r i b u t i o n s In Each of Five Weeks ®Mean R i d i t s i g n i f i c a n t l y d i f f e r e n t from 0.5 at a = .05 193 Table 43 Mean R i d i t s f o r Ketchup Brand Sales D i s t r i b u t i o n s In Each o f Four Weeks Compared to an Experimental Week Control Mean z Two Comparison R i d i t Value T a i l e d p Week 2 (Control) vs. Week 1 .385 -6.837 .000 Week 3 (Control) .379 -5.004 .000 Week 4 .431 -4.047 .000 Week 5 .352 -8.566 .000 Week 3 (Contro l ) vs. Week 1 .507 .340 n.s. Week 2 (Control) .621 5.004 .000 Week 4 .553 2.553 <.05 Week 5 .473 -1.407 n.s. 194 a ®Mean r i d i t s i g n i f i c a n t l y d i f f e r e n t from 0.5 at a = .05 195 by the p r i c e i n c e n t i v e i n Week 2. In the comparisons against Week 3 c o n t r o l , n e i t h e r the Week 1 nor the Week 5 d i s t r i b u t i o n has a s i g n i f i c a n t l y d i f f e r e n t mean r i d i t . Thus, the Week 5 d i s t r i b u t i o n f i n d i n g supports hypothesis Hg. Since the experiments with n u t r i t i o n a l information on ketchup brands showed no e f f e c t s , the f i n d i n g here t h a t the c o n t r o l d i s t r i b u t i o n i n Week 3 d i d not d i f f e r from that of pre- and post-experimental weeks gives credence to the conclusion of no information e f f e c t s i n the ex-perimental weeks with t h i s product. M&C Dinner. Table 44 and Figure 15 show th a t the mean r i d i t i n Week 1 was s i g n i f i c a n t l y d i f f e r e n t from t h a t of e i t h e r Week 2 or Week 3 c o n t r o l s f o r m&c dinner. No explanation f o r t h i s p a r t i c u l a r observation was found. The remaining r e s u l t s on t h i s product support hypothesis Hg f o r both experimental week c o n t r o l s . Neither the Week 4 nor the Week 5 mean r i d i t was s i g n i f i c a n t l y d i f f e r e n t from 0.5. Bran C e r e a l . Table 45 and Figure 16 reveal that hypothesis Hg i s supported by the r i d i t a n a l y s i s r e s u l t s on bran cereal f o r the post-experimental week comparisons versus Week 2 c o n t r o l . The mean r i d i t s f o r Weeks 1, 4 and 5 are not s i g n i f i c a n t l y d i f f e r e n t from the mean r i d i t of 0.5 f o r Week 2 c o n t r o l . Thus, the Week 2 experimental r e s u l t s on bran cereal reported i n t h i s chapter (and based on Week 2 c o n t r o l data as the reference d i s t r i b u t i o n ) are given greater v a l i d i t y . The mean r i d i t f o r Week 3 co n t r o l i s s i g n i f i c a n t l y d i f f e r e n t from the mean r i d i t s f o r every other week, i n c l u d i n g Week 2 c o n t r o l . The "noise" monitoring record f o r t h i s product (Appendix E) i n d i c a t e s t h a t an i n - s t o r e 196 Table 44 Mean R i d i t s f o r Macaroni & Cheese Dinner Brand Sales D i s t r i b u t i o n s In Each o f Four Weeks Compared to an Experimental Week Control Mean z Two-Comparison R i d i t Value T a i l e d p Week 2 (Control) vs. Week 1 .536 3.023 <.005 Week 3 (Control) .496 - .199 n.s. Week 4 .489 - .786 n.s. Week 5 .482 -1.266 n.s. Week 3 (Contro l ) vs. Week 1 .537 2.559 <.05 Week 2 (Control) .504 .198 n.s. Week 4 .494 - .385 n.s. Week 5 .487 - .768 n.s. Figure 15 Mean R i d i t s f o r Macaroni & Cheese Dinner Brand Sales D i s t r i b u t i o n s In Each o f Five Weeks 198 p r i c e i n c e n t i v e was i n e f f e c t f o r one of the cereal brands during Weeks 3, 4 and 5. This disturbance would e x p l a i n why the mean r i d i t i s greater than 0.5 f o r Week 3 co n t r o l and Weeks 4 and 5, when compared to Week 2 g c o n t r o l . However, nothing i n the "noise" monitoring record suggested a reason f o r the s i g n i f i c a n t d i f f e r e n c e i n mean r i d i t s between Week 3 c o n t r o l and Week 4 or Week 5. The p o s s i b i l i t y that the Week 3 c o n t r o l observation might r e f l e c t treatment carryover e f f e c t s from Week 2 was discounted on the grounds that the mean interpurchase i n t e r v a l f o r bran cereal i s consider-ably longer than one week (Table 40). Since no other disturbance was found to e x p l a i n the d i f f e r e n c e s i l l u s t r a t e d i n Figure 16, a t t e n t i o n focused on the Week 3 co n t r o l ob-s e r v a t i o n , which had the smallest sample s i z e among the f i v e weekly d i s -t r i b u t i o n s . Table 45 i n d i c a t e s the sample s i z e s ( u n i t s a l e s ) underlying each weekly d i s t r i b u t i o n . The Week 3 co n t r o l sample (119) i s not much l a r g e r than h a l f of the Week 2 c o n t r o l sample. Moreover, u n i t s a l e s of some brands i n the Week 3 co n t r o l data are i n the order of 5 and 6. Consequently, the Week 3 co n t r o l data were considered vulnerable to sampling e r r o r . Bross (1958:22) recommends th a t the reference d i s t r i b u t i o n chosen f o r r i d i t analyses o f treatment d i s t r i b u t i o n s be la r g e enough In the rank ordering of the seven brands by n u t r i t i v e performance f o r r i d i t a n a l y s i s , the price-reduced brand i s ranked j u s t above the two lowest-ranked brands which have a combined market share of 56% i n the Week 2 c o n t r o l d i s t r i b u t i o n o f brand s a l e s . In the absence of n u t r i t i o n a l i n f o r m a t i o n , purchase s h i f t s from these two leading brands, and from brands ranked higher i n the r i d i t a n a l y s i s h i e r a r c h y , toward the discounted brand would have the net e f f e c t of i n c r e a s i n g the mean r i d i t from i t s b a s e l i n e value of 0.5 i n Week 2 c o n t r o l . 199 Table 45 Mean R i d i t s f o r Bran Cereal Brand Sales D i s t r i b u t i o n s In Each of Four Weeks Compared to an Experimental Week Control Mean z Two- Sample Comparison R i d i t Value T a i l e d P Si z e (N) Week 2 (Control) vs. 205 Week 1 .484 - .739 n.s. 600 Week 3 (Control) .598 3.020 <.005 119 Week 4 .522 .945 n.s. 505 Week 5 .543 1.857 n.s. 521 Week 3 (Control) vs. Week 1 .388 -4.048 .000 Week 2 (Control) .402 -3.020 <.005 Week 4 .423 -2.674 <.01 Week 5 .438 -2.178 <.05 Week 2 (Control) vs. Week 3 c o n t r o l / Week 4/Week 5, Pooled .539 1.837 n.s. 1145 Figure 16 Mean Ridits for Bran Cereal Brand Sales Distributions In Each of Five Weeks © M e a n r i d i t s ignif icant ly different from 0.5 at a = .05 201 ( r e l a t i v e to treatment samples) "...to ensure t h a t the r i d i t s w i l l be s t a b l e . " In r e t r o s p e c t , these c o n s i d e r a t i o n s and the outcomes of t e s t s with Week 3 c o n t r o l suggested a more appropriate approach to the r i d i t analyses of bran cereal treatment e f f e c t s i n Week 3. A d e c i s i o n was made to com-pensate f o r the suspected sampling e r r o r i n the Week 3 reference d i s -t r i b u t i o n by pooling the data o f Week 3 c o n t r o l with the data of Week 4 and Week 5. Thus, the most l i k e l y source of "noise", the i n - s t o r e p r i c e i n c e n t i v e on one brand, during these three weeks, remained constant. The r e s u l t i n g brand s a l e s d i s t r i b u t i o n was a l a r g e r sample o f 1145 u n i t purchases and served as the reference d i s t r i b u t i o n f o r Week 3 treatment e f f e c t s . A l l bran cereal mean r i d i t s i n experimental Week 3 reported i n t h i s chapter are a c t u a l l y based on r i d i t analyses performed with t h i s pooled reference d i s t r i b u t i o n . A comparison of t h i s pooled Week 3 reference d i s t r i b u t i o n against the Week 2 c o n t r o l d i s t r i b u t i o n reveals t h a t t h e i r r e s p e c t i v e mean r i d i t s are not s i g n i f i c a n t l y d i f f e r e n t , as shown i n Table 45, although t h i s no longer c o n s t i t u t e s a t e s t of hypothesis Hg. Summary and Conclusions. With some exceptions, the product-by-product mean r i d i t s i g n i f i c a n c e t e s t s i n t h i s s e c t i o n tend to support the hypothesis that the mean r i d i t f o r a post-experimental week does not d i f f e r from the mean r i d i t value of 0.5 f o r the co n t r o l d i s t r i b u t i o n s i n experimental weeks. The f i n d i n g s on mayonnaise and m&c dinner give no reason to r e j e c t hypothesis Hg. With the products canned soup and ketchup, hypothesis Hg was supported by the f i n d i n g s on one or two of the fou r weekly comparisons. However, i n 202 other weeks the brand sa l e s d i s t r i b u t i o n s were apparently a f f e c t e d by exogenous f a c t o r s which were noted during the f i e l d experiment. F i n a l l y , the t e s t r e s u l t s on bran cereal support hypothesis with respect to weekly d i s t r i b u t i o n comparisons made against the c o n t r o l d i s t r i b u t i o n i n experimental Week 2. A p o s s i b l e anomaly was i n d i c a t e d by the mean r i d i t obtained f o r the con t r o l d i s t r i b u t i o n i n experimental Week 3. Tests showed i t to be s i g -n i f i c a n t l y d i f f e r e n t from the mean r i d i t s f o r a l l other weeks. While t h i s outcome refuted hypothesis H 5, i t suggested a sampling e r r o r i n the Week 3 c o n t r o l d i s t r i b u t i o n which could lead to l e s s r e l i a b l e r e s u l t s from the r i d i t analyses of treatment e f f e c t s i n Week 3. Therefore, the Week 3 con t r o l d i s t r i b u t i o n was pooled with the d i s t r i b u t i o n s of Weeks 4 and 5 i n order to reduce sampling v a r i a t i o n i n a newly i d e n t i f i e d reference d i s t r i b u t i o n f o r Week 3. A l l r i d i t analyses of bran cereal treatment e f f e c t s i n Week 3 reported i n t h i s chapter were based on t h i s pooled reference d i s t r i b u t i o n . A d d i t i o n a l Analyses The preceding analyses which were s p e c i f i c a l l y designed to t e s t the research hypotheses, revealed some unexpected e f f e c t s due to d i f f e r e n t experimental weeks and d i f f e r e n t t e s t products, per se. Therefore, i t was decided to explore more f o r m a l l y any d i f f e r e n c e s i n the mean r i d i t accounted f o r by product and week d i f f e r e n c e s . A n a l y t i c a l Approach. The approach which seemed most appropriate f o r i s o l a t i n g product and week e f f e c t s was to t r e a t the product and the ex-perimental week as two independent v a r i a b l e s and examine the pattern of 203 t h e i r e f f e c t s on the mean r i d i t . Towards t h i s end, the a n a l y s i s o f the mean r i d i t database proceeded along the l i n e s o f a 5 (products) x 2 (weeks) f a c t o r i a l experiment. The observations i n each of the 10 c e l l s were the mean r i d i t s r epresenting treatment e f f e c t s p e r t a i n i n g to a p a r t i c u l a r product, i n a p a r t i c u l a r week. Two-way a n a l y s i s of variance was used to i n v e s t i g a t e the j o i n t and separate e f f e c t s due to products and weeks. Following t h i s , the d i f f e r e n c e between mean r i d i t s f o r subsets of products was analyzed with the t - t e s t , i n order to i s o l a t e product c l a s s d i f f e r e n c e s . A n a l y s i s o f Variance R e s u l t s . Figure 17 p l o t s the means of the mean r i d i t s i n each of the 10 product/week c e l l s . The graph shows c l e a r d i f f e r e n c e s between products and a tendency f o r the Week 3 experimental r e s u l t s to have a lower mean r i d i t than the Week 2 r e s u l t s . Table 46 summarizes the a n a l y s i s o f variance r e s u l t s on product and week e f f e c t s , both of which are s t a t i s t i c a l l y s i g n i f i c a n t . The r e s u l t s show no i n t e r a c t i o n between weeks and products. The m u l t i p l e R 2 obtained from t h i s a n a l y s i s i n d i c a t e s t h a t d i f f e r e n c e s i n the t e s t products and weeks e x p l a i n about 21% of the variance i n the mean r i d i t data. E f f e c t s Due to Products. The t e s t s o f hypothesis H^, presented e a r l i e r , i n d i c a t e d a r e l i a b l e o v e r a l l tendency f o r the mean r i d i t to be greater than 0.5 whenever n u t r i t i o n a l information was provided at the point of purchase on canned soup, m&c dinner and bran c e r e a l . In c o n t r a s t , no such tendency was confirmed f o r the products ketchup and mayonnaise. This f i n d i n g suggested a dichotomy of product c l a s s e s c h a r a c t e r i z e d by the three products toward which consumers are apparently n u t r i t i o n a l i n f o r m a t i o n - s e n s i t i v e and two products (ketchup, mayonnaise) toward which 204 Figure 17 Mean R i d i t (Averaged Over A l l Treatments) P e r t a i n i n g to Each of Five Products and Two Experimental Weeks Product Table 46 Mean R i d i t C e l l Means and An a l y s i s of Variance of Product and Week E f f e c t s C e l l Means Product Canned M&C Bran Week Soup Mayonnaise Ketchup Dinner Cereal Week 2 .548 .519 .486 .560 .543 Week 3 .527 .497 .495 .513 .489 206 Table 46 (Continued) A n a l y s i s of Variance Source o f V a r i a t i o n SS a d.f. MS a F p Main E f f e c t s Products Weeks I n t e r a c t i o n s Products & Weeks Residual Total 3.857 2.483 1.426 5 4 1 .771 .621 1.426 3.772 3.036 6.973 .005 <.05 .01 .949 13.702 18.508 4 67 76 .237 .205 .244 1.160 n.s. a xlOO they are not. The mean r i d i t s f o r a l l information treatments with soup, m&c dinner and cereal were contrasted with the mean r i d i t s observed f o r a l l treatments w i t h ketchup and mayonnaise. The sample mean f o r t h e " n u t r i t i o n a l i n f o r m a t i o n - s e n s i t i v e " product c l a s s was .530 compared to .499 f o r the complementary c l a s s . The d i f f e r e n c e was s t a t i s t i c a l l y s i g n i f i c a n t ( t = 2.80; df = 75; p < .01, t w o - t a i l e d ) . This product c l a s s d i f f e r e n c e i n mean r i d i t s confirms the f i n d i n g s from t e s t s of hypothesis H^. There i s a r e l i a b l e tendency f o r p o i n t - o f -purchase n u t r i t i o n a l information on canned soup, m&c dinner and bran cereal to a f f e c t brand choices on these products, whereas such information does not i n f l u e n c e the choice of ketchup or mayonnaise brands. E f f e c t s Due to Weeks. The 38 mean r i d i t s computed f o r the treatments i n Week 2 had a mean value of .531. By c o n t r a s t , the mean value of the 39 mean r i d i t s obtained from treatments i n Week 3 was .504. Since t h i s d i f f e r e n c e i n means i s s t a t i s t i c a l l y s i g n i f i c a n t ( t = 2.45; df = 75; p < .05, tw o - t a i l e d ) i t demands some explanation. The question may be asked: "Why i s the mean r i d i t i n the second experimental week (Week 3) s i g n i f i c a n t l y lower than the mean r i d i t i n the f i r s t experimental week (Week 2)?" More d i r e c t l y , one might ask: "Why are the e f f e c t s o f p o i n t - o f -purchase n u t r i t i o n a l information signs weaker i n Week 3 than i n Week 2?" An explanation o f f e r e d here i s that i n Week 2 shoppers were seeing these point-of-purchase signs f o r the f i r s t time. A c e r t a i n amount o f c u r i o s i t y and responsiveness may be expected from consumers exposed to a new and h i g h l y v i s i b l e stimulus i n the s t o r e s . I t i s p o s s i b l e t h a t i n Week 3 when the signs had been up f o r the second week, part of t h i s 208 novelty had d i s s i p a t e d and fewer consumers who had seen them i n the f i r s t week responded to the information when choosing a brand. The author has discussed t h i s r e s u l t with other researchers i n the same area. Russo (1981b) and Laroche (1980) have both suggested that t h i s d i f f e r e n c e from one week to the next may represent a wearout of the sign s ' "novelty e f f e c t " i n the f i r s t experimental week. Results from a f i e l d experiment by Russo (1977) seem to show an e f f e c t s i m i l a r to the one found i n t h i s study. The f i n d i n g s from Russo's (1977) experiment i n which u n i t - p r i c e l i s t s were posted i n a supermarket to supplement the t r a d i t i o n a l u n i t - p r i c e s h e l f tags are reproduced i n Figure 18. The average percentage saved by consumers on s i x grocery products as a f u n c t i o n of d i f f e r e n t u n i t - p r i c e information formats i s -shown by the connected p o i n t s . Of i n t e r e s t i s the set of points from weeks 9 to 16 of the e x p e r i -ment which represent the e f f e c t s of posting u n i t - p r i c e l i s t s at the product s h e l v i n g . In Week 9, the l i s t s were deployed a t the point of purchase f o r the f i r s t time and apparently caused a la r g e increase i n the median savings by consumers (highest point on the graph). Note, however, the abrupt d e c l i n e i n the median savings i n the week immediately f o l l o w i n g the i n t r o d u c t i o n o f the l i s t s . Although there i s no way of being c e r t a i n that t h i s d e c l i n e i n Week 10 was due to the wearout o f novelty a s s o c i a t e d with the sign s ' i n i t i a l i n t r o d u c t i o n , i t does correspond with the change observed i n the present f i e l d experiment from Week 2 to Week 3. Summary and Conclusions. The a d d i t i o n a l analyses performed here explored the e f f e c t s on the dependent v a r i a b l e due apparently to the d i f f e r e n t t e s t products used i n the experiment. D i f f e r e n c e s i n the mean 209 Figure 18 E f f e c t s of Various U n i t - P r i c e Formats i n Russo's (1977) 20-Week In-store Experiment The savings was computed relative to price paid per unit at control store. Median is calculated over the six products. Uncon-nected points represent changes between experimental conditions. 0 2 4 6 8 10 12 14 16 18 20 WEEKS (From Russo, 1977:197, F i g . 2) 2 1 0 r i d i t a r i s i n g from d i f f e r e n t experimental weeks were a l s o analyzed. S i g n i f i c a n t d i f f e r e n c e s were found i n the mean r i d i t s obtained under treatments with d i f f e r e n t t e s t products. Information treatments with the t e s t products canned soup, m&c dinner and bran cereal c o l l e c t i v e l y had a greater e f f e c t on the mean r i d i t than treatments with mayonnaise and ketchup. I t i s assumed that consumers pay greater a t t e n t i o n to n u t r i t i o n a l information on the former c l a s s of products than on condi-ments or dressings l i k e ketchup and mayonnaise. These r e s u l t s are i n agreement with the p r o d u c t - s p e c i f i c f i n d i n g s made i n t e s t s of hypothesis A s i g n i f i c a n t d i f f e r e n c e e x i s t s i n the average values o f the mean r i d i t f o r treatments i n the f i r s t and second experimental weeks. One hypothesis f o r t h i s observed d i f f e r e n c e from one experimental week to the next i s the "novelty e f f e c t " of a f i r s t - t i m e exposure. I t i s p o s s i b l e t h a t n u t r i t i o n a l information provided i n a brand-by-cue format on signs placed conspicuously i n the supermarket a i s l e s a t t r a c t e d greater a t t e n t i o n i n the f i r s t week than i n the second week. This question remains an e m p i r i c a l one and f u t u r e research might focus on t h i s p a r t i c u l a r v a r i a b l e to t e s t the "novelty e f f e c t " hypothesis with i n - s t o r e experiments. Chapter V Summary. This chapter presented the r e s u l t s of analyses to t e s t each of the f i v e research hypotheses. The chapter commenced with with some p r e l i m i n a r y a n a l y s i s to j u s t i f y the pooling procedures across products and weeks. Next, the data a n a l y s i s s h i f t e d to t e s t s o f each hypothesis, i n t u r n . 211 Hypotheses -H^ , H^ and H^ were r e j e c t e d while hypotheses H^ and Hg were g e n e r a l l y supported by the r e s u l t s , given the a n a l y t i c methods employed. The chapter ended with some a d d i t i o n a l analyses which were warranted by the r e s u l t s i n e a r l i e r parts of the a n a l y s i s . Undoubtedly, the database assembled from t h i s e m p i r i c a l work i n the f i e l d can be e x p l o i t e d f u r t h e r and analyzed i n d i f f e r e n t ways. Further analyses by t h i s w r i t e r and h i s colleagues w i l l continue as new ideas and p o t e n t i a l v a r i a b l e s come to l i g h t to e x p l a i n more of the variance i n the sales data. Chapter VI summarizes the f i n d i n g s o f t h i s research and discusses t h e i r i m p l i c a t i o n s f o r consumer information processing and consumer p o l i c y . The r e s u l t s o f t h i s research are a l s o r e l a t e d to f i n d i n g s from s i m i l a r s t u d i e s . 212 CHAPTER VI SUMMARY AND IMPLICATIONS Chapter IV described the experimental design and the s t a t i s t i c a l methodology chosen to analyze the e m p i r i c a l data. Chapter V presented the r e s u l t s of analyses to t e s t the research hypotheses. This chapter summarizes the e m p i r i c a l outcomes and r e l a t e s them to the o r i g i n a l o b j e c t i v e s of the research e f f o r t . The p o t e n t i a l i m p l i c a t i o n s o f these f i n d i n g s are discussed and suggestions o f f e r e d f o r f u r t h e r research i n t h i s area. Summary of the Research Objectives and Procedures The primary purpose of t h i s research was to extend the f i n d i n g s of past l a b o r a t o r y s t u d i e s on consumer information l o a d . Procedures were undertaken to avoid the p o t e n t i a l a r t i f a c t s of l a b o r a t o r y experimentation by moving the experiment with information load i n t o a r e a l i s t i c consumer s e t t i n g . The research was, t h e r e f o r e , implemented u n o b t r u s i v e l y i n s i d e two cooperating supermarkets i n order to gain a b e t t e r understanding of how consumers respond to d i f f e r e n t amounts o f a s p e c i f i c type of o b j e c t i v e product i n f o r m a t i o n , when i t i s placed i n a p a r t i c u l a r format at the point of decision-making. This experiment represented a f i e l d t e s t o f a model from human information processing. Laboratory experiments i n psychology had shown that because o f the c a p a c i t y c o n s t r a i n t s o f short-term memory people can only a s s i m i l a t e a l i m i t e d amount of informa t i o n f o r use i n a d e c i s i o n -making task w i t h i n a given time p e r i o d . Excessive rates of information 213 inputs lead to l e s s e f f e c t i v e information usage by the d e c i s i o n maker, simply because the memory s t r u c t u r e cannot keep up with the processing demands imposed by these input r a t e s . This r e s u l t i s commonly r e f e r r e d to as "information overload." This i s the g i s t of the e m p i r i c a l f i n d i n g s supporting the Schroder, D r i v e r and S t r e u f e r t (1967) model of information processing. On the basis of such a model, consumer researchers t e s t e d the r e l a t i o n s h i p bet-ween load and information processing i n the consumer context. The assumption was that consumers would be s i m i l a r l y "overloaded" i f faced with a l a r g e amount of product information i n a choice-making s i t u a t i o n . Commencing with the l a b o r a t o r y work of Jacoby and h i s colleagues (Jacoby, S p e l l e r and Kohn, 1974a ;b), the f i n d i n g s from several l a b o r a t o r y studies on consumer information load are mixed. Some stu d i e s reported evidence of "information overload," while others found no such evidence. M i s s i n g from t h i s research stream were e m p i r i c a l data derived under n a t u r a l i s t i c c o n d i t i o n s i n the marketplace. Would a model of "information overload" derived i n the psychology l a b o r a t o r y be a p p l i c a b l e to brand choice d e c i s i o n s i n a commonplace shopping s i t u a t i o n ? The s t a t e of a f f a i r s as they stood p r i o r to the undertaking of the present f i e l d experiment was summarized by Day (1976:48): . ,. . People can only a s s i m i l a t e and process a l i m i t e d amount of i n f o r m a t i o n ; excess information may be dys-f u n c t i o n a l . However, there i s as y e t no c o n c l u s i v e evidence t h a t buyers are e a s i l y overloaded with information which reduces t h e i r decision-making a b i l i t y . The absence of c l e a r - c u t r e s u l t s i s a product o f : ( l ) methodological problems i n the reported experimental s t u d i e s , (2) the use of too rigorous a c r i t e r i o n o f choice accuracy (choice behavior i s more l i k e l y to r e f l e c t s a t i s f i c i n g than o p t i m i z i n g b e h a v i o r ) , (3) the f a c t t h a t the b a s i c information overload premise derives from t h e o r i e s of short-term memory, 214 and (4) the r e l i a n c e on l a b o r a t o r y experiments i n con-s t r a i n e d time and exposure s e t t i n g s . A second o b j e c t i v e of t h i s experiment was to c o n t r i b u t e to p o l i c y -o r i e n t e d research on information p r o v i s i o n . The study provided data on the behavioural e f f e c t s of d i s p l a y i n g o b j e c t i v e types of product performance cues at the p o i n t of purchase, such t h a t they were e a s i l y a c c e s s i b l e to consumers and organized i n a format a l l o w i n g d i r e c t comparisons of a l t e r -n a t i v e brands. A b a s i c input-output experimental paradigm was employed to generate the necessary data. The input v a r i a b l e s were information load and the r e l a t i v e importance of i n d i v i d u a l information cues making up a load. Those were manipulated i n the context of a f i x e d p r e s e n t a t i o n format. The outputs c o n s i s t e d of aggregate s t o r e s a l e s data on the brands of each t e s t product used i n the study. These were s t a t i s t i c a l l y analyzed to reveal whether shoppers were i n c o r p o r a t i n g the inputs i n t h e i r brand choice d e c i s i o n s and to determine whether the information had the hypo-t h e s i z e d e f f e c t s on aggregate purchase behaviour. D i f f e r e n t amounts of n u t r i t i o n a l information on the brands of several non-staple food products were p r i n t e d on point-of-purchase s t o r e signs and represented the experimental s t i m u l i . A l i t e r a t u r e review of the p a r t i c u l a r problem i n consumer behaviour i s o l a t e d i n t h i s research suggested a number of research questions about the r e l a t i o n s h i p s between information input loads and market behaviour. The research questions, which were f o r m a l i z e d i n t o hypotheses, were as f o l l o w s : 1. Does market response to information input loads reach maximum 215 at lower loads and dim i n i s h at the highest l o a d , as a r e s u l t of "information overload?" Can an i n v e r t e d U-shaped r e l a t i o n s h i p between load and market response be found, as i n the Sc h r o d e r - D r i v e r - S t r e u f e r t model o f human information processing? 2. I f an information input load i s constructed from n u t r i t i o n a l cues which are r e l a t i v e l y important (based on stated importance measures obtained from a consumer survey), w i l l the market response be greater than the response to a s i m i l a r load comprised of l e s s important n u t r i t i o n a l cues? Is there a response i n t e r a c t i o n between input l o a d and n u t r i t i o n a l cue importance? 3. W i l l a r e v e r s a l i n the l e f t - t o - r i g h t arrangement o f cues l i s t e d i n a brand-by-cue information matrix a f f e c t consumer response? 4. W i l l consumers base t h e i r brand choices on the n u t r i t i o n a l information now p r i n t e d on some food packages, i f t h a t information i s posted i n a format which f a c i l i t a t e s d i r e c t brand comparisons? 5. Does removal of the point-of-purchase i n f o r m a t i o n from the stores i n the post-experimental weeks c o i n c i d e with the re t u r n o f a product's brand s a l e s d i s t r i b u t i o n to the pre-experimental shape? Can a c a u s a l i t y between t h i s information and market response be e s t a b l i s h e d ? The hypotheses derived from these research questions were ope r a t i o n -a l l ' zed i n the f o l l o w i n g form to make them d i r e c t l y amenable to s t a t i s t i c a l t e s t i n g : H.: The mean r i d i t reaches a maximum at some lower information l o a d , and diminishes at the highest information load. Hp* The mean r i d i t at an information load c o n s i s t i n g of "high-importance" n u t r i t i o n a l cues i s greater than the mean r i d i t at the same load using "low-importance" n u t r i t i o n a l cues. 216 H^: A point-of-purchase sign l i s t i n g e i g h t cues i n decreasing order of importance, from l e f t to r i g h t , y i e l d s a greater mean r i d i t than a sign l i s t i n g the same e i g h t cues i n reverse order. H 4: The mean r i d i t f o r a product's brand s a l e s d i s -t r i b u t i o n i s greater than 0.5 when n u t r i t i o n a l i nformation i s placed i n a brand-by-cue matrix format a t the point of purchase. H^: Following the removal o f the point-of-purchase information from the s t o r e s , the mean r i d i t f o r a product's weekly brand sales d i s t r i b u t i o n remains at the experimental b a s e l i n e l e v e l o f 0.5. The mean r i d i t served as the dependent v a r i a b l e throughout these hypothesis t e s t s and i s a s t a t i s t i c which r e l a t e s the shape o f one brand sales d i s t r i b u t i o n to another. I t s choice as a dependent measure was based on the f o l l o w i n g e x p e c t a t i o n , which u n d e r l i e s a l l the hypotheses. Exposure to n u t r i t i o n a l information i n f l u e n c e s consumers' brand choices. The r e s u l t i n g brand s a l e s d i s t r i b u t i o n , - w h e n compared to a d i s t r i b u t i o n where no information was provided, i s skewed towards brands with a n u t r i t i o n a l advantage. The mean r i d i t s t a t i s t i c summarizes the extent of r e l a t i v e skewness and i t s s t a t i s t i c a l s i g n i f i c a n c e can al s o be determined. Since i t s expected value under no information e f f e c t s i s 0.5, t h i s f i g u r e provided a b a s e l i n e against which changes i n the mean r i d i t could be gauged. Hypothesized r e l a t i o n s h i p s between the mean r i d i t and the e x p e r i -mental f a c t o r s , load and cue-importance, were i n v e s t i g a t e d with the a n a l y s i s of variance and the t - t e s t . S i g n i f i c a n c e t e s t s , based on the standard normal v a r i a b l e , z, were performed on i n d i v i d u a l mean r i d i t s as part of the t e s t s o f hypotheses H^ and Hg. Summary of the Findings The major t h r u s t of t h i s study centered on the e f f e c t s of load and 217 cue importance. The f i n d i n g t h a t n e i t h e r of these f a c t o r s was s y s t e m a t i c a l l y a s s o c i a t e d with d i f f e r e n c e s i n the mean r i d i t was unexpected. However, somewhat m i t i g a t i n g these r e s u l t s was the f i n d i n g t h a t , i n c e r t a i n i n s t a n c e s , the placement o f n u t r i t i o n a l information on a product i n a more processable format apparently l e d to brand choices being made on the basis of th a t data. This study a l s o suggested t h a t whether or not n u t r i t i o n a l information w i l l be useful to consumers may depend on the type of food product i n v o l v e d . The a n a l y s i s a l s o i n d i c a t e d t h a t the general e f f e c t s of t h i s n u t r i t i o n a l i nformation were s i g n i f i c a n t l y weaker i n the second o f two weeks during which the information was made a v a i l a b l e . Hypothesis H^. The expectation was th a t the mean r i d i t would peak at some lower load and then d e c l i n e at the maximum experimental load of 8 cues as consumers avoided the most complex d i s p l a y of data i n the design and based t h e i r brand choices on other c r i t e r i a . When the mean r i d i t was p l o t t e d against information l o a d , the global maximum on that curve was a t t a i n e d at a load of 1 cue. T h e r e a f t e r , the mean r i d i t d e c l i n e d (non-monotonically) with s u c c e s s i v e l y higher loads. However, at a load o f 8 cues (the highest l o a d ) , the mean r i d i t was not s i g n i f i c a n t l y lower than the global maximum at 1 cue. In any case, the trend or pattern i n t h i s response curve was not r e l i a b l e , s i n c e the e f f e c t s o f load were found to be s t a t i s t i c a l l y non-s i g n i f i c a n t . Looking at the pattern o f aggregate brand s a l e s , one cannot d i s c e r n any d i f f e r e n c e i n shopper behaviour as a r e s u l t of d i f f e r e n t numbers o f product information cues on the si g n s . Consumers i n c o r p o r a t i n g the point-of-purchase information i n t h e i r 218 brand choices apparently were not deterred by signs p r o v i d i n g e i g h t cues, although t h i s load l e v e l has p r e v i o u s l y been associated with "overload" i n some l a b o r a t o r y studies on consumer information load' (see Table 7). The f i n d i n g s i n t h i s study do not support the Schroder-Driver-S t r e u f e r t model of information processing, s i n c e an inverted-U r e l a t i o n -ship between information load an decision-making was not observed. This f i n d i n g should be i n t e r p r e t e d with the understanding t h a t , i n the Schroder-Dri v e r - S t r e u f e r t model, load i s defined as an information input r a t e , i . e . , amount of information provided per u n i t of time. In the experiments from which t h i s model was developed, the r a t e of information input was an experimental f a c t o r over which l a b o r a t o r y subjects had no c o n t r o l . In t h i s study, consumers using the various amounts of product information provided apparently adjusted the information input r a t e by t a k i n g more time, i f necessary, to process whatever n u t r i t i o n a l information was provided. There i s no evidence i n t h i s experiment o f "information overload" i n the sense t h a t shoppers exposed to fewer cues on which to base t h e i r brand choices behaved d i f f e r e n t l y than shoppers exposed to the highest number of cues. Hypothesis H,,. Aggregate responses to the point-of-purchase signs did not depend upon the r e l a t i v e importance of i n d i v i d u a l n u t r i e n t s i n -cluded i n a s i g n . In f a c t , the tendency f o r shoppers to purchase n u t r i t i v e l y b e t t e r performing brands when subsets of the four most important n u t r i e n t s were used was apparently no greater than when subsets of the four l e a s t important n u t r i e n t s were present. The f i n d i n g s on t h i s hypothesis should be i n t e r p r e t e d c a u t i o u s l y . 219 F i r s t l y , the measures of r e l a t i v e importance were based on verbal responses obtained from consumers o f the t e s t products during an i n -dependent p e r s o n a l - i n t e r v i e w survey. I t i s q u i t e p o s s i b l e t h a t asking consumers to r a t e the importance to them of a set of n u t r i e n t s i n t h i s manner and basing measures of r e l a t i v e importance on these responses overstates the d i f f e r e n c e s i n r e l a t i v e importance between these cues. This i s e s p e c i a l l y l i k e l y when ot h e r , n o n - n u t r i t i o n a l cues are not included i n the r a t i n g task. As a r e s u l t , the true d i f f e r e n c e s i n r e l a t i v e importance between these n u t r i e n t s on which actual purchase behaviour would be based may be too small to d i f f e r e n t i a t e treatment e f f e c t s on the basis of n u t r i e n t importance. Secondly, there i s always the danger t h a t aggregated importance measures conceal i n d i v i d u a l d i f f e r e n c e s i n the importance assigned to a product cue. As was done i n t h i s research, i n d i v i d u a l responses to the cue-importance questions were aggregated across a l l survey respondents to come up with an o v e r a l l r e l a t i v e importance measure f o r the f i n a l set of eight cues. Considering the above caveats, t h i s f i e l d experiment found no d i f f e r e n c e s i n the d i s t r i b u t i o n of brand purchases which could be a t t r i -butable to d i f f e r e n c e s i n the r e l a t i v e importance of n u t r i e n t s l i s t e d on a pointrof-purchase s i g n . Hypothesis H^. No d i f f e r e n c e s i n brand choice behaviour appear to r e s u l t from a r e v e r s a l of the l e f t - t o - r i g h t arrangement o f e i g h t product cues on an information s i g n . I t was hypothesized that a brand-by-cue information matrix would have a greater impact on brand choices i f i t s e i g h t cues were arrayed 220 from l e f l > t o - r i g h t , i n decreasing order of importance, than i f those same cues were i n the reverse order. Since people c h a r a c t e r i s t i c a l l y read from l e f t to r i g h t , the cues l i s t e d i n decreasing order o f importance would be i n the " c o r r e c t h i e r a r c h i c a l order," assuming t h a t the aggregated measures of cue-importance are r e p r e s e n t a t i v e of most shoppers' im-portance o r d e r i n g s . In the reverse arrangement of the eight cues, consumers would have to search ( a t y p i c a l l y ) f o r the r e l a t i v e l y important cues on the r i g h t -hand s i d e of the information s i g n . The f i n d i n g s reveal t h a t these two d i f f e r e n t arrangements of the product cues on the signs d i s c l o s i n g e i g h t cues had no d i f f e r e n t i a l e f f e c t s on brand choices f o r any of the t e s t products. Hypothesis H^. The a n a l y s i s o f supermarket brand s a l e s data i n -d i c a t e s t h a t , f o r c e r t a i n products, consumers apparently do base t h e i r brand choices on n u t r i t i o n a l i n f o r m a t i o n , when i t i s provided i n a format which f a c i l i t a t e s interbrand comparisons. This f i n d i n g supports the contention o f many consumer researchers (e.g. Bettman, 1975; Day, 1976; Russo, 1977; W i l k i e , 1975) th a t the e f f e c t i v e n e s s of product information environments i s h i g h l y dependent on the format i n which such data i s organized f o r consumers, whichever way e f f e c t i v e n e s s may be defi n e d . In the present study, some n u t r i t i o n a l information on the brands o f c e r t a i n t e s t products was already a v a i l a b l e on i n d i v i d u a l packages of the product. Posting t h i s type of information on one point-of-purchase s i g n reduces the demands made on a consumer's short-term memory c a p a c i t y because i t f a c i l i t a t e s the interbrand comparisons required i n order to 221 a r r i v e at a brand choice. Moreover, information search and t h i n k i n g have c e r t a i n costs associated with them and these are reduced f o r the consumer c o n f r o n t i n g a s u p e r i o r information format. I t should be noted, however, th a t these f i n d i n g s were product-s p e c i f i c . Only on bran c e r e a l , macaroni & cheese dinner and canned soup was there a r e l i a b l e tendency f o r consumers to base t h e i r brand choices on the n u t r i t i o n a l data s u p p l i e d . Apparently, n e i t h e r ketchup nor mayonnaise brand choices were a f f e c t e d by the n u t r i t i o n a l information given on signs f o r these products. A number of s p e c u l a t i v e reasons can be advanced f o r these product d i f f e r e n c e s . F i r s t , the bran cereal and macaroni & cheese dinner brands (but not canned soup) are f o r t i f i e d with c e r t a i n vitamins and minerals and t h i s f a c t i s p u b l i c i z e d on the packages of the product. This p r a c t i c e may s e n s i t i z e consumers to a d d i t i o n a l n u t r i t i o n a l i n f o r m a t i o n , such as t h a t provided on the experimental s i g n s . By c o n t r a s t , n e i t h e r mayonnaise nor ketchup brands are promoted or enriched i n the manner described. Second, consumers may care l e s s about the n u t r i t i o n a l d i f f e r e n c e s between brands of ketchup and mayonnaise simply because these products are con-sumed i n smaller p o r t i o n s and are used mainly as a dressing or condiment. F i n a l l y , consumers may b e l i e v e that n u t r i t i o n a l d i f f e r e n c e s between a l t e r n a t i v e brands of ketchup and mayonnaise are small and may view the composition of competing brands as e s s e n t i a l l y the same. C e r t a i n l y , the between-brand d i f f e r e n c e s d i s c l o s e d by the n u t r i t i o n a l cues used on the experimental signs were g e n e r a l l y q u i t e s m a l l , r e l a t i v e to the other t e s t products. Hypothesis Hr. The a n a l y s i s of each t e s t product's brand s a l e s 222 d i s t r i b u t i o n i n post-experimental weeks tends to support the causal e f f e c t of the point-of-purchase information signs on consumers' brand choices.' With several exceptions, the dependent v a r i a b l e was found to l i e at i t s b a s e l i n e l e v e l f o r a product, f o l l o w i n g the permanent removal of the signs from the s t o r e s . Attempts were made to e x p l a i n the exceptions by r e f e r r i n g to the data c o l l e c t e d on extra-experimental "noise" which might have a f f e c t e d t e s t product brand sales during the five-week period i n the f i e l d . In a l l but one case, these d e v i a t i o n s from the expected shape of a product's brand s a l e s d i s t r i b u t i o n could be a t t r i b u t e d to a chain-wide promotion on a brand or an i n - s t o r e p r i c e i n c e n t i v e . Given these f i n d i n g s , some v a l i d i t y i s accorded to the conclusion t h a t , with c e r t a i n products and information treatments, consumers' brand choices were a f f e c t e d by exposure to the product i n f o r m a t i o n . A d d i t i o n a l Analyses. In the analyses which were designed to t e s t the research hypotheses, i t became apparent t h a t d i f f e r e n c e s i n the t e s t product used and d i f f e r e n c e s i n the experimental week explained more of the v a r i a t i o n i n the dependent measure than any experimental f a c t o r . F i r s t l y , separate analyses of product e f f e c t s made i t p o s s i b l e to d i s t i n g u i s h d i f f e r e n c e s i n consumer behaviour towards d i f f e r e n t food products. Confirming the e a r l i e r f i n d i n g s on hypothesis H^, consumers tended to u t i l i z e the n u t r i t i o n a l data on the point-of-purchase signs i n t h e i r brand choice d e c i s i o n s on canned soup, cereal and m&c dinner. In c o n t r a s t , the information provided on ketchup and mayonnaise tended to be ignored i n brand choices. 223 Secondly, the f i n d i n g s i n d i c a t e that consumers were l e s s responsive to the n u t r i t i o n a l information placed at the point of purchase i n the second week. Speculating on t h i s f i n d i n g , i t i s conceivable that the novelty of n u t r i t i o n a l information signs placed conspicuously i n the supermarket a i s l e s a t t r a c t e d consumer a t t e n t i o n and a correspondingly greater number of consumers used the signs i n t h e i r decision-making. In the f o l l o w i n g week, t h i s novelty had worn o f f and fewer shoppers paid . a t t e n t i o n to the data when purchasing the t e s t products. I m p l i c a t i o n s of the Findings The evidence from t h i s study of information load i n an actual market s e t t i n g i m p l i e s t h a t , w i t h i n the l i m i t e d load range experimented w i t h , consumers adapt t h e i r information processing a c t i v i t i e s to the amount of data at hand. There i s no basis f o r suspecting that consumers might have been "overloaded" with the number of product information cues which were made a v a i l a b l e to a i d i n brand choice-making. These f i n d i n g s apply to a one-time exposure, as was the o r i g i n a l i n t e n t of the research design. The f i n d i n g s represent evidence i n support of arguments by some researchers (e.g. Bettman, 1980; Haines, 1974; S t i l e s , 1974; Wilson, 1974) t h a t the actual consumption s e t t i n g i s too much u n l i k e the l a b o r a t o r y c o n d i t i o n s where the "information overload" model was developed and, t h e r e f o r e , "information overload" i s not l i k e l y to occur among buyers. S t i l e s (1974) provided e m p i r i c a l evidence that i n d u s t r i a l buyers w i l l a d j u s t the information input r a t e i n t h e i r e v a l u a t i o n of competing o f f e r s i n order not to become overloaded by a h i g h l y complex information processing 224 task. The f i n d i n g s from t h i s f i e l d experiment imply t h a t consumers f i n d ways of s i m i l a r l y a d j u s t i n g information inputs to cope with d i f f e r e n t amounts of product information i n the brand e v a l u a t i o n and choice process. In e v a l u a t i n g and drawing conclusions from these r e s u l t s , i t should be remembered that load was manipulated w i t h i n a s i n g l e format. This format i s deemed to be s u p e r i o r to the conventional method of p r o v i d i n g product information i n the r e t a i l s t o r e , which i s on i n d i v i d u a l packages. Thus, the presentation of product information i n common u n i t s i n a brand-b y - a t t r i b u t e matrix s i m p l i f i e s the brand performance comparison task f o r the consumer. Processing information i n such a format w i l l l i k e l y demand l e s s of the consumer's short-term memory ca p a c i t y than processing the same data placed on p h y s i c a l l y separate brand packages. The f i n d i n g t h a t a s i n g l e exposure of comparative product information had an immediate impact on brand choices f o r some products has policymaking and marketing i m p l i c a t i o n s with respect to format and type of i n f o r m a t i o n . F i r s t , i t i s o b v i o u s l y advantageous to the consumer to present o b j e c t i v e product performance data i n a brand-by-cue matrix format. While t h i s i s no guarantee t h a t the information w i l l be used, the f i n d i n g s here suggest that consumers w i l l apply even t e c h n i c a l information to t h e i r brand choice d e c i s i o n s . Moreover, they w i l l apparently do so as soon as the product data i s d i s p l a y e d , without needing time to l e a r n to use the d i s p l a y . Second, i t i s a l s o advantageous to the marketer r e t a i l i n g a brand which performs h i g h l y on a number of a t t r i b u t e s to place t h i s comparative data alongside t h a t of competitors' brand data i n a brand-by-cue matrix > format at the point of s a l e . This p r a c t i c e i s already common i n a d v e r t i s i n g but, apparently f o r c o n f l i c t - o f - i n t e r e s t reasons, i s not so 225 widely adopted i n the r e t a i l s t o r e . With respect to type of product i n f o r m a t i o n , i t i s c l e a r t h a t con-sumer response to a s i n g l e exposure o f n u t r i t i o n a l information on non-s t a p l e grocery foods i s l e s s than overwhelming. Whether the p r o v i s i o n of such information i n s i d e the supermarket f o r longer periods w i l l r e s u l t i n greater usage of the data i n brand choices i s an e m p i r i c a l question p r e s e n t l y being addressed by other researchers (Russo, 1981a). However, an i n t e r e s t i n g f i n d i n g was the apparent d i f f e r e n c e s i n consumer s e n s i t i v i t y to n u t r i t i o n a l data from product to product. Two of the three t e s t products (soup, m&c dinner, c e r e a l ) whose brand s a l e s d i s t r i b u t i o n s were seemingly a f f e c t e d by the point-of-purchase i n f o r m a t i o n , are already f o r t i f i e d with vitamins and minerals and t h i s f a c t i s promoted on the packages of i n d i v i d u a l brands. One of these products, macaroni & cheese dinner, co n s i s t e d of three brands: a p r i v a t e brand owned by the supermarket chain and two n a t i o n a l brands, one of which was the le a d i n g brand i n market share. The p r i v a t e brand was somewhat cheaper than the le a d i n g brand (3t per package, or about 8%). However, according to the information posted on the s t o r e signs i t was n u t r i t i v e l y i n f e r i o r to the l e a d i n g brand. Of i n t e r e s t was whether consumers would forgo the lower p r i c e o f the p r i v a t e brand and switch to a n u t r i t i v e l y s u p e r i o r n a t i o n a l brand. The f i n d i n g s i n d i c a t e t h a t consumers did pay a small premium i n return f o r a n u t r i t i v e l y b e t t e r performing food product. With the t e s t product canned soup, the s i t u a t i o n was e x a c t l y r e -versed: of the two brands, the p r i v a t e l a b e l was o s t e n s i b l y n u t r i t i v e l y s u p e r i o r to the h e a v i l y a d v e r t i z e d n a t i o n a l brand on a l l the p o i n t - o f -226 purchase s i g n s . The p r i v a t e brand was a l s o the cheaper of the two, by about 11% f o r comparable container s i z e s . Of i n t e r e s t was whether the brand image created by a h e a v i l y promoted n a t i o n a l brand could be breached by making i t appear n u t r i t i v e l y i n f e r i o r to a p r i v a t e brand using point-of-purchase i n f o r m a t i o n . The f i n d i n g s showed a r e l i a b l e tendency f o r the n u t r i t i o n a l information to erode the n a t i o n a l brand's market share as purchases s h i f t e d to the higher-performance p r i v a t e brand. F i n a l l y , the f i n d i n g t h a t shopper response to the point-of-purchase grocery product information was s i g n i f i c a n t l y weaker i n the second week of the experiment has i m p l i c a t i o n s f o r promotional s t r a t e g y with p o i n t -of-purchase information d i s p l a y s . I f there i s a "novelty e f f e c t " associated with the signs i n the f i r s t week i n which they appear, then marketers may e x p l o i t t h i s by not l e a v i n g the signs i n place f o r extended periods of time. Rather the appropriate s t r a t e g y would be to place the signs f o r a week or two, then remove them f o r a comparable p e r i o d , then replace them again, much l i k e the a d v e r t i s i n g scheduling s t r a t e g y of p u l s i n g . Research L i m i t a t i o n s I t i s recognized that t h i s f i e l d experiment has c e r t a i n l i m i t a t i o n s . L i m i t a t i o n s a r i s e from the type of data c o l l e c t e d i n the f i e l d and from the a n a l y t i c a l methods u t i l i z e d . Secondly, the experimental design has c e r t a i n shortcomings. Data and A n a l y s i s . The use of only two supermarkets of a s p e c i f i c chain i n a s i n g l e metropolitan area, i n order to c o l l e c t the d a t a , confines the g e n e r a l i z a b i l i t y of the research r e s u l t s . Inferences are based on the aggregate behaviour of consumers of grocery products exposed to n u t r i t i o n a l data on f i v e food products during a period of two weeks i n 1979. 227 The a n a l y t i c a l methods chosen were a p p l i c a b l e to aggregate purchase data. In p a r t i c u l a r , the technique o f r i d i t a n a l y s i s was used to summarize changes i n the brand s a l e s d i s t r i b u t i o n s observed i n the st o r e s . While t h i s method detects whether many i n d i v i d u a l customers s h i f t e d t h e i r brand choices to higher performing brands, i t does depend on the researcher's d e f i n i t i o n of r e l a t i v e brand performance. The issue here i s one of p r e d i c t i n g c o r r e c t l y consumers' pre-ferences, and then matching actual choices to those i n f e r r e d preferences. Given that d i r e c t i o n s of consumers' u t i l i t i e s f o r the i n d i v i d u a l n u t r i e n t s were i n f e r r e d from a consumer survey, the accuracy of t h i s experimenter's ranking of brands by n u t r i t i v e performance can be c r i t i c i z e d on the grounds that i t may not always represent an i n d i v i d u a l ' s ranking of brands by u t i l i t y . Research Design. Although care was taken to design an experiment which was appropriate to answer the research questions, a number of design concessions had to be made i n the p u r s u i t of f e a s i b i l i t y . The upper load l i m i t was set at eigh t cues. This and other load l e v e l s c l o s e l y p a r a l l e l e d the loads used i n various l a b o r a t o r y studies with consumer in f o r m a t i o n . At the same time, the maximum number of cues which could be used was l i m i t e d to the maximum number obtained from al1 cooperating manufacturers. Consequently, t h i s research could not explore the behavioural e f f e c t s of presenting higher loads to consumers on po i n t -of-purchase s i g n s . Another design l i m i t a t i o n i s imposed by the type of product i n -formation used, n u t r i t i o n a l v a l u e s , and the s p e c i f i c product c l a s s , packaged foods. A l s o , only a s i n g l e format design was employed 228 to manipulate information l o a d , thus e f f e c t s of t h i s v a r i a b l e remain unexplored i n t h i s study. F i n a l l y , i n order to c o n s t r u c t the s t i m u l i f o r the experimental f a c t o r s , i t was necessary to average the survey responses so as to come up with a s i n g l e ranking f o r r e l a t i v e cue importance. I f the assumption of r e l a t i v e l y homogeneous n u t r i e n t importances among these consumers i s suspect, the lack of f i n d i n g s on the cue-importance f a c t o r may a l s o be questioned. D i r e c t i o n s f o r Future Research A number of avenues f o r e x p l o r a t i o n of t h i s database remain. The lack of explanatory power i n the two experimental f a c t o r s , load and cue-importance, combined with the weak explanatory power due to d i f f e r e n c e s i n t e s t products and experimental weeks needs to be studied f u r t h e r . Then again, there may be a very l a r g e amount of "noise" i n the d a t a , defying e f f o r t s to f i n d new v a r i a b l e s to e x p l a i n more o f the variance i n the dependent measure. Some p r e l i m i n a r y work has begun along these l i n e s , but needs to be developed and r e f i n e d . A number of p o t e n t i a l v a r i a b l e s suggest them-se l v e s : 1 . Cue v a r i a b i l i t y ; the amount of between-brand d i f f e r e n c e s i n the brand r a t i n g s on a given cue. Can a measure be found to summarize the extent of between-brand d i f f e r e n c e s d i s c l o s e d by the data i n an information sign? W i l l t h i s v a r i a b l e e x p l a i n some of the mean r i d i t variance? 2. Cue determinance; the product of cue v a r i a b i l i t y x r e l a t i v e 229 cue importance, as measured from the survey responses. 3. The number of "negative" n u t r i e n t s included i n a p o i n t - o f -purchase s i g n , e.g., sodium, c a l o r i e s , f a t . In recent y e a r s , p u b l i c health o f f i c i a l s and the mass media have been drawing a t t e n t i o n to the f a c t t h a t c a r d i o - v a s c u l a r disease can be reduced by lowering the d i e t e t i c intake o f c e r t a i n types o f f a t s , t h a t o b e s i t y can be c o n t r o l l e d by reducing c a l o r i c i n t a k e , and that high l e v e l s of sodium i n the d i e t can c o n t r i b u t e to hypertension (see W a l l i s , 1982). Thus, these n u t r i e n t s have rece i v e d a l a r g e amount of negative p u b l i c i t y f o r several y e a r s , with the r e s u l t t h a t t h e i r d i s c l o s u r e on a sign may have drawn a corresponding amount of a t t e n t i o n from consumers, i n terms o f n u t r i e n t s to be s i n g l e d out and avoided. The r e l a t i o n s h i p between the mean r i d i t and the presence or absence o f data on these "negative" n u t r i e n t s on the signs has not y e t been i n v e s t i g a t e d . With regard to f u t u r e s t u d i e s , a research t r a d i t i o n i n f i e l d ex-perimentation to study consumer information processing and use needs to be developed. With the p r o l i f e r a t i o n of e l e c t r o n i c checkout f a c i l i t i e s i n supermarkets and other stores i t becomes e a s i e r to design f i e l d experiments and c o l l e c t data i n r e a l i s t i c b u y e r - s e l l e r environments. A s h i f t towards experimentation with product information on non-durables i s a l s o needed. F i n a l l y , the domain concerned with the c h a r a c t e r i s t i c s of o b j e c t i v e product information cues needs to be studi e d i n more d e t a i l . Besides l o a d , type and format there are other p r o p e r t i e s o f information which have been r e l a t i v e l y neglected. Best (1978) has suggested a number of these and each property warrants a t t e n t i o n from researchers i n t e r e s t e d i n how consumers process information. 230 Conclusion This d i s s e r t a t i o n has i n v e s t i g a t e d the information load aspect o f consumer information processing. A primary o b j e c t i v e was to t e s t an information processing model o f "information overload" developed i n the psychology l a b o r a t o r y and t e s t e d i n the consumer research l a b o r a t o r y , but not i n the marketplace. An experiment was conducted i n two supermarkets using o b j e c t i v e product information on several products. The f i n d i n g s do not support the "information o v e r l o a d " hypothesis but suggest t h a t consumers w i l l use such information i n t h e i r brand choices o f c e r t a i n products, when i t i s presented i n a format which f a c i l i t a t e s i n t e r - b r a n d comparisons. The f i n d i n g s are g e n e r a l l y weak and more research i s warranted to determine which aspects o f o b j e c t i v e product information are determinant i n the choice process. BIBLIOGRAPHY 232 - BIBLIOGRAPHY Asam, E.H. and.L.P. B u c k l i n (1973), " N u t r i t i o n Labeling f o r Canned Goods: A Study of Consumer Response," Journal of Marketing, 37 ( A p r i l ) , 32-37. Beales, H. et a l . (1981), "Consumer Search and P u b l i c P o l i c y , " Journal  of Consumer Research, 8 (June), 11-22. Best, R.J. (1978), "Choice Accuracy as a Function of Information Load, Choice Variance and D i s c r i m i n a t i o n Power," Working Paper, U n i v e r s i t y of A r i z o n a . Bettman, J.R. (1975), "Issues i n Designing Consumer Information Environ-ments," Journal of Consumer Research, 2 (December), 169-177. (1979), An Information Processing Theory of Consumer Choice. Reading: Addison-Wesley. and P. Kakkar (1977), " E f f e c t s of Information Presentation Format on Consumer Information A c q u i s i t i o n S t r a t e g i e s " , Journal of Consumer  Research, 3 (March), 233-240. Bross, I..D.J. (1958), "How to Use R i d i t A n a l y s i s , " BJometrics^ 14 (March), 18-38. Bymers, G. (1972), " S e l l e r - Buyer Communication: Point of View of a Family Economist," Journal of Home Economics, (February), 59-63. Byron, C. (1981), "Reining In the Regulators," Time, 117 (June 15), 52-53. Chaffee, S.H. and J.M. McLeod (1973), "Consumer Decisions and Information Use," i n Consumer Behavior: T h e o r e t i c a l Sources, S. Ward and T.S. Robertson, eds. Englewood C l i f f s , N.J.: P r e n t i c e - H a l l . C l a x t o n , J/.'D. and CD. Anderson (1980), "Energy Information at the Point of Sale: A F i e l d Experiment," i n Advances i n Consumer Research, V o l . 7, J e r r y C. Olson, ed. San F r a n c i s c o : A s s o c i a t i o n f o r Consumer Research, 277-282. Consumer Reports (1977), "Mayonnaise (And Things that Look L i k e i t ) , " 4 2 (March;), 148-151. Daly, P.A. (1976), "The Response of Consumers to N u t r i t i o n L a b e l i n g , " Journal of Consumer A f f a i r s , 10 ( W i n t e r ) , 170-178. Day, G.S. (1976), "Assessing the E f f e c t s of Information D i s c l o s u r e Requirements," Journal of Marketing, 40 ( A p r i l ) , 42-52. 233 Donnelly, J.G., J r . and M.J. Et z e l (1973), "Degrees o f Product Newness and e a r l y T r i a l , " Journal o f Marketing Research, 10 (August), 295-300. Doyle, P. and B.Z. Gidengil (1977), "A Review of In-Store Experiments," Journal of R e t a i l i n g , 53 (Summer), 47-61. Ferguson, G.A. (1976), S t a t i s t i c a l A n a l y s i s i n Psychology and Education, Fourth E d i t i o n , New York: McGraw-Hill. F l e i s s , J.L. (1981), S t a t i s t i c a l Methods f o r Rates and P r o p o r t i o n s , Second E d i t i o n , New York: John Wiley & Sons. Friedman, M.P. (1972), "Consumer Responses to Unit P r i c i n g , Open Dating, and N u t r i e n t L a b e l i n g , " i n Proceedings, T h i r d Annual Conference, A s s o c i a t i o n f o r Consumer Research, M. Venkatesan, ed. Chicago: A s s o c i a t i o n f o r Consumer Research, 361-369. (1977), "Consumer Use of Informational Aids i n Supermarkets," Journal of Consumer A f f a i r s , 11 ( 1 ) , 78-89. Goodwin, S. and M. Etgar (1980), "An Experimental I n v e s t i g a t i o n of Com-pa r a t i v e A d v e r t i s i n g : Impact of Message Appeal, Information Load, and U t i l i t y of Product C l a s s , " Journal of Marketing Research, 17 (May), 187-202. Gorman, W.P. (1975), "The Frightened Consumer?," Journal of R e t a i l i n g , 51 (Summer), 31-37. Haines, G.H. (1974), "A Prologue to Four Papers on Information Search," i n Buyer/Consumer Information Processing, G.D. Hughes and M.L. Ray, eds. ^Chapel H i l l : U n i v e r s i t y of North C a r o l i n a Press. Health and Welfare Canada (1977), Health P r o t e c t i o n and Food Laws, Ottawa. Henion, K.E. (1972), "The E f f e c t of E c o l o g i c a l l y Relevant Information on Detergent S a l e s , " Journal of Marketing Research, 9 (February), 10-14. Hutten, R.B., D.L. Mc N e i l l and W.L. W i l k i e (1978), "Some Issues i n Designing Consumer Information Studies i n P u b l i c P o l i c y , " i n Advances i n Con- sumer Research, V o l . 5, H.K. Hunt, ed. Ann Arbor: A s s o c i a t i o n f o r Consumer Research, 131-137. Jacoby, J . , R.W. Chestnut and W. Silberman (1977), "Consumer Use and Comprehension of N u t r i t i o n Information," Journal of Consumer  Research, 4 (September), 119-128. D.E. S p e l l e r and C.A. Kohn (1974a), "Brand Choice Behavior as a Function o f Information Load: R e p l i c a t i o n and Extension," Journal of Consumer Research, 1 (June), 33-42. 234 D.E. S p e l l e r and C.A. Kohn (1974b), "Brand Choice Behavior as a Function of Information Load," Journal of Marketing Research, 11 (February), 63-69. Jones, L.V. and D.W. Fiske (1953), "Models f o r Testing the S i g n i f i c a n c e of Combined R e s u l t s , " Psychological B u l l e t i n , .50 ( 5 ) , 375-378. K e n d a l l , K . W . and I. Fenwick (1979), "What Do You Learn Standing i n a Supermarket A i s l e ? , " i n Advances i n Consumer Research, V o l . 6, W i l l i a m L. W i l k i e , ed. . Miami Beach: A s s o c i a t i o n f o r Consumer Research, 153-160. Keppel, G. (1973), Design and A n a l y s i s : A Researcher's Handbook, Englewood C l i f f s , N.J.: P r e n t i c e - H a l l . Laroche, Michel (1980), Personal Communication. Lenahan, R.J., et a l . (1973), "Consumer Reaction to N u t r i t i o n a l Labels on Food Products," Journal of Consumer A f f a i r s , 7 ( 1 ) , 1-12. L i e f e l d , J . (1976), "Product Information Preference of Disadvantaged  Consumers," Ottawa: Consumer Research Council Canada. Martinsen, C S . and J . McCullough (1977), "Are Consumers Concerned About Chemical P r e s e r v a t i v e s i n Food?," Food Technology, 70 (September), 56-59. McEwen, W.J. (1978), "Bridging, the Information Gap," Journal of Consumer  Research, 4 (March), 247-251. M i l l e r , G.A. (1956), "The Magical Number Seven, Plus or Minus Two: Some L i m i t s on our Capacity f o r Processing Information," P s y c h o l o g i c a l Review, 63, 81-97. M i l l e r , J.A. (1977), "Federal Trade Commission A c t i v i t i e s Related to Consumer Information," Journal of Consumer P o l i c y , 1 ( W i n t e r ) , 62-76. Montgomery, G.F. (1977), "Product Technology and the Consumer," S c i e n t i f i c American, 237 ( 6 ) , 47-53. Murray, B.K. (1977), "Testing Consumers' Knowledge of Food and N u t r i t i o n , " Food Product Development, 11 ( J u l y - August), 15-17. Myers, J.H. and M.I. A-pert (1977), "Semantic Confusion i n A t t i t u d e Research: S a l i e n c e vs. Importance vs. Determinance," i n Advances  i n Consumer Research, V o l . 4, W i l l i a m D. P e r r e a u l t , J r . , ed. A t l a n t a : A s s o c i a t i o n f o r Consumer Research, 106-110. Nayak, P. and L . J . Rosenberg (1975), "Does Open Dating of Food Products B e n e f i t the Consumer?," Journal of R e t a i l i n g , 51 ( 2 ) , 10-20. 235 Nie, N.H. et a l . (1975), SPSS: S t a t i s t i c a l Package f o r the S o c i a l Sciences, 2nd E d i t i o n . New York: McGraw-Hill. Payne, J.W. (1976), " H e u r i s t i c Search Processes i n Decision Making," i n Advances i n Consumer Research, V o l . 3, B.B. Anderson, ed. C i n c i n n a t i : A s s o c i a t i o n f o r Consumer Research, 321-327. Peterson, E. et a l . (1978), "The Agency f o r Consumer P r o t e c t i o n : Pros and Cons," Journal of Home Economics, (January), 12-17. Peterson, R.A. (1977), "Consumer Perceptions as a Function o f Product C o l o r , P r i c e and N u t r i t i o n L a b e l i n g , " i n Advances i n Consumer  Research, V o l . 4, W.D. P e r r e a u l t , J r . , ed. A t l a n t a : A s s o c i a t i o n f o r Consumer Research, 61-63. Quelch, J.A. (1978), "Behavioral and A t t i t u d i n a l Measures of the R e l a t i v e Importance of Product A t t r i b u t e s : The Case of Cold Breakfast C e r e a l s , " Report No. 78-109. Cambridge, Mass.: Marketing Science I n s t i t u t e . R u d e l l , F. (1979), Consumer Food S e l e c t i o n and N u t r i t i o n Information. New York: Praeger P u b l i s h e r s . Russo, J.E. (1974), "More Information i s B e t t e r : A Reevaluation of Jacoby, S p e l l e r and Kohn," Journal of Consumer Research, 1 (December), 68-72. (1977), "The Value of Unit P r i c e Information," Journal of Marketing Research, 14 (May), 193-201. (1978), "Eye F i x a t i o n s Can Save the World: A C r i t i c a l E valuation and a Comparison Between Eye F i x a t i o n s and Other Information Processing Methodologies," i n Advances i n Consumer Research, V o l . 5, H.K. Hunt, ed. Ann Arbor: A s s o c i a t i o n f o r Consumer Research, 561-570. (1981a), P r i v a t e Correspondence. (1981b), Personal Communication. (undated), "A Measure of Total N u t r i t i o n f o r I n d i v i d u a l Foods." G. K r i e s e r and S. M i y a s h i t a (1975), "An E f f e c t i v e D i s play of Unit P r i c e Information," Journal of Marketing, 39 ( A p r i l ) , 11-19. and L.D. Rosen (1975), "An Eye F i x a t i o n A n a l y s i s of M u l t i - A l t e r n a t i v e Choice," Memory and C o g n i t i o n , 3 (May), 267-276. Scammon, D.L. (1977), "Information Load and Consumers," Journal o f Consumer  Research, 4 (December), 148-155. 236 Schrayer, D.W. (1978), "Consumer Response to N u t r i t i o n L a b e l i n g , " Food  Technology, 32 (December), 42-45. Schroder, H.M. (1971), "Conceptual Complexity and P e r s o n a l i t y O r g a n i z a t i o n , " i n P e r s o n a l i t y Theory and Information Processing, H.M. Schroder and P. Suedfeld, eds. New York: Ronald Press. , M.J. D r i v e r and S. S t r e u f e r t (1967), Human Information Processing. New York:Holt, Rinehart & Winston. S c o t t , W.A. and M. Wertheimer (1962), I n t r o d u c t i o n to Psychological  Research. New York: John Wiley & Sons, Inc. S f e g e l , S. (1956), Nonparametric S t a t i s t i c s f o r the Behavioral Sciences. New York: McGraw-Hill. Simon, H.A. (1974), "How Big i s a Chunk?," Science, 183 (February) 482-488. Snedecor, G.W. and W.G. Cochran (1967), S t a t i s t i c a l Methods, 6th ed. Ames, Iowa: Iowa State U n i v e r s i t y Press. S t a n l e y , T.J. (1977), " N u t r i t i o n a l Information and Preferences f o r Breakfast C e r e a l s , " Journal of Consumer A f f a i r s , 11 ( 2 ) , 121-126. S t i g l e r , G.J. (1961), "The Economics of Information," Journal of P o l i t i c a l  Economy, 69 (June), 312-225. S t i l e s , G.W. (1974), "Determinants of the I n d u s t r i a l Buyer's Level of Information Processing: O r g a n i z a t i o n s , S i t u a t i o n s , and I n d i v i d u a l D i f f e r e n c e s , " i n Buyer/Consumer Information Processing, G.D. Hughes and M.C. Ray, eds. Chapel H i l l : U n i v e r s i t y of North C a r o l i n a Press. S t r e u f e r t , S. (1970), "Complexity and Complex Decision Making," Journal of Experimental S o c i a l Psychology, 6, 494-509. , P. Suedfeld and M.J. D r i v e r (1965), "Conceptual S t r u c t u r e , Information Search, and Information U t i l i z a t i o n , " Journal of P e r s o n a l i t y and S o c i a l Psychology, 2 (November), 736-740. St. Marie, S. (1978), "Consumer A f f a i r s : P o t e n t i a l s f o r P r o f e s s i o n a l i s m , " Journal of Home Economics, (January), 18-20. Suedfeld, P. and R.L. Hagen (1966), "Measurement of Information Complexity: I. Conceptual S t r u c t u r e and Information Pattern as Factors i n Information Processing," Journal of P e r s o n a l i t y and S o c i a l Psycho- l o g y , 4 ( 2 ) , 233-236. Summers, J.O. (1974), "Less Information i s B e t t e r ? , " Journal of Marketing Research, 11 (November), 467-468. 237 T r a y l o r , M.B. (1981), "Comment on 'An Experimental I n v e s t i g a t i o n o f Comparative- A d v e r t i s i n g : Impact o f Message Appeal, Information Load, and U t i l i t y o f Product C l a s s , ' " Journal o f Marketing Research, 18 (May), 254-255. Tversky, A. (1972), " E l i m i n a t i o n by Aspects: A Theory o f Choice," Ps y c h o l o g i c a l Review, 79, 281-299. Venkatesan, M. (1977), " P r o v i d i n g N u t r i t i o n a l Information to Consumers," Paper presented at the s p e c i a l NSF/MIT Conference "Consumer Research f o r Consumer P o l i c y , " Cambridge, Mass.: J u l y , 28-29. Wall S t r e e t Journal (1978), June 9. W a l l i s , C. (1982), " S a l t : A New V i l l a i n ? " Time, 119 (March 15), 66-76. W i l k i e , W.L. (1975), "How Consumers Use Product Information: An Assess-ment of Research i n R e l a t i o n to P u b l i c P o l i c y Needs." Report prepared f o r National Science Foundation, Stock # 038-000-00237-6. (1974), "Analysis of E f f e c t s of Information Load", Journal of Marketing Research, 11 (November), 462-466. and P.W. F a r r i s (1976), "Consumer Information Processing: Per-spectives and I m p l i c a t i o n s f o r A d v e r t i s i n g , " Cambridge, Mass.: Marketing Science I n s t i t u t e , (August), Report No. 76-113. Wilson, D.T. (1974), "Models of Organization Buying Behaviour: Some .Observations," i n Buyer/Consumer Information P r o c e s s i n g , G.D. Hughes and M.L. Ray, eds. Chapel H i l l : The U n i v e r s i t y o f North C a r o l i n a Press, 136-141. Winer, B.J. (1971), S t a t i s t i c a l P r i n c i p l e s i n Experimental Design, 2nd e d i t i o n . New York: McGraw-Hill. APPENDIX A -REPLICAS OF "SIGN CONTROLS" AND POINT-OF-PURCHASE INFORMATION SIGNS CONSTRUCTED FOR EACH TREATMENT / PRODUCT C R E A M - O F - M U S H R O O M S O U P B R A N D S TOWN H O U S E C A M P B E L L ' S o M A C A R O N I & C H E E S E D I N N E R B R A N D S TOWN H O U S E K R A F T D E L U X E K R A F T D I N N E R o R E A L M A Y O N N A I S E ' B R A N D S B E S T F O O D S K R A F T TOWN H O U S E TOMATO K E T C H U P B R A N D S H E I N Z TOWN H O U S E B R A N - T Y P E B R E A K F A S T C E R E A L B R A N D S K E L L O G G ' S A L L -K E L L O G G ' S B R A N B U D S K E L L O G G ' S B R A N F L A K E S K E L L O G G ' S R A I S I N B R A N N A B I S C O B R A N C R U N C H I E S N A B I S C O 100% B R A N P O S T B R A N F L A K E S P E A N U T B U T T E R B R A N D S E M P R E S S (CHUNK STYLE) E M P R E S S (CREAMY SMOOTH) E M P R E S S (OLD-FASHIONED) K R A F T (SMOOTH) M c C O L L ' s S K I P P Y (CHUNKY) S K I P P Y (CREAMY) S U N N Y J I M (OLDE FASHIONED CREAMY) 0 0 CREAM-OF-MUSHROOM SOUP N u t r i t i o n a l Information Approximate Amount of Calcium In 100 Grams of Packaged Product* Brands Calcium mg TOWNHOUSE 60 CAMPBELL'S 22 1 * data obtained from manufacturer Treatment "1/high" CREAM-OF-MUSHROOM SOUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS CALCIUM PROTEIN VITAMIN B2 POTASSIUM mg g mg ng TOWN HOUSE 60 2.2 0.2 102 CAMPBELL'S 22 1.7 0.1 60 ii * data obtained from manufacturer Treatment "4/high" CREAM-0F-MUSHR00M SOUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF CALCIUM AND PROTEIN IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS CALCIUM PROTEIN mg 9 TOWN HOUSE 60 2.2 CAMPBELL'S 22 1.7 * data obtained from manufacturer Treatment "2/high" CREAM-OF-MUSHROOM SOUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT" CALCIUM PROTEIN VITAMIH B, POTASSIUM PHOSPHORUS CALORICS FAT SODIUM ng g mg mg "9 9 ' ™g TOWN HOUSE 60 2.2 0.2 102 57 71 3.7 792 CAMPBELL'S 22 1.7 0.1 60 31 115 8.7 923 • d a t a o b t a i n e d from mwmfaeturer Treatment " 8/high" CREAM-OF-MUSHROOM SOUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF SODIUM IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS SODIUM mg TOWN HOUSE 792 CAMPBELL'S 923 1° * data obtained from manufacturer T r e a t m e n t " 1 / l o w " CREAM-OF-MUSHROOM SOUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS SODIUM FAT CALORIES PHOSPHORUS mg g mg TOWN HOUSE 792 3.7 71 57 CAMPBELL'S 923 8.7 115 31 * data obtained from manufacturer T r e a t m e n t "4/1ow" CREAM-OF-MUSHROOM SOUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF SODIUM AND FAT IN 100 GRAMS OF PACKAGED PRODUCT * BRANDS SODIUM FAT mg 9 TOWN HOUSE 792 3.7 CAMPBELL'S 923 8.7 2° * data obtained from manufacturer T r e a t m e n t " 2 / l o w " CREAM-OF-MUSHROOM SOUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRANS OF PACKAGED PRODUCT* lOIUH FAT CALORIES PHOSPHORUS POTASSIUM VITAMIN 0 ; PROTEIN CALCIUM mg 9 mg mg mg g "9 TOWNHOUSE 792 3.7 71 57 102 0.2 2.2 60 CAMPBELL'S 923 8.7 115 31 60 0.1 1.7 22 1 d a t a o b t a i n e d from manufacturer T r e a t m e n t " 8 / l o w 1 REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF POLY-UNSATURATED FATS IN 100 GRAMS OF PACKAGED PRODUCT* POLY-BRANDS UNSATURATED FATS 9 BEST FOODS 21.7 KRAFT N/A** TOWN HOUSE 21.6 * d a t a o b t a i n e d f r o m m a n u f a c t u r e r ** d a t a n o t a v a i l a b l e T r e a t m e n t " 1 / h i g h ' REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BEST FOODS KRAFT TOWN HOUSE POLY-UNSATURATED FATS _ 2 _ 21.7 N/A" 21.6 1.1 1.1 1.3 TOTAL FATS 79 81 75 1.7 1.1 1.9 * data obtained from manufacturer data not available T r e a t m e n t " 4 / h i g h " REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF POLY-UNSATURATED FATS AND PROTEIN IN 100 GRAMS OF PACKAGED PRODUCT* POLY-BRANDS UNSATURATED PROTEIN FATS 9 9 BEST FOODS 21.7 1.1 KRAFT N/A** 1.1 TOWN HOUSE 21.6 1.3 * d a t a o b t a i n e d f r o m m a n u f a c t u r e r ** d a t a n o t a v a i l a b l e T r e a t m e n t " 2 / h i g h " REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT" POLV-BRANDS UNSATURATED PROTEIN TOTAL CARBO- SATURATED CALORIES SOOIUH SUGARS FATS FATS HYDRATE FATS 9 9 9 9 9 mg g BEST FOODS 21.7 1.1 79 1,7 5.6 710 590 N/A" KRAFT N/A*" 1.1 81 1.1 N/A" 718 516 N/A" TOWN HOUSE 21.6 1.3 75 1.9 2.6 703 561 1.1 • data obtained from manufacturer " data not available T r e a t m e n t " 8 / h i g h " ro ro REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF SUGARS . IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS SUGARS 9 BEST FOODS N/A** KRAFT N/A** TOWN HOUSE 1 .1 I" * data obtained from manufacturer ** data not available Treatment "1/low" REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS SUGARS SODIUM CALORIES SATURATED FATS _ BEST FOODS N/A" 590 710 5.6 KRAFT N/A" 516 718 N/A" TOWN HOUSE 1.1 561 703 2.6 • data obtained from arum facturer *• dita not a*oiIable Treatment "4/1ow" REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF SUGARS AND SODIUM IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS SUGARS SODIUM g mg BEST FOODS N/A** 590 KRAFT N/A** 516 TOWN HOUSE 1 . 1 561 2° * data obtained from manufacturer ** data not available Treatment "2/low" REAL MAYONNAISE NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS SUGARS SODIUM CALORIES SATURATED CARBO- TOTAL PROTEIN POLY-UNSATURATE! FATS HYDRATE FATS FATS 9 mg 9 9 9 9 9 BEST FOODS N/A" 590 710 5.6 1.7 79 1.1 21.7 KRAFT N/A" 516 718 N/A" 1.1 81 1.1 N/A" TOWN HOUSE 1.1 561 703 2.6 1.9 75 1.3 21.6 • data obtained from manufacturer data not available Treatment "8/low" TOMATO KETCHUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF IRON IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS IRON mg HEINZ 0.1 TOWN HOUSE 1.7 1 * d a t a o b t a i n e d from manufacturer T r e a t m e n t " 1 / h i g h " TOMATO KETCHUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS IRON PROTEIN CALORIES POTASSIUM mg g m 9 HEINZ 0.1 2 117 488 TOWN HOUSE 1.7 2 106 465 it * d a t a o b t a i n e d from m a n u f a c t u r e r T r e a t m e n t " 4 / h i g h " TOMATO KETCHUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF IRON AND PROTEIN IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS IRON PROTEIN mg g HEINZ 0.4 2 TOWN HOUSE 1.7 2 I 2 * d a t a o b t a i n e d from, m a n u f a c t u r e r I T r e a t m e n t " 2 / h i g h " TOMATO KETCHUP NUTRITIOHAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT" IRON PROTEIN CALORIES POTAS  I UN PHOSPHORUS SOOIUH CARBO-HYDRATE HEINZ 0.1 2 117 188 39 1288 27 0.0 TOWN HOUSE 1.7 2 106 165 36 1079 26 0.2 * data obtained from manufacturer T r e a t m e n t " 8 / h i g h " TOMATO KETCHUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF FAT IN IOO GRAMS OF PACKAGED PRODUCT* BRANDS FAT 9 HEINZ 0.0 TOWN HOUSE 0.2 1° * d a t a o b t a i n e d f r o m m a n u f a c t u r e r T r e a t m e n t " 1 / l o w " TOMATO KETCHUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS FAT CARBO- SODIUM PHOSPHORUS HYDRATE g g mg mg HEINZ 0.0 27 1288 39 TOWN HOUSE 0.2 26 1079 36 * d a t a o b t a i n e d f r o m m a n u f a c t u r e r T r e a t m e n t " 4 / l o w " TOMATO KETCHUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF FAT AND CARBOHYDRATE IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS FAT CARBO-HYDRATE 9 " 9 HEINZ 0.0 27 TOWN HOUSE 0.2 26 2° * d a t a o b t a i n e d f r o m m a n u f a c t u r e r T r e a t m e n t " 2 / l o w ' TOMATO KETCHUP NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS FAT CARBO- SODIUM PHOSPHORUS POTASIUM CALORIES PROTEIN IRON HYDRATE HEINZ 0.0 27 1288 39 188 117 2 0.1 TOWN HOUSE 0.2 26 1079 36 165 106 2 1.7' * data obtained from manufacturer T r e a t m e n t " 8 / l o w " MACARONI & CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF IRON IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE" BRANDS IRON mg TOWN HOUSE 0.1 KRAFT DELUXE 3.6 KRAFT DINNER 1.8 1 * ca lculated from, or obtained from manufacturers' data T r e a t m e n t " 1 / h i g h " MACARONI & CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE* BRANDS IRON PROTEIN VITAMIN B2 VITAMIN B, mg 9 mg mg TOWN HOUSE 0.1 1.8 0.1 ' 0.2 KRAFT DELUXE 3.6 6.9 1.5 0.1 KRAFT DINNER 1.8 5.3 0.7 0.1 * ca lcula ted from, or obtained from manufacturers* data T r e a t m e n t " 4 / h i g h " MACARONI 8 CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF IRON AND. PROTEIN IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE* I RON PROTEIN mg g TOWN HOUSE 0.1 1.8 KRAFT DELUXE 3.6 6.9 KRAFT DINNER 1.8 5.3 2 * c a l c u l a t e d from, or obtained from manufacturers' data T r e a t m e n t " 2 / h i g h " BRANDS MACARONI 8 CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE" IRON PROTEIN VITAMIN B- VITAMIN B. CALCIUM VITAMIN PHOSPHORUS FAT ' NIACIN TOWN HOUSE 0.1 1.8 0.1 0.2 21 2.6 51 10.8 KRAFT DELUXE 3.6 6.9 1.5 0.1 111 2.9 208 1.5 KRAFT DINNER 1.8 5.3 0.7 0.1 72 2.7 111 9.2 * calculated from, or obtained from manufacturers1 data T r e a t m e n t " 8 / h i g h " MACARONI S CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF FAT IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE* BRANDS FAT 9 TOWN HOUSE 10.8 KRAFT DELUXE 4.5 KRAFT DINNER 9.2 1° * calculated from, or obtained from manufacturers' data T r e a t m e n t " 1 / l o w " MACARONI & CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE* BRANDS FAT PHOSPHORUS VITAMIN CALCIUM NIACIN g mg mg mg TOWN HOUSE 10.8 51 2.6 21 " KRAFT DELUXE 1.5 208 2.9 114 KRAFT DINNER 9.2 111 2.7 72 * calculated from, or obtained from manufacturers' data T r e a t m e n t " 4 / l o w " MACARONI & CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF FAT AND PHOSPHORUS IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE* BRANDS FAT PHOSPHORUS g mg TOWN HOUSE 10.8 54 KRAFT DELUXE 4.5 208 KRAFT DINNER 9.2 111 2° * calculated from, or obtained \ from manufacturers' data T r e a t m e n t " 2 / l o w " MACARONI I CHEESE DINNER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PRODUCT, WHEN PREPARED ACCORDING TO DIRECTIONS ON PACKAGE * FAT PHOSPHORUS VI TAN IN CALCIUM VITAMIN B. VITAMIN B, PROTEIN IROH NIACIN TOWN HOUSE KRAFT DELUXE KRAFT DINNER 10.8 51 2.6 • 21 0,2 0.1 1.8 0.1 1,5 208 2.9 111 0.1 1.5 6.9 3.6 9.2 111 2.7 72 0.1 0.7 5.3 1.8 • calculated from, or obtained from manufacturers* data T r e a t m e n t " 8 / l o w 1 BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF FOOD FIBRE IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS FOD FIBRE (In alphabetical order) g KELLOGG'S ALL-BRAN 28 KELLOGG'S BRAN BUDS 23 KELLOGG'S BRAN FLAKES 10 KELLOGG'S RAISIN BRAN 8.8 NABISCO BRAN CRUNCHIES 9.3 NABISCO 100. BRAN 25 POST BRAN FLAKES N/A" 1 * data obtained from manufacturer *• data not available Treatment "1/high" BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT" BRANDS FOD PROTEIN VITAHIN B CALCIUM FIBRE NIACIN (in alphabetical order) g 9 "9 "9 KELLOGG'S ALL-BRAN 28 12 21 80 KELLOGG'S BRAN BUDS 23 12 21 89 KELLOGG'S BRAN FLAKES 10 10 21 17 KELLOGG'S RAISIN BRAN 8.8 8.8 21 53 NABISCO BRAN CRUNCHIES 9.3 10.3 21 18 NABISCO 1007. BRAN 25 11.3 23 87 POST BRAN FLAKES N/A" 10.1 21 13 k • data obtained from manufacturer data not available Treatment "4/high" BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF FOOD FIBRE AND PROTEIN IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS (in alphabetical order) FOD FIBRE g PROTEIN 9 KELLOGG'S ALL-BRAN 28 12 KELLOGG'S BRAN BUDS 23 12 KELLOGG'S BRAN FLAKES 10 10 KELLOGG'S RAISIN BRAN 8.8 8.8 NABISCO BRAN CRUNCHIES 9.3 10.3 NABISCO 1001 BRAN 25 11.3 POST BRAN FLAKES N/A" 10.1 2 * data obtained from manufacturer data not available Treatment "2/high" BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS FOD PROTEIN VITAMIN B CALCIUM MAGNESIUM POTASIUM PHOSPHORUS SODIUM (In alphabetical FIBRE NIACIN order) 9 g mg mg mg mg mg »g KELLOGG'S 2 8 n 2\ 80 395 715 930 623 ALL-BRAN KELLOGG'S 23 12 21 89 282 765 817 298 BRAN BUDS KELLOGG'S 1 0 w n m m 3 5 2 n50 6 9 7 BRAN FLAKES KELLOGCS 8 8 8 g 2 j 5 3 1 6 J 3 5 G „ 7 6 JQQQ RAISIN BRAN I ^ B i r m C c R A N 9 ' 3 1 0 • 3 2 1 m 1 6 6 1 , 9 5 m l m CRUNCHIES N A B I S C 0 25 11.3 23 87 131 1165 1120 620 1001 BRAN FLAKES'""1 N / A " 1 0 , 1 1 2 1 1 , 5 2 2 1 5 1 , 3 1 , 2 1 6 1 8 • data obtained from manufacturer data not available Treatment "8/high" ro oo BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF SODIUM IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS SODIUM (in alphabetical order) mg KELLOGG'S ALL-BRAN 623 KELLOGG'S BRAN BUDS 298 KELLOGG'S BRAN FLAKES 697 KELLOGG'S RAISIN BRAN 1000 NABISCO BRAN CRUNCHIES 610 NABISCO 100Z BRAN 620 POST BRAN FLAKES 618 1° • data obtained from manufacturer Treatment "1/low" BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS (in alphabetical order) SODIUM mg PHOSPHORUS POTASIUM mg MAGNESIUM mg KELLOGG'S ALL-BRAN 623 930 715 395 KELLOGG'S BRAN BUDS 298 ' 817 765 282 KELLOGG'S BRAN FLAKES 697 150 352 212 KELLOGG'S RAISIN BRAN 1000 176 358 163 NABISCO BRAN CRUNCHIES 610 161 195 166 NABISCO 100Z BRAN 620 1120 1165 131 POST BRAN FLAKES 618 121 513 221 " data obtained fron manufacturer Treatment "4/low" BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF SODIUM AND PHOSPHORUS IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS SODIUM PHOSPHORUS (in alphabetical order) mg mg KELLOGG'S ALL-BRAN 623 930 KELLOGG'S BRAN BUDS 298 817 KELLOGG'S BRAN FLAKES 697 150 KELLOGG'S RAISIN BRAN 1000 176 NABISCO BRAN CRUNCHIES 610 161 NABISCO 1001 BRAN 620 1120 POST BRAN FLAKES 618 121 • data obtained from manufacturer Treatment "2/low" BRAN-TYPE BREAKFAST CEREAL NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS (in alphabetical order) KELLOGG'S ALL-BRAN KELLOGG'S BRAN BUDS KELLOGG'S BRAN FLAKES KELLOGG'S RAISIN BRAN NABISCO BRAN CRUNCHIES NABISCO 100Z BRAN POST BRAN FLAKES 623 298 697 1000 610 620 618 PHOSPHORUS POTASIUM MAGNESIUM CALCIUM VITAMIN B PROTEIN FOD NIACIN FIBRE mg mg mg mg mg g g 930 715 817 765 150 352 176 358 161 195 1120 1165 121 513 395 282 212 163 166 131 221 89 17 53 18 87 13 21 21 21 21 21 23 21 12 28 12 23 10 10 8.8 8.8 10.3 9.3 11.3 25 10.1 N/A" Treatment "8/low data obtained from manufacturer data not available II PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF PROTEIN IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS "<>"'» (In alphabetical order) 9 EMPRESS (CHUNK STYLE) 25 EMPRESS (CREAMY SMOOTH) 25 EMPRESS (OLD-FASHIONED) 26 KRAFT (SMOOTH) 23 McCOLL's 28 SKIPPY (CHUNKY) 28 SKIPPY (CREAMY) 28 SUNNY JIM (OLDE FASHIONED CREAMY) 26 1 " data obtained from manufacturer Treatment "1/high" PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS PROTEIN I RON V1TAKI  VITAMIN (in alphabetical order) N1ACIK 0 mg mg mg 2 0.1 15 EMPRESS (CREAMY SMOOTH) 25 2 0.1 15 2 0.1 15 2 0.1 11 McCOLL's . . . 28 2 0.1 16 28 2 0.1 13 28 2 0.1 13 26 2 0.1 16 It * data obtained from manufacturer Treatment "4/high" PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF PROTEIN AND IRON IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS PROTEIN IRON (in alphabetical order) " EMPRESS (CHUNK STYLE) 25 2 EMPRESS (CREAMY SMOOTH) 25 2 EMPRESS (OLD-FASHIONED) 26 2 KRAFT (SMOOTH) 23 2 MCCOLL'S 28 2 SKIPPY (CHUNKY) 28 2 * SKIPPY (CREAMY) 28 2 SUNNY J I M (OLDE FASHIONED CREAMY) 26 2 Treatment "2/high" PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT* BRANDS PROTEIN I RON VITAMIN B ; Jj™J" POTASSIUH PHOSPHORUS FAT CALORIES (in alphabetical order) g 1119 mg mg mg mg g EMPRESS (CHUNK STYLE) 25 2 0.1 15 625 385 19 581 EMPRESS (CREAMY SMOOTH) 25 2 0.1 15 625 385 19 581 EMPRESS (OLD-FASHIONED) 26 2 0.1 15 638 393 50 583 KRAFT (SMOOTH) 23 2 0.1 11 611 314 19 582 McCOLL's 28 2 0.1 16 670 107 19 580 SKIPPY (CHUNKY) 28 2 0.1 13 670 360 52 590 SKIPPY (CREAMY) 28 2 0.1 13 670 370 50 590 SUNNY JIM ( o i o E FASHIONED CREAHY) ., 26 2 . 0.1 16 670 107 18 598 8 • data obtained from manufacturer Treatment "8/high" £ o PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNT OF CALORIES IN }O0 GRAMS OF PACKAGED PRODUCT' BRANDS CALORIES (in alphabetical order) EMPRESS (CREAMY SMOOTH) 581 582 580 SUNNY JIM (OLDE FASHIONED CREAMY) 598 1° • data obtained from manufacturer Treatment "1/low" PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS CALORIES FAT PHOSPHORUS POTASSIUM (in alphabetical order) 9 mg mg 19 385 625 581 19 385 625 50 393 638 19 311 611 19 107 670 SKIPPY (CHUNKY) 52 360 670 50 370 670 SUNNY JIM (OLDE FASHIONED CREAMY) . 18 107 670 L° • data obtained from manufacturer Treatment "4/low" PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF CALORIES AND FAT IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS CALORIES FAT (in alphabetical order) g 19 19 50 19 19 52 50 SUNNY JIM (OLDE FASHIONED CREAMY) 598 18 2° " data obtained from anufacturer Treatment "2/low" PEANUT BUTTER NUTRITIONAL INFORMATION APPROXIMATE AMOUNTS OF NUTRIENTS IN 100 GRAMS OF PACKAGED PRODUCT' BRANDS CALORIES FAT PHOSPHORUS POTASSIUH VITAMIN VITAMIN 9. IRON PROTEIN NIACIN (in alphabetical order) g mg mg mg mg 9 581 19 385 625 15 0.1 2 25 581 19 385 625 15 0.1 2 • 25 583 50 393 638 15 p . l 2 26 582 19 311 611 11 0.1 2 23 580 19. 107 670 16 0.1 2 28 SKIPPY (CHUNKY) 590 52 360 670 13 0.1 2 28 590 50 370 670 13 0.1 2 28 SUNNY JIM (OLDE FASHIONED CREAHY) 598 18 107 670 16 0.1 2 26 8° data obtained from ma nufcturer Treatment "8/low" APPENDIX B NUTRITIVE COMPOSITION DATA OBTAINED FROM MANUFACTURERS APPENDIX B NUTRITIVE COMPOSITION DATA OBTAINED FROM MANUFACTURERS Table B - l N u t r i t i v e Composition Data Obtained From Manufacturers o f Canned Cream-of-Mushroom Soup Per IQOg o f Brand Nutr i e n t A B C a l o r i e s 71 115 P r o t e i n g 2.2 1.7 Fat g 3.7 8.7 Carbohydrate g 7.6 7.5 Calcium mg 60 22 Phosphorus mg 57 31 Iron mg 0.1 0.4 Sodium mg 792 923 Potassium mg 102 60 Vitamin A IU 67 79 Thiamine mg 0.02 0.01 R i b o f l a v i n ( B 2) mg 0.16 0.08 Nia c i n mg 0.5 0.6 Brand A: Town House Brand B: Campbell's Table B-2 N u t r i t i v e Composition Data Obtained From Manufacturers of Tomato Ketchup Per lOOg of Brand N u t r i e n t A B C a l o r i e s 117 106 P r o t e i n g 1.8 1.7 Fat 9 0.0 0.2 Carbohydrate g 26.8 26.4 Calcium mg 48 17 Phosphorus mg 39 36 Iron mg 0.4 1.7 Potassium mg 488 465 Sodium mg 1288 1079 Brand A: Heinz Brand B: Town House Table B-3 N u t r i t i v e Composition Data of Prepared Macaroni & Cheese Dinner Obtained From Manufacturers Per lOOg of Brand Nu t r i e n t A a B C a l o r i e s 194 163 193 P r o t e i n g 4.8 6.9 5.3 Fat g 10.8 4.5 9.2 Carbohydrate g 22.3 23.2 22.1 N i a c i n mg 2.6 2.87 2.70 R i b o f l a v i n (B 2 ) mg .39 1:48 . 0.67 Thiamine (B^) mg .23 0.44 0.36 Calcium 21 114 ,72 Phosphorus mg 54 208 111 Iron mg .38 3.56 1.8 a V a l u e s f o r t h i s brand c a l c u l a t e d by author, using data supplied by manufacturer and f o l l o w i n g d i r e c t i o n s f o r preparations on package. Brand A: Town House Brand B: K r a f t Deluxe Brand C: K r a f t Dinner Table B-4 N u t r i t i v e Composition Data Obtained From Manufacturers of Mayonnaise Per IQOg of Brand N u t r i e n t A B C C a l o r i e s 710 718 703 P r o t e i n 9 1.1 1.1 1.3 Poly-unsaturated Fats g 21.7 N/Aa 21.6 Saturated Fats g 5.6 N/Aa 2.6 Total Fats g 78.8 80.6 74.9 Carbohydrate g 1.7 1.1 1.9 Sugars g N/Aa N/Aa 1.4 Sodium mg 590 516 561 a T h i s data was not a v a i l a b l e from manufacturer. Brand A: Best Foods Brand B: K r a f t Brand C: Town House 257 Table B-5 N u t r i t i v e Composition Data Obtained From Manufacturers of Bran-Type Breakfast- Cereal Nutrients Per lOOg of Brand A B C D E F G C a l o r i e s 257 286 315 284 339 257 339 P r o t e i n 9 12 12 10 8.8 10.3 11.3 10.4 Fat g 2.0 1.8 1.6 1.4 , 0.6 1.6 2.1 Carbohydrate g 76 80 80 78 83.3 82.6 77.9 Fi bre g 28 23 10 8.8 9.3 25 N/Aa Starch g 24 24 55 45 56.5 32.9 N/Aa Sugars g 23 22 14 24 17.5 24.7 N/Aa Magnesium mg 395 282 212 163 166 431 221 Cal c i um mg 80 89 47 53 48 87 43 Phosphorus mg 930 847 450 476 461 1120 421 Iron mg 14 14 18 14 14.3 14.3 21.4 Potassium mg 715 765 352 358 495 1165 543 Sodium mg 623 298 697 1000 640 620 618 Thiamine mg 2.1 2.1 2.1. 2.1 2.2 2.2 2.1 R i b o f l a v i n mg 3.6 3.6 3.6 3.6 3.6 3.6 3.6 N i a c i n mg 21 21 21 21 21.4 22.5 21.4 This data not a v a i l a b l e from manufacturer Brand A: Kellogg's A l l - B r a n Brand B: K e l l o g g 1 s Bran Buds Brand G: Kellogg.'s Bran Flakes Brand D: Kellogg's R a i s i n Bran Brand E: Nabisco Bran Crunchies Brand F: Nabisco 100% Bran Brand G: Post Bran Flakes 258 Table B-6 N u t r i t i v e Composition Data Obtained From Manufacturers o f Peanut Butter Per lOOg of Brand Nutr i e n t A B C D E F G H I C a l o r i e s 582 598 590 582 590 582 584 583 580 P r o t e i n g 23.1 26.2 28.4 25.0 28.3 25.3 25.2 25.7 27.9 Total Fat 9 48.8 48.0 49.8 49.0 51.7 49.1 49.1 50.1 49.3 Total Carbohydrate g 22.5 15.7 15.9 21.0 14.2 20.8 19.9 19.5 17.2 Ni a c i n mg 13.9 15.7 12.9 16 12.7 16.1 14.7 14.9 15.7 R i b o f l a v i n ( B 2 ) mg 0.11 0.13 0.07 0.12 0.07 0.12 0.12 0.12 0.13 Thiamine mg 0.26 0.13 0.07 0.30 0.07 0.30 0.12 0.12 0.13 Calcium mg 68 63 40 68 35 69 59 60 63 Phosphorus mg 344 407 370 370 360 373 385 393 407 Iron mg 1.7 2.0 1.6 2.0 1.7 2.0 2.1 2.1 2.0 Magnesium mg 152 173 170 200 170 202 178 183 185 Sodium mg 451 1000 470 550 470 496 570 581 607 Potassium mg 644 670 670 620 670 625 625 638 670 Brand A: K r a f t Smooth Brand F: S q u i r r e l Crunchy Brand B: Sunny Jim Brand G: Empress Creamy/Chunk Brand C: Skippy Creamy Brand H: Empress Old Fashioned Brand D: S q u i r r e l Smooth Brand I: McColl's Brand E: Skippy Super Chunk APPENDIX C SURVEY QUESTIONNAIRE AND STORE PATRONAGE RESPONSES 260 SURVEY QUESTIONNAIRE GOOD MORNING (EVENING). MY NAME IS I'M SURVEYING HOUSEHOLDS FOR THE UNIVERSITY OF BRITISH COLUMBIA SHOPPER RESEARCH. THIS RESEARCH IS SPONSORED BY THE DEPARTMENT OF CONSUMER AND CORPORATE AFFAIRS IN OTTAWA. DO YOU GENERALLY DO THE GROCERY SHOPPING FOR YOUR HOUSEHOLD? ( i f " n o " , ask to see t h a t p e r s o n , and read i n t r o d u c t i o n from beg inn ing) I'D LIKE YOU TO ANSWER SOME QUESTIONS ABOUT GROCERY SHOPPING AND CONSUMER INFORMATION ON GROCERY FOODS. ( i f respondent r e f u s e s t o c o - o p e r a t e , thank respondent and l e a v e . Then go to Quest ion ( ? ) and complete t h i s q u e s t i o n y o u r s e l f . Do t h i s be fo re making the next i n t e r v i e w . Use a f r e s h s e t o f q u e s t i o n n a i r e s f o r each attempted i n t e r v i e w — even i f on 1y Quest ion (? ) has been comple ted . Always a l t e r n a t e between white and pink set o f q u e s t i o n n a i r e s , from house to house) 1 SURVEY QUESTIONNAIRE U.B.C. SHOPPER RESEARCH T) (Start with product on top sheet and ask question (A) for product on this and a l l following sheets. Then return to question (jf) on top sheet. If any product is purchased less often than once in A months (16 weeks), cross i t out and go to next product. If a l l products have been deleted in this way, terminate interview after asking question (c)) HOW OFTEN DO YOU BUY (name product in question (5) on this sheet)? (go to next product) (I) (Skip any products you have crossed out. If this paper is wh i te (pink), hand respondent whi te_(pink) l i s t . After a l l e l i g i b l e products have been covered, go to question (f) ) @ (1) I'D LIKE YOU TO THINK YOU ARE SHOPPING FOR (name product). ON THIS LIST ARE SOME NUTRITIONAL QUALITIES OF (name product). A PERSON MIGHT COMPARE DIFFERENT (name product) BRANDS ON THESE NUTRITIONAL QUALITIES BEFORE BUYING. SUPPOSE YOU WERE GIVEN SUCH INFORMATION ON ALL THE BRANDS, WHICH SINGLE ONE OF THESE WOULD BE MOST IMPORTANT IN HELPING YOU DECIDE ON A BRAND TO BUY7~7enter "10" opposite R's choice - i f R cannot reply or declares none to be meaningful, enter zeros in a l l the blanks and go to next product) PEANUT BUTTER VITAMIN B 2 LOTS OF LITTLE POTASSIUM LOTS OF LITTLE PROTEIN LOTS OF LITTLE SODIUM LOTS OF LITTLE CALORIES LOTS OF LITTLE IRON LOTS OF LITTLE PHOSPHORUS LOTS OF LITTLE VITAMIN NIACIN LOTS OF LITTLE FAT LOTS OF LITTLE SINCE (name R's choice) IS THE MOST IMPORTANT ONE TO YOU, LET'S GIVE IT A SCORE OF "10". NOW, FOR EACH OF THE REMAINING NUTRITIONAL QUALITIES, I WILL ASK THE FOLLOWING. HOW IMPORTANT IS EACH QUALITY IN HELPING YOU DECIDE WHICH BRAND TO BUY? LET'S USE A SCORE FROM ZERO TO 10. ZERO MEANS IT HAS NO IMPORTANCE TO YOU. TEN MEANS IT IS JUST AS IMPORTANT ASHjilme i n i t i a l choice). YOU CAN USE THE SAME SCORE MORE THAN ONCE IF YOU FEEL THAT TWO OR MORE NUTRITIONAL QUALITIES ARE EQUALLY IMPORTANT TO YOU. (read out nutrients, one at a time, from top to bottom and enter scores above) (?) (2) LET'S GO BACK OVER THESE NUTRITIONAL QUALITIES. IF YOU WERE CHOOSING A BRAND OF (name product) TO BUY, WOULD YOU, YOURSELF, GENERALLY WANT THE BRAND WHICH HAS LOTS OF OR LITTLE OF EACH OF THESE? (read out nutrients, one at a time, from top to bottom and c i r c l e appropriate word above. If respondent says neither, enter any comments opposite that nutrient) (B) (3) WHICH BRAND OF (name product) DO YOU USUALLY BUY? (If R mentions more than one brand, record them a l l in the order mentioned. Go to next product) SURVEY QUESTIONNAIRE  U.B.C. SHOPPER RESEARCH (Xs (Start with product on top sheet and ask question (A^ for product on this and a l l following sheets. Then return to question (By on top sheet. If any product is purchased less often than once in k months (16 weeks), cross i t out and go to next product. If a l l products have been deleted in this way, terminate interview after asking question (c)) HOW OFTEN DO YOU BUY (name product in question on this sheet)? (go to next product) (B) (Skip any products you have crossed out. If this paper is wh i te (pink), hand respondent wh i te _(p i nk) l i s t . After a l l e l i g i b l e products have been covered, go to ques t i on (C^ ) ( ? (1) I'D LIKE YOU TO THINK YOU ARE SHOPPING FOR (name product). ON THIS LIST ARE SOME NUTRITIONAL QUALITIES OF (name product). A PERSON MIGHT COMPARE DIFFERENT (name product) BRANDS ON THESE NUTRITIONAL QUALITIES BEFORE BUYING. SUPPOSE YOU WERE GIVEN SUCH INFORMATION ON ALL THE BRANDS, WHICH SINGLE ONE OF THESE WOULD BE MOST IMPORTANT IN HELPING YOU DECIDE ON A BRAND TO BUY?~[e~nter "10" opposite R's choice - i f R cannot reply or declares none to be meaningful, enter zeros in a l l the blanks and go to next product) TOMATO KETCHUP _SUGAR _CARB0HYDRATE _PH0SPH0RUS _FAT J RON _P0TASSIUM _PR0TE I N _S ODIUM CALORIES LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE SINCE (name R's choice) IS THE MOST IMPORTANT ONE TO YOU, LET'S GIVE IT A SCORE OF "10". NOW, FOR EACH OF THE REMAINING NUTRITIONAL QUALITIES, I WILL ASK THE FOLLOWING. HOW IMPORTANT IS EACH QUALITY IN HELPING YOU DECIDE WHICH BRAND TO BUY? LET'S USE A SCORE FROM ZERO TO 10. ZERO MEANS IT HAS NO IMPORTANCE TO YOU. TEN^MEANS IT IS JUST AS IMPORTANT AS~rn7me i n i t i a l choice). YOU CAN USE THE SAME SCORE MORE THAN ONCE IF YOU FEEL THAT TWO OR MORE NUTRITIONAL QUALITIES ARE EQUALLY IMPORTANT TO YOU. (read out nutrients, one at a time, from top to bottom and enter scores above) 0 (2) LET'S GO BACK OVER THESE NUTRITIONAL QUALITIES. IF YOU WERE CHOOSING A BRAND OF (name product) TO BUY, WOULD YOU, YOURSELF, GENERALLY WANT THE BRAND WHICH HAS LOTS OF OR LITTLE OF EACH OF THESE? (read out nutrients, one at a time, from top to bottom and c i r c l e appropriate word above. If respondent says neither, enter any comments opposite that nutrient) (?) (3) WHICH BRAND OF (name product) DO YOU USUALLY BUY? ( i f R mentions more than one brand, record them alt in the order mentioned. Go to next product) , SURVEY QUESTIONNAIRE U.B.C. SHOPPER RESEARCH (Start with product on top sheet and ask question (A) for product on this and a l l following sheets. Then return to question (B; on top sheet. If any product is purchased less often than once in k months (16 weeks), cross i t out and go to next product. If a l l products have been deleted in this way, terminate interview after asking question (x)) HOW OFTEN DO YOU BUY (name product in question (fT on this sheet)? (go to next product) (Skip any products you have crossed out. If this paper is wh i te (pink), hand respondent white (pink) l i s t . After a l l e l i g i b l e products have been covered, go to question (C_,< ) (1) I'D LIKE YOU TO THINK YOU ARE SHOPPING FOR (name product). ON THIS LIST ARE SOME NUTRITIONAL QUALITIES OF (name product). A PERSON MIGHT COMPARE DIFFERENT (name product) BRANDS ON THESE NUTRITIONAL QUALITIES BEFORE BUYING. SUPPOSE YOU WERE GIVEN SUCH INFORMATION ON ALL THE BRANDS, WHICH SINGLE ONE OF THESE WOULD BE MOST IMPORTANT IN HELPING YOU DECIDE ON A BRAND TO BUY?~Tenter "10" opposite R's choice - i f R cannot reply or declares none to be meaningful, enter zeros in a l l the blanks and go to next product) BRAN-TYPE BREAKFAST CEREAL _CAL0RIES _CALCIUM _MAGNESIUM _P ROTE IN _PH0SPH0RUS _F00D FIBRE _P0TASSIUM JODIUM VITAMIN NIACIN LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE SINCE (name R's choice) IS THE MOST IMPORTANT ONE TO YOU, LET'S GIVE IT A SCORE OF "10". NOW, FOR EACH OF THE REMAINING NUTRITIONAL QUALITIES, I WILL ASK THE FOLLOWING. HOW IMPORTANT IS EACH QUALITY IN HELPING YOU DECIDE WHICH BRAND TO BUY? LET'S USE A SCORE FROM ZERO TO 10. ZERO MEANS IT HAS NO IMPORTANCE TO YOU. TEN_ MEANS IT IS JUST AS IMPORTANT AS~Tname i n i t i a l choice). YOU CAN USE THE SAME SCORE MORE THAN ONCE IF YOU FEEL THAT TWO OR MORE NUTRITIONAL QUALITIES ARE EQUALLY IMPORTANT TO YOU. (read out nutrients, one at a time, bottom and enter scores above) from top to (2) LET'S GO BACK OVER THESE NUTRITIONAL QUALITIES. IF YOU WERE CHOOSING A BRAND OF (name product) TO BUY, WOULD YOU, YOURSELF, GENERALLY WANT THE BRAND WHICH HAS LOTS OF OR LITTLE OF EACH OF THESE? (read out nutrients, one at a time, from top to bottom and c i r c l e appropriate word above. If respondent says neither, enter any comments opposite that nutrient) (3) WHICH BRAND OF (name product) DO YOU USUALLY BUY? (If R mentions more than one brand, record them a l l in the order mentioned. Go to next product) SURVEY QUESTIONNAIRE  k U.B.C. SHOPPER RESEARCH (Start with product on top sheet and ask question (A) for product on this and a i l following sheets. Then return to question (?) on top sheet. If any product is purchased less often than once in h months (16 weeks), cross i t out and go to next product. If a l l products have been deleted in this way, terminate interview after asking question (c)) HOW OFTEN DO YOU BUY (name product in question (T on this sheet)? (go to next product) (Skip any products you have crossed out. If this paper is wh i te (pink), hand respondent white (pink) l i s t . After a l l e l i g i b l e products have been covered, go to question [C_, ) (1) I'D LIKE YOU TO THINK YOU ARE SHOPPING FOR (name product). ON THIS LIST ARE SOME NUTRITIONAL QUALITIES OF (name product). A PERSON MIGHT COMPARE DIFFERENT (name product) BRANDS ON THESE NUTRITIONAL QUALITIES BEFORE BUYING. SUPPOSE YOU WERE GIVEN SUCH INFORMATION ON ALL THE BRANDS, WHICH SINGLE ONE OF THESE WOULD BE MOST IMPORTANT IN HELPING YOU DECIDE ON A BRAND TO BUY?~Te"nter "10" opposite R's choice - i f R cannot reply or declares none to be meaningful, enter zeros in a l l the blanks and go to next product) MAYONNAISE-TYPE SALAD DRESSING PROTEIN LOTS OF POLY-UNSATURATED LOTS OF "FATS SATURATED FATS TOTAL FATS CALORIES _S0DIUM SUGARS CARBOHYDRATE LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LOTS OF LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE LITTLE SINCE (name R's choice) IS THE MOST IMPORTANT ONE TO YOU, LET'S GIVE IT A SCORE OF "10". NOW, FOR EACH OF THE REMAINING NUTRITIONAL QUALITIES, I WILL ASK THE FOLLOWING. HOW IMPORTANT IS EACH QUALITY IN HELPING YOU DECIDE WHICH BRAND TO BUY7 LET'S USE A SCORE FROM ZERO TO 10. ZERO MEANS IT HAS NO IMPORTANCE TO YOU. TEN_ MEANS IT IS JUST AS IMPORTANT AS~rname i n i t i a l choice). YOU CAN USE THE SAME SCORE MORE THAN ONCE IF YOU FEEL THAT TWO OR MORE NUTRITIONAL QUALITIES ARE EQUALLY IMPORTANT TO YOU. (read out nutrients, one at a time, from top to bottom and enter scores above) (2) LET'S GO BACK OVER THESE NUTRITIONAL QUALITIES. IF YOU WERE CHOOSING A BRAND OF (name product) TO BUY, WOULD YOU, YOURSELF, GENERALLY WANT THE BRAND WHICH HAS LOTS OF OR LITTLE OF EACH OF THESE? (read out nutrients, one at a time, from top to bottom and c i r c l e appropriate word above. If respondent says neither, enter any comments opposite that nutrient) (3) WHICH BRAND OF (name product) DO YOU USUALLY BUY? (If R mentions more than one brand, record them a l l in the order mentioned. Go to next product) r SURVEY QUESTIONNAIRE U.B.C. SHOPPER RESEARCH (A^ (Start with product on top sheet and ask question (A) for product on this and a l l following sheets. Then return to question (B) on top sheet. If any product is purchased less often than once in 4 months (16 weeks), cross i t out and go to next product. If a l l products have been deleted in this way, terminate interview after asking question (£)) HOW OFTEN DO YOU BUY (name product in question (T on this sheet)? (go to next product) cT (i) (Skip any products you have crossed out. If this paper is whi te (pink), hand respondent whj_te (pink) l i s t . After a l l e l i g i b l e products have been covered, go to question {tj ) I'D LIKE YOU TO THINK YOU ARE SHOPPING FOR (name product). ON THIS LIST ARE SOME NUTRITIONAL QUALITIES OF (name product). A PERSON MIGHT COMPARE DIFFERENT (name product) BRANDS ON THESE NUTRITIONAL QUALITIES BEFORE BUYING. SUPPOSE YOU WERE GIVEN SUCH INFORMATION ON ALL THE BRANDS, WHICH SINGLE ONE OF THESE WOULD BE MOST IMPORTANT IN HELPING YOU DECIDE ON A BRAND TO BUY?~7enter "10" opposite R's choice - i f R cannot reply or declares none to be meaningful, enter zeros in a l l the blanks and go to next product) MACARONI 8 CHEESE DINNER IRON LOTS OF LITTLE FAT LOTS OF LITTLE VITAMIN B1 LOTS OF LITTLE CARBOHYDRATE LOTS OF LITTLE VITAMIN B 2 LOTS OF LITTLE CALORIES LOTS OF LITTLE PHOSPHORUS LOTS OF LITTLE VITAMIN NIACIN LOTS OF LITTLE PROTEIN LOTS OF LITTLE CALCIUM LOTS OF LITTLE SINCE (name R's choice) IS THE MOST IMPORTANT ONE TO YOU, LET'S GIVE IT A SCORE OF "10". NOW, FOR EACH OF THE REMAINING NUTRITIONAL QUALITIES, I WILL ASK THE FOLLOWING. HOW IMPORTANT IS EACH QUALITY IN HELPING YOU DECIDE WHICH BRAND TO BUY? LET'S USE A SCORE FROM ZERO TO 10. ZERO MEANS IT HAS NO IMPORTANCE TO YOU. TEN MEANS IT IS JUST AS IMPORTANT ASlnTme i n i t i a l choice). YOU CAN USE THE SAME SCORE MORE THAN ONCE IF YOU FEEL THAT TWO OR MORE NUTRITIONAL QUALITIES ARE EQUALLY IMPORTANT TO YOU. (read out nutrients, one at a time, from top to bottom and enter scores above) (2) LET'S GO BACK OVER THESE NUTRITIONAL QUALITIES. IF YOU WERE CHOOSING A BRAND OF (name product) TO BUY, WOULD YOU, YOURSELF, GENERALLY WANT THE BRAND WHICH HAS LOTS OF OR LITTLE OF EACH OF THESE? (read out nutrients, one at a time, from top to bottom and c i r c l e appropriate word above. If respondent says neither, enter any comments opposite that nutrient) (3) WHICH BRAND OF (name product) DO YOU USUALLY BUY? (If R mentions more than one brand, record them a l l in the order mentioned. Go to next product) SURVEY QUESTIONNAIRE 6 U.B.C. SHOPPER RESEARCH (A (Start with product on top sheet and ask question (A ) for product on this and a i l following sheets. Then return to question (?) on top sheet. If any product is purchased less often than once in M months (16 weeks), cross i t out and go to next product. If a l l products have been deleted in this way, terminate interview after asking question (c)) HOW OFTEN DO YOU BUY (name product in question (?) on this sheet)? (go to next product) (?) (Skip any products you have crossed out. If this paper is wh i te (pink), hand respondent wh i te (pi nk) l i s t . After a l l e l i g i b l e products have been covered, go to question ) (? (1) I'D LIKE YOU TO THINK YOU ARE SHOPPING FOR (name product). ON THIS LIST ARE SOME NUTRITIONAL QUALITIES OF (name product). A PERSON MIGHT COMPARE DIFFERENT (name product}. BRANDS ON THESE NUTRITIONAL QUALITIES BEFORE BUYING. SUPPOSE YOU WERE GIVEN SUCH INFORMATION ON ALL THE BRANDS, WHICH SINGLE ONE OF THESE WOULD BE MOST IMPORTANT IN HELPING YOU DECIDE ON A BRAND TO BUY?-[e"nter "10" opposite R's choice - i f R cannot reply or declares none to be meaningful, enter zeros in a l l the blanks and go to next product) CANNED CREAM-OF-MUSHROOM SOUP FAT LOTS OF LITTLE VITAMIN B„' LOTS OF LITTLE POTASSIUM LOTS OF LITTLE CALORIES LOTS OF LITTLE SODIUM LOTS OF LITTLE CARBOHYDRATE LOTS OF LITTLE PHOSPHORUS LOTS OF LITTLE CALCIUM LOTS OF LITTLE PROTE1N LOTS OF LITTLE SINCE (name R's choice) IS THE MOST IMPORTANT ONE TO YOU, LET'S GIVE IT A SCORE OF "10". NOW, FOR EACH OF THE REMAINING NUTRITIONAL QUALITIES, I WILL ASK THE FOLLOWING. HOW IMPORTANT IS EACH QUALITY IN HELPING YOU DECIDE WHICH BRAND TO BUY? LET'S USE A SCORE FROM ZERO TO 10. ZERO MEANS IT HAS NO IMPORTANCE TO YOU. TEN_ MEANS IT IS JUST AS IMPORTANT ASlnTme i n i t i a l choice). YOU CAN USE THE SAME SCORE MORE THAN ONCE IF YOU FEEL THAT TWO OR MORE NUTRITIONAL QUALITIES ARE EQUALLY IMPORTANT TO YOU. (read out nutrients, one at a time, from top to bottom and enter scores above) (?) (2) LET'S GO BACK OVER THESE NUTRITIONAL QUALITIES. IF YOU WERE CHOOSING A BRAND OF (name product) TO BUY, WOULD YOU, YOURSELF, GENERALLY WANT THE BRAND WHICH HAS LOTS OF OR LITTLE OF EACH OF THESE? (read out nutrients, one at a time, from top to bottom and c i r c l e appropriate word above. If respondent says neither, enter any comments opposite that nutrient) (?) (3) WHICH BRAND OF (name product) DO YOU USUALLY BUY? (If R mentions more than one brand, record them a l l in the order mentioned. Go to next product) AT WHICH STORE DO YOU USUALLY SHOP FOR YOUR GROCERIES? OO YOU SHOP ANYWHERE ELSE? ( i f a Safeway store is mentioned, ask location) LAST OF ALL, I WOULD LIKE TO ASK YOU FOR SOME INFORMATION ON YOUR HOUSEHOLD. COULD YOU TELL ME HOW MANY PERSONS THERE ARE IN YOUR HOUSEHOLD? COULD YOU TELL ME THE OCCUPATION OF THE HEAD OF YOUR HOUSEHOLD? (hand respondent demographic guide sheet and ask for code number of categories on:) PLEASE INDICATE THE LANGUAGE YOU FIRST SPOKE AND STILL UNDERSTAND. COULD YOU INDICATE YOUR AGE CATEGORY? COULD YOU INDICATE YOUR EDUCATIONAL LEVEL, FROM AMONG THESE CATEGORIES? PLEASE INDICATE YOUR TOTAL FAMILY INCOME, BEFORE TAXES AND DEDUCTIONS, IN THE LAST 12 MONTHS FROM AMONG THESE CATEGORIES. MAY I ASK FOR YOUR PHONE NUMBER SO THAT THE RESEARCH PROJECT DIRECTORS CAN VERIFY THAT I DID MAKE A CALL AT THIS HOUSE? (ci rcle sex) (thank respondent and leave. Then complete BLOCK NO., DATE and TIME in Question (p) , before next interview) (Non-Response Observations -- To be completed after leaving and before next interview. Complete this section for a l l interview refusals and for a l l refusals to Question (5) . Complete this section also for a l l interviews terminated after Question (A) was followed by Question © ) EST. AGE SEX CH ILD (REN) PRESENT? YES NO EST. ETHNICITY FROM FEATURES AND/OR ACCENT BLOCK NUMBER DATE TIME 268 Table C - l Summary of Responses to Question C of Survey Questionnaire Regarding Store Patronage Frequency Grocery Store of Mention Experimental Store 1 Patronage Area: Experimental Store 1 28 Super Valu (11th & Lonsdale) 17 Woodwards 6 Produce C i t y 4 Safeway (Lynn V a l l e y M a l l ) 3 Stongs 2 Super Value (Capilano M a l l ) 2 Experimental Store 2 Patronage Area: Experimental Store 2 17 IGA (Main & 14th) 15 Super Valu (Fraser & 17th) 11 Woodwards (Oakridge) 10 Safeway (25th & Oak) 4 Safeway (Kingsway & Knight) 3 Safeway (Broadway & Commercial) 2 Produce C i t y 2 Buy Low (32nd & V i c t o r i a ) 2 Shaws (Cambie & 22nd) 2 Safeway (Cambie & 39th) 1 Kwiksave 1 Pay-Low 1 APPENDIX D DAILY CODING SHEETS FOR ITEM MOVEMENT DATA AND REPLICA OF A SHELF TAG 270 A P P E N D I X D D A I L Y C O D I N G S H E E T S FOR I T E M MOVEMENT DATA T C P " : DATE: pkg size g o r ml $ per un i t TIME E HOCK 1 TI ME BL OCK 2 K ATE I ITE U.P.C. r n n t r START 9:00 END 2:15 JNITS SOLD kg o r L J SOLD TART END JNITS SOLD kg ar L SOLD S END ) PM Ivut 3:00 6:00 CAMPBELL'S M.SOUP 28k 632 1101 261 CAMPBELL'S M.SOUP 540 632 1101 263 TOWN HOUSE M.SOUP 2ok 582 0037 020 EMPRESS CREAMY SM. 2kO 582 0046 622 EMPRESS CREAMY SM, 680 582 0046 638 EMPRESS CREAMY SM. 907 58? fW4R 628 EMPRESS CREAMY SM. 1360 582 0046 646 EMPRESS CHUNK STYLE ksk 582 0046 602 FMPRFSS CHUNK STYLE 136O 582 0046 606 EMPRESS OLD FASH 907 582 0046 630 KRAFT SMOOTH 500 681 0008 421 KRAFT CRUNCHY 500 681 0008 422 McCOLL'S 680 562 2010 003 SKIPPY CREAMY 907 626 4609 420 SKIPPY CHUNKY 907 626 4609 427 SUNNY JIM OLD FASH 750 716 5003 265 BEST FOODS MAYO 227 626 4606 120 BEST FOODS MAYO 500 626 4606 141 BEST FOODS MAYO 750 626 4606 151 KRAFT MAYO 500 681 0004 031 KRAFT MAYO 750 681 0004 032 TOWN HOUSE 682 58? 0047 228 271 S^ORE: DATE: pkg i ze % o r t ml $ per jn i t I ME B LOCK 1 TI <E BL OCK 2 N ATE ITE U.P.C. CODE -START 9:00 END I 2:15 JN ITS SOLD kg or L ' SOLD TART 3:00 END t 6:00 JNITS SOLD kg o r L SOLD 9 END PM HEINZ KETCHUP • 313 570 0000 305 HEINZ KETCHUP 568 570 0000 307 HFINZ KETCHUP 909 570 0000 309 HEINZ KETCHUP 2840 570 0000 324 TOWN HOUSE KETCHUP 313 582 0049 038 TOWN HOUSE KETCHUP 568 58? 0049 046 TOWN HOUSE KETCHUP 909 582 0049 052 KRAFT DELUXE M&C 297 681 0005 865 KRAFT DINNER M&C 206 681 0005 861 TOWN HOUSE M&C 206 582 0050 290 KELLOGG ALL-BRAN 425 641 0000 144 KELLOGG ALL-BRAN 575 641 0000 145 KELLOGG BRAN BUDS 525 641 0000 071 KELLOGG BRAN FLAKE! 400 R41 0000 065 ] \ L— L* tw v»s \J \i ill i I— • KELLOGG BRAN FLAKE! 600 641 0000 066 KELLOGG RAISIN BR. 525 641 0000 086 KELLOGG RAISIN BR. 800 641 0000 089 NABISCO BRAN CRUNC • 500 580 2501 144 NABISCO 100% BRAN 450 580 2501 128 POST BRAN FLAKES 400 661 8803 890 CUSTOMER COUNT SIGN 0N/16/ENTH R 1 272 REPLICA OF A SHELF TAG APPENDIX E NOISE" MONITORING RECORDS ON TEST PRODUCTS CREAM-OF-MUSHROOM SOUP *HU/H*Zfni page | I — -275 at DL 10 Y. a > 3 tr s :4 2 3 5 5 1 1 g g: 0,1 S t 33 a ) 5 "8— U J 3 UJ < z z o 2 2 4> >.* £4 3 : a8 V-i £ IEEE CO, <-J[ fc u o uJj O 0 3 ^ aaj i 2 J S Pel ViJ M l IA 1 ly1 — 276 MACARONI & CHEESE DINNER APfill/VAY Page J of _ S"ToR6T KG MALL mm ML i AT At nfj 4 4 4 BLCfCJk 1 G A aoJA n MS A.PWM I II WJSlS I cM i I2J> WIT I7i'rl c l c iaow c T T 2<J 6W P£M/v/! K-S do Iky ^ 2 bt ntaps 2(J27 It. 17 c C HZ ti VAN PJT MM 33 r ooT-ar-iT^l v V N tm C I C ^2 278 BRAN CEREAL STO*£: 0 3 7 K&M M L 0 « Y J ^ p a g e [ of_2___ APPENDIX F DETERMINATION OF OVERALL NUTRITIVE PERFORMANCE RANKS IN TREATMENTS WITH MAYONNAISE AND KETCHUP BRANDS Table F - l Brand Performance Ranks (R^) by Nutrient Ratings of Mayonnaise Brands Nutrient ( L i s t e d , From L e f t to Right, In Order of Decreasing Importance) Poly-unsaturated Total Carbo- Saturated Brand Code Fats P r o t e i n Fats. hydrate Fats C a l o r i e s Sodium Sugars (Rj) (R 2) (R 3) (R 4) (R 5) (R 6) (R 7) (R 8) A B C 1 3 2 2 . 5 2 . 5 . 1 2 3 1 2 1 3 2 3 1 2 3 1 3 1 2 aNo ranking p o s s i b l e because of missing r a t i n g s on two brands. Table F-2 Determination of Overall N u t r i t i v e Performance Ranks (R or R) of Mayonnaise Brands In Each Treatment, Using Table F - l Data Treatment  "1/high" "2/high" "4/high" "8/high" " " " l / l o w " "2/low" T / W " "8/low" Brand ;2 4 7 7 7 Code R , IR. R IR. R £R R~ R a R £ R R £ R R i=5 i = l A 1 3.5 2 7.5- 2 14.5 2 - 3 7 2 b 14.5 2 B 3 5.5 3 9.5 3 16.5 3 1 7 3 b 16.5 3 C 2 3.0 1 7.0 1 11.0 1 - 2 4 1 11.0 1 No ranking of brands was p o s s i b l e on t h i s treatment because of missing r a t i n g s f o r the cue "sugars" on two brands. b T i e i n rank sums was broken by Brand A's higher rank on t h i s treatment's most important cue. Table F-3 Brand Performance Ranks (R.) by Nutrient Ratings of Tomato Ketchup Brands Nutrient ( L i s t e d , From L e f t to Right, In Order of Decreasing Importance) - - - Carbo-Brand Code Iron P r o t e i n C a l o r i e s Potassium Phosphorus Sodium hydrate Fat (R x ) (R 2 ) (R 3 ) - (R 4 ) (R 5 ) (R 6 ) (R 7 ) ( R 8 ) A 2 1.5 2 1 1 2 2 1 B 1 1.5 1 2 2 1 1 2 IX) oo C O Table F-4 Determination of Overall Nutritive.Performance Ranks (R or R) of Tomato Ketchup Brand In Each Treatment, Using Table F-3 Data "1/high" "2/high" "4/high" "8/high" "1/low" "2/low" "4/1ow" "8/low" Brand 2 _ 4 _ 8 _ 8 _ 8 _ 8 Code R IR. R IR. R £R. R R .£ R. -R £ R. R I R,- R 1 1 1 i=7 1 i=5 1 i = l 1 A 2 3.5 2 6.5 2 12.5 2 1 3 2 a 6 l a 12.5 2 B 1 2.5 1 5.5 1 11.5 1 2 3 l a 6 2 a 11.5 1 a T i e i n rank sums was broken by r e f e r r i n g to each brand's rank on t h i s treatment's most important cue. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0095612/manifest

Comment

Related Items