UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The use of habit-change strategies in demarketing: reducing excessive discretionary consumption Gallagher, Katherine 1994

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

Item Metadata

Download

Media
831-ubc_1995-983779.pdf [ 5.56MB ]
Metadata
JSON: 831-1.0088868.json
JSON-LD: 831-1.0088868-ld.json
RDF/XML (Pretty): 831-1.0088868-rdf.xml
RDF/JSON: 831-1.0088868-rdf.json
Turtle: 831-1.0088868-turtle.txt
N-Triples: 831-1.0088868-rdf-ntriples.txt
Original Record: 831-1.0088868-source.json
Full Text
831-1.0088868-fulltext.txt
Citation
831-1.0088868.ris

Full Text

THE USE OF HABIT-CHANGE STRATEGIES IN DEMARKETING:REDUCING EXCESSIVE DISCRETIONARY CONSUMPTIONbyKATHERINE GALLAGHERB.A., McGill University, 1978M.B.A., The University of British Columbia, 1982A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Faculty of Commerce and Business Administration)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIADecember 1994© Katherine Gallagher, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)LLtDepartmen- of (Ju&rrL..The University of British ColumbiaVancouver, CanadaDate Q&Lc i’?/’1(DE-6 (2188)ABSTRACTAccording to the Bruntland Commission, sustainable development requires consumers inindustrialized nations to reduce significantly their consumption of resources. This research bringsa new perspective to the reduction of discretionary consumption, using both theoretical and empiricalapproaches.Demarketing programs have often been unable to achieve sustained reductions inconsumption. It is argued here that they have incorrectly treated demand reduction as a variation onthe usual marketing problem of building demand, when it is (1) more complex than typical marketingproblems, and (2) essentially similar to clinical habit change problems.The dissertation reviews the literature on habits and automated processes, introduces theconcept of “habit-likefl behavior, and argues that reducing discretionary consumption can often beframed as a habit-change problem.The Prochaska and DiClemente (1984) Revolving Door Model of Behavior Change (RDM)describes how people change habitual behaviors in clinical situations. Study 1, an energyconservation (cold water laundry washing) survey (n=340), using a decisional balance framework,indicated that the RDM generalizes to demarketing situations and that it is consumers’ perceptionsof the importance of disadvantages, not advantages, that influence consumption reductions.The research develops new theory to explain habit-like behavior changes. Based on previoustheory and findings on automated processes, it is proposed that changing habit-like behavior proceedsin three steps: de-automation, volitional behavior change, and consolidation. Study 2 was alaboratory experiment (n= 117) in which two demarketing approaches (the traditional approach andthe habit-change approach) competed in two situations (when the consumption behavior targeted forchange was under volitional control, and when it was habit-like). Contrary to expectations, apersuasive message supplemented by limited practice of the new behavior was more effective whenIIthe old behavior was volitional than when it was habit-like, suggesting that the disadvantages ofchanging are more evident to people whose behavior is habit-like.There are two important practical implications: that (1) segmentation based on the RDMstages of change may be more powerful than other approaches; and (2) it is more important toaddress disadvantages of reducing consumption than to emphasize advantages.111TABLE OF CONTENTSABSTRACTLIST OF TABLESxLIST OF FIGURES XiVACKNOWLEDGEMENTS xvDEDICATION xviCHAPTER 1THE ISSUEGrowth, Consumption, and MarketingSustainability, Conservation, and DemarketingOverview811141718Summary 38CHAPTER 4A COMPREHENSIVE MODEL OF HABITUAL BEHAVIOR CHANGE.Why a Psychotherapeutic Model9Development of the ModelThe Linear Version of the ModelPrecontemplationContemplationActionMaintenance13468CHAPTER 2MARKETING APPROACHES TO DECREASING CONSUMPTIONDemarketingReverse MarketingConventional MarketingOther VariablesSummaryCHAPTER 3HABITS AND HABIT-LIKE BEHAVIORHabits: A Brief IntroductionHistorical OverviewDefinitions and Descriptions of HabitsAutomatic ProcessesHabit-like BehaviorDoes Demarketing Attempt to Change Habit-like Behavior?Habit-like Behavior in Marketing and Demarketing22232325303234353940414343434444ivDemarketing Insights from the Linear Model 44The Revolving Door Model of Behavior Change 46Demarketing Insights from the Revolving Door Model 46A Spiral Model of the Stages of Change 48Ten Change Processes 49Consciousness Raising 50Self-Reevaluation 51Social (or Environmental) Reevaluation 51Self-Liberation 51Social Liberation 52Counter-Conditioning 52Stimulus Control 53Contingency (or Reinforcement) Management 53Dramatic Relief 53Helping Relationships 54Integration of Stages and Processes of Change 54Summary 57CHAPTER 5STUDY 1: HYPOTHESES AND DESIGN 58General Approach 58The Original Decisional Balance Study 59Hypotheses 60Methodological Approach 64Criteria for the Target Behavior \65Object of a Previous Demarketing Program 65Focus on Changes in Usage Behavior 65Accuracy in Self-Reports 66Behavioral Context: Cold Water Washing 67Sample 68Procedure 69Questionnaire 70Qualifiers 70Stages of Change Items 70Decisional Balance Items 71Operational Hypotheses 73Summary 77CHAPTER 6STUDY 1: ANALYSIS AND RESULTS 78Assignment to Stages 78Decisional Balance Measures 79Deletion of Items 80Principal Component Analysis 81Reliability of the Advantages and Disadvantages Scales 87Cross Sectional Comparisons 87Treatment of Missing Data 87Overview of the Cross Sectional Comparisons 88Difference Scores 89Hypothesis 1 . . 91vHypothesis 2a 92Hypothesis 2b 93Hypothesis 2c 93Hypothesis 3a 94Hypothesis 3b 95Hypothesis 3c 96Hypothesis 4a 97Hypothesis 4b 97Hypothesis 4c 98Hypothesis 5a 99Hypothesis 5b 100Hypothesis 5c 101Hypothesis 6a 102Hypothesis 6b 102Hypothesis 6c . 103Summary of Cross Sectional Results . . 104Summary 106CHAPTER 7STUDY 1: DISCUSSION 107Assignment to Stages of Change 108Differential transience 108Low interest level 109Methodological weaknesses 110A strategy for deciding among the explanations 111Decisional Balance Measures 112Cross Sectional Comparisons 113Issues Arising from Study 1 117Issue 1: Definitions of the Stages of Change 118Issue 2: Dynamic Nature of the Model 123Issue 3: Fuzzy Borders 124Issue 4: The Need for Theory 125Summary 126CHAPTER 8CHANGING COMPLEX, HABIT-LIKE BEHAVIORS 127Changing Habit-Like Behavior: An Overview of the Literature 128Habit Reversal 128Pre-Emptive Habitual and Habit-Like Behavior Change 131Decomposing the Habit Change Problem 133De-Automation 134Automated processes and habit-like behaviors 135Description 136Reliable anticipation 138Motivation 140Volitional Behavior Change: The Fishbein and Ajzen Approach 141The Theory of Reasoned Action in a Demarketing Context 143Consolidation 144viCHAPTER 9STUDY 2: HYPOTHESES AND DESIGNHypothesesMethodological Approach .Criteria for the Target BehaviorOverview of the ProcedurePre-manipulation lists..ManipulationsDistractionPost-manipulation listsQuestionnaireDebriefingDesignOperational HypothesesPretestsProcedureRandomizationPreliminariesPre-manipulation listsManipulationsDistractionPost-manipulation listsQuestionnaireReleaseDebriefingManipulations176177177177177Relationships Among the Components of Habitual Behavior ChangeAn Additional ComplicationSummary145146146147148151152153154154154154154154154155159159159160160161161161162162162162162162163163Persuasion + Behavior ManipulationPersuasion ManipulationControlSampleCHAPTER 10STUDY 2: ANALYSIS AND RESULTSOverview of the DataA Note on TerminologyManipulation CheckHypotheses Concerning Groups with Less Premanipulation Task Experience.Assumption 1AAssumption lBHypothesis 1AHypothesis lBHypothesis 1CHypotheses Concerning Groups with More Premanipulation Task ExperienceAssumption 2AAssumption 2BHypothesis 2A..164..166• . 168169• . 172• 173• . 173• . 173174viiHypothesis 2B 178Hypothesis 2C 179Hypothesis 2D 180Hypotheses Concerning the Relative Effectiveness of the Manipulations AcrossDifferent Levels of Premanipulation Task Experience . 180Hypothesis 3A 181Hypothesis 3B 181Summary of the Experimental Results 182CHAPTER 11STUDY 2: DISCUSSION 184The Experimental Procedure 185Little Pre-Manipulation Practice Groups . . 186Methodological Issues 186Theoretical Issue 189Extensive Pre-Manipulation Practice Groups 191Methodological Issues 192Theoretical Issue 192Relative Effectiveness of the Manipulations Across Pre-Manipulation Levels 193Summary 194CHAPTER 12CONCLUSIONS 197Summary of the Dissertation Research. 197Contributions 201Theoretical Contributions 201Methodological Contribution . . 203Practical/Managerial Implications 204Limitations 204Theoretical Limitations 204Limitations of Study 1 205Limitations of Study 2 205Future Research Directions 206Conclusion 208APPENDIX 1BC HYDRO TRACKING SURVEY Wave 7 209APPENDIX 2MATERIALS FOR PARTICIPANTS IN STUDY 2 225APPENDIX 2a: Information Sheet for Participants with Little PremanipulationPractice Receiving the Persuasion Manipulation and Participants with LittlePremanipulation Practice Receiving No Manipulation (Control) 226APPENDIX 2b: Information Sheet for Participants with Little PremanipulationPractice Receiving the Persuasion + Behavior Manipulation 228APPENDIX 2c: Information Sheet for Participants with Extensive PremanipulationPractice Receiving the Persuasion Manipulation 230APPENDIX 2d: Information Sheet for Participants with Extensive PremanipulationPractice Receiving the Persuasion + Behavior Manipulation 232APPENDIX 2e: Consent Form for All Participants 234vi”APPENDIX 2f: Instructions for All Participants.236APPENDIX 2g: Script for the Videotaped Demonstration for All Participants . . . . 238APPENDIX 2h: Questionnaire for All Participants 242APPENDIX 3:CATEGORIES 246APPENDIX 3a: Premanipulation Categories Used for Little PremanipulationPractice Groups 247APPENDIX 3b: Premanipulation Categories Used for Extensive PremanipulationPractice Groups 248APPENDIX 3c: Postmanipulation Categories Used for All Groups 249REFERENCES 250ixLIST OF TABLESTable 2-1Review of Demarketing Research 20Table 3-1Comparison of Conventional Marketing and Demarketing . 37Table 4-1Processes of Change 50Table 6-1Assigmnent of Respondents to Stages of Change 79Table 6-2“Disagree” and “Don’t Know” Responses in the Decisional Balance Items 80Table 6-3Factor Eigenvalues 83Table 6-4Rotated and Interpreted Factor Matrix 83Table 6-5Rotated and Interpreted Factor Matrix When Two Factors Are Extracted . 85Table 6-6Advantages of Cold Water Washing Scale 86Table 6-7Disadvantages of Cold Water Washing Scale 86Table 6-8Means and Standard Deviations of the Advantages and Disadvantages Scales by Stage ofChange 89Table 6-9Ordered Mean Difference Scores By Stage of Change 91Table 6-10Disadvantages and Advantages Scores in the Precontemplation Stage 92Table 6-11Disadvantages and Advantages Scores in the Contemplation Stage . . 92Table 6-12Advantages Scores in the Contemplation and Precontemplation Stages 93xTable 6-13Disadvantages Scores in the Contemplation and Precontemplation Stages 94Table 6-14Disadvantages and Advantages Scores in the Action Stage 95Table 6-15Advantages Scores in the Action and Contemplation Stages 96Table 6-16Disadvantages Scores in the Action and Contemplation Stages 97Table 6-17Disadvantages and Advantages Scores in the Maintenance Stage 97Table 6-18Advantages Scores in the Maintenance and Action Stages 98Table 6-19Disadvantages Scores in the Maintenance and Action Stages 99Table 6-20Disadvantages and Advantages Scores in the Relapse Stage 100Table 6-21Advantages Scores in the Relapse, Maintenance, and Action Stages 101Table 6-22Disadvantages Scores in the Relapse, Maintenance, and Action Stages 102Table 6-23Disadvantages and Advantages Scores in the Termination Stage 102Table 6-24Advantages Scores in the Termination and Maintenance Stages 103Table 6-25Disadvantages Scores in the Termination and Maintenance Stages 104Table 6-26Summary of Cross Sectional Results 105Table 7-1Operational Definitions of the Stages of Change 119Table 7-2Markov Model: Likely Transition Probabilities 123Table 10-1Experimental Procedure and Design 155xiTable 10-1Experimental Procedure and Design 165Table 10-2Mean Number of Times Participants in Each Group Used Both Sides of the Paper at VariousPoints in the Experiment 167Table 10-3Number of Participants Who Switched from Using One Side of the Paper to Using BothSides 167Table 10-4Attitude Toward Using Both Sides of the Paper 169Table 10-5Student-Newman-Keuls Multiple Range Test for Attitude Toward Using Both Sides ofthe Paper 170Table 10-6Attitude Toward Using Both Sides of the Paper 171Table 10-7Analysis of Variance: Effectiveness of Manipulations in Producing FavorableAttitudes 171Table 10-8Comparison of the Average Number of Times Participants in the Control and ShortPersuasion Groups Used Both Sides of the Paper for Lists Immediately Followingthe Videotape 174Table 10-9Comparison of the Average Number of Times Participants in the Control and Short BehaviorGroups Used Both Sides of the Paper for Lists Immediately Following the Videotape 175Table 10-10Comparison of the Average Number of Times Participants in the Short Persuasion and ShortBehavior Groups Used Both Sides of the Paper for Lists Immediately Following theVideotape 175Table 10-11Comparison of the Average Number of Times Participants in the Long Persuasion GroupUsed Both Sides of the Paper for Lists Following the Videotape 178Table 10-12Average Number of Times Participants in the Long Behavior Group Used Both Sides ofthe Paper for Lists Immediately Following the Videotape 179Table 10-13Comparison of the Average Number of Times Participants in the Long Behavior GroupUsed Both Sides of the Paper for Lists Following the Videotape 179xliTable 10-14Comparison of the Average Number of Times Participants in the Long Persuasion and LongBehavior Groups Used Both Sides of the Paper for the Final Ten Lists Followingthe Videotape 180Table 10-15Comparison of the Number of Times Participants in the Short Persuasion and LongPersuasion Groups Used Both Sides of the Paper in the Lists Immediately Followingthe Videotape 181Table 10-16Comparison of the Number of Times Participants in the Short Behavior and Long BehaviorGroups Used Both Sides of the Paper in the Lists Immediately Following the Videotape . . . 182XIIILIST OF FIGURESFigure 4-1The Linear Model 43Figure 4-2The Revolving Door Model of Behavior Change 47Figure 4-3The Spiral Model of Behavior Change 49Figure 4-4Processes and Stages of Change 55Figure 6-1Scree Plot 84Figure 6-2Mean Advantages and Disadvantages Scores by Stage of Change. . . . 90Figure 8-1Proposed Habitual Behavior Change Process 134Figure 8-2Relationship Between Automated Processes and Habit-like Behaviors . 135Figure 8-3Hierarchy of Automated Processes and Habit-like Behaviors 136Figure 8-4Overview of the Theory of Reasoned Action 143xivACKNOWLEDGEMENTSI extend my thanks to my Research Supervisor, Richard W. Pollay, and the members of myResearch Committee, Charles B. Weinberg and Clifton Lee Gass, for their guidance during the courseof this research. Their high-standards were a continuing challenge. I hope that they are pleased withthis work, to which they made such important contributions. I also had a very helpful andconstructive Examining Committee, whose other members were Lawrence W. Green, Gerald 3. Gorn,Robert E. Sparks, and Ida E. Berger (Queen’s University). They provided valuable advice andperspectives, for which I am grateful.I would also like to thank Caroline Helbig, who “starred” in the manipulations videotapes,and Brian Tiller, who edited them. Both suffered through inumerable versions and takes with thegreatest of patience and good humor.My collegues at Memorial University of Newfoundland were unfailingly cooperative as Iprogressed through this research.On a more personal note, I would like to acknowledge several people: Jane Londervile andDave Downie made lunchtimes the high point of my UBC days; Laurenda Daniells helped me retainsome perspective; Ellen Mcintosh, Judy Cumby, and Dale Foster made me laugh, as well as helpingin practical ways. My closest friends, Joanne Lapin and Wanda Parsons, were always willing tolisten to me and provide encouragement. My parents, Grace and Paul Gallagher, and my brothersand their wives, Stephen and Donna Gallagher, Ed Gallagher and Teresa Megli (to whose memorythis work is dedicated), and Peter Gallagher, were all unbelievably supportive. I thank all of thesepeople. I truly could not have done it without them. Finally, I would like to thank my husband, JeffParsons, whose patience, tolerance, good judgement and advice never failed, and whose capacity forencouragement and support was never exhausted.xvIn Memory ofTeresa Suzanne Megli(1958 1994)xviCHAPTER 1THE ISSUEFrom space, we see fthe Earth as] a small and fragile ball dominated not by human activity andedWce but by a pattern of clouds, oceans, greenery and soils. Humanity’s inability to fit its doingsinto thatpattern is changing planetary ystems,fundamentally. Many such changes are accompaniedby life-threatening hazards. This new reality, from which there is no escape, must be recognized—and managed. (World Commission on Environment and Development 1987: 1)For most of this century, in much of the world, growth has been a key economic goal.Economic growth has brought increases in the standard of living and improvements in the quality oflife, particularly for people in the industrialized nations of the world. But economic growth is notwithout cost; in fact, it is becoming obvious that it can come at a very high price, especially withrespect to the environment.Concern about the impacts of economic growth upon the environment started to developduring the 1960s and 1970s (e.g., Carson 1962, Club of Rome 1972, Schumacher 1973). Air andwater pollution were regularly in the news; in many jurisdictions, regulations were put in place tolimit the harm done by industrial byproducts. By the early 1980s, concern about the environmenthad expanded and deepened with the recognition that ecological stress brought on by industrialdevelopment — degradation of soils, water systems, the atmosphere, and forests— was having anunfavorable impact on economic prospects (e.g., World Commission on Environment andDevelopment 1987). Now, in the 1990s, the economic effects are undeniable, even in prosperous12countries like Canada: for instance, groundfish stocks off Atlantic Canada, once so abundant thatthe sea was “swarming with fish, which can be taken not only with net, but in baskets let down witha stone,” (Environment Canada 1991: 8-4) are so depleted that the Northern Cod fishery has beenclosed at least until the late 1990s, throwing tens of thousands of Atlantic Canadians out of work.In 1983, the General Assembly of the United Nations called upon Gro Harlem Brundtland,former Prime Minister of Norway, to chair a special independent commission to propose long termenvironmental strategies for achieving sustainable development by the year 2000. The Commission’swidely discussed final report, Our Convnon Future, emphasizes the importance of reducingdiscretionary consumption:Sustainable development requires the promotion of values thatencourage consumption standards that are within the bounds of theecologically possible and to which all can reasonably aspire. (WorldCommission on Environment and Development 1987: 44)Canada supports the recommendations of the Brundtland Commission and recognizes our country’srole:Canadians continue to place greater stresses on the environment thanindividuals in most of the rest of the world. Progress is evident inthe reduction, reuse, and recycling of some materials, but ourlifestyle continues to be based on high levels of consumption.[Ojur attitudes and actions need to change greatly if weare to successfully build a sustainable Canada for future generations.We will have to become not only better stewards of the environment,but better managers of our own actions and their impacts on theEcosphere. (Environment Canada 1991: 27-1)The question is how to change our actions, how to reduce consumption to sustainable levels.No one piece of research can hope to answer a question of such complexity. Neither can a singlediscipline. Only the combined efforts of people working on various aspects of the problem frommultiple points of view can provide the needed answers. This dissertation research attempts toaddress, from a marketing point of view, the question of how to reduce discretionary consumption.It does not attempt to solve the whole problem; it is just one brick in the foundation required to buildan understanding of the problem. This research proposes and investigates one new approach to3obtaining sustained reductions in discretionary consumption— treating excessive discretionaryconsumption as a bad habit that needs to be changed. The research reviews literature ondemarketing, habits, and habit change; it borrows from psychotherapy a model of behavior changeand tests its applicability to a specific demarlceting problem, it also develops theory on habit change,and in a controlled experiment, tests some hypotheses derived from the theory.Before describing the research in more detail, a brief and simplified account of some of thefactors that have led to the environmental problems we now face is in order.Growth, Consumption, and MarketingEver since the Industrial Revolution, in which production capacity was vastly increased, thenormal situation has been one of goods oversupply. For example, in 1910, American factories madeseven times as much pig iron as had been produced thirty years earlier in 1880, as well as nine timesas much paper, fourteen times as much cottonseed oil, and nearly four times as many railroad carsto transport all the new goods made from these basic materials. In contrast, during that thirty yearperiod, the population of the United States did not quite double (Strasser 1989).Expansion of demand has been critical to the continued prosperity of industrial economies,and it has been the job of marketing to help develop and nurture new demand. All that pig iron,paper, cottonseed oil, and all those railroad cars found buyers and consumers, and it was marketingthat got the goods from the producers to the users.Thorstein Veblen’s wry observation that “invention is the mother of necessity” might be asuitable motto for marketers, as an unprecedented range and variety of goods and services have beenproduced and marketed in the last few decades. Thus we in North America have enjoyed a“Consumer Society” for most of this century.4Sustainability, Conservation, and DemarketingAssuming that the World Commission on Environment and Development is correct in itsassessment of the issues, and in its general approach to dealing more effectively with environmentalconcerns, then excessive consumption will have to be moderated by conservation. Marketing can andshould play a key role in achieving sustainable development. However, there is a popularmisconception that, since marketing is part of the problem, it cannot be part of the solution. Sucha view sells marketing short. In fact, sustainable development will not be possible without effectivemarketing.The American Marketing Association defines marketing as:the process of planning and executing the conception, pricing,promotion, and distribution of ideas, goods, and services to createexchanges that satisfy individual and organizational objectives.(Bennett 1988: 115)From a managerial point of view, marketing is much more than getting goods from producers toconsumers; it is a matter of demand management: “not only ... finding and increasing demand, butalso ... changing or even reducing it” (Kotler, McDougall and Armstrong 1988: 11). Mostprofessional and academic attention has been devoted to understanding how and why demandincreases— not surprising, given the prevailing economic conditions in the Western world for thepast century. Now that economic and environmental conditions have changed, though, it isappropriate that some attention turn to marketing’s capacity to reduce demand.The process of demand reduction is called demarketing by many academics and demand sidemanagement by many marketing practitioners, and reducing demand is not as uncommon a goal asmight be expected. For instance:• Electric utilities encourage energy conservation in order to delay or avoid expensiveinvestment in hydroelectric or thermal generating facilities.5• Canadian public health officials work to persuade people to take care of minor ailments likethe “common cold” themselves, rather than going to a doctor, in order to keep health carecosts under control.• Health professionals urge North Americans to eat less fat in order to decrease the incidenceof heart disease and some cancers.• Municipalities want citizens to produce less solid and liquid waste, to reduce the load on theinfrastructure.• Environmentalists try to convince consumers in industrialized countries to reduce their overallconsumption of goods in order to keep the planet habitable.In addition, both permanent and temporary, planned and unplanned product shortages can and dooccur, and they must be managed properly in order to mitigate compromising customer satisfactionand firm survival. On a broader scale, economies sometimes face widespread product shortages,through, for instance, war or economic sanctions, as Rwandans, Bosnians, Serbs, Somalis, and Iraqishave recently experienced; or failure of distribution systems, as citizens of the former Soviet Bloccountries have faced lately. In such situations, failure to reduce demand in an orderly and equitablemanner has contributed to economic and social chaos.For reasons as vast as the need for a global commitment to sustainable development to thoseas specific as one firm’s need to address a temporary resource shortage, the processes involved indemarketing are important to understand. Unfortunately, the study of demarketing has largely beenneglected in recent years. Consequently, we do not really know how to design effective demarketingprograms. This dissertation research is intended to increase understanding, primarily from atheoretical point of view, of how certain types of consumption can be successfully reduced.6OverviewThe following chapters explore both theoretical and practical demarketing issues. Chapter2 begins with a review of past strategies aimed at decreasing discretionary consumption, the resultsof which have often been disappointing. It argues that a reason for this lack of effectiveness may bethat marketing and demarketing are trying to do very different things. A metaphor is employed toelucidate the difference: marketing is like asking consumers to develop new, “good” habits, whiledemarketing is like asking them to change existing “bad” habits. Furthermore, changing a bad habitis more difficult to do than developing a good habit, for reasons discussed in the chapter.Chapter 3 examines the question of what a habit is, and whether discretionary consumptioncan be thought of as a habit. It reviews the literature on habits and proposes a new defmition ofhabits. It also discusses the extent to which discretionary consumption of the type that demarketingprograms aim to reduce can be called habitual or habit-like.The remainder of the dissertation explores empirically and theoretically the implications ofconceiving of demarketing as a habit-change problem. Chapters 4, 5, 6, and 7 investigate the utilityof borrowing a model of behavior change developed in a psychotherapeutic context. Chapter 4outlines a model of behavior change developed in a psychotherapeutic setting, an area not typicallyassociated with marketing. It also discusses some of the implications of the model for demarketing.Chapter 5 presents hypotheses and describes the design of an exploratory study intended to explorethe degree to which the psychotherapeutic model of behavior change provides a reasonable and usefuldescription of behavior change in a demarketing context. The results of that study are presented inChapter 6 and discussed in Chapter 7.Chapters 8, 9, 10, and 11 address the demarketing problem from a more theoretical point ofview. Chapter 8 turns to the topic of habit change strategies, deriving theory on what ought to benecessary to change habit-like behavior, based on the cognitive properties of habit-like behaviors.Chapter 9 presents hypotheses based on the theory developed in Chapter 8, and describes the design7of a laboratory study that tests those hypotheses. The results of the laboratory study are presentedin Chapter 10 and discussed in Chapter 11.The final chapter, Chapter 12, reviews the dissertation research, discusses its limitations;identifies theoretical, methodological, and practical implications; draws conclusions; and suggestsfuture research options.CHAPTER 2MARKETING APPROACHES TODECREASING CONSUMPTIONThe changes in human attitudes that we callfor depend on a vast campaign of education, debate, andpublic participation. This campaign must start now fsustainable human progress is to be achieved.(World Commission on Environment and Development 1987: 23)Although the usual emphasis in marketing research and practice is on increasing demand, theissue of decreasing demand has not been ignored. This chapter begins with a review of literaturerelating to the concept and techniques of demarketing, especially as they relate to energyconservation. It then moves on to an examination of possible reasons for the dearth of consistentlyeffective demarketing programs.DemarketingKotler and Levy (1971) coined the term demarketing in the early 1970s, when the conceptof marketing was being broadened beyond its then narrow scope as the art and science of buildingsales volume through the use of product, price, place (i.e., distribution), and promotion (i.e.,communications) variables. They defined demarketing as:that aspect of marketing that deals with discouraging customers ingeneral or a certain class of customers in particular on either atemporary or permanent basis (Kotler and Levy 1971: 75)89Demarketing is sometimes confused with countermarketing. Like demarketing, countermarketing isused to decrease demand, but only in situations where the product is inherently unwholesome;demarketing is used when there is nothing wrong with the product per se, but demand is judged tobe excessive (though not necessarily by consumers). Consequently, tobacco products have been thesubject of countermarketing programs, as tobacco use has been linked to various serious healthproblems. On the other hand, energy conservation programs fall into the demarketing category:energy, far from being unwholesome, is necessary to sustain human life; it is the excessive use ofenergy that is discouraged. Similarly, the health professions’ campaigns to convince NorthAmericans to reduce the amount of fat in their diets is an example of demarketing, as some fat —though much less than North Americans typically consume — is necessary in a healthy diet. Onceagain, it is the excessive consumption of fat that is undesirable, not the fat itself.Kotler and Levy identified four types of demarketing: general demarketing, required whenthe goal is to reduce total demand; selective demarketing, used to discourage demand from certaincustomer segments; ostensible demarketing, in which the appearance of discouraging demand isactually a device for increasing it; and unintentional demarketing, where inept marketing programsaimed at increasing demand actually drive customers away. Subsequent research efforts have notfocused on ostensible or unintentional demarketing, as neither is aimed at demand reduction, andmight more properly be called pseudo demarketing. Rather, they concentrate on what might be calledtrue demarketing problems.Table 2-1, which appears at the end of this chapter, gives a brief description of the purpose,approach, and, where applicable, results of demarketing studies appearing in the literature from 1971,when Kotler and Levy first defined and described demarketing, to the present. The table presentsthe studies in chronological order, to show how the literature has developed over time. During the1970s, a decade in which concerns about energy supplies were particularly high, most studies focusedon strategic responses to shortages outside the firm’s control (Blankenship and Holmes 1974, Cravens101974, Cullwick 1975, Hanna, Kizilbash, and Smart 1975, Kotler 1974, Shama 1978). Kotler andLevy’s basic prescriptions — raise prices, restrict consumer access to the product, reduce productquality, and cut back on promotion — were elaborated upon. However, no empirical verification ofthe effectiveness of this approach was undertaken. (Shama 1978 did provide empirical data, but itspurpose was to identify strategic options that had been used, not to assess their efficacy.) The interestin strategies to cope with supply constraints continued into the 1980s (Bellur 1980, Dadzie 1989,Papadopoulos 1983). In the late 1970s and through the 1980s, the demarketing literature broadenedto include the effects of demarketing programs on consumer attitudes (Frisbie 1980, Kelley andScheewe 1975, Reddy 1989) and societal concerns about ecology (Shapiro 1978), the new economicworld order, and changing consumer values (Van Dam 1978). During the 1980s, empirical work wasrelatively more common than it had been in the previous decade (Dadzie 1989, Frisbie 1980), butthe purpose was still primarily descriptive, rather than aimed at understanding the underlyingprocesses.In terms of types of demarketing, the table shows that most authors proposed generaldemarketing strategies (Bellur 1980, Blankenship and Holmes 1974, Cravens 1974, Dadzie 1989,Frisbie 1980, Hanna, Kizilbash, and Smart 1975, Kelley and Scheewe 1975, Kotler 1974, Monroeand Zoltners 1979, Papadopoulos 1983, Shama 1978, Shapiro 1978), although some investigatedselective demarketing (Cullwick 1975, Reddy 1989).The next two sections of this chapter give detail on the types of strategies recommended bydemarketing experts. The first section reviews literature on what is referred to in this research as“reverse marketing”; the second reviews literature on the use of “conventional marketing” techniquesin demarketing contexts.11Reverse MarketingThe assumption underlying many demarketing prescriptions is that reduction in demand canbe accomplished by raising prices, and reducing product quality, service, promotion, and convenience(a strategy advocated in Kotler and Turner 1981). Based on an economic model, this “reversemarketing” is a plausible and intuitively appealing approach to the problem of reducing discretionarydemand. A fundamental concept in microeconomics is that changes in price will affect the quantitydemanded: lowering prices increases the quantity demanded, and raising prices lowers the quantitydemanded. Similarly, reducing product quality, service, promotion, and convenience should lowerthe quantity demanded. Unfortunately, however, reverse marketing has not always been as effectiveas might be expected.Consider the case of energy conservation — per capita consumption reduction — an areaidentified by the World Commission on Environment and Development (1987) as a key to sustainabledevelopment. Energy conservation, to which a variety of reverse marketing strategies have beenapplied, is one of the few well-documented demarketing problems. For almost two decades,government, the energy industry, and the environmental movement, faced with unstable and depletingenergy supplies, have been actively encouraging conservation. While recent improvements in energyefficiency and a shift in the economy toward less energy intensive sectors have limited increases inconsumption, results on a per capita basis have been disappointing.Two reverse marketing techniques, raising prices and restricting supplies, have been used toencourage consumers to reduce their demand for energy. In their review of energy conservationpolicies and programs, Ritchie and McDougall (1985) suggest that the reason these strategies lackeffectiveness, at least in the short run, is that demand for energy in the short term is relativelyinelastic. They conclude that without dramatic price increases, only minimal reductions in energyuse can be expected. In the same vein, Stern’s (1986) review of the energy conservation literatureindicates that people do not always notice economic signals because their information about the costs12associated with various energy-use actions is incomplete and systematically inaccurate. He calls theeconomic assumption that price increases will yield energy savings “simplistic”. Other authors(Claxton 1989, Heberlein and Warriner 1983) suggest that for short periods following significantprice increases, consumers take active steps to reduce their demand for energy; however, as theybecome accustomed to higher prices, they return to old consumption patterns. Reverse marketingtechniques other than pricing are even less successful. For example, consumer resistance torestrictions of supply (such as odd-even days for gasoline purchases) is high, and increases with theseverity of the restrictions (Ritchie and McDougall 1985).Why does reverse marketing not yield aggregate decreases in consumption? One possibilityis that there is no further potential for conservation. This seems unlikely. In fact, some researcherssuggest that conservation efforts could reduce consumer consumption by as much as 40% (Gray andvon Hippel 1981, Sawhill 1979, Stern and Gardner 1981).Another possibility is that the incentives to conserve are not strong enough, given incomedistribution in North America. Ritchie and McDougall (1985) conclude that high-income householdsare likely not to change their consumption patterns in response to price increases; middle-incomehouseholds are likely to maintain their energy use behavior, but invest in energy efficient appliancesand cars; while low-income households, lacking the financial resources to invest in energy efficiency,will change their lifestyles to curtail energy use. And the middle-income strategy of reducingconsumption by investing in energy efficient plant, rather than by attempting to decrease wastefulbehavior, may actually be counterproductive, leading to increases in consumption. It may be thatpeople have the perception that investment in conservation devices frees them from having to takeany other actions to conserve, including conservation actions they may already have undertaken.Claxton (1989) provides a vivid example. Although gasoline consumption per car decreased almost18% in the U.S. between 1973 and 1982, and gasoline efficiency improved by 12%, actual gasolineconsumption increased slightly (1.1 %)— due to a per capita increase in the number of cars. Of13course, this is correlational data, so no causal inferences can properly be drawn; nevertheless, it isclear that investment in efficiency does not necessarily coincide with conservation.Several researchers in economics (e.g., Etzioni 1986, Deaton and Muellbauer 1980, Katona1975) suggest that “habit persistence” may account for the observed short run insensitivity ofconsumption to changes in price (several examples of which are given by Thurow 1983). Katonadescribes habit persistence this way:Habitual behavior occurs much more frequently [than problemsolving behavior] and, whatever the origin of our habits, they do notexhibit the major features of rational behavior. People act as theyhave acted before under similar circumstances, without deliberatingand choosing. Routine procedures and the application of rules ofthumb by consumers as well as businessmen [and/or businesswomen]exclude the weighting of alternatives (Katona 1975: 218-219).Most explanations for habit persistence relate to various financial and temporal costs associated withfinding good substitutes for the commodity in question; here, of course, that would be energy.However, demarketing energy involves reducing discretionary demand, not simply converting demandfor one type of energy to demand for another — unless the switch causes a net decrease in energyconsumed.Deaton and Muellbauer (1980) offer one explanation for habit persistence that does seem tofit the energy conservation situation: tastes may be affected by previous consumption experience.These authors suggest that people do not look ahead to future effects when making certain types ofdecisions; rather, they look back at what they have done before. Thus, consumers are more likelyto be driven by their past patterns of energy consumption than by their anticipation of theconsequences those consumption patterns might have. Possibly this occurs because of the relativesizes of North American households and the economies within which they operate. Each householdis so small that it can safely ignore the feedback from its actions through group actions back to itsown preferences.14Etzioni (1986) suggests another explanation for habit persistence, based on the observationthat rational behavior requires considerable effort and high costs: information on each of thealternatives must be collected and used in decision making. In contrast, habitual consumptionbehavior requires no cognitive work, preparation, development, or adaptation. Consequently, routinechanges in prices, wages, tax incentives and the like are unlikely to have an effect because it is justtoo costly, from both cognitive and financial points of view, to make the behavioral adjustments thata narrow economics-based definition of “rationality” would demand.To summarize, the literature related to reverse marketing suggests that the disappointingresults of these techniques may be a matter of habit persistence, or the perception that investment inconservation devices frees the investor from having to take any other actions to conserve, or simplythat household income in North America is so high that the conservation incentives offered so far aresimply not strong enough. The reasons are not perfectly understood, but these suggestions, especiallywhen taken together with what we know about the results of conventional marketing techniques onthe same problem, which will be discussed in the next section, provide insight into the uniquechallenges demarketing problems offer.Conventional MarketingConventional marketing programs, based on the assumption that behavior change is the resultof changes in relevant attitudes, have also been used in demarketing situations, notably in energyconservation. Numerous educational and promotional campaigns have been designed to raiseawareness of the need for energy conservation and to increase knowledge of appropriate conservingbehaviors. Results have been mixed, a sort of “good news, bad news” situation.The good news is that there has been a marked change in awareness and attitudes— NorthAmericans are aware of the energy problem and have positive attitudes toward conservation (Kellerand McDougall 1981, Milstein 1977, Mittelstaedt 1991, Olsen 1981). Furthermore, they report15engaging in a wide range of conservation activities (Claxton 1989, Cook and Berrenberg 1981,McDougall, Claxton, Ritchie and Anderson 1981).The bad news is that the aggregate energy consumption data do not show an appreciable percapita decrease (Claxton 1989), despite technical improvements that increase energy efficiency inmany products. In other words, people seem to think energy conservation is a good idea, they wantto do it, they think they are doing it, but whatever they are doing is not enough, or it is not beingdone consistently enough by large enough numbers of consumers to affect significantly the overalldemand for energy. This discrepancy between attitudes and behavior is not confined to energy use;it is also evident in other environmental issues such as solid waste, air quality, and recycling(Mittelstaedt 1991).Attitude-behavior inconsistency is not a problem unique to demarketing situations; it hasplagued behavior change theorists for years. Wicker (1969), in a review of studies attempting to linkattitudes and behavior, concluded that “it is considerably more likely that attitudes will be unrelatedor slightly related to overt behaviors than that attitudes will be closely related to actions” (Wicker1969: 68).Once again, the question is why? The explanations are well known and summarized conciselyand coherently by Fishbein and Ajzen (1975). They outline two different explanatory approaches.The “other variables” explanation is based on the assumption that attitudes are related to behavior,but that additional variables, acting either independently or as moderators, must be considered ifbehavior is to be predicted accurately. For instance, beliefs about the efficacy of detergents in coldwater might mediate the relationship between attitudes concerning cold water washing and the actualbehavior; they might also affect the behavior directly. In fact, Fishbein and Ajzen propose that theattitude-behavior relationship is always mediated, by “behavioral intention,” an individual’s subjectiveestimation of the probability that he or she will perform the behavior.16Fishbein and Ajzen also outline a “measurement” explanation. They propose that three majorfactors influence the magnitude of the relationship between intention and behavior. The first iscorrespondence in levels of specificity: the degree to which the intention is measured at the samelevel of specificity as the behavior to be predicted. Thus a person’s performance of a particularbehavior, say washing this week’s laundry in cold water, is determined by his or her intention toperform that particular behavior at that particular point in time, rather than his or her more generalintention to conserve energy.The second measurement-related factor influencing the magnitude of the relationship betweenintention and behavior is the stability of the intention. The longer the time interval between themeasurement of the intention and the performance of the behavior, the greater the chance that theindividual will receive new information that could affect his or her intention. Frequently, thebehavior can occur only after a sequence of other behaviors has been executed. The greater thenumber of intervening steps, the lower the intention-behavior correlation will be. For example,someone might intend to do the next dishwasher load only when the machine is full, if that personthen impulsively invited guests for supper, he or she might run the dishwasher half full in order toclear it before the guests arrive, thereby seeming to be inconsistent; actually the problem is that theintention was measured too far in advance of the performance of the behavior.Both the “measurement” and “other variables” explanations of attitude-behavior inconsistencyare well known and well understood, and have been for years. In fact, many undergraduate socialpsychology textbooks (e.g., Baron and Byrne 1984) devote several paragraphs to this issue. Thiswidespread understanding of the pitfalls of attitude-based behavior change programs, plus the verylarge number of energy conservation programs that have been designed and implemented by capableand competent professionals, suggest that it is unlikely that these explanations can completely accountfor the consistently disappointing results in terms of long term curtailment (as opposed to efficiency)behaviors. In fact, there is another explanation.17A third factor influencing the magnitude of the relationship between intention and behaviorhas often been overlooked, but it may be critical in understanding the failure of attitudes to predictbehavior in a demarketing context like energy conservation. This third factor is volitional control:attenuation in the intention-behavior relationship may be due to lack of ability or resources, or it maybe due to “force of habit.” The latter possibility is particularly relevant.Fishbein and Ajzen highlight the difference between habitual behaviors and those undervolitional control:Although a person may intend to do one thing, by “force of habit”he [or she] does something else. Before leaving home, a person mayintend to try a new route to his [or her] office, but later finds himself[or herself] driving along the same route he [or she] takes every day.In fact, many well-learned skills (e.g., playing the piano, driving acar) are performed almost automatically without much consciouseffort. Most behaviors of interest to social scientists, however, donot involve such automatic sequences of motor responses. Instead,investigators attempt to predict a person’s decisions, participation invarious activities, purchasing behavior, voting for political office,and interactions with other people. We have argued that these kindsof behaviors are under volitional control and thus can be predictedfrom the person’s intentions. (Fishbein and Ajzen 1975: 371)(Emphasis added.)Likewise, Janis and Mann (1977) explicitly exclude from their study of decision making the “endlessround of minor, routinized decisions a person faces when he looks over a menu in a restaurant, [or]selects a movie” (Janis and Mann 1977: 4) in favor of more substantial decision making.Thus, to the extent that the behavior of interest is like a habit, conventional behavior changetheories alone are likely to be inapplicable. If demarketing problems such as energy conservationinvolves changing habitual or habit-like behavior, then it will be necessary to look beyond socialpsychological theories of behavior change that assume volitional control.Other VariablesThis chapter has argued that habits or habit-like behaviors may play an important part in thesuccess or failure of demarketing programs. Consequently, habits and habit-like behaviors are the18focus of the theory and empirical work developed in this dissertation research. Note, however, thatother variables may also affect significantly the power of demarketing programs. There is, forinstance, a considerable body of research on altruism and prosocial behavior (e.g., Clark 1991) thatmay offer insight: it seems reasonable to expect that the degree of altruism (or, conversely,selfishness) may influence individuals’ propensity to reduce discretionary consumption. Selfishpeople would presumably be less willing than altruistic people to reduce their discretionaryconsumption, unless the benefits of doing so accrued to themselves, directly or indirectly. Similarly,the literature on time orientation (e.g., Block 1990, Hoch and Loewenstein 1991) may suggest otheruseful variables, such as the way individuals discount time. People who strongly value the presentover the future may be unwilling to make sacrifices now to prevent future hardships, even if theythemselves will have to bear those hardships.While these variables (and others) are interesting and potentially powerful in understandingdemarketing, they are beyond the scope of this research.SummaryAlthough marketers usually seek to increase demand for their products and services,sometimes they want to reduce demand — to demarket. During the 1970s, mostly in response to theEnergy Crisis, a small literature on demarketing developed, most of which is applied in its focus andprescriptive in nature. As Table 2-1 indicates, empirical validation was apparently not a priority.The best-documented demarketing problem is that of energy conservation. Two kinds ofstrategies, “reverse marketing,” based on microeconomic theory, and “conventional marketing,” basedon social psychological theories of attitude and behavior change, have been used in attempts to reducedemand for energy. Reviews suggest that both types of strategies have been relatively ineffective,at least in the short run. This is somewhat surprising, since it is not obvious why either type ofstrategy should be less effective when the aim is curbing rather than building demand. However,19closer examination of the theories on which those strategies are based reveals that a key assumptionis not being met: that the behaviors being influenced be volitional. If a consumption behavior ishabitual or habit-like, those theories may not apply. This may explain some of the disappointingresults of both reverse marketing and conventional marketing strategies aimed at reducingdiscretionary consumption, such as the insensitivity of demand for energy to price increases, and theattenuated attitude-behavior relationship in energy conservation.The next chapter examines the feasibility of viewing the demarketing problem as analogousto a habit change problem, and then elaborates on some of the ramifications of this conceptualizationof the problem.20Table 2-1Review of Demarketing ResearchKotler and Levy 1971Hanna, Kizilbash, and Smart1975Kelley and Scheewe 1975Definition and description ofdemarketingQuestioning whether themarketing concept willsurvive a sellers’ marketDevelopment of strategyguidelines for firms facingraw material and/or productshortagesExamination of the role ofmarketing during a period ofshortageDevelopment of strategyguidelines for firms facingexcess demandDevelopment of strategyguidelines for firms facingraw material shortagesDescription of effect ofshortages on consumerdecision makingDevelopment of strategyguidelines for firms facingshortages, inflation andrecessionExamination of the role ofmarketing in a world ofincreased ecological concernand resource shortagesClassification andprescriptions for managerialaction based on anecdotalevidence; no empiricalverificationInformed speculation on theimplications of shortages onthe buyer/seller relationship;no empirical verificationIdentification of marketingstrategy situations based oncombination of resourceconstraints and trend inmarket demand; no empiricalverificationStrategy prescriptions basedon literature review andinformed speculation; noempirical verificationIdentification of marketingstrategy options based onmarketing objectives andmarketing mix priorities; noempirical verificationIdentification of marketingstrategy options by marketingmix element; no empiricalverificationLiterature review ofbehavioral economicsIdentification of marketingstrategy options by marketingmix element, based onsurveys of marketingmanagers (n= 104) andconsumers (n = 969)Discussion of theramifications of a “conserversociety”Article Purpose Approach/ResultsBlankenship and Holmes 1974Cravens 1974Kotler 1974Cullwick 1975Shama 1978Shapiro 197821Article Purpose Approach/ResultsVan Dam 1978Monroe and Zoltners 1979Bellur 1980Frisbie 1980Papadopoulos 1983Dadzie 1989Reddy 1989Examination of the role ofmarketing in a world ofresource shortages, newconsumer values, and a neweconomic world orderExamination of criteria forpricing the product lineduring periods of scarceresourcesExamination of theeffectiveness of various policyoptions to curtail gasolineconsumptionExamination of the role ofdemarketing in dealing withthe energy crisisExamination of the majortypes of demarketingsituations as well as theparameters that will determineappropriate demarketingstrategyExamination of how firms inselected African firms applydemarketing in a sellers’market environmentRecommendation thatemployers can reduce healthcare costs by demarketingemployee health benefitsDiscussion of theramifications of changes inthe business environment;discussion of how cooperativeaction by marketers canaddress the wastemanagement problemDevelopment of an analyticalmodel showing that profitsare largest when “contributionper resource unit” ismaximized; no empiricalverificationRegression model suggests adecrease in automobileweight, elimination of gasguzzling gadgets (e.g., airconditioning), and eitheravoiding higher prices orinstituting a two-price policyExploratory study (n = 303)on attitudes concerning theenergy crisis and energy-related consumption behavioryielded four psychographicprofiles after factor analysisDevelopment of frameworkfor shortage analysisSurvey of selected ThirdWorld firms (n= 188) foundthree forms of sellers’markets, each using differenttypes of demarketingstrategiesList of advantages anddisadvantages of traditionalcost-reducing methods anddemarketing methods; noempirical verificationCHAPTER 3HABITS ANDHABIT-LIKE BEHA VIORLiving standards that go beyond the basic minimum are sustainable only if conswnption standardseverywhere have regard for long term sustainabiity. Yet many of us live beyond the worldecological means, for instance in our patterns of energy use. Perceived needs are socially andculturally determined, and sustainable development requires the promotion of values that encourageconswnption standards that are within the bounds of the ecologically possible and to which all canreasonably aspire. (World Commission on Environment and Development 1987: 44)The previous chapter’s review of demarketing programs and their underlying assumptionsoffered a number of possibilities to explain why most programs seem unable to reduce discretionaryconsumption significantly and consistently over time. Among the possibilities is the idea that thedemarketing strategies typically used (i.e., reverse and conventional marketing strategies) assumevolitional control on the part of the consumer when, in fact, it is not clear whether discretionaryconsumption is always volitional. If and when it is habitual, those strategies may not be the mostappropriate. That possibility is explored more fully in this chapter and the next.This chapter examines the question of what a habit is, and whether discretionary consumptioncan be thought of as a habit. It begins with an historical overview of the literature on habits and adiscussion of the definitions and descriptions which have appeared in general reference dictionaries,economic theory, and psychological research. It then proposes a new definition of habits. Since thatdefinition highlights the role of automaticity in habitual behavior, there is also a summary of research2223on automatic processes, highlighting the relative nature of automaticity. It is then suggested that theconcept of a habit is also relative, and a definition of habit-like behavior is provided. The chapterconcludes with a discussion of the extent to which discretionary consumption of the type thatdemarketing programs aim to reduce can be characterized as habit-like.Habits: A Brief IntroductionAs Fishbein and Ajzen (1975) and Janis and Mann (1977) point out, social psychologists havebeen interested mostly in volitional behavior, behavior that is the result of conscious choice. Thestudy of what would commonly be thought of as habits — consistent patterns of action (or thought),acquired by frequent repetition and demanding little or no attention — has not been a priority, despitethe fact that habits govern much of our daily lives.Most people have many routines that they follow without much thought, such as brushingtheir teeth after meals, getting to and from work every day, answering phone messages aftermeetings, watching the evening news on television, doing the laundry and grocery shopping. In thenormal course of events, they rarely, if ever, consider alternative courses of action, and pay little orno attention to these activities. To the extent that such routines involve frequently repeated behaviorsrequiring little attention (unless there are disruptions to the expected stream of behavior andconsequences, in which case they may receive considerable attention), they can be called habits.Historical OverviewIn the early days of social psychology, habits were considered by prominent researchers tobe a central issue. Over 100 years ago, William James, in his classic work, Principles ofPsychology(1890), described the origins and consequences of habits, and distinguished between habits andconscious decisions. John Dewey, in the preface to his book, Human Nature and Conduct (1922),stated that ‘an understanding of habit and of different types of habit is the key to social psychology.”24By 1932, Knight Dunlap, in Habits: Their Making and Unmaking, was able to reflect on 50 yearsof “rapid growth in the study of habit formation” (Dunlap 1932: viii). Behaviorist psychologists likeC.L. Hull (1943) considered habits and their formation to be a very important issue. Socialpsychologists, however, were much less interested in habits. But by the late 1920s and early 1930s,when Thurstone and Likert started to publish their seminal work on attitudes (Likert 1932, Thurstoneand Chave 1929), the focus in social psychology was shifting away from habits (Mixon 1980). Thestudy of attitudes has dominated social psychology since the 1950s (1’esser and Shaffer 1990); duringthis period, social psychological research on habits has been, until recently, virtually non-existent.Lately, however, habit has re-entered the picture. Researchers examining the attitude-behavior relationship have been questioning how often people rationally consider alternatives indeciding what course of action to take. Tesser and Shaffer (1990), in their review of recent literatureon attitudes and attitude change, cite research showing that people frequently behave on the basis of“automatically activated attitudes”. Ronis, Yates and Kirscht (1989), in a review of the influenceson repeated health-related behaviors (i.e., repeated behaviors that have clear positive or negativeeffects on health), conclude that repeated behaviors are usually determined largely by habits ratherthan by attitudinal variables, although attitudes are central to the formation and modification of habits.In an empirical study, Mittal (1988) found that habits (both pro- and anti-intentional) are significantlyrelated to seat belt use, even after the effects of attitude are taken into account. On the theoreticalside, Triandis (1979) explicitly included habits in his theoretical framework explaining the relationshipof attitudes, values, and “other acquired behavioral dispositions” to behavior.Researchers in economics are also paying increasing attention to habits, due mostly to theircloser examination of two issues, rationality and the independence of behavior over time. Thediscipline of economics focuses on “rational human behavior,” where rational behavior is defined as“nothing more than action well-suited to achieve your goals, within your limitations and capacities”(Hirshleifer 1976: 7). Habits, particularly “bad” habits, have traditionally been thought of as25“irrational” and therefore outside the domain of economics. However, Becker and Murphy (1988)were able to show analytically that addiction (the most extreme form of a bad habit) is, in fact, arational phenomenon. In addition, economists have been re-examining the assumption that choicestoday are independent of choices in the past. Becker (1992) points out that, despite the fact that thisassumption was criticized as early as 1965, it discouraged economists from grappling with significantissues such as habits, addictions, and traditions until relatively recently.Part of the problem for both psychologists and economists may also be that a satisfactorydefinition of “habits” is elusive.Definitions and Descriptions of HabitsGiven that most people would claim that they are familiar with both the concept and theexperience of habits, it is surprisingly difficult to find or develop a definition that is broad enoughto encompass the many instances of behaviors that people would call “habits,” yet specific enoughto be operationalized in a research setting.Dictionary definitions capture many of the connotations associated with the everydaydefinition of habits. The Random House Dictionary of the English Language (1983) lists no less thanseven relevant definitions:habit (hab’it), n. 1. customary practice or use: Daily bathing is anAmerican habit. 2. a particular custom, practice, or usage: thehabit of shaking hands. 3. compulsive need, inclination, or use;addiction: liquor habit; drug habit. 4. a dominant or regulardisposition or tendency; prevailing character or quality: She has ahabit of looking at the bright side of things. 5. Slang. an addictionto narcotics (often prec. by the). 6. an acquired behavior patternregularly followed until it has become almost involuntary: the habitof looking both ways before crossing the street. 7. mental characteror disposition: a habit of mind. (Random House Dictionary of theEnglish Language 1983: 634)These definitions accomplish their primary purpose, to provide general information on meaning,pronunciation, spelling, usage, and etymology, but they present problems for empirical research.26Words and phrases like “customary”, “compulsive”, “regular”, and “almost involuntary” areimprecise, context-specific, and subjective. In short, they are difficult to operationalize. Severalresearchers have attempted to overcome these problems by crafting definitions in terms reflecting theperspectives of the research traditions within which they work.From the consumer behavior point of view, Assael (1992) offers this definition:Habit can be defined as repetitive behavior resulting in limitation orabsence of (1) information seeking and (2) evaluation of alternativechoices. Learning leads to habitual purchasing behavior if theconsumer is satisfied with the brand over time. After repetitivepurchases, the consumer will buy the brand again with littleinformation seeking or brand evaluation. (Assael 1992: 79).This is a useful definition, as it not only emphasizes the role of repetition in habits, but alsohighlights an important attribute of habitual behavior, that it limits consideration of other alternatives.As long as the system in which the habit is operating remains stable, this limitation is nodisadvantage, but if circumstances change, a habit can reduce the individual’s effectiveness inresponding. People often stick to the old ways of doing things when new alternatives may besuperior (although when challenged, they usually have a rationalization). For instance, in may areas,commuters cling to their cars when taking the bus would save them both time and money. One ofthe reasons may be that they have a habit of driving to and from work, developed when trafficcongestion was not as severe, bus lanes were rare, and downtown parking rates were not as high, andthey never seriously consider public transit (though, when challenged, they may provide an elaboraterationalization, in which “freedom” probably figures prominently). So, this definition points out the“stickiness” of habitual behavior, and attributes it to lack of consideration of alternatives. This is auseful insight, but it does not capture the full character of habits. Other definitions, from other pointsof view, provide additional information.Becker (1992) presents an economic definition: habitual behavior displays “a positive relationbetween past and current consumption” (Becker 1992: 328). For most goods, consumption at timet will lead to reduced consumption at time t,,, as long as the time intervals are short and the27quantities consumed are large enough. (Note that “short” and “large enough” share the imprecision,context-specificity, and subjectivity of the dictionary definition.) For instance, half an hour afterdrinking a litre of water, most people would have little desire to drink more. But for many goods,when t, and t+, are not close together, consumption at time t, stimulates increased rather thandecreased consumption of the good at time 1,. Under these circumstances, a habit is said to beoperating. For instance, while most people would not want another bowl of cereal immediately afterhaving eaten one, they might want one the next morning. When this occurs every day (i.e., for everyi and where the interval between t and i÷, is one day), then the person can be said to have ahabit of eating cereal for breakfast. Likewise, while one might not want a heavy brunch everymorning, it could become a habit to have it every Sunday morning (i.e., the interval between t. andt, is one week).Becker’s (1992) definition is useful for two reasons. First, it highlights the fact that forhabits, current consumption is not independent of past consumption. Second, it specifies the natureof the nonindependence, allowing economists to explore the implications of habitual behavior. Forexample, Becker and Murphy (1988) are able to demonstrate that addiction, the strongest form ofhabit, previously regarded in economics as completely irrational, can actually be quite rational; andBecker, Grossman, and Murphy (1991) have looked at the effect of price on consumption in rationaladdiction. The problem with this definition is that, while it may be adequate to obtain analyticalresults, it is not very useful for empirical research, as the relevant time intervals would vary by bothgood and individual.Becker makes another contribution, by distinguishing between “good habits” and “bad habits.”We make this distinction regularly in everyday life — good habits, such as regular exercise, are thosethat we believe will lead to benefits such as better health and longer life; and bad habits, such aswasteful use of energy, will have detrimental effects in the future, such as higher bills and28environmental degradation. (Surprisingly, none of the psychological literature reviewj had anythingto say about good and/or bad habits.) Becker (1992: 328) says:Habits are harmful or “bad” if greater present consumption lowersfuture utility, as in the detrimental effects on future health of heavysmoking or drinking. Similarly, habits are beneficial if greaterpresent consumption raises future utility; regular swimming orregular church attendance may be examples.There is considerable variability in the definitions used in psychology. Some authors usedefinitions similar to those given in general purpose dictionaries. Mixon (1980) describes habits as“physical, emotional and cognitive ways of behaving” (Mixon 1980: 178); “a characteristic way aperson behaves” (Mixon 1980: 183); and “organized, skill-like ways of behaving” (Mixon 1980:184). Verville (1988) defines a habit as “a learned way of behaving: a pattern of acting, thinking,or feeling that has become routine” (Verville 1988: 3). These characterizations of habits may beadequate for some purposes, but they have limited utility in an empirical setting, as they would bedifficult to operationalize. Fortunately, a number of authors provide additional detail.A concise and coherent definition is provided by The Penguin Dictionary of Psychology(Reber 1985), in which habitual behavior is defined as “patterns of activity that have, throughrepetition, become automated, fixed and effortlessly carried out” (Reber 1985: 314). Note that therequirement for repetition is consistent with the economic definition of habits. Other definitions aremore descriptive. Triandis (1979) defines habits as:situation-specific sequences that are or have become automatic, sothey occur without self-instruction. The individual is not usually“conscious” of these sequences. . . . Habits require learning .Certain cognitive schemata [are habits] . . . They are elicited bypatterns of stimuli, and a consequence of their activation oftenincludes a complex sequence of behavior. (Triandis 1979: 204)Mittal (1988) elaborates on the unconscious nature of habits, referring to them as recurrent orrepeated behavior that occurs without awareness: “awareness is the discriminating characteristic”(Mittal 1988: 997). Based on this requirement, Mittal argues that “habits” such as brushing one’steeth are not necessarily habits at all, because although the behavior occurs repeatedly, and one can29develop stable patterns of activity (and this is certainly necessary for a behavior to be considered ahabit), “if the behavior recurs with awareness, it must be deemed as being driven by intentions, orself-instruction, not habit” (Mittal 1988: 997). According to this definition, then, habit-drivenbehavior presupposes a lack of awareness of plans and self-instructions about the act and of the actitself— during the actual performance of the behavior. Consequently, wearing a seat belt would beclassified as being a habit if one were unaware of putting on the seat belt at the moment it is beingput on. Subsequent awareness of the behavior is allowed. In fact, for Mittal ‘s study, ex-postawareness is necessary, as the measures of habit depend on self-reports.The definition of habits provided by Ronis, Yates, and Kirscht (1989) is largely in line withboth Mittal (1988) and Triandis (1979):Habits are the results of automatic cognitive processes. Suchprocesses develop by extensive repetition. Automatic processes areso well-learned that they do not require conscious effort. Automaticprocesses are unintentional. They are typically set in motion bystimulus cues. Automatic processes can go on simultaneously withother cognitive processes without any interference. We are oftenunaware of automatic processes. (Ronis, Yates, and Kirscht 1989:219)Unlike Mittal (1988), however, these authors would not exclude teeth-brushing from the habitcategory. “Most adults, for example, brush their teeth at one or more specific times during the day,without a daily consideration of whether or when to do it” (Ronis, Yates, and Kirscht 1989: 220).Here, the lack of awareness applies to the decision to brush one’s teeth, rather than the actual act ofdoing so.The distinction between the behavior and the decision to behave is important. It suggests thathabits may be the result of automation in either decision making or behavior (or both). Thus thedomain of habits includes not just habitual behavior, but habits of thought and decision. To reflectthis, the definition of habits proposed in this research, Definition 3.1, is based on the Reber (1985)definition, but explicitly recognizes that habits are not confined to behaviors, although they may be30manifest as behavior. In addition, it is compatible with the economic definition of habits, as ithighlights the role of repetition in the establishment of habits.Definition 3.1A habit is a pattern of activity (behavior, decision, perceptual bias,etc.) that has, through extensive repetition, become automated, fixed,and effortless, and occurs without conscious awareness orconsideration of alternatives.At the core of a habit is a set of “automated activities.” In order to understand habits, it is necessaryto understand these activities which are called “automatic processes” in cognitive psychology.Automatic Processes’Cognitive psychologists have shown that the execution of a complex behavior requires thecoordination of many component subskills (LaBerge and Samuels 1974). Given the capacityconstraints on the human cognitive system, and the fact that many complex behaviors require bothsimultaneous and sequential processing and/or performance of various component subskills, it is clearthat most complex behaviors could not be executed if each process required attention. Some of thecomponent processes therefore become “automatic.” Automatic processes, in contrast to “controlled”or “effortful” processes, occur without intention2, without necessarily demanding attention orconscious awareness, and without interfering with other processing (Hasher and Zacks 1979, LaBerge1984, Posner and Snyder 1975). They are triggered by an event (which may or may not be thesubject of conscious awareness), they are difficult to suppress once triggered, they run to completion,and they do not provide the individual with any new information (Schneider and Shiffrmn 1977,Shiffrmn and Schneider 1977). In the meantime, attention can be focused elsewhere, on controlled1Note that most research on automatic processes studies the microprocesses of perception (seeLaBerge 1981 for a review), but researchers have adopted the general ideas to understand morecomplex behaviors and emotions (Greenberg and Safran 1984, Ronis, Yates, and Kirscht 1988).2Note that the lack of intention refers to process rather than goals. For instance, a skillful righthanded tennis player faced with a ball coming to his or her left intends to return the ball (goal) butdoes not intend to use his or her backhand in a particular way (process); he or she just does it.31processes. Automaticity, then, refers to specific properties of performance: tasksthat can beperformed quickly, effortlessly, and relatively autonomously are thought to be automatic(Logan1985).Some automatic processes are thought to be innate (e.g., encoding the frequencies, spatiallocations, and time of events), and do not change much with practice; others (e.g., in reading,knowing that a certain pattern of letters makes a word with a certain meaning) are acquired gradually,over a period of time, through practice (Hasher and Zacks 1979).Most of the time, automating recurring processes is an effective response to the fact that weface a complex environment equipped with a cognitive system that has severecapacity constraints(Kahneman 1973). However, by their nature, automatic processes lack flexibility, so they cannoteasily be adapted to take account of changing circumstances, even when there is a conscious effortto inhibit them. Consequently, making changes to an automated process is difficult, as the nextchapter will detail.Automaticity is not an absolute; it is a relative concept. It is viewed asbeing on acontinuum, although there is some disagreement as to the nature of the continuum. Some researchersconceive of the continuum as having two endpoints, automatic processing and controlled processing(Hasher and Zacks 1979, Shiffrin and Schneider 1977). In other words, the more automated theprocess is, the less control the individual has over it. Other researchers suggest thatautomaticity isa continuum with no discrete endpoints, and that automatic and controlled processes are not opposites.Logan (1985) points out that skilled performers (such as experienced pilots) are usually able tocontrol their performance better than unskilled performers (such as novice pilots), even though theirperformance is likely to be more highly automated, probably because skilled performers are betterable to anticipate future responses. Logan’s observation is not necessarily inconsistent with the notionthat automatic and controlled processes are opposites, at least at the component subskill level. Eachsubskill involved in a complex skill could become more automated and less controlled as practice32accumulates, while the complex skill itself could remain a controlled process. In other words, theremay be processes within processes, and at each level the various processes may be more automatedor more controlled, but more automation at one level does not imply more automation at the nexthigher level of organization.Since this research defines a habit as a set of automated activities, the idea that there is acontinuum between automatic and controlled processes has implications for the way that habits areconceived of in this research.Habit-like BehaviorDefinition 3.1 (habit), while clearer and more consistent than many of the definitionspreceding it, is still difficult to operationalize if the terms used to describe it, such as “automated,”“fixed,” and “effortless,” are absolutes. It would be difficult to demonstrate that a behavior ascomplex as the consumption behaviors that are the subject of this research is, in fact, automated,fixed, and effortless. However, since a habit is an automated activity, and automation is a relativeconcept, then it follows that habits are a relative concept. Furthermore, just as it seems that thereis a continuum between automatic and controlled processes, it seems reasonable, by analogy, to posita continuum between habitual behavior and volitional behavior. To make explicit the distinctionbetween the absolute and relative concepts of habits, the following definition is offered:Definition 3.2Habit-like behavior is behavior that has been sufficiently well-practiced that its execution requires little or no attention.This definition has both advantages and disadvantages when compared to the more traditionalDefinition 3.1. The main disadvantage is that it is not clear cut: we cannot say that a particularbehavior is or is not a habit; being able to make that distinction would be helpful in empiricalresearch. A clear advantage is that this definition is both intuitively plausible, and consistent withfindings from research on concepts and concept formation, which indicate that when people decide33whether or not something (e.g., brushing one’s teeth after eating) belongs to a class (e.g., habit) theydo not usually appeal to a list of necessary and sufficient conditions that clearly define categorymembership; instead they take a more “probabilistic” approach. People have in their minds a concept(e.g., habit) that is represented in terms of properties that are only characteristic or probable, ratherthan defining (Medin and Smith 1984). Thus membership in the “habit” category would be on acontinuum, from more habit-like (and less volitional) to less habit-like (and more volitional), wherebehaviors that are more habit-like possess more characteristic properties than do less habit-likebehaviors.What are these characteristic properties? The definition of habit-like behavior and theprevious discussion of automatic processes suggests four properties:Property 1:Habit-like behavior requires practice to develop.Since a habit-like behavior is a complex of relatively automatic processes, and automatic processes(with the exception of innate automatic processes, such as those involved in perception) are acquiredthrough practice, a behavior can become habit-like only with practice.Property 2:Habit-like behavior takes time to develop.This follows directly from Property 1. Practice takes time. A new, volitional behavior graduallybecomes more habit-like as it is repeated over time.Property 3:A habit-like behavior is initiated by one or more “triggers”.Since habit-like behaviors are a complex of relatively automated processes, and automated processesrequire activation, it follows that habit-like behaviors also have triggers that activate the componentautomated activities.Property 4:Changing habit-like behavior requires effort and attention.34Because habit-like behavior is a relatively automated activity, it is relatively inflexible. Oncetriggered, the automatic processes tend to run to completion. However, depending on the degree ofautomation, overriding the automatic process is possible, given sufficient attention and effort (Rasherand Zacks 1979, Shiffrmn and Schneider 1977, Logan 1985).Does Demarketing Attempt to Change Habit-like Behavior?This chapter began by suggesting that discretionary consumption may be more habit-like thanvolitional, the implication being that behavior change strategies that are appropriate for volitionalbehavior may be less effective in altering habit-like behavior. A review of the literature on habitsand their underlying automatic processes followed. An important issue remains unresolved: whetherthe types of consumption behaviors that demarketing programs seek to modify can reasonably becharacterized as habit-like.While counterexamples can probably be found, most demarketing programs, though they mayuse strategies that assume volitional control, do attempt to change behaviors that seem to fit thepatterns attributed to habit-like behavior. Consider, for example, programs aimed at curtailinghousehold energy consumption. They urge consumers to turn off unnecessary lights, to lowerthermostat settings in the winter and to limit the use of air conditioning in the summer, to run thedishwasher only when it is full, and to wash laundry in cold water. In each of these situations, thepattern of events is consistent with that associated with automatic processes and habits: there is atrigger, such as noticing that ambient light levels are dropping, feeling cooler than is comfortable,realizing that there won’t be enough clean dishes to set the table for the next meal, or knowing thatit is washday, which sets off in the individual a stream of behavior which does not demand muchattention to perform competently, but which accomplishes superordinate goals such as physicalcomfort. Furthermore, that stream of behavior is consistent from one performance to the next.35It seems reasonable, therefore, to conclude that much discretionary consumption is, in fact,habit-like. This does not deny that there are instances of discretionary consumption that are relativelyvolitional. But the volitional behavior change problem is relatively well-understood. Changingbehavior that is complex and habit-like is not as well understood. Consequently, this research focuseson reducing habit-like discretionary consumption.What, then, are the implications of framing the demarketing problem as a habit-changeproblem? The next section begins to address that question.Habit-like Behavior in Marketing and DemarketingThe preceding discussion of differences between volitional behaviors and habit-like behaviors,as well those between conventional marketing programs and demarketing programs does not implythat habit-like behavior is of concern only in demarketing; it is important in both marketing anddemarketing. However, the focus is different. Many marketing programs are aimed at developinghabit-like behavior, while many demarketing programs are aimed at changing existing habit-likebehaviors. This is a crucial difference.At the most basic level, marketing is the facilitation of transactions — exchanges in whichthe customer pays a price (monetary and/or nonmonetary) in order to receive a package of benefits(the product) that is intended to satisfy his or her wants or needs (Houston and Gassenheimer 1987).Much research effort has been devoted to understanding the nature of and processes underlying thesetransactions. Of the many marketing constructs that have been used to explain various aspects ofmarketing transactions, two are particularly applicable to this research: habitualpurchasing behaviorand the related concept of brand loyalty.Habitual purchasing behavior occurs when a customer has settled on a regular brand,preferably due to brand loyalty, rather than mere inertia. For instance, someone might buy aparticular brand of soap regularly, without engaging in extensive search for and evaluation of36information. Such habitual purchasing behavior benefits both the marketer and the consumer, becauseit eliminates the search and evaluation process, which is expensive for both the organization (whichmust ensure that the needed information is available at appropriate times and places), and for thecustomer (who must spend valuable time and energy acquiring and processing the information).Marketing efforts, especially for packaged goods, are often directed towards encouraging habitualpurchasing behavior — getting consumers to form the habit of buying a particular brand. Heavyadvertising and point of purchase support are used to remind customers of the brand; intensivedistribution ensures that the customer will see the brand frequently. The importance of habitualpurchasing and brand loyalty is reflected in the attention it receives in university textbooks. Mostconsumer behavior textbooks include a lull chapter (e.g., Assael 1992, Engel and Blackwell 1982)or a major section within a chapter (e.g., Peter and Olson 1993) on the topic.In most conventional marketing situations, the customer’s perception is that most of the costsand benefits associated with a transaction are known, and will accrue in a known time frame, usuallythe short run: customers think they know what they are getting, and they know how much it coststhem. For instance, a bottle of wine has a monetary cost, and may also have some nonmonetarycosts, such as the aggravation involved in finding a parking spot in a busy liquor store parking lot.If the wine is consumed with a good meal in pleasant company, the benefits include nourishment ofboth the body and the soul. Clearly, in cases like this, most of the costs and benefits associated withthe transaction are perceived to occur during, just before, or just after consumption of the product.If the consumer enjoyed the bottle of wine, he or she is more likely to buy that brand again. Themarketer’s goal is to ensure that consumers do enjoy the wine, and that they purchase the brand overand over, each time with a little less information seeking and evaluation of other brands. Thisrepresents the development of a habit-like behavior.In contrast to conventional exchanges like the purchase of a bottle of wine, many demarketingexchanges involve more than developing habit-like behaviors; before a new habit can be developed,37an old habit in the same behavioral domain must be extinguished. That is often difficult, becausethere may not be many reinforcing benefits associated with breaking old habits. Table 3-1 highlightssome of the important differences between conventional marketing and demarketing with respect tohabit-like behavior. In contrast to what it takes to develop new habits, breaking old habits ofteninvolves convincing consumers:• to pay what they view as a significant price in the short run (because breaking habits iseffortful),• to continue to pay that price indefinitely (since new habits often take considerable time tobecome consolidated),• to pay the price despite uncertain returns (as there are no guarantees that the hoped-for resultwill actually occur),• to pay the price even though most of the benefits can only be enjoyed well into the future(because it takes a long time to undo the damage already done),• to pay the price even when people other than those making the sacrifice will enjoy thebenefits (since in many cases demarketing involves public goods).Table 3-1Comparison of Conventional Marketing and DemarketingDimension Conventional Marketing DemarketingMarketing task Increase demand Decrease demandDesired consumer behavior Consume more of something Consume less of somethingdesirable desirableConsumer perception of cost Short run OngoinghorizonConsumer perception of Short and/or long run Long runbenefit horizonConsumer confidence that High Lowpersonal benefits will exceedcostsBenefit realized Personal Personal or Social38For example, encouraging adults to increase their levels of physical activity (i.e., demarketingsloth) as a preventative health measure means persuading people to change their daily routines byadding a new activity, which many of them will never really enjoy, for the uncertain benefit of betterhealth and longer life, which arguably benefits the health care system as much as it does them. Itis not as easy a sell as the relatively clear cut offers in conventional marketing.Demarketing is therefore often a more ambitious undertaking than is conventional marketing.The demarketer faces a more difficult task on several dimensions. The next chapters examineapproaches to changing habit-like behaviors.SummaryThis chapter has explored the possibility that demarketing programs are aimed at changingbehavior that is more habit-like than volitional. It provided an historical overview of the literatureon habits and reviewed critically the definitions and descriptions of habits. In order to integratevarious views and to resolve some of the problems with earlier definitions, a new definition wasproposed. Since automaticity is a key component of the new definition of habits, the chapter alsooffered a brief review of research on automatic and controlled processing. That review highlightedthe notion that automaticity is relative, not an absolute. It was then argued that, by analogy, habitsare also relative. Hence, the concept of habit-like behavior was introduced, defined, and described.The chapter then argued that demarketing does, in fact, deal with habit-like behaviors. The chapterconcluded by discussing some of the implications of framing the demarketing problem as a habitchange problem. The following chapters look at changing complex, habit-like behaviors.CHAPTER 4A COMPREHENSIVE MODEL OFHABITUAL BEHAVIOR CHANGEThe time has come to break out ofpastpatterns. Attempts to maintain social and ecological stabilitythrough old approaches to development and environmental protection will increase instability.Security must be sought through change. (World Commission on Environment and Development1987, p.22)Chapter 2 showed that demarketing strategies based on existing marketing theory and methodsare, at best, inconsistently successful. It suggested that one reason for such disappointing resultsmight be that traditional marketing approaches are well suited to changing volitional behavior, butexcessive discretionary consumption is often more like a habit than a volitional behavior. Chapter3 explored the implications of this notion. Among other things, it argued that most marketingprograms are aimed at developing habit-like behavior (i.e., habitual purchasing behavior and brandloyalty), while most demarketing programs are aimed at changing existing habit-like behaviors. And,as most people know from experience, it is usually easier to develop a new habit than to change anexisting habit. This chapter begins to address the next logical question: How can existing habit-likebehaviors be changed?A logical first step in attempting to answer this question is to refer to the existing literatureon changing habitual behavior. Reviewing the literature on habits reveals that, while there are plentyof suggestions as to how to change various specific habits, these appear to have developed on an ad3940hoc basis, according to the particular characteristics of the specific habit. Each habit-change strategyseems to be driven by its own, habit-specific theory or theories. For instance, strategies to rid peopleof the nail-biting habit are based on theories of why people bite their nails, while strategies to helppeople change their eating habits are based on theories of why people develop certain eating habits.Such ad hoc approaches are so habit-specific that they tend not to generalize well.It should be possible, however, to look across the many studies to determine whether thereare basic, common principles underlying the various ad hoc habit change strategies. Only oneresearch team seems to have attempted this. James 0. Prochaska and Carlo C. DiClemente and theircolleagues (henceforth referred to as “P&D”) have, since the early 1980s, been attempting tounderstand why and how people change, especially with respect to certain kinds of habit-likebehaviors.This chapter outlines the P&D model of behavior change. The model was developed in apsychotherapeutic setting, an area not typically associated with marketing, so the chapter begins withan explanation of why psychotherapeutic theory is an appropriate starting point for understandingbetter the demarketing problem. It continues with a description of P&D’s Revolving Door Model ofBehavior Change (e.g., Prochaska and DiClemente 1984) and its variations. It also discusses someof the implications of the model for demarketing.Why a Psychotherapeutic Model?Several disciplines study behavior change in humans. Marketing has traditionally drawn onpsychology and economics to understand how people behave in their roles as consumers.Psychotherapy, “the treatment of emotional or mental disorders by psychological methods,”3 is afield that has attracted little interest from researchers in marketing. But marketing and psychotherapyare not as far apart as they might initially seem.3Accord ing to Webster’s ii New Riverside Dictionary.41Marketing and psychotherapy share a common foundation discipline, psychology. Likemarketing, psychotherapy has a more applied focus than psychology. Both marketing andpsychotherapy share the goal of encouraging people to change their behaviors in ways that will makethem become happier and more productive members of society. As well, psychotherapy oftenaddresses problems that arise from habitual behaviors — exactly the kind of behavior that is ofinterest in the case of demarketing.Since business marketing has, by definition, a commercial focus, its emphasis is on the useof products as the route to happiness. Public and nonprofit marketing has a broader, less commercialperspective, because the primary organizational goal is not profit. Psychotherapy, based on a medicalmodel, assumes that people sometimes need help in changing behavior, and that such help is notsolely available through the prefabricated products and services.It is clear, therefore, that there are important parallels between the demarketing problem andmany psychotherapeutic problems. It makes sense that some therapeutic insights would betransferable to the demarketing context, although adaptation to the constraints of the market wouldno doubt be necessary. What follows is a description of one psychotherapeutic model that may offera better understanding of the processes necessary to reduce habit-like consumption.Development of the ModelWithin psychotherapy, there are many approaches to accomplishing behavior change. In fact,there are over 200 therapies, each advocating a different set of methods (Prochaska and DiClemente1984). Each therapy is better suited to some situations than others, so many therapists do not alignthemselves with a single therapeutic school, but use an eclectic group of therapies, tailored to eachtreatment situation. In recent years, there have been calls by leaders in the field for integrativemodels based on commonalities across effective therapies (e.g., Goldfried 1980).42The P&D model of behavior change (e.g., Prochaska and DiClemente 1984, Prochaska,DiClemente, and Norcross 1992) is one response to this need. It is a comprehensive model thatdescribes the behavior change process as a recursive sequence of stages, each with a different patternof attitudes, behavioral intentions, and behaviors. It deals both with when people change and howthey change.To identify the stages, the methodological approach taken by P&D has been to ask peopleinvolved in both self-initiated and professionally-assisted behavior change questions concerningrelevant attitudes, behavioral intentions, and behavior. The resulting data has been submitted to bothprincipal component analysis and cluster analysis to yield a series of stages of change. Versions ofthe stage model have been thoroughly tested in the context of smoking cessation (DiClemente andProchaska 1982, DiClemente et al. 1991, Pallonen et al. 1990, Prochaska 1985, Prochaska andDiClemente 1983, Prochaska and DiClemente 1984, Prochaska et al. 1988, Velicer et al. 1985) andhave also been validated in several other areas: alcoholism, drug abuse, sexual dysfunction, maritalproblems, and psychic distress (McConnaughy, Prochaska and Velicer 1983, MeConnaughy et al.1989, Prochaska and DiClemente 1982, Prochaska and DiClemente 1984, Prochaska and DiClemente1986a, Prochaska and DiClemente 1986b). The range of applications in which therapies based onthis approach have been successful suggests a robustness that bodes well for its extension todemarketing situations.Note that, although the P&D model is derived from research on addictions, it would beincorrect to infer that there is an assumption underlying this dissertation research that we are“addicted” to excessive consumption. The use of an addiction-based model merely reiterates that oneof the reasons it is difficult to get sustained reductions in discretionary consumption is that there aremany habit-like behaviors associated with consumption. Addiction research is relevant here for tworeasons. First, an addiction is an extreme form of a habit. Second, addictions comprise a categoryof habits on which considerable research has been focused. Furthermore, it seems reasonable to43assume that a change strategy that works for an extreme habit (i.e., an addiction) will probably workfor a less extreme habit, or a habit-like behavior.The Linear Version of the ModelIn their early research (Prochaska and DiClemente 1982, 1983), P&D identified foursequential stages of change: precontemplation, contemplation, action, and maintenance, as shownin Figure 4-1. Each stage represents a unique pattern of attitudes, behavioral intentions, andbehaviors.Figure 4-1The Linear ModelPREc0NTEMPLATI0N- ComEMPLATIoN-, AcTIoN - MAINmNANcEPrecontemplation. In the precontemplation stage, people are unaware that they have aproblem, and they are not thinking seriously of changing. Their lack of awareness may be due toignorance, or it may be that they are “in denial.” Since moving forward from precontemplation islikely to involve giving up something, as well as risking failure, defensiveness is a natural reaction.It is often manifested in a desire for others to change.Contemplation. Contemplation is the stage at which people realize that they should change.They try to understand the problem, its causes, and what they can do about it. They weigh the prosand cons of changing their behavior, keenly aware of the benefits from the behavior that ought to bechanged, as well as the effort and energy it will take to make a change. People in the contemplationstage often need to talk about the problem. They may seek reassurance that it can be overcome, oftenthrough reading relevant articles and books. Despite their eagerness to talk, contemplators have notmade a commitment to change. Many people remain in the contemplation stage for long periods,44thinking of changing their behavior, telling themselves they ought to change their behavior, but neverquite feeling ready to do it.Action. In the action stage, people make the commitment of time and energy to change theirovert behavior and the environmental conditions that affect their behavior. Their sense of self-efficacy is high. In addition, the changes they make often get the attention of others, so people inthe action stage tend to receive social approval, which tends to reinforce the new behavior. Action,however, is only the beginning of change, and people who fail to appreciate that are not likely toprepare themselves properly for the effort it takes to maintain change over the longer term.Maintenance. Maintenance is the stage in which people consolidate their gains, and stabilizethe new behavior. This behavioral stability is not an absence of change, but its continuation. Peoplein maintenance are making efforts to avoid slipping back into old patterns of behavior. This stageoften lasts a long time. For some people, it never ends.The goal of the change process is terminolion, a state in which there is no longer anytemptation to return to old habits. Some people are never able to truly terminate their problembehavior, even though they may never go back to their old ways, because for some reason, they haveto continue to work to maintain their new behavior patterns.Demarketing Insights from the Linear ModelThe linear version of the model offers some useful insights for demarketing. First, itprovides another explanation for the common but puzzling finding that many people have positiveattitudes toward energy conservation, but they seem not to follow through in their behavior. TheP&D linear model would predict that there will be a sizeable segment of the population (people inthe contemplation stage of change) with positive attitudes toward energy conservation, but nobehavioral manifestation of these attitudes.45A second insight is that, since the patterns of attitudes, behavioral intentions, and behaviordiffers from stage to stage, demarketing programs must be designed so that they are appropriate forthe stage that people are in. For instance, if a person is in the contemplation stage, an action-orientedprogram, no matter how good it is, is unlikely to be effective, because the contemplator is simplynot yet ready to take action.Indeed, in their research on smokers, P&D have found that professionals typically designexcellent action-oriented programs, but the vast majority of people are not in the action stage:aggregating across studies and populations, only 10-15% of smokers are prepared for action, while30-40% are in the contemplation stage, and 50-60% are in precontemplation (Prochaska, DiClementeand Norcross 1992). Consequently, the programs fail to attract many people and/or have high dropout rates:Action-oriented therapies may be quite effective with individuals whoare in the preparation or action stages. These same programs maybe ineffective or detrimental, however, with individuals inprecontemplation or contemplation stages. (Prochaska, DiClementeand Norcross 1992: 1106)This suggests yet another explanation for the disappointing results of energy conservationprograms. It is not outside the realm of possibility that some, perhaps most, energy conservationprograms are very effective — for people in a particular stage of change. If there are relatively fewpeople in that stage of change, however, the overall results would be poor.Readers familiar with the literature on marketing communication may have noticed asimilarity between the P&D linear model and the well-known response hierarchy models, such as theAIDA model (Strong 1925), the Hierarchy of Effects model (Lavidge and Steiner 1961), theInnovation-Adoption model (Rogers 1962) and the standard Communications model (e.g., Kotler andTurner 1992). Does the P&D linear model add anything? In one sense, no: the linear version ofthe model can be viewed as another variation on the same theme. In another sense, it is differentbecause it is intended to the used differently. The response hierarchy models are used primarily in46determining the communication objectives, with the overriding goal of eventually moving the targetaudience through the stages to purchase. P&D’s linear model can also be used in this manner.However, the P&D approach also entails segmenting the market by stage, and the stage of changeis the primary segmentation variable. This is a different way of thinking about the market.Another difference between the response hierarchy models and the P&D linear model is thatthe latter is merely a simplified version of a more complex model, the Revolving Door Model ofBehavior Change.The Revolving Door Model of Behavior ChangeP&D observed that a problem with the linear model is that many people do not progressthrough the stages sequentially (Prochaska and DiClemente 1984) . Some get stuck in a particularstage (usually contemplation). Some give up and return to the precontemplation stage. But the mostcommon pattern is to go into relapse by returning to the problem behavior after an initial slip. Infact, in many types of behavior change, relapse is the rule rather than the exception. However, mostpeople do not give up after relapsing; a significant majority go back into the contemplation stage.This process can happen several times. The average self-changing smoker, for example, relapsesthree times before becoming relatively free of the temptation to smoke (Prochaska and DiClemente1984). For people like this, a cyclical pattern would be a more realistic representation of the processof change. The result, a Revolving Door Model ofBehavior Change, is shown in Figure 4-2.Demarketing Insights from the Revolving Door ModelThe revolving door metaphor suggests some of the dynamics involved in changing habit-likebehavior. First, the use of the term “door” implies that there are two states, analogs to “inside” and“outside.” People on the “outside” have problem behavior (such as excessive consumption); peopleon the “inside” do not (they conserve). The goal for demarketers is to encourage people to move47Figure 4-2The Revolving Door Model of Behavior ChangeTERMINATION/ / F/ MMNTEWNOE‘ 41CONSERVATION II FEXCESSIVE CONSUMPTION 3 jPRECONTEMPLATIONfrom precontemplation into the door, through the various stages of change within the door, and thenexit, with a new, more conservative approach to consumption.Second, the “revolving” property of the door can be a problem. If a revolving door ismoving too quickly, it can be almost impossible to exit. Likewise, if people try to change tooquickly, they may overshoot the maintenance stage and relapse. This pattern, common among, forinstance, dieters, is something that demarketers should keep in mind. When an individual reduceshis or her consumption rapidly, complying immediately with the goals of a demarketing program,that individual is likely to move quickly into relapse, despite best intentions to the contrary. Habit-like behaviors are, to some extent, automated and inflexible. New, incompatible behavioral patternsneed time to become automated, too.48Third, if enough momentum is not generated, a revolving door can stall, just as people canstall in a particular stage. It is not uncommon, for example, for people to stay in the contemplationstage indefinitely, thinking that they really ought to do something about their behavior, but nevergetting around to doing it. This reinforces the notion that demarketing programs are important, butthey need to be correctly targeted.Fourth, being “in the door” is only a means to an end. People in the maintenance stage,although on the right side of the behavioral wall, are still within the revolving door. Within the door,(i.e., during the behavior change process) the new behavior requires conscious attention for consistentperformance. The ultimate goal of demarketing programs is to help people reduce their discretionaryconsumption automatically, without reminders or the requirement for conscious attention.Fifth, one of the most important insights from the Revolving Door Model is that failure isnormal, and should be expected. In fact, it may be that failure is essential for success, because theremay some things that can be learned only through failure. This implies that demarketing programsshould include components designed to give people who have reverted to excessive consumptionpatterns incentives and strategies to try once again to reduce their consumption.A Spiral Model of the Stages of ChangeA further refinement of the P&D model (Prochaska, DiClemente and Norcross 1992) addsa third dimension. In this version of the model, shown in Figure 4-3, the stages of change are shownin a spiral pattern. The spiral illustrates explicitly the finding that people who relapse do not, as theRevolving Door Model might imply, revolve endlessly in circles, regressing with each failure all theway back to where they started; instead, they can learn from their mistakes, and try differentstrategies the next time around (DiClemente, et al. 1991).The spiral model also includes, based on a re-analysis of the data that led to the developmentof the revolving door model, an additional stage of change between contemplation and action. People49Figure 4-3The Spiral Model of Behavior Changein the preparation stage are making small behavioral changes (such as reducing the number ofcigarettes they smoke in a day, or delaying their first cigarette of the day), and intend to make thebig change (abstinence from smoking or alcohol) very soon. In demarketing research, the preparationstage might be difficult to identif’, as the behavioral goal is reduction, rather than abstinence.The spiral model does offer a more complete representation of the process of behavior changethan the revolving door model does, but, because of the exploratory nature of this dissertationresearch, the simpler revolving door model will be the primary focus.Ten Change ProcessesThe Revolving Door Model seems to describe well the process of behavior change wherethere is a significant short term cost for an uncertain return that certainly will not be received untilsome time in the future, if at all (see Prochaska and DiClemente 1984 for a summary of empiricaltests of the model before 1984, and McConnaughy et al. 1989 for a more recent but lesscomprehensive review). The model’s utility is further enhanced by the investigation of how peoplePRECONTEMPLATION50progress from one stage of change to the next. P&D have investigated the processes of change:“covert and overt activities and experiences that individuals engage in when they attempt to modifyproblem behaviors,” (Prochaska, DiClemente and Norcross 1992: 1107), and the integration of thestages and processes of change.Based on principal components analysis, followed up by extensive validity and reliabilitychecks, P&D have distilled almost 400 different psychological therapies based on divergent theoreticalassumptions (hence the “transtheoretical” label), and over 130 different self-change techniques, downto ten processes (DiClemente and Prochaska 1982, Prochaska and DiClemente 1983, Prochaska andDiClemente 1984, Prochaska, DiClemente, and Noreross 1992). The ten processes of change arelisted in Table 4-1 and explained below. Note that the terminology is based in psychotherapy, butin most cases the concepts have marketing analogs.Table 4-1Processes of Change1. Consciousness raising2. Self-reevaluation3. Social (or environmental) reevaluation4. Self-liberation5. Social liberation6. Counter-conditioning7. Stimulus control8. Contingency (or reinforcement) management9. Dramatic relief10. Helping relationshipsConsciousness Raising. Consciousness raising increases the information available toindividuals so they can make more effective responses to the stimuli they encounter. Consciousnessraising can take two forms. Feedback is information given a person about the effects of his or herown actions and experiences. Education is information contained in stimuli generated by theenvironment (DiClemente and Prochaska 1982, Prochaska and DiClemente 1983, Prochaska andDiClemente 1984). Specifically, consciousness raising involves increasing information about oneself51and the problem, through observations, confrontations, interpretations, and bibliotherapy (Prochaska,DiClemente and Norcross 1992). For example, utility bills provide customers with feedback on theirresource use (although so delayed that its usefulness is doubtful), and advertisements about compactfluorescent bulbs can be classified as education.Self-Reevaluation. Self-reevaluation is a realistic appraisal of the pros and cons of tryingto overcome a problem by changing behavior. It requires the person to recognize that his or hervalues are in conflict, and that a change in behavior will cost something of value (DiClemente andProchaska 1982, Prochaska and DiClemente 1983, Prochaska and DiClemente 1984). Specifically,self-reevaluation involves assessing how one feels and thinks about oneself with respect to a problemthrough value clarification, imagery, and corrective emotional experience (Prochaska, DiClementeand Norcross 1992). For instance, some people who are considering making a serious effort toconserve energy must trade off their expectations about normal levels of ambient light and their desireto behave environmentally responsibly (assuming that is, indeed, what motivates them).Social (or Environmental) Reevaluation. Social (or environmental) reevaluation requiresan appreciation of the impact a change in behavior would have on other groups and/or the physicalenvironment. This may be achieved through empathy training and documentaries (DiClemente andProchaska 1982, Prochaska and DiClemente 1983, Prochaska and DiClemente 1984, Prochaska,DiClemente and Norcross 1992). For instance, New York State’s decision not to buy powergenerated by the James Bay II hydroelectric project in Northern Quebec was at least partly the resultof widespread social reevaluation, due to a lobbying and public relations efforts of the environmentalmovement and the Cree living in the area that would have been flooded by the new project.Self-Liberation. Self-liberation involves becoming aware of new alternatives, as well asincreasing the sense of self-efficacy with respect to making a particular option succeed. The latteris achieved partly by recognizing that merely making a commitment to a particular alternative makesit more likely to work. At the same time, there must be an appreciation of constraints that cannot52be removed (DiClemente and Prochaska 1982, Prochaska and DiClemente 1983, Prochaska andDiClemente 1984). Specifically, self-liberation involves choosing to act or believing in one’s abilityto change, which may be accomplished through decision-making therapy, New Year’s resolutions,logotherapy techniques, and commitment enhancing techniques (Prochaska, DiClemente and Norcross1992). An example of self-liberation in an energy conservation context might be a high schoolstudent who begins to realize that she can influence the behavior of other family members just byconsistently pointing out opportunities for and practicing energy conservation. Thinking that shecould singlehandedly prevent or postpone the building of new hydroelectric dams would, however,be unrealistic.Social Liberation. Social liberation involves changes in the environment that lead to a widerarray of choices for certain groups of people. Some people may benefit by becoming involved ingroups which lobby for more alternatives (DiClemente and Prochaska 1982, Prochaska andDiClemente 1983, Prochaska and DiClemente 1984). Specifically, social liberation involvesincreasing alternatives for nonproblem behaviors available in society, through advocating for rightsof repressed groups, empowering others, and policy interventions (Prochaska, DiClemente andNorcross 1992). Thus, some people may want to become involved in consumer groups that pressuremanufacturers to produce and promote energy-efficient appliances.Counter-Conditioning. Counter-conditioning changes the way people experience or respondto particular stimuli. It involves learning new responses to conditional stimuli, so that gradually thesenew responses replace the problem responses (DiClemente and Prochaska 1982, Prochaska andDiClemente 1983, Prochaska and DiClemente 1984). Specifically, counter-conditioning involvessubstituting alternatives for problem behaviors through relaxation, desensitization, assertion, andpositive self-statements (Prochaska, DiClemente and Norcross 1992). Not so many years ago, forexample, an office building with lights on all night was either ignored, or thought of as an attractiveaddition to a skyline; now, many people are appalled at the unnecessary waste inherent in such53practices, and make negative judgements about the social responsibility of firms that engage in thisbehavior. This is the result of counter-conditioning.Stimulus Control. With stimulus control, a person restructures his or her environment (orit is restructured by someone else) so that the probability of a particular conditional stimulusoccurring is significantly reduced (DiClemente and Prochaska 1982, Prochaska and DiClemente 1983,Prochaska and DiClemente 1984). Specifically, stimulus control involves avoiding or counteringstimuli that elicit problem behaviors by restructuring one’s environment (by, for instance, removingalcohol or fattening foods), avoiding high risk cues, and the use of fading techniques (Prochaska,DiClemente and Norcross 1992). Wearing warmer clothes at home in the colder months of the yearis an example of stimulus control. By dressing more sensibly, people are less likely to feel chilly,and are therefore less tempted to turn up the heat.Contingency (or Reinforcement) Management. The behaviorist school of psychology holdsthat behavior is under the control of its consequences: if a desired reinforcement is made contingenton a particular response, then the probability of making that response is increased. On the otherhand, if particular punishments are made contingent on certain responses, those responses are lesslikely to be made. The extent to which a given consequence controls behavior depends on itsimmediacy, saliency, and schedule (DiClemente and Prochaska 1982, Prochaska and DiClemente1983, Prochaska and DiClemente 1984). Specifically, contingency management involves rewardingone’s self or being rewarded by others for making changes through contingency contracts, overt andcovert reinforcement, and self-reward (Prochaska, DiClemente and Norcross 1992). The pricing ofelectricity can be thought of as contingency management. The actual price, as well as rebates forconservation, and frequency and detail of billing all contribute to reinforcing or punishing certaintypes of electricity consumption behavior.Dramatic Relief. As far back as Aristotle, it has been known that the reactions fromobserving emotional scenes can move people to change (DiClemente and Prochaska 1982, Prochaska54and DiClemente 1983, Prochaska and DiClemente 1984). Specifically, dramatic relief involvesexperiencing and expressing feelings about one’s problems and solutions through psychodrama,grieving losses, and role playing (Prochaska, DiClemente and Norcross 1992). For instance, HydroQuebec faced serious popular opposition to its James Bay II project, at least partly due to people’semotional reaction to the effect that James Bay I had on native people in the area.Helping Relationships. Helping relationships are both a precondition and a process ofchange. A helping relationship is characterized by openness, trust, warmth, and understanding. Helpcan be in the form of technical expertise, or more emotional support (DiClemente and Prochaska1982, Prochaska and DiClemente 1983, Prochaska and DiClemente 1984). Specifically, helpingrelationships involve being open and trusting about problems with someone who cares, in atherapeutic alliance, social support, or self-help group (Prochaska, DiClemente and Norcross 1992).In the case of energy conservation, an electric utility might be seen by many of its customers as ahelpful partner in their efforts to change, mostly through the provision of information.While these ten processes of change are not a complete compendium of the ways that peoplecan and do change, they are useful as a starting point for discussion and research. P&D nextexamined how the processes of change are used at different stages of change.Integration of Stages and Processes of ChangeOne of the interesting and practical findings from the work of P&D is that all ten processesof change are used to a certain extent in all the stages of change, but different processes seem to bemore commonly used at each stage. This is important, because it defines which processes have abetter chance of helping people progress from one stage to the next. Figure 4-4 (Procahska and DiClemente 1984) shows the integration of stages and processes.Confirmatory analysis has supported the ten-process model, but also identifies two secondaryfactors, labeLed experiential and behavioral (Prochaska, Velicer, DiClemente, and Fava 1988). The55Figure 4-4Processes and Stages of ChangePrecontemplation Contemplation Action, MaintenanceConsciousness raisingSelf-reevaluationSelf-liberationHelping relationshipReinforcementmanagementCounter-conditioningStimulus controlexperiential processes involve cognitive and/or affective activities. The behavioral processes involve,as the label implies, behavioral activities.Experiential Processes Behavioral ProcessesConsciousness raising Helping relationshipsDramatic relief Stimulus controlEnviromnental reevaluation Counter-conditioningSocial liberation Reinforcement managementSelf-reevaluation Self liberationBoth cross sectional and longitudinal studies have shown that, while there are differences inthe absolute frequency of the use of change processes across problems, certain processes areconsistently more useful at certain stages (Prochaska, Velicer, DiClemente, and Fava 1988).Precontemplators, as would be expected, tend not to use any of the change processes, because theysee no reason to change. People in contemplation are most open to experiential strategies:consciousness raising, dramatic relief, and environmental reevaluation. As people prepare for action,they continue to use those processes, and add behavioral processes: counterconditioning and stimuluscontrol. During the action stage, people tend to use behavioral strategies: self-liberation (willpower),56counterconditioning, and stimulus control, as well as helping relationships. People in maintenancebuild on the mix of experiential and behavioral processes that came before, especially selfreevaluation, counterconditioning, and stimulus control. As well, they engage in reinforcementmanagement (Prochaska, DiClemente, and Norcross 1993; Prochaska, Velicer, DiClemente, and Fava1988).There are two important omissions in this analysis: the processes of change that move peoplefrom precontemplation to contemplation, arid those that move people from maintenance totermination. In other words, the instigators of movement into and out of the revolving door are notincluded.The former issue, what moves people out of precontemplation, has been studied (Prochaskaand DiClemente 1984). People in the precontemplation stage are not particularly concerned orinvolved with the problem, so they do not process much information about it, nor do they spendmuch energy reevaluating themselves, or experiencing emotions about the problem, or anything else,for that matter. Prochaska and DiClemente (1984) speculate, based on their experience and ontheory, that movement is most likely due to either developmental changes or enviromnental changes.Developmental changes are those that occur due to aging. For example, the peak age for people todecide to quit smoking is 39. It is probably no coincidence that at about age 40, many people beginto think seriously about their mortality, and make some changes in their lives. Environmentalchanges refer to alterations in an individual’s external environment, such as new ordinances restrictingsmoking to certain defined areas.The second issue, concerning what moves people from maintenance to termination, is evenless well understood. What is known is that it takes time, presumably for the new behavior patternsto become habits and for the old, problem behavior patterns to be extinguished. It is unclear whetherthere are things that can be done to minimize the chances of relapse, but a healthy sense of self-57efficacy would seem to be important, so that the inevitable slips are not interpreted as irredeemablesigns of failure.SummaryIn summary, the P&D Revolving Door Model of Behavior Change suggests that people gothrough several discrete stages of change as they alter their habits. They may cycle through thestages several times before the new behavior becomes well-entrenched in their lives. Variousprocesses of change can be used to encourage movement from stage to stage, with certain processesbeing more effective at some stages than others.This approach seems well suited to demarketing problems of the sort that is the subject of thisdissertation research. The Revolving Door Model of Behavior Change describes and explains howpeople change habitual and habit-like behaviors, where the change entails significant short run costsfor uncertain returns that they have to have faith will be positive. The stages of change aspectfocuses on explaining when people make particular kinds of changes in modifying their behavior, andthe processes of change aspect addresses the issue of how people move from one stage to another.The next logical step is to test empirically whether the model generalizes to a typicaldemarketing situation. The next chapter describes the design of one such test.CHAPTER 5STUDY 1: HYPOTHESES AND DESIGNChapter 4 outlined three versions of the P&D model of behavior change and suggested thatthe model may offer a new and useful perspective on demarketing problems. Conceptually, theRevolving Door version of the model seems to apply well to demarketing. A logical next step is totest empirically whether the model generalizes to a typical demarketing situation.This chapter presents hypotheses and describes the design of a study intended to explore thedegree to which the Revolving Door Model of Behavior Change provides a reasonable and usefuldescription of behavior change in a demarketing context.General ApproachTesting the Revolving Door Model of Behavior Change in a demarketing context took placein two steps. It was first necessary to test whether people in the target market for a givendemarketing program could be successfully classified into one of the six stages of change(precontemplation, contemplation, action, maintenance, relapse, and termination). Once successfulclassification could be demonstrated, it was then possible to test whether key results from past studiesof the Revolving Door Model, particularly results concerning variables that differentiate among thestages of change, could be obtained in the context of that demarketing situation.5859The classification step was accomplished by adapting the Short Form Stages of ChangeQuestionnaire for Smoking (Prochaska undated) to a particular demarketing situation.The test of whether key results could be obtained in a demarketing context was accomplishedby adapting a study relating the stages of change to a decisional balance measure (Velicer et al. 1985)to the demarketing situation.The Original Decisional Balance StudyAccording to Decisional Balance Theory (Janis and Mann 1977), people make decisions basedon their evaluations of the gains and losses associated with the alternatives. However, it is not theabsolute amount of the gains and losses that determine whether an alternative will be chosen; rather,it is the amount of the gains and losses compared to what the person already has, or knows thatothers have, that determines the choice.Velicer, et al. (1985) describe the development of a 24-item Decisional Balance measure toassess and predict the smoking status of 960 current and former smokers. The assumption drivingthe research was that the decision to quit smoking is the result of comparisons of gains and losses.Two independent scales, the Pros of Smoking and the Cons of Smoking, were identified, and weresuccessful in differentiating among five groups representing stages of change in the quitting process.In other words, people at different stages of the quitting process assessed the relative importance ofthe pros and cons of smoking differently. For instance, people in the “immotive” (precontemplation)stage4 viewed the pros of smoking as outweighing the cons; while people in the contemplation stageviewed the pros and cons as being nearly in balance.4Velicer at al. (1985) used five groups, named differently but corresponding to the Revolving Doorstages of change: immotives (precontemplation), contemplator (contemplation), recent quitters(action), long term quitters (maintenance), and relapsers (relapse). The termination stage was notmentioned in the study.60Study 1 attempted to adapt the Velicer, et al. (1985) approach to differentiating among thestages of change based on a decisional balance framework to a demarketing context. The rationalefor the study was that, if similar results could be obtained, then confidence in the applicability of theRevolving Door Model of Behavior Change in demarketing situations would increase.HypothesesThe hypotheses are stated here in general terms; operational versions of the hypotheses arepresented after the study has been described in more detail.Velicer et al. (1985) found that, in the “immotive” (precontemplation) stage, people rated thepros of smoking as more important than the cons of smoking5. In other words, consistent with theRevolving Door Model of Behavior Change, as far as people in this stage were concerned, theimportance of the advantages of smoking (i.e., the “old” behavior) exceeded the importance of thedisadvantages by such a margin that they really had no reason to consider quitting. In a demarketingcontext, a similar finding is be expected:Hypothesis 1For people in the precontemplation stage, the importance ofdisadvantages associated with the new behavior will exceed theimportance of the advantages.Velicer et al. (1985) also found that, in the contemplation stage, people rated the importanceof the pros and cons of smoking about equally, although the cons of smoking were slightly moreimportant. The pros of smoking were still very important to them, but the cons were much moreimportant than they were for people in the immotive (precontemplation) stage. In other words, forpeople in the contemplation stage, the importance of the advantages of smoking was about the same5Velicer at al. (1985) refer to pros and cons of smoking; Janis and Mann (1977) refer to gains andlosses; in this dissertation research the elements of the comparison will be termed advantages anddisadvantages, where the advantages and disadvantages refer to the new behavior. For instance, ifa demarketing program were trying to get people to use less gasoline, an advantage of using lessgasoline might be saving money, and a disadvantage might be inconvenience.61as the importance of the disadvantages. The large increase in the importance of the cons of smokingcaused people in contemplation to seriously consider quitting. This pattern is consistent with thedescription of the contemplation stage of the Revolving Door Model of Behavior Change. In ademarketing context, similar findings are expected:Hypothesis 2aFor people in the contemplation stage, the importance ofdisadvantages associated with the new behavior will equal theimportance of the advantages associated with it.Hypothesis 2bPeople in the contemplation stage will view the advantages associatedwith the new behavior as being more important than will people inthe precontemplation stage.Hypothesis 2cPeople in the contemplation stage will view the disadvantagesassociated with the new behavior as being no more or less importantthan will people in the precontemplation stage.Concerning “recent quitters” (people in the action stage), Velicer et al. (1985) found that theimportance of the cons of smoking clearly exceeded the importance of the pros of smoking.Consistent with the Revolving Door Model of Behavior Change, people seemed to be motivated totake action when, as far as they were concerned, the disadvantages of smoking clearly outweighedthe advantages. Interestingly, the importance of both the pros and cons of smoking were less forpeople in the action stage than they were for people in the contemplation stage. It seems that actuallytaking action makes the issue of changing one’s behavior less attention-getting than does thinking thatone ought to take action. This, too, is consistent with descriptions of the dynamics of thecontemplation and action stages of change in the Revolving Door Model of Behavior Change. In ademarketing context, similar findings are expected:Hypothesis 3aFor people in the action stage, the importance of the disadvantagesassociated with the new behavior will be exceeded by the importanceof the advantages associated with it.62Hypothesis 3bPeople in the action stage will view the advantages associated withthe new behavior as being less important than will people in thecontemplation stage.Hypothesis 3cPeople in the action stage will view the disadvantages associated withthe new behavior as being less important than will people in thecontemplation stage.Velicer et al. (1985) found that long-term quitters (people in the maintenance and terminationstages), like recent quitters (people in the action stage), rated the importance of the cons of smokingas exceeding the pros of smoking. Both pros and cons of smoking were less important to people inthe maintenance stage than to people in the action stage. The latter result may be due to the fact that,the longer people have managed to keep from smoking, the less important the whole issue of smokingwould be to them. These findings are consistent with the descriptions in the Revolving Door Modelof Behavior Change. In a demarketing context, the following findings concerning the maintenancestage are expected:Hypothesis 4aFor people in the maintenance stage, the importance of thedisadvantages associated with the new behavior will be exceeded bythe importance of the advantages associated with it.Hypothesis 4bPeople in the maintenance stage will view the advantages associatedwith the new behavior as being less important than will people in theaction stage.Hypothesis 4cPeople in the maintenance stage will view the disadvantagesassociated with the new behavior as being less important than willpeople in the action stage.Velicer et al. (1985) found that people in the relapse stage viewed the pros of smoking asbeing slightly more important than the cons of smoking, which might explain why they returned tosmoking after having stopped. But for relapsers, the pros and cons of smoking were in balance, andboth were very important, much more important than they had been in the maintenance stage,reflecting a new preoccupation with smoking and/or quitting. Such a preoccupation would be natural63in people who had successfully overcome the urge to smoke for some time, only to succumb onceagain. Again, these findings are consistent with descriptions of the relapse stage of the RevolvingDoor Model of Behavior Change. In a demarketing context, similar findings are expected:Hypothesis 5aFor people in the relapse stage, the importance of disadvantagesassociated with the new behavior will equal the importance ofadvantages associated with it.Hypothesis 5bPeople in the relapse stage will view the advantages associated withthe new behavior as being more important than will people in themaintenance stage and people in the action stage.Hypothesis 5cPeople in the relapse stage will view the disadvantages associatedwith the new behavior as being more important than will people inthe maintenance stage or people in the action stage.Velicer et al. (1985) did not explicitly study people in the termination stage, although their“long term quitters” category may have included people in both the maintenance and terminationstages of change. However, the Revolving Door Model of Behavior Change suggests that people inthe termination stage would view the disadvantages of smoking as being more important than theadvantages, but since smoking is no longer a concern for people in termination, the importance ofboth would be low. In a demarketing context, this would suggest the following hypotheses:Hypothesis 6aFor people in the termination stage, the importance of disadvantagesassociated with the new behavior will be exceeded by the importanceof advantages associated with it.Hypothesis 6bPeople in the termination stage will view the advantages associatedwith the new behavior as being less important than will people in themaintenance stage.Hypothesis 6cPeople in the termination stage will view the disadvantages associatedwith the new behavior as being less important than will people in themaintenance stage.64Note that other hypotheses are also possible. For instance, the relationship between theimportance of the Advantages to people in the contemplation stage and the maintenance stage couldbe hypothesized by extension of Hypotheses 3b (relating the importance of the Advantages in theaction and contemplation stages) and Hypothesis 4b (relating the importance of the Advantages in themaintenance and action stages). However, it is the relationships within a particular stage (the “a”version of each hypothesis) and between adjacent stages (the “b” and “c” versions of each hypothesis)that are most important from a conceptual point of view. Consequently, extensional hypotheses arenot included.Methodological ApproachThe data for this study were collected in a telephone survey. The use of a survey wasconsistent with the methodology employed by Velicer et al. (1985). A telephone survey was chosenover mail or face-to-face interviews for two reasons. First, the questionnaire developed for this studywas easily administered by telephone. It required no visual aids and questions were structured so thatrespondents did not have to recall a large number of alternative responses. Second, it was cost-efficient, particularly since there was an opportunity to “piggyback” on another telephone survey.(See Barnes 1991 for a discussion of the factors that should be taken into account in choosing a datacollection method).The use of a telephone survey had a potential disadvantage, though; it required self-reportsof behavior, and self-reports are sometimes unreliable. This potential problem was one of theconsiderations that influenced the choice of the specific behavior change situation studied — coldwater laundry washing.65Criteria for the Target BehaviorThe specific behavior change situation studied, cold water laundry washing, was chosen tosatisfy three key criteria. First, the behavior had to have been the object of a demarketing program.Second, past demarketing had to have focused on changes in usage behavior, rather than changes inpurchasing behavior. Third, accurate self-reports concerning the behavior had to be relatively easyto make. The rationale for each criterion is discussed below.Object of a Previous Demarketing Program. In order to test the hypotheses, it wasessential that there be people in each of the stages of change. If people had never been asked tochange their behavior, virtually everyone surveyed would be in the precontemplation stage, and therewould be no opportunity to test hypotheses having to do with stages other than precontemplation.Consequently, the behavior chosen had to have been the object of a demarketing program.Since energy conservation programs, in which energy in its various forms is demarketed,have been widespread since the early 1970s, it was decided to focus on energy conservationbehaviors.Focus on Changes in Usage Behavior. Like many consumption reduction behaviors, energyconservation involves many separate behaviors — buying energy-efficient appliances, using the carless, turning out unnecessary lights, running the dishwasher only when it is full, deciding not to buythat hot tub, turning down the heat when no one is home, and so on. Clearly there is a large numberof behaviors that could accurately be called energy conservation. However, the focus of thisdissertation research is restricted to a subset of energy-related behaviors, those involving habits orhabit-like behaviors. To clarify the distinction, Defmition 5.1 was developed.Definition 5.1Energy conservation is behavior deliberately aimed at reducingdiscretionary consumption of energy, through changes in purchasingbehavior and/or through changes in usage behavior.66Definition 5.1.1Changes in purchasing behavior that constitute energy conservationinclude: (1) the substitution of energy-efficient products, brands, ormodels for less energy-efficient alternatives (e.g., compactfluorescent light bulbs for incandescent light bulbs, more energy-efficient appliances for less energy-efficient appliances, lined orfoam-backed draperies for unlined curtains or blinds, addinginsulation to the attic); and (2) the decision not to purchase orreplace luxury products (i.e., those not commonly regarded asnecessities) that use relatively large amounts of electricity (e.g., hottubs, home computers, VCRs, color televisions).Definition 5.1.2Changes in usage behavior that constitute energy conservationinclude substitutions of energy-conserving habits or habit-likebehaviors for less energy-conserving habits or habit-like behaviors(e.g., setting the thermostat lower at night and when no one is home,setting a lower temperature on the water heater, turning off lightswhen no one is in the room, washing only full loads of laundry,using cold water).Changes in purchasing behavior and changes in usage behavior are both important means of reducingdiscretionary consumption of energy, but this dissertation research addresses only the latter.Consequently, the behavior chosen for this study was an energy conservation behavior that involvedchanges in usage behavior.Accuracy in Self-Reports. Even if people are asked questions about changes in usagebehavior that constitute energy conservation (as defined in Definition 5.1.2), they have to considerseveral behaviors. This puts demands on memory that may compromise accuracy. To improveaccuracy, a common research strategy is to ask about specific behaviors. Even here, though, thereare potential problems.Many conservation behaviors, like turning out unnecessary lights, are not particularlyattention-getting, so people are often unable to estimate accurately the frequency with which theyengage in the behavior in question. As well, subjectivity is inherent in many of the measures usedto identify energy conserving behaviors. What does “unnecessary lights” mean? How full is a “fulldishwasher”? In addition, people may bias their responses toward what they think they should be67thinking or doing. The result of these problems is the potential for a great deal of “noise” in thedata.Consequently, the behavior chosen for this study was an energy conservation behavior thatinvolved changes in usage behavior. In addition, the behavior had to be attention-getting enough andsufficiently well-defined to encourage accurate reports by respondents.Behavioral Context: Cold Water WashingThe behavior change situation chosen for study was the switch from hot and warm waterwashing to cold water washing. It meets all three criteria discussed in the previous section. Peoplehave been encouraged for several years to save energy by switching from using hot or warm waterfor laundry to using cold water. Cold water washing is a single, well-defined usage behavior thatconserves energy. It was judged that laundry behavior is both sufficiently routine and sufficientlyattention-getting to enable people to make relatively accurate reports. Subjectivity is reduced, too,since the water temperature choices on most washing machines are clear — cold, warm, or hot.Cold water washing had an additional advantage, too. Industry research indicated that coldwater washing may be a proxy for other electricity conservation behaviors (Burr 1991). In otherwords, according to the industry research, people who do their laundry in cold water are more likelythan people who do their laundry in warm or hot water to engage in a variety of other conservationbehaviors, such as turning off unnecessary lights, using less heat in winter and less air conditioningin the summer, keeping the water heater at a lower temperature, and taking advantage of energysaving products, such as florescent light bulbs. To the extent that cold water washing is a proxy forother conservation behaviors, results from this study can be generalized to other energy conservationbehaviors.The cooperation of BC Hydro, British Columbia’s electric utility, was obtained for this study.BC Hydro had, for several years, been encouraging electricity conservation among its residential,68commercial and industrial users. Programs had been based on attitude-behavior models of behaviorchange and, despite significant investments of time and money (e.g., a large investment in the PowerSmart mass media advertising campaign supported by a regular mail-out program, rebates on energyefficient appliances such as hot water heaters and refrigerators, and extensive related research), resultshad been disappointing. Per capita consumption of electricity was not decreasing (Burr 1991).SampleThe survey was conducted by MarkTrend Research, Inc., a Vancouver-based market researchfirm. Randomly selected adult residents of each of BC Hydro’s four service districts (VancouverIsland, Lower Mainland, South Interior, and North Interior) were contacted by telephone in their ownhomes and asked to participate in a survey among residents of British Columbia about utilitycompanies. Telephone numbers were randomly selected from local directories, in proportion to thepopulation in each district. To cover new and unlisted numbers, “10” was added to each numberselected. Within sampled homes, either the male or female head of household was interviewed, inrotation. Up to three calibacks were made to minimize nonresponse bias, and supervisors checkedback with more than 10% of the respondents to ensure that interviews had taken place. MarkTrendResearch, Inc. did not provide figures on the number and types of survey (as opposed to question)nonresponse.A total of 666 randomly selected adult residents of British Columbia responded to the survey.In order to maximize accuracy of responses, only respondents who possessed all of the followingcharacteristics qualified for inclusion in this study:• primary responsibility for doing the household laundry;• knowledge of the number of loads done in a typical week;• knowledge of the number of loads done with both cold water wash and cold water rinse ina typical week;69• access to a washing machine in the home, rather than in a shared laundry room orlaundromat6;and• access to a washing machine that permitted cold water wash and cold water rinse.Of the 666 survey respondents, 340 possessed all the qualifications. The sample size for this studywas therefore 340.ProcedureThe instrument for this study was embedded in the Power Smart Tracking Study: Wave 7.After the request to participate, all respondents first answered a series of questions about advertisingby B.C. utilities, advertising by BC Hydro, participation in BC Hydro rebate programs, and attentionto the amount of electricity they used. The questions for this study followed. They began with aqualifier item. Then, qualifying respondents answered questions about present and past laundrypractices with respect to cold water washing and rinsing, as well as intentions for the future (thestages of change items); as well as the importance of various factors in the water temperature decision(the decisional balance items).All respondents then answered questions on bill payment plans, lifestyle changes due toenvironmental concerns, home and hot water heating, and compact fluorescent light bulbs.Demographic questions ended the survey.The time taken to complete the survey varied considerably, because parts of the questionnairewere answered only by qualified respondents. Most respondents who qualified for the questions usedin this study took between 15 and 20 minutes to complete the whole survey.‘This criterion was included because in laundromats and shared laundry rooms, the cost for a loadof laundry is usually the same, no matter what water temperature is used. In contrast, people withlaundry facilities in their own homes pay directly for the cost of heating water, so there is a costadvantage to using cold water rather than hot or warm water. Since the cost advantage of cold waterwashing was alluded to directly or indirectly in some of the decisional balance items, people withoutin-home laundry facilities were not included in the survey.70QuestionnaireA copy of the full Power Smart Tracking Study: Wave 7 questionnaire appears in Appendix1. Items 39 to 66 were developed specifically to test Hypotheses 1 to 6c. Those 27 items can bedivided into three groups: qualifiers, stage of change items, and decisional balance items.Qualifiers. The first group of items (Q.39 to Q.41) qualify respondents to answer the stagesof change and decisional balance items. As noted above, only respondents who possessed all of thefollowing characteristics were asked to respond to the stages of change and decisional balance items:primary responsibility for doing the household laundry; knowledge of the number of loads done ina typical week; knowledge of the number of loads done with both cold water wash and cold waterrinse in a typical week; access to a washing machine in the home, rather than in a shared laundryroom or laundromat; and access to a washing machine that permitted cold water wash and cold waterrinse.Stages of Change Items. Qualifying respondents were then asked a set of questions (Q.42to Q.47) concerning their own laundry behavior and behavioral intentions. These items weremodelled after items used to classify people in the process of quitting smoking (Prochaska undated).Operational definitions of the six stages of change are presented below.Definition 5.2Precontemplation is operationally defined as: no regular use of cold waterwash and cold water rinse presently or six months ago, and no intention touse cold water wash and cold water rinse six months in the future.Definition 5.3Contemplation is operationally defined as: no regular use of coldwater wash and cold water rinse presently or six months ago, butexpectation of regular use six months in the future.Definition 5.4Action is operationally defined as: some regular use of cold waterwash and cold water rinse at present, representing an increase oversix months ago.71Definition 5.5Maintenance is operationally defined as: some regular use of coldwater wash and cold water rinse, at the same level as six monthsago.Definition 5.6Relapse is operationally defined as: a reduction in the use of coldwater wash and cold water rinse over the last six months.Definition 5.7Termination is operationally defined as: all loads for the past sixmonths regularly done in cold water wash and cold water rinse.An unclassWed category was also included to accommodate individuals who could not remember whattheir laundry behavior was six months ago, or who did not want to speculate on what they would bedoing six months in the future.Decisional Balance Items. After the laundry behavior questions, qualifying respondents wereasked a set of attitudinal items (Q.48 to Q.66) based on the decisional balance theory of decisionmaking (Janis and Mann 1977), and adapted from those in the instrument developed by Prochaskaand DiClemente and their colleagues for a 1985 study of smokers (Velicer, et al. 1985). In thatresearch, a 24-item Decisional Balance measure was developed to assess and predict smoking status.The assumption driving the research was that decision making is the result of comparisons of gainsand losses. Two independent scales, the Pros of Smoking and the Cons of Smoking, were identified,and were successful in differentiating among five groups representing stages of change in the quittingprocess.For this study, items were written to represent the advantages and disadvantages associatedwith cold water washing and rinsing of laundry. A small convenience sample (ii = 8) of people whoregularly do laundry provided suggestions for additional items representing negative or positiveaspects of cold water washing and rinsing. Further pretests with a convenience sample of volunteers(n = 16) identified items that were obscure, unclear, irrelevant, or inaccurate. These were revisedor dropped. The initial item pool was thereby reduced to 20 items, one of which was inadvertentlydropped due to an administrative error. The final pool was therefore 19 items, each of which72described either an advantage or a disadvantage of cold water washing. A final pretest, again witha convenience sample of volunteers, confirmed that the advantages items and the disadvantages itemsrepresented equally strong arguments in favor of and against cold water washing.The ten advantage items used in the telephone survey were:Washing in cold water saves energy.Cold water is less harmful to fabrics.Washing in cold water keeps colors bright.Washing in cold water prevents clothes from shrinking.Washing in cold water saves money.Washing in cold water helps protect the environment.Detergents made especially for cold water use work as well as otherdetergents work in hot water.Cold water prevents colors from running.For all but really tough stains, cold water gets laundry clean.Doing laundry in cold water conserves hot water.The nine disadvantage items were:Some detergents dissolve better in hot water.Hot water kills germs better than cold water does.Hot water is more effective at removing stains.Laundry washed in hot water looks cleaner.Some detergents rinse out better in hot water.Laundry rinsed in hot water dries faster in the dryer.Washing in hot water keeps light colors from becoming dull anddingy looking.Some detergents work better in hot water.Hot water reduces the need for bleach and other laundry additives.73A five-point importance response format was used: “How important to you personally is itthat ... ?“ with response options ranging from “not important” (1) to “very important” (5).Respondents were also given the option of indicating that they did not agree with any given statement.Operational HypothesesThe operational hypotheses are expressed in terms of 13k and Ak, where:13k refers to the mean importance of the disadvantages of cold waterwashing to respondents in stage of change k;Ak refers to the mean importance of the advantages of cold waterwashing to respondents in stage of change k;where the valid values of k are:P = precontemplation;C = contemplation;A = action;M = maintenance;R = relapse; andT = termination.Recall Hypothesis 1:Hypothesis 1For people in the precontemplation stage, the importance ofdisadvantages associated with the new behavior will exceed theimportance of the advantages.The operational statement of this hypothesis in the context of cold water washing is:Recall Hypothesis 2a:Hypothesis 2aFor people in the contemplation stage, the importance ofdisadvantages associated with the new behavior will equal theimportance of the advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:D=ARecall Hypothesis 2b:74Hypothesis 2bPeople in the contemplation stage will view the advantages associatedwith the new behavior as being more important than will people inthe precontemplation stage.The operational statement of this hypothesis in the context of cold water washing is:A> A,,Recall Hypothesis 2c:Hypothesis 2cPeople in the contemplation stage will view the disadvantagesassociated with the new behavior as being no more or less importantthan will people in the precontemplation stage.The operational statement of this hypothesis in the context of cold water washing is:—Recall Hypothesis 3a:Hypothesis 3aFor people in the action stage, the importance of the disadvantagesassociated with the new behavior will be exceeded by the importanceof the advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:DA <AARecall Hypothesis 3b:Hypothesis 3bPeople in the action stage will view the advantages associated withthe new behavior as being less important than will people in thecontemplation stage.The operational statement of this hypothesis in the context of cold water washing is:AA<AcRecall Hypothesis 3c:Hypothesis 3cPeople in the action stage will view the disadvantages associated withthe new behavior as being less important than will people in thecontemplation stage.The operational statement of this hypothesis in the context of cold water washing is:75DA<DcRecall Hypothesis 4a:Hypothesis 4aFor people in the maintenance stage, the importance of thedisadvantages associated with the new behavior will be exceeded bythe importance of the advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:Recall Hypothesis 4b:Hypothesis 4bPeople in the maintenance stage will view the advantages associatedwith the new behavior as being less important than will people in theaction stage.The operational statement of this hypothesis in the context of cold water washing is:AM < AARecall Hypothesis 4c:Hypothesis 4cPeople in the maintenance stage will view the disadvantagesassociated with the new behavior as being less important than willpeople in the action stage.The operational statement of this hypothesis in the context of cold water washing is:DM < DARecall Hypothesis 5a:Hypothesis 5aFor people in the relapse stage, the importance of disadvantagesassociated with the new behavior will equal the importance ofadvantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:DR = ARRecall Hypothesis 5b:76Hypothesis 5bPeople in the relapse stage will view the advantages associated withthe new behavior as being more important than will people in themaintenance stage and people in the action stage.The operational statement of this hypothesis in the context of cold water washing is:AR > AM and AR > AARecall Hypothesis 5c:Hypothesis 5cPeople in the relapse stage will view the disadvantages associatedwith the new behavior as being more important than will people inthe maintenance stage or people in the action stage.The operational statement of this hypothesis in the context of cold water washing is:DR > DM and DR > DARecall Hypothesis 6a:Hypothesis 6aFor people in the termination stage, the importance of disadvantagesassociated with the new behavior will be exceeded by the importanceof advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:Recall Hypothesis 6b:Hypothesis 6bPeople in the termination stage will view the advantages associatedwith the new behavior as being less important than will people in themaintenance stage.The operational statement of this hypothesis in the context of cold water washing is:AT<AMRecall Hypothesis 6c:Hypothesis 6cPeople in the termination stage will view the disadvantages associatedwith the new behavior as being less important than will people in themaintenance stage.The operational statement of this hypothesis in the context of cold water washing is:77DT<DMSummaryThis chapter described a study designed to test the applicability of the Revolving Door Modelof Behavior Change in the context of a specific demarketing problem: encouraging people to use cold(rather than hot or warm) water for laundry washing. The study had two components. The first partwas designed to test whether people in the target market for a given demarketing program could besuccessfully classified into one of the six stages of change (precontemplation, contemplation, action,maintenance, relapse, and termination). The second part was designed to test whether the stages ofchange could be related to a decisional balance measure as predicted by several hypotheses based onthe Revolving Door Model of Behavior Change and the results of a previous study (Velicer et al.1985) on smoking.The next chapter describes the data analysis and results of Study 1.CHAPTER 6STUDY 1: ANALYSIS AND RESULTSThe last chapter described a study designed to test the applicability of the Revolving DoorModel of Behavior Change in the context of a specific demarketing problem: encouraging peopleto use cold (rather than hot or warm) water for laundry washing. This chapter describes the dataanalysis and results of Study 1.The analysis proceeded in three stages. First, respondents were classified into the stages ofchange, based on their responses to the group of behavior and behavioral intention items describedin Chapter 5. Then, the 19 decisional balance measures, also described in Chapter 5, were submittedto a principal component analysis to yield two scales, the Advantages of cold water washing and theDisadvantages of cold water washing. Finally, the hypotheses developed in Chapter 5 were tested,using cross sectional comparisons of the mean Advantages and Disadvantages scores by stage ofchange.Assignment to StagesOf the 340 qualifying respondents, 18 (5.3%) were unclassified due to missing data on oneor more of the classification questions. Table 6-1 shows the distribution of the remaining 322respondents across the stages of change.7879Table 6-1Assignment of Respondents to Stages of ChangeStage Number Percent7Precontemplation 103 32Contemplation 15 5Action 30 9Maintenance 81 25Relapse 18 6Termination 75 23Total 322 100The sample was not evenly distributed across the stages of change. Most respondents were in theprecontemplation, maintenance and termination stages; together, these three stages account for justover 80% of respondents.Decisional Balance MeasuresRecall that in this dissertation research, the purpose of the decisional balance measures wasto provide initial validation for the stages of change classification in a demarketing context. In theirstudies of smoking cessation, Velicer, et al. (1985) found significant differences in decisional balanceamong people in different stages of change. If the same pattern of results (as predicted in Hypotheses1 to 6C, presented in Chapter 5) were to occur in the cold water washing survey, it would provideinitial empirical evidence that the Revolving Door Model of Behavior Change generalizes to ademarketing context.Respondents were asked to rate each of the 19 decisional balance items in terms of itsimportance to them in their own decision about what water temperature to use for doing laundry.They used a five-point rating scale, where “1” meant “not important” and “5” meant “veryimportant.” Respondents were also given the option of indicating that they did not agree with the7Percent of classified respondents.80statement. In addition, some respondents said that they did not know how important an item was tothem.Deletion of Items. The first step in the analysis of the decisional balance measures was toexamine responses to each item to determine whether any items had a high level of “disagree” and/or“don’t know” responses, as many such responses for a particular item would indicate that the itemwas poorly worded, or incorrect, or otherwise problematic. Any item to which 20%8 or more ofrespondents said they disagreed with the statement or didn’t know how important it was to them waseliminated from further analysis. Table 6-2 shows each of the 19 items in the questionnaire andindicates for each the number of “don’t agree” responses, the number of “don’t know” responses, andthe decision to include or delete. (Items included in further data analysis are boldfaced.)Table 6-2“Disagree” and “Don’t Know” Responses in the Decisional Balance ItemsItem Disagree DK Total % Decision48. Washing in cold water savesenergy. 8 5 13 4 include49. Some detergents dissolve betterin hot water. 17 34 51 15 include50. Hot water kills germs betterthan cold water does. 33 11 44 13 include51. Cold water is less harmful tofabrics. 16 17 33 10 include52. Washing in cold water keepscolors bright. 25 18 43 13 include53. Washing in cold water preventsclothes from shrinking. 10 5 15 4 include54. Hot water is more effective atremoving stains. 65 11 76 22 delete8There is no theory to guide the selection of a cut-off point; it is a matter of judgement. Here, arelatively conservative approach was taken, on the grounds that the scales that were to be the resultof the principal component analysis were not an end in themselves; they were a means to investigatethe applicability of the Revolving Door Model of Behavior Change in a demarketing situation. Themore “disagree” and “don’t know” responses an item had, the more error would be introduced.81Item Disagree DK Total % Decision55. Washing in cold water savesmoney. 8 7 15 4 include56. Laundry washed in hot water lookscleaner. 79 16 95 28 delete57. Some detergents rinse out better inhot water. 58 48 106 31 delete58. Washing in cold water helpsprotect the environment. 35 42 77 23 delete59. Detergents made especially forcold water work as well as otherdetergents work in hot water. 27 60 87 26 delete60. Laundry rinsed in hot water driesfaster in the dryer. 83 45 128 38 delete61. Cold water prevents colors from 11 15 26 8 includerunning.62. For all but really tough stains,cold water gets laundry clean. 43 18 61 18 include63. Doing laundry in cold waterconserves hot water. 4 5 9 3 include64. Washing in hot water preventslight colors from becoming dulland dingy looking. 68 18 86 25 delete65. Some detergents work better in 22 25 47 14 includehot water.66. Hot water reduces the need forbleach and other laundry additives. 82 24 106 31 deleteOf the eight items deleted, six (items 54, 56, 57, 60, 64, 66) came from the list of disadvantages ofcold water washing; two (items 58, 59) came from the list of advantages of cold water washing. Theremaining pool included 11 items, eight of which (items 48, 51, 52, 53, 55, 61, 62) were written asadvantages, and three of which (items 49, 50, and 51) were written as disadvantages.Principal Component Analysis. The remaining 11 decisional balance items were submittedto principal component analysis to determine whether they represented two independent latentconstructs (advantages and disadvantages of cold water washing). Only data from respondents whoprovided an importance response for each and every decisional balance item were included in the82principal component analysis (i.e., data from respondents who answered “disagree” or “don’t know”to any item were excluded). Data from 173 of the 340 respondents were therefore submitted toprincipal component analysis.To test whether factor analysis was appropriate for these data, the Kaiser-Meyer-Olkin (KMO)measure of sampling adequacy was computed. The KMO measure is an index for comparing theobserved correlation coefficients to the partial coefficients. If the KMO measure is small, correlationsbetween pairs of variables cannot be explained by the other variables, so factor analysis of thevariables is inappropriate:Kaiser (1974) characterizes measures in the 0.90’s as marvelous, inthe 0.80’s as meritorious, in the 0.70’s as middling, in the 0.60’s asmediocre, in the 0.50’s as miserable, and below 0.5 as unacceptable.(Norusis/SPSS Inc. 1992: 59)The value of the overall KMO statistic for these data was 0.79. Since the KMO measure is close to0.80, it was reasonable to proceed with the principal component analysis.One of the important decisions in principal component analysis is how many components (orfactors) to retain. There are several approaches to this issue. The most common approach is theKaiser-Guttman Rule. According to this heuristic, the “true” number of components is equal to thenumber of eigenvalues greater than one. Steiger (1989) explains the rationale for this “mechanical”approach:Since the size of an eigenvalue represents the variance “accountedfor” by a principal component, and since the mean eigenvalue of acorrelation matrix is one, the components corresponding toeigenvalues greater than one are “above averaget’ in terms ofvariance accounted for. (Steiger 1989: 13-14)For these data, three factors had eigenvalues greater than one (see Table 6-3).83Table 6-3Factor EigenvaluesPercent of CumulativeFactor Eigenvalue Variance Percent1 3.67 33.4 33.42 2.04 18.5 51.93 1.15 10.4 62.4The rotated and interpreted factor matrix including all three components with eigenvaluesgreater than one is presented in Table 6-4. Together, the three components account for 62.4% ofthe total variance of the decisional balance items.Table 6-4Rotated and Interpreted Factor MatrixIncluding All Factors with Eigenvalues Greater Than One.Factor 1 Factor 2 Factor 3Item Advantages Advantages Disadvantages(Fabric) (Financial)Cold water prevents colors fromrunning. .79 .11 .10Washing in cold water keeps colorsbright. .73 .18 .02Washing in cold water prevents clothesfrom shrinking. .66 .21 .06Cold water is less harmful to fabrics. .62 .42 .15For all but really tough stains, coldwater gets laundry clean. .60 .21 -.42Washing in cold water saves money. .26 .83 -.02Washing in cold water saves energy. .14 .83 -.03Doing laundry in cold water conserveshot water. .24 .75 -.02Some detergents dissolve better in hotwater. -.05 .04 .86Some detergents work better in hotwater. -.03 -.03 .86Hot water kills germs better than coldwater does. .30 -.03 .5884Note that factors (components) 1 and 2, while independent, are both interpretable asadvantages of cold water washing, while factor (component) 3 is interpretable as disadvantages ofcold water washing.Another approach to determining the number of factors is to examine the scree plot (a plotof the total variance associated with each factor). The scree plot usually shows a distinct breakbetween the steep slope of the factors that account for a large portion of the total variance and thegradual tailing off of the remaining factors. The idea underlying this approach is that, once theeigenvalues level off, the components are no longer distinct, and will be more difficult to interpret(Steiger 1989). The “true” number of factors is the number of factors in the steep slope. The screeplot for these data is shown in Figure 6-1.Figure 6-1Scree Plot3.673 *EIGENV 2.040 *ALUES 1.148 *.801 *.585 * *.406 * * *.295 * *.000 I I I I I I I I1 2 3 4 5 6 7 8 9 10 11Based on the scree test, since the third factor is part of the gradual tailing off of the scree,the “true” number of factors is two. When only two components are extracted in a principalcomponent analysis with varimax rotation, the results are well defined. The factor matrix is85presented in Table 6-5. Together the two components account for 5 1.9% of the total variance in thedecisional balance items. This result compares favorably with the Velicer et al (1985) 24-itemdecisional balance instrument for smoking, which accounted for 40% of the total variance.Table 6-5Rotated and Interpreted Factor MatrixWhen Two Factors Are ExtractedFactor 1 Factor 2Item Advantages DisadvantagesWashing in cold water saves money. .74 -.04Cold water is less harmful to fabrics. .74 .18Doing laundry in cold water conserves hot water. .68 -.05Washing in cold water keeps colors bright. .67 .07Cold water prevents colors from running. .66 .16Washing in cold water saves energy. .65 -.07Washing in cold water prevents clothes from shrinking. .63 .10For all but really tough stains, cold water gets laundryclean. .60 -.38Some detergents work better in hot water. -.05 .86Some detergents dissolve better in hot water. -.02 .85Hot water kills germs better than cold water does. .19 .60The result of the two-factor principal components analysis with varimax rotation was well-defined, indicating that the two components are, in fact, independent. As expected, the firstcomponent is made up of items involved with the advantages of cold water washing and rinsing (andwas therefore named Advantages); the second component is made up of items involved with theadvantages of hot water washing. Since the advantages of hot water washing are unavailable whenlaundry is done in cold water, the second component can also be interpreted as the disadvantages ofcold water washing (and was therefore named Disadvantages).Another approach to deciding on how many factors to include involves examining the residualcorrelations. Ideally, all the residual correlations will be below .05, and most will be around .01(Steiger 1989). In these data, 50.0% (28) of the residual correlations are greater than .05 for the86three-component model, and 52.0% (29) of the residual correlations are greater than .05 for the two-component model. Clearly, neither model is ideal.Based on interpretation of the factors, inspection of the scree plot, the modest loss (10.4%)in explained variance in a two-component model versus a three-component model, and the minimalincrease (2.0%) in the number of residual correlations greater than .05 in a two-component versusa three component model, the two-component model was adopted.A related issue is the criterion for inclusion of an item in a given component. Here, theapproach taken was to include all items with loadings greater than one-half of the largest loading forthat component. Based on the results of the principal component analyses, all 11 remaining itemswere included. Two separate scales were formed, the Advantages of Cold Water Washing scale witheight items, and the Disadvantages of Cold Water Washing scale with three items. The two scalesappear in Tables 6-6 and 6-7.Table 6-6Advantages of Cold Water Washing Scale1. Washing in cold water saves money.2. Cold water is less harmful to fabrics.3. Doing laundry in cold water conserves hot water.4. Washing in cold water keeps colors bright.5. Cold water prevents colors from running.6. Washing in cold water saves energy.7. Washing in cold water prevents clothes from shrinking.8. For all but really tough stains, cold water gets laundry clean.Table 6-7Disadvantages of Cold Water Washing Scale1. Some detergents work better in hot water.2. Some detergents dissolve better in hot water.3. Hot water kills germs better than cold water does.A respondent’s score on each scale was the unweighted average of his or her responses to the itemson the scale.87Reliability of the Advantages and Disadvantages Scales. Cronbach’s coefficient alpha wasused to assess the reliability of each scale. Again, only data from respondents who provided animportance response for each and every one of the eleven final decisional balance items were includedin the analysis (i.e., data from respondents who answered “disagree” or “don’t know” to any of thefinal decisional balance items were excluded). Data from 224 respondents were therefore includedin the reliability analyses. Alpha for the Advantages scale was .84; it was .71 for the Disadvantagesscale. Reliabilities of .50 to .60 are sufficient for the early stages of research (Nunnally 1967, Peter1979); these results are well above those levels.Cross Sectional ComparisonsTreatment of Missing Data. To this point in the analysis, all missing data were dealt withthrough “listwise deletion of cases”: only data from respondents who provided an importanceresponse for each and every one of the eleven final decisional balance items were included in theanalysis. In other words, all data from respondents who answered “disagree” or “don’t know” to any(even one) of the final decisional balance items were excluded. Listwise deletion of cases reducesthe sample size considerably, but since the purpose of the analyses to this point was to purifymeasures, this conservative approach was appropriate.From this point forward, a different approach to missing data was taken, in order to preservestatistical power. In the cross sectional analyses that follow, the Advantages and Disadvantagesscores used are the scale value for each respondent, not the respondent’s responses to individual scaleitems. In other words, for a respondent who supplied an importance score of each of the eightAdvantage items, the Advantages score would be the unweighted average of the responses to the eightitems. Similarly, for a respondent who supplied an importance score for each of the threeDisadvantage items, the Disadvantages score would be the unweighted average of the three items.88If a respondent answered “disagree” or “don’t know” to, say, one of the Advantages scaleitems, deleting all the data from that respondent because of one missing importance rating would beunnecessarily conservative, especially since a scale value is used, rather than ratings of individualitems. Using the unweighted mean of the remaining seven scale items would preserve therespondent’s data without biasing his or her score (although the variance would be reduced).Consequently, the approach taken here was to use the unweighted average of the scale itemsfor which there is an importance score as long as the number of items for which an importance scoreis missing is less than half of the total number of items for that scale. Consequently, for theAdvantages scale, the scale score is the unweighted average of the eight Advantages items if therespondent provided an importance score for each of the eight items; otherwise it is the unweightedaverage of the seven, six, or five items for which the respondent provided an importance score. Datafrom respondents who provided four or fewer importance scores for the Advantages items were notused in the following analyses. Similarly, for the Disadvantages scale, the scale score is theunweighted average of the three Disadvantages items if the respondent provided an importance scorefor each of the three items; otherwise, it is the unweighted average of the two items for which therespondent provided an importance score. Data from respondents who provided only one importancescore for the Disadvantages items were not used in the following analyses.Overview of the Cross Sectional Comparisons. Once respondents had been assigned to thestages of change based on their behavior and behavioral intentions, and their scores on the twodecisional balance scales (Advantages and Disadvantages of cold water washing) had been calculated,the analysis turned to an examination of the pattern of the decisional balance measures across stages.Table 6.8 presents the means and standard deviations for the six stages of change on the two scales.89Table 6-8Means and Standard Deviations of the Advantages and Disadvantages Scales by Stage of ChangeStage Scale n Mean SDPrecontemplation Advantages 96 3.63 0.91Disadvantages 92 3.97 1.00Contemplation Advantages 14 3.82 1.01Disadvantages 15 4.21 0.78Action Advantages 30 4.24 0.61Disadvantages 25 3.50 1.07Maintenance Advantages 81 4.24 0.63Disadvantages 76 3.22 1.28Relapse Advantages 17 3.93 1.18Disadvantages 18 3.24 1.39Termination Advantages 75 4.39 0.76Disadvantages 67 2.87 1.38Overall Advantages 313 4.05 0.86Disadvantages 293 3.45 1.27The number of responses on the Advantages and Disadvantages scales differ in each stage, due todifferent numbers of “disagree” or “don’t know” responses on the two scales.Figure 6-2 presents the means arranged by stage of change.Difference Scores. The difference between the Disadvantages and the Advantages scores wascalculated. Positive values of the difference score would mean that respondents rated theDisadvantages of cold water washing as more important than the Advantages; negative values wouldmean that respondents rated the Advantages of cold water washing as more important than theDisadvantages. Table 6-9 shows the mean difference scores for each stage, in ascending order. Itindicates that respondents in the termination, maintenance, action, and relapse stages rated theAdvantages of cold water washing as more important than the Disadvantages, while respondents inthe precontemplation and contemplation stages rated the Disadvantages of cold water washing as moreimportant than the Advantages. This pattern is what would be expected, according to the RevolvingDoor Model of Behavior Change (refer to Figure 4-2). Respondents in precontemplation andcontemplation are using hot water (excessive discretionary consumption), and consistent with their90Figure 6-2Mean Advantages and Disadvantages Scores by Stage of Changebehavior, they rate the Disadvantages of cold water washing as more important. Respondents intermination, maintenance, and action are all using cold water (conservation), and consistent with theirbehavior, they rate the Advantages of cold water washing as more important. Respondents in relapseare presently using hot water, but have used cold water in the past; on average, they rate theAdvantages of cold water washing as more important than the Disadvantages, but not by a largemargin. Typical of people in relapse, their behavior is not consistent with their ratings on thedecisional balance measure.548Cl)00C0E2C01Advantages± Disadvantages0P C A M R TStage of Change91n677625178714p<.0001Table 6-9Ordered Mean Difference Scores By Stage of ChangeStage Mean SD FTermination -1.50 1.45 17.94Maintenance -0.99 1.23Action -0.68 1.13Relapse -0.55 1.13Precontemplation 0.31 1.31Contemplation 0.41 1.33The hypotheses developed in Chapter 5 examine the cross sectional relationships in greaterdetail.Hypothesis 1. The statement of Hypothesis 1 in theoretical terms is:For people in the precontemplation stage, the importance ofdisadvantages associated with the new behavior will exceed theimportance of the advantages.The operational statement of this hypothesis in the context of cold water washing is:D > AHypothesis 1 was tested by comparing the mean Disadvantage score to the mean Advantage score forrespondents in the precontemplation stage. As Table 6-10 shows, the mean Disadvantage score forrespondents in the precontemplation stage was 3.97, while the mean Advantage score for the9Recall that the operational hypotheses are expressed in terms of Dk and Ak, where:Dk refers to the mean importance of the disadvantages of cold water washingto respondents in stage of change k;Ak refers to the mean importance of the advantages of cold water washing torespondents in stage of change k;where the valid values of k are:P = precontemplation;C = contemplation;A = action;M = maintenance;R = relapse; andT = termination.92precontemplation group was 3.63. A t-test indicates that, as hypothesized, the Disadvantages scoreis greater than the Advantages score for respondents in the precontemplation stage (t = -2.23, p.01).Table 6-10Disadvantages and Advantages Scores in the Precontemplation StageDecisional Balance Measure Mean SD t p a-tail)Disadvantages 3.97 1.00 -2.23 .01Advantages 3.63 0.91Based on this result, Hypothesis 1 is supported.Hypothesis 2a. The statement of Hypothesis 2a in theoretical terms is:For people in the contemplation stage, the importance ofdisadvantages associated with the new behavior will equal theimportance of the advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:Hypothesis 2a was tested by comparing the mean Disadvantage score to the mean Advantage scorefor respondents in the contemplation stage. As Table 6-11 shows, the mean Disadvantage score forrespondents in the contemplation stage was 4.21, while the mean Advantage score for theprecontemplation group was 3.82. A t-test indicates that the Disadvantages score is not significantlydifferent from the Advantages score for respondents in the contemplation stage (t13 = -1.14, p =.27).Table 6-11Disadvantages and Advantages Scores in the Contemplation StageDecisional Balance Measure Mean SD t pDisadvantages 4.21 0.78 -1.14 .27Advantages 3.82 1.01Based on this result, Hypothesis 2a is supported.93Hypothesis 2b. The statement of Hypothesis 2b in theoretical terms is:People in the contemplation stage will view the advantages associatedwith the new behavior as being more important than will people inthe precontemplation stage.The operational statement of this hypothesis in the context of cold water washing is:A> AHypothesis 2b was tested by comparing the mean Advantages score for respondents in thecontemplation stage with the mean Advantages score for respondents in the precontemplation stage.As Table 6-12 shows, the number of respondents with Advantages scores in the precontemplationstage (96) was much larger than the number of respondents with Advantages scores in thecontemplation stage (14), so the Levene Test for Homogeneity of Variance was performed.According to the Levene Test, the variances of the Advantages scores in the two groups are notsignificantly different (F11 = .31, p .58), which satisfies the homogeneity of variance assumptionfor the t-test.The mean Advantages score for respondents in the contemplation stage was 3.82, while themean Advantages score for respondents in the precontemplation stage was 3.63. A t-test indicatesthat the two values are not significantly different (t1 = -0.73, p = .23). This result does notsupport Hypothesis 2b.Table 6-12Advantages Scores in the Contemplation and Precontemplation StagesStage Mean SD n t p (1-tail)Contemplation 3.82 1.01 14 -0.73 .23Precontemplation 3.63 0.91 96Based on this result, Hypothesis 2b is rejected.Hypothesis 2c. The statement of Hypothesis 2c in theoretical terms is:People in the contemplation stage will view the disadvantagesassociated with the new behavior as being no more or less importantthan will people in the precontemplation stage.94The operational statement of this hypothesis in the context of cold water washing is:D =Hypothesis 2c was tested by comparing the mean Disadvantages score for respondents in thecontemplation stage with the mean Disadvantages score for respondents in the precontemplation stage.As Table 6-13 shows, the number of respondents with Disadvantages scores in the precontemplationstage (92) was much larger than the number of respondents with Disadvantages scores in thecontemplation stage (15), so the Levene Test for Homogeneity of Variance was performed.According to the Levene Test, the variances of the Advantages scores in the two groups are notsignificantly different (F1105 = 1.49, p = .23), which satisfies the homogeneity of varianceassumption for the t-test.The mean Disadvantages score for respondents in the contemplation stage was 4.21, whilethe mean Advantages score for respondents in the precontemplation stage was 3.97. A t-test indicatesthat, as hypothesized, the two values are not significantly different (t105 = -0.88, p = .38).Table 6-13Disadvantages Scores in the Contemplation and Precontemplation StagesStage Mean SD n t pContemplation 4.21 0.78 15 -0.88 .38Precontemplation 3.97 1.00 92Based on this result, Hypothesis 2c is supported.Hypothesis 3a. The statement of Hypothesis 3a in theoretical terms is:For people in the action stage, the importance of the disadvantagesassociated with the new behavior will be exceeded by the importanceof the advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:DA <AAHypothesis 3a was tested by comparing the mean Disadvantage score to the mean Advantage scorefor respondents in the action stage. As Table 6-14 shows, the mean Disadvantage score for95respondents in the action stage was 3.50, while the mean Advantage score for the action group was4.24. A t-test indicates that, as hypothesized, the Disadvantages score is significantly less than theAdvantages score for respondents in the action stage (t = 3.00, p < .01).Table 6-14Disadvantages and Advantages Scores in the Action StageDecisional Balance Measure Mean SD t p a-tail)Disadvantages 3.50 1.07 3.00 < .01Advantages 4.24 0.61Based on this result, Hypothesis 3a is supported.Hypothesis 3b. The statement of Hypothesis 3b in theoretical terms is:People in the action stage will view the advantages associated withthe new behavior as being less important than will people in thecontemplation stage.The operational statement of this hypothesis in the context of cold water washing is:AA <AHypothesis 3b was tested by comparing the mean Advantages score for respondents in the action stagewith the mean Advantages score for respondents in the contemplation stage.The number of respondents with Advantages scores in the action stage (30) was larger thanthe number of respondents with Advantages scores in the contemplation stage (14), so the LeveneTest for Homogeneity of Variance was performed. According to the Levene Test, the variances ofthe Advantages scores in the two groups are significantly different (F142 = 5.13, p = .03), whichdoes not satisfy the homogeneity of variance assumption for the t-test. However, the t-test is robustto violations of homogeneity of variance assumption when n1= 2• Consequently, a random samplewas taken of 14 of the 30 respondents in the action stage, making the number of responses in theaction and contemplation stages equal.As Table 6-15 shows, the mean Advantages score for respondents in the action stage was4.17, while the mean Advantages score for the random sample of respondents in the contemplation96stage was 3.82. A t-test indicates that the two values are not significantly different (t26 = -1.06, p= .15). This result does not support Hypothesis 3b.Table 6-15Advantages Scores in the Action and Contemplation StagesStage Mean SD n t p U-tail)Action 4.17 0.67 14 -1.06 .15Contemplation 3.82 1.01 14Based on this result, Hypothesis 3b is rejected.Hypothesis 3c. The statement of Hypothesis 3c in theoretical terms is:People in the action stage will view the disadvantages associated withthe new behavior as being less important than will people in thecontemplation stage.The operational statement of this hypothesis in the context of cold water washing is:DA<DcHypothesis 3c was tested by comparing the mean Disadvantages score for respondents in the actionstage with the mean Disadvantages score for respondents in the contemplation stage. As Table 6-16shows, the number of respondents with Disadvantages scores in the action stage (25) was larger thanthe number of respondents with Disadvantages scores in the contemplation stage (15), so the LeveneTest for Homogeneity of Variance was performed. According to the Levene Test, the variances ofthe Disadvantages scores in the two groups are not significantly different (F138 = 2.31, p = .14),which satisfies the homogeneity of variance assumption for the t-test.The mean Disadvantages score for respondents in the action was 3.50, while the meanDisadvantages score for respondents in the contemplation stage was 4.21. A t-test indicates that, ashypothesized, the mean Disadvantage score for respondents in the action stage is less than the meanDisadvantages score for respondents in the contemplation stage (t38 = 2.24, p .02).97Table 6-16Disadvantages Scores in the Action and Contemplation StagesStage Mean SD n t p (1 -tail)Action 3.50 1.07 25 2.24 .02Contemplation 4.21 0.78 15Based on this result, Hypothesis 3c is supported.Hypothesis 4a. The statement of Hypothesis 4a in theoretical terms is:For people in the maintenance stage, the importance of thedisadvantages associated with the new behavior will be exceeded bythe importance of the advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:DM<AMHypothesis 4a was tested by comparing the mean Disadvantage score to the mean Advantage scorefor respondents in the maintenance stage. As Table 6-17 shows, the mean Disadvantage score forrespondents in the maintenance stage was 3.22, while the mean Advantage score for the maintenancegroup was 4.24. A t-test indicates that, as hypothesized, the Disadvantages score is significantly lessthan the Advantages score for respondents in the action stage (t75 = 7.00, p < .001).Table 6-17Disadvantages and Advantages Scores in the Maintenance StageDecisional Balance Measure Mean SD t p a-tail)Disadvantages 3.22 1.28 7.00 < .001Advantages 4.24 0.63Based on this result, Hypothesis 4a is supported.Hypothesis 4b. The statement of Hypothesis 4b in theoretical terms is:People in the maintenance stage will view the advantages associatedwith the new behavior as being less important than will people in theaction stage.98The operational statement of this hypothesis in the context of cold water washing is:AM <AAHypothesis 4b was tested by comparing the mean Advantages score for respondents in themaintenance stage with the mean Advantages score for respondents in the action stage. As Table 6-18 shows, the number of respondents with Advantages scores in the action stage (81) was much largerthan the number of respondents with Advantages scores in the maintenance stage (31), so the LeveneTest for Homogeneity of Variance was performed. According to the Levene Test, the variances ofthe Advantages scores in the two groups are not significantly different (F = 0.69, p = .41),which satisfies the homogeneity of variance assumption for the t-test.The mean Advantages score for respondents in the maintenance stage was 4.24, while themean Advantages score for respondents in the action stage was also 4.24. A t-test indicates that thetwo values are not significantly different (t1 = .02, p = .49). This result does not supportHypothesis 4b.Table 6-18Advantages Scores in the Maintenance and Action StagesStage Mean SD n t p (1 -tail)Maintenance 4.24 0.63 81 0.02 .49Action 4.24 0.61 30Based on this result, Hypothesis 4b is rejected.Hypothesis 4c. The statement of Hypothesis 4c in theoretical terms is:People in the maintenance stage will view the disadvantagesassociated with the new behavior as being less important than willpeople in the action stage.The operational statement of this hypothesis in the context of cold water washing is:DM < DAHypothesis 4c was tested by comparing the mean Disadvantages score for respondents in themaintenance stage with the mean Disadvantages score for respondents in the action stage. As Table996-19 shows, the number of respondents with Disadvantages scores in the maintenance stage (76) waslarger than the number of respondents with Disadvantages scores in the action stage (25), so theLevene Test for Homogeneity of Variance was performed. According to the. Levene Test, thevariances of the Advantages scores in the two groups are not significantly different (F1 = 2.00, p= .16), which satisfies the homogeneity of variance assumption for the t-test.The mean Disadvantages score for respondents in the maintenance stage was 3.22, while themean Disadvantages score for respondents in the action stage was 3.50. A t-test indicates that themean Disadvantage score for respondents in the maintenance stage is not significantly different fromthe mean Disadvantages score for respondents in the action stage (tm, = 0.99, p = .16). This resultdoes not support Hypothesis 4c.Table 6-19Disadvantages Scores in the Maintenance and Action StagesStage Mean SD n t p (1-tail)Maintenance 3.22 1.28 76 0.99 .16Action 3.50 1.07 25Based on this result, Hypothesis 4c is rejected.Hypothesis 5a. The statement of Hypothesis 5a in theoretical terms is:For people in the relapse stage, the importance of disadvantagesassociated with the new behavior will equal the importance ofadvantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:DR=ARHypothesis 5a was tested by comparing the mean Disadvantage score to the mean Advantage scorefor respondents in the relapse stage. As Table 6-20 shows, the mean Disadvantage score forrespondents in the relapse stage was 3.24, while the mean Advantage score for the relapse group was3.93. A t-test indicates that the Disadvantages score is significantly less than the Advantages score100for respondents in the relapse stage (t16 = 2.02, p = .03). This result does not support Hypothesis5a.Table 6-20Disadvantages and Advantages Scores in the Relapse StageDecisional Balance Measure Mean SD t p (1 -tail)Disadvantages 3.24 1.39 2.02 .03Advantages 3.93 1.18Based on this result, Hypothesis 5a is rejected.Hypothesis 5b. The statement of Hypothesis 5b in theoretical terms is:People in the relapse stage will view the advantages associated withthe new behavior as being more important than will people in themaintenance stage and people in the action stage.The operational statement of this hypothesis in the context of cold water washing is:AR > AM and AR > AHypothesis 5b was tested by comparing the mean Advantages score for respondents in the relapsestage with the mean Advantages scores for respondents in the maintenance and action stages.The number of respondents with Advantages scores in the relapse stage (17) was smaller thanthe number of respondents with Advantages scores in the maintenance stage (81) or the action stage(30), so the Levene Test for Homogeneity of Variance was performed. According to the LeveneTest, there are significant differences in the variances of the Advantages scores in the three groups(F21 = 4.80, p = .0 1), which does not satisfy the homogeneity of variance assumption for analysisof variance. However, analysis of variance is robust to violations of homogeneity of varianceassumption when n, = n2 ... = ... n (Glass and Hopkins 1984). Consequently, a random samplewas taken of 17 of the 30 respondents in the action stage, and 17 of the 81 respondents in themaintenance stage, making the number of responses in the three stages equal.As Table 6-21 shows, the mean Advantages score for respondents in the relapse stage was3.93, while the mean Advantages score for the random samples of respondents in the action and101maintenance stages were 4.28 and 4.27, respectively. A oneway analysis of variance indicates thatthe three values are not significantly different (F2, = .92, p = .40). This result does not supportHypothesis 5b.Table 6-21Advantages Scores in the Relapse, Maintenance, and Action StagesStage Mean SD n F pRelapse 3.93 1.18 17 0.92 .40Maintenance 4.27 0.59 17Action 4.28 0.70 17Based on this result, Hypothesis 5b is rejected.Hypothesis 5c. The statement of Hypothesis 5c in theoretical terms is:People in the relapse stage will view the disadvantages associatedwith the new behavior as being more important than will people inthe maintenance stage or people in the action stage.The operational statement of this hypothesis in the context of cold water washing is:DR > DM and DR > DAHypothesis 5c was tested by comparing the mean Disadvantages score for respondents in the relapsestage with the mean Disadvantages scores for respondents in the maintenance and action stages.The number of respondents with Disadvantages scores in the relapse stage (18) was smallerthan the number of respondents with Disadvantages scores in the maintenance stage (76) or the actionstage (25), so the Levene Test for Homogeneity of Variance was performed. According to theLevene Test, there are no significant differences in the variances of the Disadvantages scores in thethree groups (F2116 = 1.34, p .27), which satisfies the homogeneity of variance assumption foranalysis of variance.As Table 6-22 shows, the mean Disadvantages score for respondents in the relapse stage was3.24, while the mean Disadvantages scores for respondents in the action and maintenance stages were3.50 and 3.22, respectively. A oneway analysis of variance indicates that the three values are notsignificantly different (F2116 = .49, p = .62). This result does not support Hypothesis 5c.102Table 6-22Disadvantages Scores in the Relapse, Maintenance, and Action StagesStage Mean SD ii F pRelapse 4.83 1.10 18 0.49 .62Maintenance 4.82 1.01 76Action 5.04 0.84 25Based on this result, Hypothesis 5c is rejected.Hypothesis 6a. The statement of Hypothesis 6a in theoretical terms is:For people in the termination stage, the importance of disadvantagesassociated with the new behavior will be exceeded by the importanceof advantages associated with it.The operational statement of this hypothesis in the context of cold water washing is:DT<ATHypothesis 6a was tested by comparing the mean Disadvantage score to the mean Advantage scorefor respondents in the termination stage. As Table 6-23 shows, the mean Disadvantage score forrespondents in the termination stage was 2.87, while the mean Advantage score for the terminationgroup was 4.39. A t-test indicates that, as hypothesized, the Disadvantages score is significantly lessthan the Advantages score for respondents in the termination stage (t = 8.42, p < .001).Table 6-23Disadvantages and Advantages Scores in the Termination StageDecisional Balance Measure Mean SD t pDisadvantages 2.87 1.38 8.42 <.001Advantages 4.39 0.76Based on this result, Hypothesis 6a is supported.Hypothesis 6b. The statement of Hypothesis 6b in theoretical terms is:People in the termination stage will view the advantages associatedwith the new behavior as being less important than will people in themaintenance stage.The operational statement of this hypothesis in the context of cold water washing is:AT < AM103Hypothesis 6b was tested by comparing the mean Advantages score for respondents in the terminationstage with the mean Advantages score for respondents in the maintenance stage. As Table 6-24shows, the number of respondents with Advantages scores in the maintenance stage (81) was largerthan the number of respondents with Advantages scores in the termination stage (75), so the LeveneTest for Homogeneity of Variance was performed. According to the Levene Test, the variances ofthe Advantages scores in the two groups are not significantly different (F1154 = 0.36, p = .55),which satisfies the homogeneity of variance assumption for the t-test.The mean Advantages score for respondents in the termination stage was 4.39, while themean Advantages score for respondents in the maintenance stage was 4.24. A t-test indicates thatthe two values are significantly different (t154 = -1.32, p = .09) at the a = .10 level of significance,but, contrary to Hypothesis 6b, the Advantages score for respondents in the termination stage wasgreater than the Advantages score for respondents in the maintenance stage. This result does notsupport Hypothesis 6b.Table 6-24Advantages Scores in the Termination and Maintenance StagesStage Mean SD n t p a-tail)Termination 4.39 0.76 75 -1.32 .09Maintenance 4.24 0.63 81Based on this result, Hypothesis 6b is rejected.Hypothesis 6c. The statement of Hypothesis 6c in theoretical terms is:People in the termination stage will view the disadvantages associatedwith the new behavior as being less important than will people in themaintenance stage.The operational statement of this hypothesis in the context of cold water washing is:DT<DMHypothesis 6c was tested by comparing the mean Disadvantages score for respondents in thetermination stage with the mean Disadvantages score for respondents in the maintenance stage. As104Table 6-25 shows, the number of respondents with Disadvantages scores in the maintenance stage(76) was larger than the number of respondents with Disadvantages scores in the termination stage(67), so the Levene Test for Homogeneity of Variance was performed. According to the LeveneTest, the variances of the Advantages scores in the two groups are not significantly different (F1141= .78, p = .38), which satisfies the homogeneity of variance assumption for the t-test.The mean Disadvantages score for respondents in the termination stage was 2.87, while themean Disadvantages score for respondents in the maintenance stage was 3.22. A t-test indicates that,at the a = .10 level of significance, the mean Disadvantage score for respondents in the terminationstage is significantly less than the mean Disadvantages score for respondents in the maintenance stage(t141 = 1.57, p = .06). This result supports Hypothesis 6c at the .10 level of significance.Table 6-25Disadvantages Scores in the Termination and Maintenance StagesStage Mean SD n t p (1-tail)Termination 4.54 1.09 67 1.57 .06Maintenance 4.82 1.01 76Based on this result, Hypothesis 6c is supported weakly.Summary of Cross Sectional Results. Chapter 5 developed six hypotheses, each of which,with the exception of Hypothesis 1, has three parts. The first part of the hypotheses, Hypothesis Na,where N is the number of the hypothesis, predicts the relationship between the importance of theAdvantages and Disadvantages of cold water washing for respondents in a given stage. As Table 6-26 shows, this part of the hypotheses was supported for five of the six stages of change:precontemplation, contemplation, action, maintenance, and termination. It was not supported for therelapse stage.105D> A3.97 > 3.63p < .05supportedD = A4.21 > 3.82n.s.supportedDA<AA3.50 < 4.24p < .01supportedDM<AM3.22 < 4.24p < .01supportedDR=AR3.24 < 3.93p < .05rejectedDT<AT2.87 < 4.39p < .01supportedA > A,3.82 > 3.63n.s.rejectedAA <A4.17 > 3.82n.s.rejectedAM < A,4.24 = 4.24n.s.rejectedAR > AM, AA3.93 < 4.27,4.28n.s.rejectedA <AM4.39 > 4.24p < .10rejectedD=D4.21 > 3.97n.s.supportedDA<Dc3.50 < 4.21p < .05supportedDM<DA3.22 < 5.50n.s.rejectedDR > DM, DA(3.24 > 3.22) < 3.50n.s.rejectedDT<DM2.87 < 3.22p < .10supportedimportance of the Advantages of cold water washing for respondents in a given stage of change andthose in the preceding stage(s). As Table 6-26 shows, this part of the hypotheses was rejected foreach of the stages of change.The third part of the hypotheses, Hypothesis Nc, predicts the relationship between theimportance of the Disadvantages of cold water washing for respondents in a given stage of changeand those in the preceding stage(s). As Table 6-26 shows, this part of the hypotheses was supportedfor some stages of change: the relationship between contemplation and precontemplation, betweenaction and contemplation, and between termination and maintenance.Table 6-26Summary of Cross Sectional ResultsN Relationship Hypothesis Na Hypothesis Nb Hypothesis Ncn/a n/a123456HypothesizedActualSigncanceResultHypothesizedActualSignificanceResultHypothesizedActualSignificanceResultHypothesizedActualSigncanceResultHypothesizedActualSigncanceResultHypothesizedActualSigncanceResultThe second part of the hypotheses, Hypothesis Nb, predicts the relationship between the106The next chapter discusses these results.SummaryThis chapter has described the data analysis and results of Study 1. First, respondents wereclassified into the stages of change. The 322 respondents for whom enough data was available topermit classification were not evenly distributed across the stages. Over 80% of respondents werein either the precontemplation, maintenance, or termination stages. Respondents in thecontempLation, action, and relapse stages were less numerous.The second task for analysis was to purify the decisional balance measures. Eight of theoriginal 19 decisional balance items were deleted due to unacceptably high rates of “disagree” or“don’t know” responses. The remaining 11 items were submitted to a principal component analysis,which yielded two scales, the Advantages of cold water washing and the Disadvantages of cold waterwashing.Finally, the hypotheses developed in Chapter 5 were tested, using cross sectional comparisonsof the mean Advantages and Disadvantages scores by stage of change. Eight of the 16 hypothesesdeveloped in Chapter 5 were supported.Chapter 7 examines the implications of the results reported here.CHAPTER 7STUDY 1: DISCUSSIONThe previous two chapters described the design, analysis, and results of an exploratory studyintended to test the applicability of the Revolving Door Model of Behavior Change in the context ofa specific demarketing situation: energy conservation through the use of cold (rather than hot orwarm) water for laundry washing. On the basis of their answers to a series of behavioral andbehavioral intention questions, survey respondents were classified into one of six stages of change.Then, their importance ratings of a number of advantages and disadvantages of cold water washingwere used to construct decisional balance scales. Finally, the decisional balance scales were analyzedacross stages to determine whether the relationships hypothesized, which were based on resultsobtained in a previous empirical test of the Revolving Door Model of Behavior Change, actuallyoccur in the energy conservation context. As Chapter 6 indicated, results were mixed.This chapter discusses theoretical, methodological, and practical issues arising from Study1. The organization of this chapter parallels that of the previous chapter: it examines first issuesrelated to the classification of respondents to the stages of change, then discusses theoperationalization of the decisional balance scales, and finally turns to an examination of theimplications of the findings regarding the hypotheses.107108Assignment to Stages of ChangeThere were no a priori expectations regarding the distribution of respondents across the stagesof change, since this exploratory study was the first in an energy conservation situation to attemptto classify people into the stages of change. However, previous findings in other contexts suggestthat some variation in the number of respondents per stage should be expected. For instance,McConnaughy et al (1989) found that the number of subjects in eight homogeneous client profilescorresponding to the precontemplation, contemplation, action, and maintenance stages of changevaried from 18 to 70 (n = 293); and DiClemente et al (1991), in an analysis of the precontemplation,contemplation and preparation stages of change, found that the number of subjects per stage variedfrom 166 to 794 (n 1466); and Velicer et al (1985) found that the number of subjects in the fivegroups they studied (immotive, contemplator, relapser, recent quitters, and long-term quitters) variedfrom 116 to 225 (n = 960).The results of the laundry study likewise indicate that the distribution of respondents acrossthe stages of change was very uneven. As Table 6-1 shows, the number of respondents per stageranged from 15 to 103. This represents a relatively larger range than those in the studies cited above(here the ratio of largest to smallest n was almost 8:1, while in the DiClemente et al (1991) study theratio of largest to smallest n was almost 5:1). While variability in the number of respondents acrossthe stages of change was not unexpected, it does raise the question why there was so much variability,and what accounts for the pattern of variability. There are a number of possible explanations.Differential transience. One explanation for the variability in the number of respondentsacross the stages of change is that people may spend more time in some stages than others, at leastfor cold water washing. In other words, people may stay in precontemplation for a long time, butonce they move into contemplation, they may generally decide fairly quickly to take action. Oncein the action stage, they may move along rapidly to maintenance, where they stay a long time.Relapse might also be relatively brief, possibly because relapsers in the conservation context may109more commonly exit into precontemplation, rather than moving into contemplation again.Termination is, of course, a stable state, by definition; once there, people stay. A longitudinal studythat tracks individuals’ movement through the stages would be the best way to test this explanation.If, indeed, some stages of change are more transient than others, then it is important tounderstand why this is, if effective demarketing strategies are to be found. It may be that thetransience is the result of previous marketing programs. For instance, marketing programs by BCHydro, the environmental movement, and/or the makers of detergents formulated for use in coldwater may have focused their efforts on action-oriented strategies, encouraging people to dosomething — to use cold water for their laundry. This strategy would encourage contemplators tomove into action, and people in action to move into maintenance. If the Revolving Door Model ofBehavior Change is appropriate in this situation, however, such strategies would be lost onprecontemplators, who don’t see any need to even think about using cold water for their laundry.On the other hand, it may be that, for conservation behaviors in general, or perhaps for thisconservation behavior in particular, some stages inherently take more time to travel through.In either case, priority should be placed on processes of change that move people fromprecontemplation to contemplation, because once people get into contemplation, they are likely tomove relatively quickly into action and then maintenance. Priority should also be placed onpreventing people from slipping into relapse from maintenance, particularly if relapse usually leadsto precontemplation. Unfortunately, as indicated in Chapter 4, the transitions from precontemplationto contemplation, and from maintenance into relapse are the least well understood in the RevolvingModel of Behavior Change. More research is clearly needed in this area.Low interest level. A second possible explanation for the uneven distribution of respondentsacross the stages of change is that the behavior studied has “had its day,” and is no longer an issuereceiving attention in most people’s lives. During the energy crisis of the 1970s, consumers wereencouraged to undertake a variety of energy conservation behaviors, including cold water washing.110Many did switch to cold water washing. Some found that cold water washing was just as satisfactoryas warm or hot water washing; others did not. In recent years, relatively little attention has been paidto energy conservation, particularly cold water washing. Thus it could be that people have some timeago decided where they stand vis a vis cold water washing: they either do it or they don’t, and theygive little thought to it one way or the other.If this explanation is valid, it would produce a distribution weighted towardsprecontemplators, who do not use cold water for laundry and have no plans to change, andmaintainers and terminators, who do use cold water for laundry and have been doing so for sometime. In fact, precontemplation, maintenance, and termination are the three most numerous groups.If this explanation is valid, then, from a practical point of view, priority should be placed onmaking cold water washing an issue, so that there will be some movement from the precontemplationstage into the Revolving Door. This implication is consistent with those developed based on thepossibility that the uneven distribution of respondents across stages might be due to people spendingless time in some stages than others.Methodological weaknesses. A third possible explanation is that the uneven distribution ofrespondents across the stages of change is due to methodological problems. For instance, note thelarge number of respondents in the precontemplation stage and small number (15) of respondents inthe contemplation stage. Given that many people have positive attitudes towards energy conservation(even though their behavior may not reflect those attitudes), and that cold water washing is a goodway to conserve energy, a relatively large number of respondents would be expected in thecontemplation stage. Instead, contemplators make up the least numerous group.It is possible that the operational definitions used in this research, though based on measuresvalidated in other behavioral domains, were too strict for conservation behavior, leading to distortionsin assignment of respondents to stages. The operational definitions focused only on cold waterwashing: respondents who might have previously rinsed with hot or warm water and now used cold111water for rinsing, and respondents who might now use warm water rather than hot would both beclassified as being in the precontemplation stage, despite the fact that they had made behavioralchanges that conserve energy. Thus some respondents, who may actually have been in thecontemplation stage or even the action stage with respect to the real issue of conserving energy whiledoing the laundry, may have been inaccurately classified into precontemplation because theirconservation behavior did not fit the strict definition used in this research.Note that the questions used to classify respondents into the stages of change were based oncriteria developed and validated for smoking cessation. While both smoking cessation strategies andenergy conservation programs are aimed at decreasing consumption, there is an important differencein the goal. Smoking cessation strategies are aimed at reducing consumption of tobacco products tothe point of eliminating it completely; energy conservation programs are aimed only at reducingdiscretionary consumption. By modeling the classification items on a situation in which the goal iselimination of consumption, this study may not have been sensitive enough to modest, but real,reductions in consumption of hot water. This lack of sensitivity may have inflated the number ofprecontemplation classifications.Similarly, a six month horizon was chosen, again based on existing measures. While theredoes not seem to be any obvious reason why six months is an inappropriate horizon for cold waterwashing, a different period may have been more appropriate.A strategy for deciding among the explanations. Three possible explanations have beenproposed to account for the variability in the number of respondents per stage of change: it may bethat people pass through some of the stages more quickly than others, or that the particular behaviorexamined in this experiment has had its day, or that the uneven distribution of respondents across thestages is an artifact of overly strict operational definitions. One or more of these explanations maybe valid. Future research to account for the uneven distribution of respondents across stages shoulduse a longitudinal study. By tracking individuals over time it would be possible to determine whether112people tend to spend more time in some stages than others. It would also be possible to determinewhether people tend to stay in the same stage, which would suggest that cold water washing is a deadissue.To overcome the limitations of the strict operational definitions of the stages of change, itwould also be desirable to use a more realistic and multi-dimensional approach to assigningrespondents to the stages of change. This will be discussed more fully later in this chapter.Decisional Balance MeasuresRecall that of the 19 original items, eight were deleted due to high levels of “disagree” or“don’t know” responses. Principal component analysis on the remaining 11 items yielded twoindependent scales, one containing eight items describing advantages of cold water washing, and theother containing three items describing disadvantages of cold water washing. The principalcomponent analysis did not yield a single bipolar scale, with the items describing the advantages ofcold water washing receiving loadings of one sign and the items describing disadvantages receivingloadings of the opposite sign.The emergence of two orthogonal components suggests that the ratings of the advantages ofcold water washing are evaluated separately from the disadvantages. This finding provides moreempirical support for a key aspect of the Janis and Mann (1977) theory of decision making, namelythat decision making should be modeled as a comparative process, rather than as an absolute one.(Recall that the Velicer et al (1985) study, on which Study 1 was modeled, used Janis and Mann’sdecisional balance theory to determine whether the Pros and Cons of Smoking could differentiateamong five stages of change). Janis and Mann developed decisional balance theory to apply toimportant decisions, such as career choice, but the decision to use cold water for the laundry is arelatively unimportant one, so this result suggests that decisional balance theory generalizes beyondthe important life decisions.113The content of the decisional balance scales represents a subset of the concerns about andreasons for using cold water washing identified in interviews and pretests with convenience samplesof consumers. The Advantages scale contains items concerning financial and fabric-related reasonsto use cold water for laundry. The Disadvantages scale contains items concerning performance-related reasons not to use cold water. Together, the two scales assess the salience of an individual’sreasons for switching to cold water washing compared to the salience of the motives to continue touse hot water.Cross Sectional ComparisonsThe cross sectional comparisons examined differences between the Advantages andDisadvantages scores within the stages of change, as well as changes in the Advantages andDisadvantages scores between stages of change.Figure 6-2 displays the mean Advantages and Disadvantages scores by stage of change. Thepattern is largely consistent with the Revolving Door Model of Behavior Change. In theprecontemplation stage, when the Revolving Door Model says that people are engaged exclusivelyin the old behavior (using hot or warm water for laundry washing) and have no intention of changingtheir behavior (to the use of cold water), precontemplation respondents to this survey, as expected,rated the Disadvantages of cold water washing as significantly more important than the Advantages(supporting Hypothesis la).The Revolving Door Model says that in the contemplation stage, people have not changedtheir behavior, but have formed a serious intention to do so. In the Velicer et al (1985) survey, onwhich the hypotheses for this study were based, subjects in the contemplation stage viewed the Prosand Cons of smoking as being nearly in balance (i.e., not significantly different from each other).Similarly, in the laundry survey there were no significant differences between the Disadvantages ofcold water washing and the Advantages as rated by respondents in the contemplation stage (supporting114Hypothesis 2a). However, the lack of significance may be attributable more to lack of power dueto the small number of respondents (14) in the contemplation stage than to a genuine lack ofdifference.The low number of respondents in the contemplation stage may also have led to the lack ofa significant difference in both the Advantages and Disadvantages between the contemplation andprecontemplation groups (leading to lack of support for both Hypothesis 2a and 2b). Although it isnot statistically significant, there is an increase in the importance of both the Advantages andDisadvantages of cold water washing between the precontemplation and contemplation stages’°. Anincrease in the salience of the disadvantages of reducing discretionary consumption is consistent withthe assertions made in Chapter 3 about how people experience breaking old habits (as opposed todeveloping new ones). (See Table 3-1 and the accompanying discussion). A longitudinal studycould verify this explanation if it found that, as a respondent moves through the contemplation stage,the importance of the Advantages increases while the importance of the Disadvantages decreases.In the action stage, the Revolving Door Model says that people have recently changed theirbehavior, are feeling good about the change, and are relatively confident about their ability to persistwith the new behavior. Velicer et al (1985) found that, for recent quitters, the cons of smokingoutweighed the pros, but both were of less importance than for subjects in contemplation. Similarly,in the laundry study, the Advantages of cold water washing were rated as significantly moreimportant than the Disadvantages (supporting Hypothesis 3a). The change from the balance betweenthe Advantages and Disadvantages in the contemplation stage to the clear superiority of theAdvantages in the action stage was due to a significant decrease in the importance of theDisadvantages (supporting Hypothesis 3b) rather than any change in the importance of the Advantages(a finding that does not support Hypothesis 3a). This suggests that respondents in the action stageare no longer as acutely aware of the disadvantages of the behavior change. Without additional‘°See Churchill (1991) (page 774) for a discussion of the difference between statistical significanceand practical significance.115research, the reasons for the shift in the relationship between the Advantages and the Disadvantagescan only be a matter of speculation, but perhaps experience with cold water washing allayed someof the concerns respondents had with cold water washing.The Revolving Door Model says that in the maintenence stage, people are working tocontinue with the behavior change. As people gain experience with the new behavior, the initialenthusiasm wears off, and maintaining the behavior change is more effortful. Velicer et al (1985)found that, for long term quitters, the cons of smoking outweighed the pros, but both the pros andcons of smoking were of less importance than they had been in the previous stage. In the laundrystudy, as expected, respondents in the maintenance stage rated the Advantages of cold water washingas more important than the Disadvantages (supporting Hypothesis 4a). However, only theDisadvantages decreased in importance, and not significantly (failing to support Hypothesis 4b); theAdvantages remained at the same level of importance (failing to support Hypothesis 4a). Thesefindings suggest that, as people gain experience with cold water washing, the Advantages remainsalient, while the Disadvantages may become less salient.The pattern of the Advantages and Disadvantages to this point in the analysis suggests thatthe Disadvantages of cold water washing loom large in peoples’ minds if they are not using coldwater; once they start using it, however, their worries seem to diminish, and continue to do so asthey gain more experience with cold water washing.The Revolving Door Model suggests that people who leave maintenance move either torelapse or to termination. Those who return to the old behavior after a period of engaging in the newbehavior are said to be in relapse. Velicer et al (185) found that the smokers in relapse were likecontemplators: the pros and cons of smoking were in balance. In the laundry study, while theAdvantages and Disadvantages of cold water washing were in balance for respondents in thecontemplation stage, they were not in balance for respondents in relapse (failing to supportHypothesis 5a). It was expected that both the Advantages and the Disadvantages would be more116salient to respondents in the relapse stage than in the maintenance or action stages, due to theirreversion to the old behavior after some success with the new behavior, but there were no significantdifferences on either scale between relapse and either maintenance or action (failing to support bothHypothesis 5b and Hypothesis Sc).People who eventually integrate the new behavior into their lives are said to be intermination. The Revolving Door Model predicts that in the termination stage, the importance of theadvantages of the new behavior will exceed the disadvantages, and that the importance of both theadvantages and the disadvantages will decrease compared to the maintenance stage, because intermination there is no longer any competition from the old behavior. In the laundry study,respondents in the termination stage rated the importance of the Advantages higher than theimportance of the Disadvantages (supporting Hypothesis 6a). Also, the importance of theDisadvantages was lower than for respondents in the maintenance stage (supporting Hypothesis 6c).However, contrary to expectations (Hypothesis 6b), the importance of the Advantages actuallyincreased. This implies that when cold water washing becomes a new habit-like behavior, thedisadvantages associated with it become less salient, while the advantages become even more salient.Another possibility is that many of the people who wash and rinse all their laundry in cold water holdstrong convictions about environmental and energy issues, so they would be convinced of theimportance of the advantages of cold water washing.Examining the results of Study 1 on a stage-by-stage basis, then, reveals that, while theresults of the Velicer et al (1985) study were not completely reproduced, there was clearlyconvergence at several points. In addition, even where there was no convergence, it could be arguedthat the findings of Study 1 were not inconsistent with the Revolving Door Model of BehaviorChange, but merely with the pattern of results found in the Velicer et al (1985) study.It should also be noted that there were no significant differences between Advantages foradjacent stages. In contrast, there were significant differences between the Disadvantages between117contemplation and action, and between maintenance and termination. A larger number of respondentsin the contemplation stage may also have revealed a significant difference between precontemplationand contemplation. Together, these results suggest that the salience of the advantages of cold waterwashing does not change much as people move through the stages of change. What does change isthe importance of the disadvantages. It is not that people who use hot or warm water are unawareor unconvinced of the advantages of changing to cold water washing; it is just that they are concernedabout the disadvantages. This finding suggests that promotional efforts aimed at reducingdiscretionary consumption of electricity through a switch from hot or warm water to cold shouldfocus on addressing people’s concerns about the disadvantages of the change.Issues Arising from Study 1The purpose of Study 1 was exploratory: to get an indication whether the Revolving DoorModel of Behavior, which was developed in a clinical setting to explain and predict the changebehavior of people with harmful habits such as smoking, might also be applicable in a demarketingcontext. The results suggest that, in the case of cold water washing, the Revolving Door Model doesnot fit the data perfectly, but it does offer some promise.Transfering the Revolving Door Model from clinical settings to a demarketing context wasnot particularly difficult, but it did identify some problems and raise some questions for future study.The first concerns the definitions used to classify respondents into the stages of change. Thedefinitions were successful in the sense that most respondents (almost 95%) were classified, but, asnoted in this chapter’s section on the classification of respondents, the operational definitions mayhave been too rigid. A more flexible and multidimensional approach to defining the stages may beuseful. Consequently, to aid future work in this area, a new set of definitions is proposed below.118A second issue arising from Study 1 is difficulty of testing a model of behavior change usingthe “snapshot” of a cross sectional study. This issue is also explored below, as is a third, relatedissue, which concerns the extent to which adjacent stages have meaningful borders.A final issue is the need for theory in this area. This, too, is discussed below.Issue 1: Definitions of the Stages of ChangeRecall Chapter 5’s presentation of the operational definitions of the stages of change, whichwere based on a protocol used to classify people in the process of quitting smoking (Prochaskaundated; see also DiClemente et al 1991):Definition 5.2Precontemplation is operationally defined as: no regular use of cold waterwash and cold water rinse presently or six months ago, and no intention touse cold water wash and cold water rinse six months in the future.Definition 5.3Contemplation is operationally defined as: no regular use of coldwater wash and cold water rinse presently or six months ago, butexpectation of regular use six months in the future.Definition 5.4Action is operationally defined as: some regular use of cold waterwash and cold water rinse at present, representing an increase oversix months ago.Definition 5.5Maintenance is operationally defined as: some regular use of coldwater wash and cold water rinse, at the same level as six monthsago.Definition 5.6Relapse is operationally defined as: a reduction in the use of coldwater wash and cold water rinse over the last six months.Definition 5.7Termination is operationally defined as: all loads for the past sixmonths regularly done in cold water wash and cold water rinse.These operational definitions worked well, in the sense that assignment of respondents to the stagesof change based on their answers to the behavioral and behavioral intention items was119straightforward. All but 5.3% of respondents were successfully classified into one of the stages ofchange.The operational definitions worked well because they are both mutually exclusive andcollectively exhaustive. Table 7-1 shows that Definitions 5.2 to 5.7 are mutually exclusive. Inaddition, the definitions were collectively exhaustive for respondents who provided adequate data.All respondents who provided adequate data were classified; 5.3% of the sample did not provide coldwater usage data for one or more of the time periods required and were therefore impossible toclassify.Table 7-1Operational Definitions of the Stages of ChangeUse of Cold Water (CW)Six Months Ago Presently Six Months HenceStage (t = -1) (t = 0) (t +1)Precontemplation CW1 = 0 CW3 = 0 CW1 = 0Contemplation CW1 = 0 CW = 0 CW÷1 > 0Action CW4 > 0 CW0 = CW4 + x, x > 0 n.a.Maintenance CW, > 0 CW = CW4 n.a.Relapse CW1 >0 CW0= CW1-y,y >0 n.a.Termination CW4 = z, z = all CW0 = z, z = all n.a.While the operational definitions are mutually exclusive and collectively exhaustive, theycapture only the behavioral dimension of the differences among the stages of change. Since P&Dprovide much richer descriptions of the stages of change (e.g., Prochaska and DiClemente 1984,Prochaska, DiClemente, and Norcross 1992), the unidimensional approach used in the survey, whileeasy to administer, may have reduced the accuracy of the classifications. However, the only way todetermine whether accuracy suffers significantly when a unidimensional classification strategy is usedis to compare its results with a richer, multidimensional classification scheme.120Published studies by P&D supply the salient characteristics of people in the various stages.For instance, the following quote is typical of the vivid but informal descriptions of the stages ofchange provided by P&D.Precontemplation is the stage at which there is no intention to changebehavior in the foreseeable future. Many individuals in this stage areunaware or underaware of their problems. As G. K. Chestertononce said, “It isn’t that they can’t see the solution. It is that theycan’t see the problem.” Families, friends, neighbors, or employees,however, are often well aware that the precontemplators haveproblems. When precontemplators present for psychotherapy, theyoften do so because of pressure from others. Usually they feelcoerced into changing the addictive behavior by a spouse whothreatens to leave, an employer who threatens to dismiss them,parents who threaten to disown them, or courts who threaten topunish them. They may even demonstrate change as long as thepressure is on. Once the pressure is off, however, they often quicklyreturn to their old ways. . . Resistance to recognizing or modifyinga problem is a hallmark of precontemplation. (Prochaska,DiClemente, and Norcross 1992: 1103)Such descriptions, while offering a good intuitive “feel” for the stages of change, provide neithersystematic, multidimensional descriptions nor a clear delineation of what differentiates one fromanother. For instance, the transition from contemplation to action is marked by a change in behavior,but the transition from action to maintenance is more an internal matter: the action stage is like a“honeymoon” period, when the new less excessive behavior is relatively easy to perform andrelatively rewarding, simply because it is so new that it is attention-getting; the maintenance stageoccurs when persevering with the new behavior becomes relatively effortful and relatively lessrewarding.A fuller, multidimensional, systematic description of the characteristics of each stage isneeded. McConnaughy et al (1989) have partially addressed this issue with their Stages of ChangeScales, which use a series of 32 items (eight items for each of the first four stages) that taprespondents’ beliefs and attitudes, rather than relying strictly on behavior and behavioral intentionsto classify respondents into the precontemplation, contemplation, action, and maintenance stages. The121operational definitions that underlie the items are not made explicit, and items are not provided forthe relapse or termination stages.In an effort to fill the kinds of definitional gaps that have been outlined above, the followingconceptual and operational definitions are proposed for use in future research, based on thedescriptions of the stages of change provided by P&D supplemented by descriptions and findingsfrom research on relapses and their prevention. Each definition covers several dimensions: behavior,behavioral intentions, attitude toward the old behavior, and attitude toward the new behavior. Inaddition, for stages of change in which there has been a change in behavior (i.e., in action,maintenance, relapse, and termination), the definition also covers beliefs about lapses. The conceptof a lapse is important in the recent relapse literature (e.g., Marlatt and George 1990, Marlatt andGordon 1985) and is defined for the purposes of this research in Definition 7.1.Definition 7.1A lapse is a single instance, usually unintentional, of a return to theold pattern of behavior.Definitions 7.2 to 7.7 describe the six stages of change.Definition 7-2In the precontemplation stage, people are engaging in the oldbehavior. They have no intention of changing their behavior. Theirattitude toward the old behavior is favorable, while their attitudetoward the new behavior is unfavorable.Definition 7-3In the contemplation stage, people continue to engage in the oldbehavior. They have formed a serious intention to change, but havenot yet acted on it. Their attitude toward the old behavior isfavorable, but not as favorable as it was in precontemplation. Theirattitude toward the new behavior is favorable.Definition 7-4In the action stage, people begin to engage in the new behavior in asignificant (to them) way; they either do not have lapses or do notnotice them. Their intention is to continue to engage in the newbehavior. Their attitude toward the old behavior is unfavorable.Their attitude toward the new behavior is very favorable. Theybelieve that they are unlikely to lapse.122Definition 7-5In the maintenance stage, people continue to engage in the newbehavior; they have lapses and they notice them. They return to thenew behavior after each lapse. Their intention is to continue toengage in the new behavior. Their attitude toward the old behavioris unfavorable. Their attitude toward the new behavior is favorable,but less favorable than it was in the action stage. They believe thatlapses are inevitable and forgivable, and can be overcome.Definition 7-6In the relapse stage, people revert to the old behavior after anescalating series of lapses. They have no serious intention ofreturning to the new behavior in the near future. Their attitudetoward the old behavior is favorable. Their attitude toward the newbehavior is less favorable than it was in the maintenance stage. Theybelieve that lapses represent failure. They have given up (at leasttemporarily) trying to change.Definition 7-7In the termination stage, people consistently engage in the newbehavior; they have few, if any, lapses. They intend to continue toengage in the new behavior. Their attitude toward the old behavioris unfavorable. Their attitude toward the new behavior is favorable.They believe that lapses occur only in unusual circumstances, and areconfident that, in the event of a lapse, they will be able to return tothe new behavior.While the behavior and behavioral intention measures may be adequate proxies for otherdimensions of the stages of change in smoking behavior, they may be inadequate when the stages ofchange are applied to new areas. For instance, the behavioral classification scheme used in Study1 would assign a person who had been doing, say, one third of his or her laundry in cold water forone month to the action stage. However, if that person started to forget occasionally — and thereforeto feel a bit annoyed with him- or herself— to turn the washing machine’s temperature dial back tocold-wash-cold-rinse, then that person should properly be classified as being in the maintenance stage.An instrument based on the multidimensional definitions presented here might increase confidencethat people are being classified appropriately.The multidimensional definitions proposed here should be further elaborated to take intoaccount the research advances expressed in the Spiral Model (Prochaska, DiClemente, and Norcross1992). Given that people gain experience as they cycle through the stages of change, their attitudes123and intentions, and their perceptions of the relative importance of the advanatges and disadvantagesof change, may be different the first time they go through a given stage from the second time or thenth time.Issue 2: Dynamic Nature of the ModelThe Revolving Door Model of Behavior Change describes the stages people go through inchanging behavior. Since people move through these stages over time, and not everyone is the same,different people may move through a given stage at different rates.To cope with this problem, the Revolving Door Model can be expressed mathematically asa two-state Markov model, as shown in Table 7-2. Given valid and reliable measures of the stagesof change, transition probabilities can be estimated by repeated measures of the same individuals ina longitudinal study. Furthermore, the use of appropriate change processes in a demarketing programshould selectively increase some of the transition probabilities. This, too, could be tested, probablyin a field experiment.Table 7-2Markov Model: Likely Transition ProbabilitiesToP C A M R TPm’ Ppc 0 0 0 0C Pca 0 0 0A 0 0 p p 0 0From M 0 0 0 p p,,,. PmtR Pr,, Prc 0 0 p,,. 0T 0 0 0 0 0 1Expressing the Revolving Door Model of Behavior Change as a Markov model would make itpossible to identify where the demarketing emphasis should be put, determine whether a demarketingprogram (or any change process) is effective at moving people through the stages of change, how124effective it is at each transition, and over what transitions it is most effective. Ideally, the procedurewould involve, first, measuring the baseline transition probabilities. This would require alongitudinal study. The probability associated with a given transition would be the proportion ofpeople who make the transition. It would be prudent to take that measurement several times to geta mean value for each transition probability. These mean values would identify any bottlenecks(relatively low transition probabilities). Bottleneck transitions would be a logical target fordemarketing programs. For instance, if the probability that a person would move from contemplationinto action were significantly lower than the probability that a person would move fromprecontemplation into contemplation, then demarketing programs should be designed to get peoplewho are thinking about a change to actually do it, rather than attempting to get people to considerchanging.Once demarketing programs were operating, their effectiveness could be measured bycomparing the mean transition probabilities before demarketing with the mean transition probabilitiesafter. Other things being equal, a significant increase in any transition probability would indicate thatdemarketing was effective in moving people between those stages; no significant difference wouldindicate that demarketing was ineffective; and a significant decrease would indicate that thedemarketing did more harm than good.Issue 3: Fuzzy BordersAnother, related issue is that the stages of change are discrete approximations of a continuousprocess. People do not, at a particular moment, instantaneously change from, say, action tomaintenance. Moreover, there is change within a stage. A person who has just entered, say, theaction stage from the contemplation stage has much less experience with the new behavior than heor she will later, when he or she is about to leave the action stage for the maintenance stage. Thiscontinuity makes it difficult to identify a point at which a person crosses from one stage to the next.125McConnaughy et al (1989) use a continuous measure, rather than a discrete measure like theone used in this study. Their 32-item scale is made up of eight items for each of the following stagesof change: precontemplation, contemplation, contemplation, and action. Subjects responded to all32 items. The resulting data were submitted to a cluster analysis, which yielded eight clusters, eachof which had a different profile. For instance, the Contemplation Cluster is described as:above average on Contemplation, below average onPrecontemplation, well below average on Action, and about averageon Maintenance. (McConnaughy et al 1989: 499-500)While the discrete approach to measurement is appropriate for exploratory research such as thatconducted in Study 1, the continuous approach to measurement should be seriously considered forfuture research in this area. The Stages of Change Scale developed by McConnaughy et al (1989)should, therefore, be expanded to include items for relapse and termination. In addition, theoperational definitions of the stages of change should be made explicit.Issue 4: The Need for TheoryThe Revolving Door Model of Behavior Change is an empirically-driven model. Put verysimply, P&D gathered data on behavior, behavioral intentions, attitudes, and other dimensions forpeople involved in changing their behavior. They then subjected this data to principal componentanalyses and cluster analyses, which yielded the stages of change. These analyses have brought orderto a very confused and confusing area.The Revolving Door Model of Behavior Change is primarily a descriptive model. P&D’sprocesses of change11 explain when and how the Revolving Door works, but more theory is neededto explain why the process works this way. The next part of this dissertation begins to address theneed for more theory concerning changing habitual and habit-like behavior.11See Chapter 4 for a summary of P&D’s research in this area.126SummaryThis chapter has discussed some of the theoretical, methodological, and practical issuesarising from Study 1. The chapter began with an examination of possible reasons for the unevendistribution of respondents across the stages of change. Three possible explanations were discussed:that, on average, people pass through some stages more quickly than others; that cold water washingis a behavior about which most people have made up their minds; and that there were methodologicaldifficulties.The chapter then turned to a brief discussion of some of the implications of the results of theanalysis of the decisional balance items. It was pointed out that the emergence of two orthogonalcomponents, rather than one component on which items loaded either positively or negatively,suggests that respondents to this survey made laundry water temperature decisions based on acomparison of the advantages and disadvantages associated with cold water washing.Following that was a discussion of the implications of the cross sectional comparisons.Although the Revolving Door Model of Behavior Change did not fit the data perfectly, there wereenough points of agreement to suggest that the Revolving Door Model does offer some promise.The chapter concluded with an examination of some of the problems with Study 1 and someof the questions it raised for future research, specifically: the need for a more flexible andmultidimensional approach to defining the stages of change (to that end a new set of definitions wasproposed); the effects of the dynamic nature of the stages of change (the expression of which wassuggested through a two-state Markov model); the implications of the fact that the Revolving DoorModel is a discrete approximation of a continuous process (the primary implication being thatcontinuous measures are most appropriate); and that there is a need for more theory in the area ofhabitual and habit-like behavior change. The next part of the dissertation begins to develop and testsome theory in that area.CHAPTER 8CHANGING COMPLEXHABIT-LIKE BEHA VIORSRecent improvements in energy efficiency and a shift towards less energy-intensive sectors have helpedlimit consumption. But the process must be accelerated to reduce per capita consumption. . . (WorldCommission on Environment and Development 1987: 59)It has been argued in this dissertation that the demarketing task — to encourage people toreduce their discretionary consumption of certain good products — is essentially a matter ofencouraging people to change their consumption habits. Habits and habit-like behaviors pervadehuman lives. At any given time, it seems that most people have at least a few habits they would liketo change, or are trying to change, or have tried to change. In fact, in North America, the commondetermination to break bad habits has become ritualized in the annual “New Year’s Resolutions.”Given the level of concern with and amount of effort devoted to changing habits and habit-likebehaviors, there are surprisingly few comprehensive theories on the subject.One exception is the work of Prochaska and DiClemente and their colleagues (reviewed inChapter 4), whose Revolving Door Model of Behavior Change, a test of which has been describedand discussed in Chapters 5 to 7, does seem to hold some promise for understanding the dynamicsof behavior change in demarketing contexts. However, since P&D’s model is essentially descriptiveand based on empirical observation of thousands of people attempting to change such habit-like127128behaviors as smoking and eating, even it cannot properly be called a general, truly theory-drivenapproach to changing habits and habit-like behavior. The problem is that, even though there may bea correlation between certain behavior change strategies and changes in habit-Like behaviors, withoutunderlying causal explanations we cannot be confident that it really was the behavior change strategiesthat produced the observed changes.This chapter begins to fill the theory gap. It starts with a brief review of the psychologyliterature related to habit change that reveals that, while there are programs applicable to variousspecific habits or habit-like behaviors, there seem to have been few attempts to gain a more generalunderstanding of the underlying processes. The chapter then turns to developing theory on changinghabits and habit-like behaviors, based on what is known about the nature of habits and habit-likebehavior (reviewed in Chapter 3).Changing Habit-Like Behavior: An Overview of the LiteratureThe clinical psychology literature contains many articles on changing such habits as hair-pulling (e.g., Qureshi 1990), lip-biting (e.g., Ellis 1986), nail-biting (e.g., Leung and Robson 1990,Silber and Haynes 1992), nose-picking (e.g., Pianta and Hudson 1990), and thumb-sucking (e.g.,Friman and Leibowitz 1990, Luciano, Vilchez, and Herruzo 1992). As well, there are extensiveliteratures (both scientific and popular) on changing eating habits, and on overcoming various extremehabits such as addictions to tobacco, alcohol, and pharmaceuticals. In most cases, the changestrategies are specific to the habit at issue; there seem to have been few attempts to identify and takeadvantage of commonalities across the habit change situations. An exception is the habit reversalmethod (Azrin and Nunn 1973).Habit Reversal. Azrin and Nunn (1973) propose a set of procedures to overcome nervoushabits and tics. The procedures are based on assumptions about the source of nervous habits and tics,which Azrin and Nunn suggest begin as normal reactions to extreme events or personal traumas. For129some reason, the behavior persists beyond the injury or trauma and becomes part of normalmovement. Azrin and Nunn then propose that habit reversal requires several “methods of treatment”:The client should learn to be aware of every occurrence of the habit.Each habit movement should be interrupted so that it no longer ispart of a chain of normal movements. A physically competingresponse should be established to interfere with the habit.Social reinforcement should be reversed or eliminated. (Azrin andNunn 1973: 620)There are several procedures required to accomplish habit reversal. To begin with, theindividual needs to become aware of every occurrence of the habit. To that end, he or she recordsthe incidence of the habit, and a validating report is obtained from an appropriate observer. Inaddition, the individual is required to describe in detail the physical manifestations of the habit, andis taught to detect each instance of the habit, as well as its earliest sign. The individual is also madeaware of situations in which the habit is likely to occur, by recalling past performances of the habit.Once the individual is aware of the habit, a competing response is developed. The competingresponse is a behavior that is incompatible with the habit, can be maintained for several minutes (untilthe urge to perform the habitual behavior passes), raises the individual’s awareness of both the habitand the competing response, and is both socially inconspicuous and compatible with normal ongoingactivities. The exact nature of the competing response, of course, depends on the habit.Azrin and Nunn also emphasize the importance of motivation:Preliminary efforts with other clients had indicated that little successwould result if the client was only casually interested in eliminatinghis habit. Sufficiently strong motivation was indicated if the adultclient sought out the treatment himself, rather than being urged to doso by others. (Azrin and Nunn 1973: 624)They suggest a number procedures to maintain and increase motivation to change. One is to reviewwith the individual, in detail, the inconveniences, embarrassment, and suffering that have resultedfrom the habit. Another procedure is to increase social support for efforts to change through positivecomments by family and friends.130To ensure that the habit reversal is not confined to counselling sessions, but also generalizesto everyday situations, the individual imagines situations in which the habit is likely to occur,imagines that he or she detects a habit movement, and practices the competing response.The habit reversal method was initially tested on 12 subjects with various nervous habits andtics, including head jerking, shoulder jerking, head-shaking, tongue-pushing, lisping, eye-lashpicking, thumb-sucking, and fingernail biting. The results were impressive: performance of the habitwas reduced by at least 90% for all subjects, and for ten of the 12 subjects, the habit was completelyabsent by the third week of treatment. (A modified procedure was subsequently developed for nail-biting.) Since publication of the Azrin and Nunn (1973) paper, the habit reversal technique, andvariations on it, have been employed a number of times (e.g., Christensen and Sanders 1987, Weicket al 1988, Azrin and Peterson 1990) with mixed success.There is clearly some convergence between Azrin and Nunn’s habit reversal procedure,particularly in the area of motivation, and P&D’s Revolving Door Model of Behavior Change. Azrinand Nunn (1973) point out that the individual who seeks treatment on his or her own initiative ismore likely to be successful in overcoming the habit than people who seek treatment because theyare urged to do so by others. This observation echoes P&D’s distinction between theprecontemplation stage, in which people are not convinced that there is a real need for them to changetheir behavior, even though others might be strongly urging them to change, and the contemplationstage, in which people become convinced that they ought to change. In addition, Azrin and Nunnadvocate enhancing motivation through a detailed review of the inconveniences, embarrassment, andsuffering that have resulted from the habit. In the context of Study 1 of this dissertation, thisprocedure would correspond to increasing awareness and salience of the disadvantages associated withthe old behavior. Similarly, Azrin and Nunn suggest increasing social support for efforts to changethrough positive comments by family and friends. In the context of Study 1, this procedure wouldcorrespond to increasing the awareness and salience of the advantages of the new behavior.131The convergence between the Revolving Door Model and the habit reversal procedure canbe taken as an indication that there is some understanding of the habit change process. However,while the P&D Revolving Door Model of Behavior Change describes what happens as habitual andhabit-like behavior changes, and Azrin and Nunn’s habit reversal procedure explains how to makeit happen, neither approach explains why the habitual behavior change process happens as it does, norwhy a procedure such as habit reversal might work. Such explanations might have more thantheoretical value; they might suggest other techniques that might also be effective, even in othercontexts. The remainder of this chapter is devoted to developing theory to explain why and howhabitual and habit-like behavior changes, based on what is known about the characteristics of habitualand habit-like behavior.Pre-Emptive Habitual and Habit-Like Behavior ChangeRecall, from Chapter 3, that a habit is a pattern of activity that has, through extensiverepetition, become automated, fixed, and effortless, and occurs without conscious awareness orconsideration of alternatives (Definition 3.1), and habit-like behavior is a pattern of behavior that hasbeen so well practised that its execution requires little or no attention (Definition 3.2). Habit-likebehavior has a number of properties, one of which is that it is initiated by one or more “triggers” —stimuli that activate the automated processes that characterize habit-like behavior. It stands to reasonthat if the trigger(s) can be removed, the habitual or habit-like behavior would not occur, asProposition 8.1 indicates.Proposition 8.1Inhibition of the trigger(s) that initiate a habitual or habit-likebehavior will prevent the performance of that habitual or habit-likebehavior.Inhibiting the trigger(s) of a habitual or habit-like behavior would have two requirements.First, it would be necessary to identify the trigger(s). There may be multiple triggers for a singlehabitual or habit-like behavior in a single individual, any one of which could elicit the behavior. For132instance, someone who needs and wants to reduce the amount of fat in his or her diet might, despiteintentions to the contrary, be triggered to eat ice cream by boredom, or loneliness, or the need topamper him- or herself. In addition, the trigger(s) may vary from one person to the next. Forinstance, another person, also intending to cut clown on dietary fat might be triggered to eat ice creamby seeing an ice cream store on a hot day, or by the fact that it is time for dessert and “we alwayshave ice cream for dessert.”Since a single individual may be triggered to perform a habit or habit-like behavior by morethan one stimulus, and since different individuals may be triggered by different stimuli, successfulinhibition of a trigger-inhibition strategy would need, at minimum, the ability to identify and targetsegments based on trigger. For many habitual and habit-like behaviors, individual and individualizedattention would be desirable to implement the inhibition strategy successfully. Such opportunities arelikely to be scarce, especially for demarketers, who often do not have direct and personal access tothe people whose behavior they are attempting to change.Identifying the trigger(s) would not, by itself, be sufficient to prevent instances of the habitualor habit-like behavior. The second requirement for inhibition of the habitual or habit-like behavioris gaining control of the trigger. This would often be difficult. For instance, low blood sugar isoften a trigger for eating sugary food high in fat. Demarketing dietary fat might therefore involvehelping people avoid low blood sugar, but since different people have different metabolisms anddifferent eating habits, a different strategy would be desirable for each person, for maximumeffectiveness. The delivery of such idiosyncratic strategies would usually be impractical.Even if the difficulties associated with identifying and controlling the trigger(s) could beovercome, the pre-emptive approach has a significant flaw: it only prevents the performance of thehabit or habit-like behavior; it does nothing to change it. Should the circumstances that allow controlover the trigger(s) change, people would likely revert to the habitual or habit-like behavior, because133nothing has been done to inhibit the causal relationship between trigger and habit, unless the trigger-habit connection requires regular reinforcement to maintain its strength.The preceding arguments suggest that the pre-emptive approach to changing habits (i.e.,changing habitual and habit-like behavior by preventing exposure to stimuli that trigger suchbehaviors) is, in most cases, neither practical for demarketing nor particularly desirable. The nextsection examines a different approach to the habit change problem.Decomposing the Habit Change ProblemEarlier chapters of this dissertation have made a distinction between volitional and habitualbehavior. Volitional behavior is the result of conscious choice, while a habit is a pattern of activitythat has, through extensive repetition, become automated, fixed, and effortless, and occurs withoutconscious awareness or consideration of alternatives (Definition 3.1). Although changing volitionalbehavior is not necessarily easy, it is relatively well understood by social psychologists. The samecannot be said for habitual behavior. Given the remarkable efficiency and flexibility of humancognitive systems, though, it seems unlikely that the two kinds of behavior change are completelyseparate. Rather, it is probable that there are overlaps between the two. It is therefore proposed herethat every successful habit change includes volitional behavior change. (Note that this does not implythat every volitional behavior change is a part of a larger habit change.) It is further proposed thatthe volitional behavior change portion of the habit change is preceded by a process of de-automation,and followed by a process of consolidation. The former process brings the habit to consciousawareness and under conscious control, so that volitional behavior change is possible, while the latterprocess automates the new behavior so that, once again, it can be performed without demanding toomany cognitive resources. Once this progression, shown in Figure 8-1, is understood, the habitualbehavior change problem, and hence the demarketing problem, becomes more tractable.134Figure 8-1Proposed Habitual Behavior Change ProcessDE-AuT0MATI0N -, V0Lm0NAL BEHAVIOR CHANGE - CONSOLIDATIONHence, Proposition 8.2:Proposition 8.2Successful habit change involves three components, which alwaysoccur in the same order. First, the individual must become awareof specific occurrences of the habit and bring it under consciouscontrol. Then, volitional behavior change can occur. if the changeis to persist, the new behavior must be automated.CorollaryIf the new behavior is not automated, it will be extinguished, and theoriginal habit will re-assert itself.De-AutomationSince habits and habit-like behaviors are relatively automated processes that run withoutawareness or conscious control, the first step in changing such behaviors is to bring them underconscious control (Definition 8.1).Definition 8.1De-automation is the process of bringing under conscious control theautomated processes that underlie habit-like behaviors.Both bringing and keeping the habit under conscious control can be difficult, even under the best ofcircumstances. Even people who are truly determined to change a habit often find it difficult to doso. And demarketers face an even more difficult task. Lacking control over the people whosebehavior they seek to change, demarketers have to persuade them to bring certain of their ownhabitual and habit-like behaviors under conscious control.It is proposed below that de-automation requires that people know precisely what behavior(s)to change, that they can reliably anticipate the performance of the behavior(s), and that they haveenough motivation to invest the cognitive energy necessary to bring the habit under conscious control.135Before developing those propositions, however, it is necessary to make explicit a key assumptionabout the relationship between automated processes and habit-like behaviors.Automated processes and habit-like behaviors. As Chapter 3 explained, habits and habit-like behaviors are the results of automated processes. This notion is formalized in Assumption 8.1:Assumption 8.1Habits and habit-like behaviors are the results of automatedprocesses.Consider, for instance, a university student taking notes in class. For most undergraduate students,note-taking is clearly a habit-like behavior. It is likely made up of several automated processes, suchas: one to pick out cues that indicate which points are more important, one to organize the notes(e.g., using headings, under which each related point appears on a new line), one to actually do thewriting, and one to keep track of what other people in the class are doing, as illustrated in Figure 8-2. None of these automated processes requires attention; the student pays attention to the content ofthe class.Figure 8-2Relationship Between Automated Processes and Habit-Like BehaviorsAP1 AP2 AP3 •.. AP •.. APIt is also assumed in this research that the automated processes that combine to form arelatively complex habit-like behavior such as note-taking are themselves combinations of lower-levelautomated processes. For instance, writing is an automated process used in note-taking, but writingitself is made up of several automated processes, such as: one for the formation of each individual136letter, one to connect the letters, one to spell, one to separate letters in one word from letters inanother, and so on. Hence Assumption 8.2:Assumption 8.2The automated processes that combine to form a habit-like behaviorare themselves combinations of lower-level automated processes.It is further assumed that, in the hierarchy of automated processes and habit-like behaviors,a given automated process at one level is not necessarily dedicated to a single automated process orhabit-I ike behavior at a higher level. Returning to the note-taking student, writing is one of theautomated processes making up the note-taking habit, but it may also be an automated process makingup a different habit, such as writing letters home. Hence Assumption 8.3:Assumption 8.3Any automated process may be a constituent process in an automatedprocess or habit-like behavior at a higher level of organization.Expressing Assumption 8.1 to 8.3 graphically, Figure 8-3 illustrates possible relationships amonghabit-like behaviors and automated processes at various levels of complexity.Figure 8-3Hierarchy of Automated Processes and Habit-Like Behaviorsap1 ap2 ap3 apk apAP1 AP2 AP3 AP APH1 H2 H3 H1 HDescription. Given Assumption 8.1, that habit-like behaviors are the results of automatedprocesses, it follows that changing the habit-like behavior requires control over the constituentautomated processes. In order to gain control over those processes, it is necessary to know exactly137which processes are involved. For instance, because the note-taking habit is made up of severalautomated processes, merely telling the student to “take better notes” would be of limited value. Thestudent needs to know what exactly the problem is. Does he or she focus on the wrong points, oris the problem slow writing? Similarly, in a demarketing context, exhortations to “save energy” oreven “turn off unnecessary lights” are too vague. There are other, related problems.In some cases, people may want and intend to change their behavior, but doing so mayrequire changes in specific habits that are so ingrained that they may not even realize that they havethem. For instance, the irony of a student newspaper editorial arguing against logging is lost on itsauthors. They are devoted to using their persuasive skills to saving trees but use a medium thatdestroys them, even though other options are available.Alternatively, the behavior change may require changing habits people are aware they havebut do not realize are related to the intended behavior change. For instance, someone might knowthat he or she eats eggs for breakfast every day, but not realize that that habit is inconsistent withreducing dietary fat. It should therefore be a priority for demarketers to describe exactly what habitsor habit-like behaviors that they want people to change.Hence Proposition 8.3:Proposition 8.3To change a habit-like behavior, an individual needs a specific anddetailed description of the habit-like behavior and relevant underlyingautomated processes. This is a necessary but not sufficient conditionfor change.Consider, for instance, a demarketing program designed to reduce energy consumption by persuadingpeople not to use electrical devices unnecessarily. For the program to achieve maximumeffectiveness, people need to be told exactly what that means. They need to know what electricaldevices the request refers to, what use would be classified as necessary (as opposed to unnecessary),and what specific behaviors (including, where possible, triggers) they should change. Such ademarketing program might, for example, point out that many people walk into their offices and138reflexively turn on the overhead lights even though there is plenty of light from the window. Withoutan explicit description like this, people who sincerely want to reduce their consumption of energymight not even notice that opportunity, simply because walking into their offices and turning on thelights is an automated process and therefore not subject to conscious awareness. A specific, detaileddescription of the habit-like behavior and its relevant underlying automated processes can bring themto conscious awareness.The requirement for a detailed description of the specifics of the habit-like behavior changeplaces a heavy burden on demarketers, as it requires cataloguing the habit-like behavior, includingits underlying automated processes and their triggers. Accomplishing this may be complicatedbecause a single habit-like behavior (e.g., turning on a light) can be the result of different automatedprocess (e.g., walking into the room, starting to read something that requires concentration), eachtriggered by a different stimulus. Communicating the complete catalog of underlying automatedprocesses is likely to be impractical for most demarketers, so typical examples of behavior shouldbe identified and communicated.Given the preceding discussion, it is not surprising that, in their explication of the habit-reversal method they developed for overcoming nervous habits and tics, Azrin and Nunn (1973)emphasize that the first step is for the client “to describe the details of the movement to thecounsellor, using a mirror, if necessary, while [re-enactingj several instances of the typicalmovement” (Azrin and Nunn 1973: 622). Note that having the client describe the habit to thecounsellor ensures that the client really does know what the habit is, and is therefore superior tocommunication in the other direction, but the latter option is the one most usually available todemarketers.Reliable anticipation. Another condition for successful de-automation is the ability toreliably anticipate performance of the habit-like behavior that is at issue. Since habit-like behaviorsare complexes of automated processes, and automated processes are triggered by certain stimuli, it139should really be necessary only to anticipate the appearance of the triggering stimuli. There is plentyof anecdotal evidence to indicate that anticipation of habitual and habit-like behaviors is oftendifficult. In trying to break habits, people frequently complain that they have already performed thebehavior before they even notice it, such as a dieter who suddenly realizes that he or she is halfwaythrough a chocolate bar. Hence Proposition 8.4:Proposition 8.4To change a habit-like behavior, an individual needs to be able toreliably anticipate its performance. This is a necessary but notsufficient condition.CorollarySince a habit-like behavior’s underlying automated processes aretriggered by certain stimuli, its reliable anticipation amounts toreliable anticipation of the triggering stimuli.It is therefore important that the description of the habit-like behavior include the triggers.If people are aware that a particular stimulus triggers the habit-like behavior they are trying tochange, it may be possible for them to avoid the trigger, and thereby prevent the performance of thehabit-like behavior. For instance, if the trigger to turning up the thermostat is feeling cold, thenwearing warmer clothes may prevent unnecessary consumption of energy.As well, situations in which the trigger is likely to arise should be described, so that, ifpossible, they can be avoided. For instance, if mid-afternoon hunger pangs are the trigger forsomeone to eat a chocolate bar at work, then eating a larger lunch or bringing a healthier snack toeat before the hunger pangs usually arise may prevent the hunger pangs that trigger the consumptionof an unnecessary chocolate bar.Azrin and Nunn (1973), in their habit-reversal procedure, recommend that people trying toovercome nervous habits and tics be given practice in detecting the earliest sign of a habit movement,and that they learn to become aware of situations in which the habit is likely to occur by having theperson “recall all situations, persons, and places where the habit was likely to occur and having himdescribe how the habit was performed in each of those situations” (Azrin and Nunn 1973: 623).140Motivation. Both P&D (e.g., Prochaska and DiClemente 1984) and Azrin and Nunn (1973)emphasize the importance of motivation in changing habitual and habit-like behavior. Change is,according to these researchers’ experience, unlikely to occur unless there is considerably more thana casual interest in the change. However, neither team of researchers provides an explanation fortheir conclusion.From the cognitive point of view adopted in this dissertation research, the need for motivationcan be explained as a result of scarcity of cognitive resources. Kahneman (1973) points out thathumans have limited processing capacity. When there is competition for processing capacity, it islikely that if something can run automatically, it will, because automatic processing does not useattentional capacity, so attention can be directed elsewhere. The tendency to use automated processescan be overridden — if the motivation to override automation is great enough. Hence Proposition8.5:Proposition 8.5To change a habit-like behavior, an individual must have enoughmotivation to override the underlying automated processes. This isa necessary but not sufficient condition.Motivation to override the underlying automated processes would occur if the person becomesconvinced that the problem has personal relevance (Janis and Mann 1977, Ronis, Yates and Kirscht1989). This could be done either by increasing the importance of the advantages associated withchanging the habit-like behavior or increasing the importance of the disadvantages associated withnot changing the habit-like behavior. Depending on the degree of automation, it may be enough todraw attention to the problem, and make suggestions for behavior change; or it may be necessary tomanipulate the consequences of the habit-like behavior so that they are intense enough to attractattention. In the domain of energy conservation, for example, purchasing behavior is likely to be lessautomated than actual consumption behavior: it is easier to get people to pay attention to what kindof light bulb they are buying than to how often they switch it on and off.141In summary, then, de-automation is the first step in changing habitual and habit-like behavior.Since such behavior is a complex of automated processes, it can be difficult to overcome. Acognitive analysis of the problem suggests that people need a specific and detailed description of thehabit-like behavior and its underlying automated processes, including the triggering stimuli, and theyneed to be able to anticipate when such triggers are likely to occur. Finally, they need to bemotivated to make the effort to overcome the automation so that volitional behavior change can occur.Volitional Behavior Change: The Fishbein and Ajzen ApproachOnce de-automation has been achieved, and the relevant behaviors are under consciouscontrol, then the problem of changing habit-like behavior reverts to the traditional behavior changeproblem that social psychologists have been working on for years.Although various credible volitional behavior change theories exist, the Fishbein and Ajzen(1975) Theory of Reasoned Action is one of the more widely accepted. This theory posits thatbehavior (B) is determined by a person’s intention to perform it (I), and the more specific theintention, the more likely it is to be an accurate predictor of behavior.Intentions are, in turn, determined by the person’s attitudes (i.e., general feeling offavorableness or unfavorableness) toward the behavior (AB), as well as by the associated subjectivenorm (i.e., the person’s perception that people important to him or her think he or she should orshould not perform the behavior in question) (SN,. The relative importance of the individual’sattitude and subjective norm (w1 and w2 respectively) may vary across behaviors and individuals’2.Expressed mathematically, the theory at this level is:B I = (A8)w1 + (SN)w2‘w1 and w2 are empirically determined, and may be influenced by variations in behavior, object,situation, and time, as well as by individual difference variables. Their relative influence may alsobe affected by behavior change strategies.142Both the attitude and the subjective norm components of the theory can be further analyzed. Aperson’s attitude toward the behavior is determined by his or her salient beliefs about it, and can beestimated by summing the products of the individual’s evaluation of each of the behavior’sconsequences (e1) and the strength of his or her belief that performing the behavior will lead to thatconsequence (b,):AB = E1 b1e,The relative influence of each belief will be determined by one’s evaluations of each of thoseconsequences.Similarly, the subjective norm is determined by the perceived expectations of specific referentindividuals or groups (ba) and by the person’s motivation to comply with those expectations (mi):SN = E b1nAccording to the Fishbein and Ajzen theory, any attempt to induce behavior change mustalways be directed at one or more of the individual’s beliefs. Beliefs can be influenced two ways,by active participation and by persuasive convnunications. In active participation, a person is placedin a situation where he or she can personally observe that a behavior has a particular consequence.Assuming that the person actually perceives the behavior-consequence association, this is a powerfulbehavior change strategy, as people rarely doubt information received through their own senses (i.e.,e will be high). Unfortunately, it is frequently impractical to arrange for people to experience orobserve personally the behavior-consequence relationship. In such cases, persuasive communicationscan be used — the person is told that the behavior in question has a particular consequence. Withpersuasive communications, the major problem is to ensure that the person believes thecommunication linking the behavior and the consequence (i.e., the danger is that e will be low).143The Theory of Reasoned Action in a Demarketing ContextThe kind of behavior change sought in demarketing, reduction in excessive discretionaryconsumption, can be difficult to get, even when de-automation has been successful. Part of thereason is that the behavior change process is complex, with many intervening processes, each ofwhich is likely to dilute the potential effect. As Figure 8-4 shows, changes in beliefs affect attitudesand/or subjective norms, which, in turn, affect behavioral intentions. Finally, behavioral intentionsinfluence behavior. The effect of a change in any given belief is therefore likely to be small.Figure 8-4Overview of the Theory of Reasoned ActionBELIEFS - A’rrrruDEs AND SUBJECTIVE NoRMS - INTENTIONS - BEHAvIoRAnother part of the reason for the difficulty in getting reductions in consumption is thatdemarketing asks people to give up something desirable for a set of benefits that are not well-defined,as outlined in Table 3-1. Consequently, demarketing strategies have to rely mostly on persuasivecommunications, which are generally less effective than the alternative, active participation.In spite of these obstacles, though, many people do decide to try to consume less: they goon diets, they cut down on passive diversions like television viewing, they turn down the thermostat,they decline to take a bag when they make a purchase. If this were a typical marketing problem,such progress might be perceived as success, and rightly so. Often the biggest problem inconventional marketing situations is to convince potential users to try the product; once they havesampled it, they are likely to continue using it, because most products satisfy some immediate need.In the case of demarketing, however, sampling the “product” is not likely to induce repeated use,because the benefits are not immediately evident, and because of the habit-like nature of theconsumption behavior.Recall that the reason for automation in cognitive and other processes is that cognitiveresources are limited and automation permits processes to run in parallel, rather than serially. This144capacity constraint remains. It is therefore unlikely that adequate resources will be consistentlyavailable to permit the “reasoned action” proposed by volitional behavior theories, particularly whenthe benefits of behavior change are unclear, as they often are in demarketing situations.Consequently, in instances when there is competition for controlled processing capacity, there islikely to be reversion to the habit-like behavior. People go off their diets, they gradually resume theirusual television viewing habits, they forget to turn down the thermostat, and they don’t notice untilit is too late that they have been given an unnecessary bag with their purchase.Since it is clearly impractical to expect that people will be willing or able to make a consciousdecision to consume less every time there is an opportunity to do so, the new behavior must beautomated and the old habit extinguished. In other words, the bad habit must be replaced with agood habit.ConsolidationThe final step in changing habit-like behaviors is to consolidate the new behavior so that itbecomes habit-like. Automating the new behavior may be where demarketing has tended to fallapart. Marketing programs are generally good at getting people to try to consume less — they goon diets, they watch less television, they turn off lights — but somehow, in many people, thesebehavior changes never seem to get beyond the novelty stage. If the new behavior is not consolidated(see Definition 8.2), the gains achieved in de-automation and volitional behavior change are quicklyextinguished.Definition 8.2Consolidation involves repetition of the new behavior until it hasbecome so well-practised (i.e., automated) that there is no advantagein terms of cognitive resource use to reverting to the old habitualbehavior.In order to foster consolidation, demarketers must provide people with both the motivationand the means to engage in volitional behavior change many times, preferably in a variety of settings.145In other words, they have to make it worth people’s while to override their natural inclination toconsume excessively, until the new style of consumption is habit-like, at which time it will notrequire attention, or conscious awareness, and, once triggered, will run to completion.The process of consolidation will be affected not only by repetition itself, but by theinformation generated by each repetition of the act. This information will affect the belief structurethat ultimately drives the volitional behavior. For instance, as a dieter becomes accustomed to eatingless, hunger pangs may subside, so the belief that eating less causes discomfort will be changed.Conversely, if someone carefully turns off unnecessary lights but sees only a minimal change in theelectric bill, then his or her beliefs about the economic value of energy conservation will be changednegatively. Demarketers must therefore, where possible, ensure that when people make the effortto consume less, their belief structure is altered in a way that makes future volitional acts more likely.Only when the new pattern of behavior has become habit-like and the old pattern of behaviorhas been extinguished can the habitual behavior change be said to be complete. Each of the threecomponents of habitual behavior change, de-automation, volitional behavior change, and consolidation— must operate for effective demarketing.Relationships Among the Components of Habitual Behavior ChangeThe three components of habitual behavior change operate consecutively, but also iteratively.Cognitive resources are sufficiently scarce and the environment sufficiently complex and challengingthat it is almost certain that people will revert occasionally or frequently to the old behavior patterns,despite their best intentions. In situations where there is competition for processing resources,anything that can be processed in parallel using minimal resources will be. Thus it is the rule ratherthan the exception to observe lapses among people seeking to reduce their consumption. They go offtheir diets, and sometimes they just forget that they intended to insist on less packaging.146An Additional ComplicationIn addition to the challenges inherent in each of the component processes, there is anotherpotential complication. Since the habitual behavior change process is actually a sequence of threedistinct components, operating differently, to different purposes, it is not inconceivable that they maysometimes act to counterpurposes. In other words, the same item of persuasive content may workwell for people in one component, but may actually be detrimental if used for people in anothercomponent. For instance, dramatic information may be useful in de-automation and even volitionalbehavior change, because it attracts attention, but when consolidation is happening, dramatic newinformation, even if it is supportive, is disruptive. An example of this complication sometimes occursin weight loss. Some people lose a lot of weight quickly when they go on an extreme diet, such asa liquid diet, but that kind of weight loss and that style of eating are not maintainable for long. Manyof these people then regain the lost weight because they have no maintainable skills to consolidate.SummaryThis chapter began with a brief review of the psychology literature related to habit changethat revealed that, while there are programs applicable to various specific habits or habit-likebehaviors, there have been few attempts to gain a more general understanding of the underlyingprocesses. The chapter then turned to developing theory on changing habits and habit-like behaviors,based on what is known about the nature of habits and habit-like behavior (reviewed in Chapter 3).It was proposed that changing habit-like behavior proceeds in three steps: de-automation, volitionalbehavior change, and consolidation. Explanations from a cognitive psychology point of view wereoffered for each component of the proposed habit-change process.The next chapter presents the hypotheses and design of a study that tests some of thetheoretical concepts proposed here.CHAPTER 9STUDY 2: HYPOTHESES AND DESIGNThe thesis of this dissertation is that demarketing is not just a variation on the usual marketingproblem, but a problem that is (1) more complex than the typical marketing problem, and (2) similarto the habit change problems studied in clinical psychology. In other words, marketing is like askingconsumers to develop new, “good” habits; while demarketing is like asking consumers to changeexisting “bad” habits; and changing a bad habit is more difficult to do than developing a good habit.Chapter 8 proposed that habit change proceeds in three steps: de-automation of the old habit-like behavior, volitional change of that behavior, and consolidation of the new behavior. This chapterdevelops hypotheses and describes the design of a study that allows two approaches to demarketingexcessive discretionary consumption — i.e., wasteful behavior— to compete directly. The firstapproach is a persuasion-based “traditional marketing” approach, and the second is a behavior-based“habit change” approach, as proposed in Chapter 8. Each is tested in two situations: when thewasteful behavior is new; and when the wasteful behavior is well-entrenched, and has become habitlike. It is expected that if the wasteful consumption behavior and/or the consumption situation isrelatively new (i.e., not yet habit-like), then persuasion-based behavior change strategies will workjust as well as behavior-based change strategies. However, if the wasteful consumption behaviorand/or the consumption situation has become habit-like, then persuasion-based strategies will not147148achieve reductions in consumption; behavior-based strategies will be more effective, though they maybe subject to decay.HypothesesThe hypotheses are stated here in general terms; operational versions of the hypotheses arepresented after the study has been described in more detail.There are three sets of hypotheses. The first set (1A, 1B, and 1C) concerns behavior changewhen a behavior has not been repeated extensively enough to be considered habit-like. Under thesecircumstances, both persuasion-based and habit-based strategies should be effective in changingbehavior (1A and 1B). However, the reinforcement provided by the behavioral reinforcement in thepersuasion-plus-behavior should be more effective in sustaining behavior change (1C). Thesehypotheses are not the central concern of the dissertation. They merely reiterate the predictions ofthe Theory of Reasoned Action (1A) and those of learning theory (lB and 1C). They are presentedto demonstrate that the experimental procedure produces expected results for volitional behavior (i.e.,behavior that is the result of a conscious decision), so that the results produced by the same procedurefor habit-like behavior can be attributed solely to the difference between the two types of behavior.The following hypotheses assume that there is an undesirable current behavior and a desirablealternative behavior that accomplishes the same terminal goal for the consumer (where resource useis instrumental to that terminal goal). Then:Hypothesis 1AIf the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then a persuasive message that produces(or activates) a (more) positive attitude toward the desirable behaviorwill induce the desirable behavior in the post-manipulation period.149Hypothesis lBIf the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will induce the desirable behavior in the post-manipulation period.Hypothesis 1CIf the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then as experience with the situationaccumulates, the behavior change induced by a manipulation thatboth induces (or activates) a (more) positive attitude toward thedesirable behavior and reinforces the positive attitude through limitedpractice of the desirable behavior will be sustained better than thebehavior change induced by a persuasive message that produces (oractivates) a (more) positive attitude toward the desirable behavior.In other words, if the consumption behavior and/or the consumption situation is relatively new,consumption is still volitional, so in the period immediately following the manipulation, traditionalbehavior change strategies based on persuasion to change attitudes should work (1A) just as well asbehavior change strategies based on habit change strategies (1B). Repeated post-manipulationexposure to the situation should generate habit-like behavior, so the habit-based strategies arepredicted to have a superior effect.The second set of hypotheses (2A, 28, 2C, and 2D) concerns behavior change when abehavior has been repeated extensively enough to be considered habit-like. When consumption hasbecome habit-like, merely using a persuasive message to change attitudes should be relativelyineffective (2A), since the theory driving such strategies assumes volitional behavior. In contrast,reinforcing positive attitudes with limited practice (the habit-based strategy) should be effective inachieving behavior change (2B), although interference from the habit-like undesirable behavior islikely to cause some decay in the incidence of the desirable behavior (2C). However, even takinginto account this decay, the habit-based strategy should be more effective than the attitude-onlystrategy (2D). These hypotheses are the central concern of the study.150As with the first set, this second set of hypotheses assumes that there is an undesirable currentbehavior and a desirable alternative behavior that accomplishes the same terminal goal for theconsumer (where resource use is instrumental to that terminal goal). Then:Hypothesis 2AIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then a persuasive message that produces (oractivates) a (more) positive attitude toward the desirable behaviorwill not induce the desirable behavior in the post-manipulationperiod.Hypothesis 2BIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will induce the desirable behavior in the post-manipulation period.Hypothesis 2CIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then as experience with the situation accumulates,behavior change induced by a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will exhibit some decay.Hypothesis 2DIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then as experience with the situation accumulates,behavior change induced by a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will be greater than behavior change (if any)following a manipulation that produces (or activates) a (more)positive attitude toward the desirable behavior.In other words, once consumption has become “habitually” excessive, persuasion-based strategies willnot achieve reductions in consumption. Instead, strategies that are based on habit-change should bemore effective, though they may be subject to decay.The third set of hypotheses concerns the relative effectiveness of a persuasive message alone(the traditional marketing approach) and the persuasive message reinforced by limited practice of the151desirable behavior (the habit-change approach) depending on whether the undesirable behavior hadbeen practised enough to become habit-like.If the undesirable behavior has been practised so much that it has become habit-like, thenchanging that behavior should require more than just a persuasive message; behavioral reinforcementshould also be necessary (3A). On the other hand, if the undesirable behavior has not been practisedenough to have become habit-like, then a persuasive message alone should be enough to effectbehavior change. Reinforcing the persuasive message with behavior would be unnecessary overkill(3B).Again, this third set of hypotheses assumes that there is an undesirable current behavior anda desirable alternative behavior that accomplishes the same terminal goal for the consumer (whereresource use is instrumental to that terminal goal). Then:Hypothesis 3AA manipulation that merely induces (or activates) a (more) positiveattitude toward the desirable behavior will be less effective if theundesirable behavior has been extensively repeated (i.e., has becomehabit-like) than if the undesirable behavior has not been extensivelyrepeated.Hypothesis 3BA manipulation that both induces positive attitudes to ward thedesirable behavior and reinforces that behavior with limited practicewill be equally effective whether or not the undesirable behavior hasbeen extensively repeated (i.e., has become habit-like).Methodological ApproachThe survey in Study 1 had given some indication that a habit-based model, the RevolvingModel of Behavior Change, provides a reasonable description of changes that occur in a consumptionsituation typical of those demarketing programs seek to change, energy conservation. Exploringfurther the connection between demarketing and habit-like behavior called for a more controlledstudy. The data for this study were therefore collected in a laboratory experiment. The experimental152situation was designed to simulate the critical aspects of demarketing problems, while controlling,as much as possible, for the effects of other variables, thereby increasing internal validity.To summarize briefly, participants were asked to complete a task requiring the use of aresource. Their attention was focused on the task; the resource was just a means to an end. Theirinitial resource use was measured. Then the manipulations were delivered. Participants wereexposed either to a persuasive message urging them to reduce their consumption of the resource (thetraditional marketing approach) or to the same persuasive message reinforced with limited practiceof the desired conservation behavior (the habit-change approach). Participants’ post-manipulationconsumption of the resource was measured.Criteria for the Target BehaviorParticipants in the experiment were led to believe that they were involved in a study on howthe organization of requests for information affects memory for that information. The actual variableof interest, however, was their consumption of paper in the completion of the memory task. Thisconsumption behavior possesses many of the key characteristics of behaviors that are the focus ofdemarketing problems.First, in many discretionary consumption situations, people become accustomed to using theresource wastefully. While there may originally have been good reasons for using the resourceswastefully, those good reasons have become obsolete (at least from the demarketer’s point of view).In this experiment, participants quickly became used to using only one side of the paper, rather thanboth sides.Second, in many discretionary consumption situations, the resource is not of central concernto the participant; it is merely a means to an end. In fact, it may be taken completely for granted.For example, most people do not think about how much energy they are using when they turn uptheir thermostats a few degrees; they just want to get warm. In this experiment, a cover story was153used to lead participants to believe that they were involved in a study of how the organization ofrequests for information affects memory for that information; the paper was just a resource theyneeded to complete the tasks assigned.Third, in many discretionary consumption situations, changing the behavior is costly, thoughnot necessarily in financial terms. For example, keeping the house a few degrees cooler than usualis not as comfortable. In this experiment, participants were under some time pressure, they wereinstructed to keep their work organized chronologically, and they knew that they would have to beable to locate specific lists quickly; this was not as easy to do when two sides of the paper are usedinstead of one.Fourth, in many discretionary consumption situations, changing the behavior often hasuncertain returns that will accrue far in the future, frequently to people other than those making theeffort. For example, keeping the house a few degrees cooler means that people are uncomfortablenow so that their descendants may be less uncomfortable in the future. In this experiment, there wasno clear benefit (with respect to the task being performed) to using two sides of the paper, other thanit was the “right” thing to do.Finally, in many discretionary consumption situations, the reasons for changing the behaviorare not immediately evident. For example, there are so many rivers in B.C. that could accommodatehydroelectric dams that many people have trouble believing that there is any need for BritishColumbians to conserve electricity. In this experiment, there was plenty of paper in full view; it wasclear that there was no immediate danger of running out. The only reason participants wouldconserve paper is because it is the “right” thing to do.Overview of the ProcedureAfter an introductory videotape that set up the cover story, the experiment proceeded in sixsteps: first, participants completed an initial set of lists; then, manipulations were delivered; after154that, there was a distraction; this was followed by the completion of another set of lists; aquestionnaire was then filled out. Finally, participants were thoroughly debriefed by mail.Pre-manipulation lists. In an initial set of trials (either a short set or a long set), participantsmade lists of items in various categories (e.g., countries in South America). They were instructedto put the number of the list in the upper left hand corner of the page and to keep the lists in order.After the lists were completed, participants were asked to place specific lists on the top of their pileof lists.Manipulations. Within a videotaped lecture on network models of memory, participantsreceived either a persuasive manipulation, a persuasion plus behavior manipulation, or nomanipulation (Control).Distraction. Participants were shown an interesting 10 minute videotape (an award-winninghumorous cartoon documentary on home safety).Post-manipulation lists. Participants were be asked to make lists of items in 30 categories.Again, they were instructed to put the number of the list in the upper left hand corner of the pageand keep the lists in order. After the lists were completed, participants were asked to count thenumber of items in certain specific lists and record the count in the bottom left hand corner of thepage on which the list appeared.Questionnaire. Participants filled out a short questionnaire containing demographic andattitudinal information.Debriefing. Participants were thoroughly debriefed in writing. Special attention was givento explaining the deception and why it was judged to be necessary.DesignThe design was simple. As Table 9-1 shows, each participant was to be randomly assignedto one of the six groups. Participants would receive either little premanipulation practice of making155one-sides lists (five lists) or extensive premanipulation practice (30 lists). Their base rate of resourceuse could thereby be measured. The manipulations were then delivered. A distraction was thenprovided, after which participants continued to make lists. Their post-manipulation rates of resourceuse could then be measured.Table 10-1Experimental Procedure and Design’3Little Pre-Manipulation Repetition of the Undesirable BehaviorControl R U1 O 02 03 QPersuasion R U1 04 P 05 06 QPersuasion + Behavior R U1 07 P B 08 09 QExtensive Pre-Manipulation Repetition of the Undesirable BehaviorControl R U2 010 012 QPersuasion R‘2 0)3 P 0J4 QPersuasion + Behavior R‘2 0)6 P B 0J7 0 QThe dependent variables concerned how successful the various manipulations were at inducingparticipants to use both sides of the paper.Operational Hypotheses. Recall that the first set of hypotheses concerns the Little Pre-ManipulationPractice of the Undesirable Behavior groups. It was assumed that before the manipulations, all threetreatment groups (Persuasion + Behavior, Persuasion, and Control) would be doing the tasks usingonly one side of the paper. Confirming this assumption required:Ala: O,=O4O7013 where: R refers to the randomization procedure;Uk refers to the amount of pre-manipulation practice of the undesirable behavior:k 1 for little premanipulation practice (five lists),k = 2 for extensive premanipulation practice (30 lists);P refers to the persuasive manipulation;B refers to the behavioral reinforcement of the persuasive manipulation;O refers to the nth observation of the dependent variable; andQ refers to the administration of the questionnaire (attitude and demographic data).156Moreover, it was assumed that the task itself did not induce the desirable behavior. Confirming thisassumption required:Aib: O7=Og=Og=ORecall Hypothesis 1A:Hypothesis 1AIf the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then a persuasive message that produces(or activates) a (more) positive attitude toward the desirable behaviorwill induce the desirable behavior in the post-manipulation period.The operational statement of this hypothesis is:Hia: 05 > 02 and 02 = 0Recall Hypothesis 1B:Hypothesis lBIf the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will induce the desirable behavior in the post-manipulation period.The operational statement of this hypothesis is:Hib: 08>O2andO=0A stronger statement of Hib is:0205This states that attitude-based and habit-based strategies are equally effective in changing volitionalbehaviors.Recall Hypothesis 1C:157Hypothesis 1CIf the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then as experience with the situationaccumulates, the behavior change induced by a manipulation thatboth induces (or activates) a (more) positive attitude toward thedesirable behavior and reinforces the positive attitude through limitedpractice of the desirable behavior will be sustained better than thebehavior change induced by a persuasive message that produces (oractivates> a (more) positive attitude toward the desirable behavior.The operational statement of this hypothesis is:Hic: 03> 06The second set of hypotheses concerns the High Pre-Manipulation Practice of the UndesirableBehavior groups. As with the first set of hypotheses, it was assumed that before manipulations, allthree treatment groups (Persuasion + Behavior, Persuasion, and Control) would be doing the tasksusing only one side of the paper. Confirming this assumption requires:A2a: 010 = 013 = 03 0Moreover, it was assumed that the task itself did not induce the desirable behavior. Confirming thisassumption requires:A2b: 016 = 07 = 0 = 0Recall Hypothesis 2A:Hypothesis 2AIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then a persuasive message that produces (oractivates) a (more) positive attitude toward the desirable behaviorwill not induce the desirable behavior in the post-manipulationperiod.The operational statement of this hypothesis is:ma: 014 = = 0Recall Hypothesis 2B:158Hypothesis 2BIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will induce the desirable behavior in the post-manipulation period.The operational statement of this hypothesis is:H2b: 011>0Recall Hypothesis 2C:Hypothesis 2CIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then as experience with the situation accumulates,behavior change induced by a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will exhibit some decay.The operational statement of this hypothesis is:H2c: > 012Recall Hypothesis 2D:Hypothesis 2DIf the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then as experience with the situation accumulates,behavior change induced by a manipulation that both induces (oractivates) a (more) positive attitude toward the desirable behavior andreinforces the positive attitude through limited practice of thedesirable behavior will be greater than behavior change (if any)following a manipulation that produces (or activates) a (more)positive attitude toward the desirable behavior.The operational statement of this hypothesis is:H2d: 012 > 015The third set of hypotheses concerned the relative effectiveness of persuasion alone versuspersuasion with behavioral reinforcement depending on whether the undesirable behavior had beenextensively practised.159Recall Hypothesis 3A:Hypothesis 3AA manipulation that merely induces positive attitudes toward thedesirable behavior will be less effective if the undesirable behaviorhas been extensively repeated (i.e., has become habit-like) than if theundesirable behavior has not been extensively repeated.The operational statement of this hypothesis is:H3a: 014 < ORecall Hypothesis 3B:Hypothesis 3BA manipulation that both induces positive attitudes toward thedesirable behavior and reinforces that behavior with limited practicewill be equally effective whether or not the undesirable behavior hasbeen extensively repeated (i.e., has become habit-like).The operational statement of this hypothesis is:H3b: 08 = 07PretestsSeveral pretests were done. Pretest 1 (n = 5) confirmed that there was not an existingtendency to use both sides of the paper for the experimental lists. Pretest 1 participants made 20 listswithout receiving any instructions on whether to use one or both sides. All Pretest 1 participantsused one side of the paper.Pretests 2 through 6 (total n = 26) were used to make sure the procedure ran smoothly.Several minor changes in procedure resulted.ProcedureRandomization. The experiment was run in groups. Random assignment of participants togroups was achieved by listing the participants (who had signed up in advance) in alphabetical order,by surname and secondarily by given name(s), then assigning the first participant on the list to the160first group, the second to the second group, and so on through the groups until each participant hasbeen assigned to a group.Preliminaries. Each group met in a different room and all groups ran simultaneously.Participants in the Persuasion + Behavior groups each received an information sheet (see Appendix2), a consent form (see Appendix 2), and four manila envelopes, clearly labelled 1, 2, 3, and 4.Participants in the Persuasion and Control groups each received an information sheet (see Appendix2), a consent form (see Appendix 2), and three manila envelopes, clearly labelled 1, 2, and 3. Inaddition, each participant was supplied with a large pad of 3Hx5u paper and a pen. Plenty of sparepads were available and prominently displayed at the front of the room (on the administrator’s desk).At the beginning of the session, the administrator read through the information sheet with theparticipants and then asked the participants to sign the consent form. After the participants had achance to do so, the administrator collected the signed consent forms.After all the signed consent forms were collected, the administrator distributed the instructionsheet (see Appendix 2), and then read through it with the participants, who then had an opportunityto ask questions about the instructions before beginning the task.Pre-manipulation lists. The administrator read out a randomly-ordered list of categories,numbered consecutively, at the rate of one every 30 seconds (see Appendix 2). As the administratorread each category, he/she projected it on a screen. (The list of categories were printed on overheadtransparencies, and sequentially revealed as they were read.) Participants wrote the number of thecategory in the upper left hand corner of each page, and then listed items that fit into the category.The Little Pre-manipulation Repetition groups completed five lists (2.5 minutes); the Extensive Premanipulation Repetition groups did 30 lists (15 minutes). After every 10 lists, the administratorinstructed participants to count the number of items in certain specific lists of that ten (e.g., 1, 6, 8)and record the number of items in the bottom left hand corner of the page on which the list appeared.(The purpose of this was to increase the perceived cost of changing to making lists on both sides of161the paper.) After all the lists had been completed, the administrator instructed participants insert thelists into envelope 1.Manipulations. The manipulations were delivered via videotape. The script, with variationsindicated for each condition, appears in Appendix 2. The videotapes were all made from a singlemaster. Since the Persuasion + Behavior manipulation uses more words than the Persuasionmanipulation, which itself uses more words than the Control, the Persuasion + Behavior version wastaped as the master; appropriate deletions were made for the other two versions.Participants watched the video. Participants in the Persuasion + Behavior groups also made15 lists during the showing of the video, on paper on which the category names had been preprintedon both sides of the paper (i.e., Persuasion + Behavior participants made 15 lists during the video,using both sides of the paper). Persuasion + Behavior participants then inserted the 15 lists they hadmade during the showing of the video into envelope 2.Distraction. To minimize the carryover of rote behavior and to determine whether thePersuasion and Persuasion + Behavior manipulations had been internalized enough to persist overa short interval, a distraction was provided. Participants watched a 10 minute award-winning,humorous animated documentary on an unrelated topic (household safety).Post-manipulation lists. The administrator again read out categories at the rate of one every30 seconds. In order to maintain consistency with the cover story, each post-manipulation categoryhad some connection with the one before. See Appendix 2 for the list of post-manipulationcategories. All participants made a total of 30 lists (15 minutes). Again, after every 10 lists, theadministrator instructed participants to count and record the number of items in certain specific listsof that ten (e.g., 1, 6, 8). After all the lists were completed, the administrator instructed participantsto all the lists into the envelope (envelope 3 for the Persuasion + Behavior participants, and envelope2 for the Persuasion and Control participants).162Questionnaire. The administrator asked participants to complete the questionnaire (seeAppendix 2), put it into the remaining envelope when they were finished.Release. Participants were thanked for their participation and paid by the administrator.Debriefing. Participants were thoroughly debriefed in a package mailed to their residences.The “process debriefing” technique was used, in which the experimental procedures are explainedto participants thoroughly, including reasons for any deception.ManipulationsIt was important that the manipulations not be perceived by the participants as instructionsthat would carry over to the post-manipulation part of the session. All manipulations were bedelivered during the videotaped lecture. The lecture used 15 categories to explain how the networkmodel of memory might work in the present situation. The first category was “Zoo animals;” thesecond, “Endangered species” and the third, “Environmental problems.”Persuasion + Behavior Manipulation. Participants were instructed to open envelope 3, inwhich they found 8 sheets of paper, with the 15 categories preprinted with one category on each sideof each sheet. The person appearing in the video instructed participants to make the lists referredto. On the third category, the demonstrator said, “That reminds me . . . We use a lot of paper inthis study, since we have to have each list on a separate page. If you use both sides of the paper,that would save some paper and some trees. Whether you use one side or two makes no differenceto the study, but it would be good for the environment.”Persuasion Manipulation. Participants watched the video, but did not make lists. On thethird category, the demonstrator said, “That reminds me . . . We use a lot of paper in this study,since we have to have each list on a separate page. If you use both sides of the paper, that wouldsave some paper and some trees. Whether you use one side or two makes no difference to the study,but it would be good for the environment.”163Control. Participants watched the video, but the person appearing in the video neitherinstructed participants to make lists, nor did she say anything about the use of paper in this study.SampleStudent participants were recruited from the Faculty of Business at Memorial University ofNewfoundland. The experimenter visited class meetings for required courses in all Classes, invitingparticipation and handing around signup sheets. Reminder posters were put up the day before andthe day of the experiment.Participation was voluntary and occurred outside of class time, during a university-wide noclass period. Each participant was paid $5.00 per hour. Participation was anonymous andconfidential, in the sense that there was no way to match the data with the individual supplying it.A total of approximately 175 students signed up to participate in the experiment. Based onexperience with pretests, a 25 to 30 percent dropoff rate was expected. In fact, 117 (67%) of thestudents who signed up to participate actually presented themselves at the appointed place and time.CHAPTER 10STUDY 2: ANALYSIS AND RESULTSRecall that in Study 2, each participant was randomly assigned to either a control group orone of four treatment groups. Participants practised a task (making lists) either a little (five lists) orextensively (30 lists), using a resource (paper) wastefully (by using only one side). Participants wereled to believe that the variable of interest was how long and interesting their lists were. Thenparticipants saw a videotape concerning the task, in which the manipulations were inserted for thefour treatment groups. Treatment participants were encouraged, either in a persuasive message, orin a persuasive message that was reinforced through limited practice, to complete the task using lessof the resource (by using both sides of the paper rather than just one side). Participants thencontinued with the task (for 30 more lists).Each participant’s resource use was observed at three points during the experiment: onceafter the initial set of practice trials; and twice after the manipulations, once immediately followingthe manipulation (the first ten lists post-manipulation) and once after some time had elapsed (the finalten of 30 lists post-manipulation). The dependent variable was the number of times participants usedboth sides of the paper in the set of lists being observed.The procedure and design are represented symbolically in Table 10-1:164165Table 10-1Experimental Procedure and Design14Little Pre-Manipulation Repetition of the Undesirable BehaviorControl R U1 02 03 QPersuasion R U1 04 P 05 05 QPersuasion + Behavior R U1 07 P B°8 09 QExtensive Pre-Manipulation Repetition of the Undesirable BehaviorControl R U2 010 011 012 QPersuasion R U2 0)3 P 0J4 0)5 QPersuasion + Behavior R U2 0J5 P B 0)7 018 QSince the dependent variable was the number of times participants used both sides of the paper in theset of lists being observed, 0 reports, for each participant, the number of pieces of paper with listson both sides, for a given set of trials. For instance, 01 reports, for the five initial practice lists, thenumber of pieces of paper with lists on both sides. If a participant used only one side of the paper,then there would be no pieces of paper with lists on both sides, and the value of 0 would be 0. If,however, the participant had consistently used both sides of the paper, then there would be two piecesof paper with lists on both sides, and one piece of paper with a list on one side, and the value of 0would be 2. Similarly,°2 reports on the first ten lists after the videotape. If a participant used onlyone side of the paper for all ten lists, then there would be no pieces of paper with lists on both sides,and the value of°2 would be 0; if the participant consistently used both sides of the paper, then therewould be five pieces of paper with lists on both sides, and the value of 02 would be 5; and if the‘4where: R refers to the randomization procedure;Uk refers to the amount of pre-manipulation practice of the undesirable behavior:k = 1 for little premanipulation practice (five lists),k 2 for extensive premanipulation practice (30 lists);P refers to the persuasive manipulation;B refers to the behavioral reinforcement of the persuasive manipulation;0,, refers to the nth observation of the dependent variable; andQ refers to the administration of the questionnaire (attitude and demographic data).166participant used both sides of the paper for some of the first ten lists after the videotape and only oneside for other lists, then the value of 02 would be greater than 0 and less than 5.The possible range for 0 therefore depended on the number of lists on which it is based.For the initial set of practice trials in the Little Pre-Manipulation Repetition groups, (i.e., 01, 04, and07), the possible range was between 0 and 2. For the initial set of practice trials in the ExtensivePre-Manipulation Repetition groups, (i.e., 0, 0)3, and the possible range was between 0 and15. In all groups, the possible range for the first ten lists post-manipulation (i.e., 02, 0, 08, 011,014, and 0)7) was 0 to 5, as it was in the final ten of 30 lists post-manipulation (i.e., 03, 06, 09, 012,0)5, and 018).Overview of the DataIn order to provide a quick overview of the data, the design and procedure schematic isreproduced in Table 10-2, with the means of each observation added. Note that the design called forsix groups (two control and four treatment groups) but, given the number of volunteers available, itwas judged appropriate to drop one control group in order to increase the number of participants,therefore statistical power, in the remaining groups.Note that the groups sizes were very similar. Tests of the data gathered for this studyincluded the t-test and analysis of variance (ANOVA), which are robust to violations of homogeneityof variance as long as group sizes are equal. In this study, group sizes were not exactly equal, butthey were close enough that the effect on a was negligible (Glass and Hopkins 1984: 238-240).Consequently, tests of homogeneity of variance were not performed in subsequent analyses.167Table 10-2Mean Number of Times Participants in Each Group Used Both Sides of the Paper at VariousPoints in the ExperimentLittle Pre-Manipulation Repetition of the Undesirable BehaviorControl R U1 0 02 03n = 22 0.00 0.00 0.00Persuasion R U1 04 P 05 06n = 23 0.00 0.87 0.87Persuasion + Behavior R U1 07 P B 08 09n = 24 0.00 2.92 2.92Extensive Pre-Manipulation Repetition of the Undesirable BehaviorControl R U2 010 012n=0 n/a n/a n/aPersuasion R U2 013 P 014 015n = 23 0.00 0.87 0.80Persuasion + Behavior R U2 0)6 P B 0)7 °18n = 25 0.72 0.80Another way to get an overview of the results is to look at how many participants in eachgroup switched from using one side of the paper to using both sides. Table 10-3 provides thatinformation.Table 10-3Number of Participants Who Switched from Using One Side of the Paper to Using Both SidesAmount of Pre-ManipulationRepetition of theUndesirable BehaviorLittleExtensivePersuasionControl Persuasion +Behavior0 4 14n/a 5* 4*** Of the five participants in this group who switched from using one side of the paper to using bothsides, three switched for all pairs of lists, one did so after the fifth pair of lists (of 15), and the otherdid so only for the first ten pairs of lists (of 15).** Of the four participants in this group who switched from using one side of the paper to using bothsides, three switched for all pairs of lists, and one did so after the second pair of lists (of 15).Treatment168A Note on Terminology“Little Pre-Manipulation Repetition of the Undesirable Behavior” and “Extensive PreManipulation Repetition of the Undesirable Behavior” are accurate descriptions of the type ofbehavior participants engaged in before seeing the videotape containing the behavior changemanipulations. These descriptions are, however, lengthy and awkward when they are repeated manytimes. Consequently, a shorthand terminology was adopted here. Groups with “Little PreManipulation Repetition of the Undesirable Behavior” are referred to as “Short” groups; and groupswith “Extensive Pre-Manipulation Repetition of the Undesirable Behavior” are referred to as “Long”groups. Likewise, the “Persuasion” and “Persuasion + Behavior” manipulations are referred to as“Persuasion” and “Behavior.”These changes result in group names that are less accurately descriptive, but much lessunwieldy:• Short Persuasion refers to the group with “Little Pre-Manipulation Repetition of theUndesirable Behavior,” followed by the “Persuasion” manipulation.• Short Behavior refers to the group with “Little Pre-Manipulation Repetition of theUndesirable Behavior,” followed by the “Persuasion + Behavior” manipulation.• Long Persuasion refers to the group with “Extensive Pre-Manipulation Repetition of theUndesirable Behavior,” followed by the “Persuasion” manipulation.• Long Behavior refers to the group with “Extensive Pre-Manipulation Repetition of theUndesirable Behavior,” followed by the “Persuasion + Behavior” manipulation.The Control group in this experiment had “Little Pre-Manipulation Repetition of the UndesirableBehavior,” and no behavior change manipulation.169Manipulation CheckThe persuasive and behavioral manipulations were intended to produce more favorableattitudes towards using both sides of the paper in making the lists. For the manipulations to bejudged successful, participants in the experimental groups should have more favorable attitudestoward using both sides of the paper than would participants in the control group.Attitude to using both sides of the paper was measured by summing the three attitudesubscales. (“In this study, using two sides of the paper is:” . . . good/bad; wise/foolish;beneficial/harmful.)Table 10-4 shows the mean and standard deviation of the summed attitude measure for eachgroup.Table 10-4Attitude Toward Using Both Sides of the PaperGroup Mean SD n’5Control 14.18 4.44 22Short Persuasion 16.57 3.63 23Short Behavior 17.74 3.92 23Long Persuasion 17.86 2.52 21Long Behavior 19.00 2.21 24Entire Sample 17.10 3.76 113A oneway ANOVA revealed that there were significant differences among the means (F =6.37, p = .0001). A planned contrast confirmed that participants in the experimental groups hadmore positive attitudes towards using both sides of the paper than did participants in the control group(t = 4.41, p < .0001). On the basis of this result, it seems reasonable to conclude that the‘5The t-test and analysis of variance are robust to violations of the homogeneity of varianceassumption when the group sizes are equal. In this study, group sizes are not exactly equal, but theyare close enough that the effect on c will be negligible (Glass and Hopkins 1984: 238-240).Consequently, tests of homogeneity of variance will not be performed in subsequent analyses of thesedata.170experimental manipulations were successful in producing more favorable attitudes towards using bothsides of the paper.A second issue concerning the experimental manipulations is whether some were moreeffective than others in producing favorable attitudes toward using both sides of the paper. If so,differences in the behavior produced by the manipulations could be attributed at least partly to theirdifferential success in changing attitudes. This issue was addressed two ways.First, a post-hoc multiple comparison test was employed. Since no hypotheses had beendeveloped concerning the relative effectiveness of the manipulations in changing attitudes, plannedcontrasts were not appropriate. The Student-Newman-Keuls procedure, generally the preferredmethod of doing post-hoc multiple comparisons (Glass and Hopkins 1984), was therefore used todetermine which pairs of groups were significantly different at the a = .05 level of significance. TheStudent-Newman-Keuls method is one of several multiple comparison techniques that uses thestudentized range statistic, q, to compare each mean with each and every other mean (Glass andHopkins 1984). The procedure yields subsets of means that do not differ significantly. Table 10-5shows the resulting subsets for the experimental data16.Table 10-5Student-Newman-Keuls Multiple Range Test forAttitude Toward Using Both Sides of the PaperSUBSET 1Group ControlMean 14.18SUBSET 2Group Short Persuasion Short Behavior Long Persuasion Long BehaviorMean 16.57 17.74 17.86 19.00‘The statistical package used to analyze these data, SPSSpc+, does not provide the value of thestudentized range statistic, q. For that reason, q is not reported here.171According to this test, there was a significant difference between the control group and theexperimental groups, but there were no significant differences among the four experimental groupsin terms of attitude towards using both sides of the paper in this study.A two-factor ANOVA was the second approach used to test whether the experimentalmanipulations were differentially effective in producing more favorable attitudes. Table 10-6 showsthe same data as Table 10-4, but set up for the two-factor ANOVA. Table 10-7 shows the resultsof the ANOVA.Table 10-6Attitude Toward Using Both Sides of the PaperAmount of Persuasive Manipulation Behavior ManipulationPre-manipulationPractice Mean SD n Mean SD nless (5 lists) 16.57 3.63 23 17.74 3.92 23more (30 lists) 17.86 2.52 21 19.00 2.21 24Table 10-7Analysis of Variance:Effectiveness of Manipulations in Producing Favorable AttitudesSource of Variation df MS F pPractice 1 36.99 3.71 .057Treatment 1 30.47 3.06 .084Practice x Treatment 1 00.01 < .01 .981The results in Table 10-7 are consistent with those of the multiple comparison. Both tests indicatethat the attitudes generated by the manipulations did not differ significantly among the experimentalgroups.Another issue related to the manipulation check concerns the role that concomitant variablesmight play in participants’ behavior. Analysis of covariance (ANCOVA) was used to examine theeffects of the following concomitant variables on participants’ behavior after the manipulations: sex,age, size of home community, academic degree sought, and importance of environmental issues.Three ANCOVAs were performed. The first examined the effect of the concomitant variables on172whether participants used one or both sides of the paper in the first ten lists post-manipulation. Noneof the concomitant variables had a significant effect (for sex, F = 1.31, p = 0.25; for age, F =0.O8,p = 0.78; for size of home community, F = O.Ol,p = 0.78; for degree sought, F = 2.19,p 0.14; and for importance of environmental issues, F = 0.61; p = 0.44).The second ANCOVA examined the effect of the concomitant variables on whetherparticipants used one or both sides of the paper in the final ten of the 30 post-manipulation lists.Again, none of the concomitant variables had a significant effect (for sex, F = 1.70, p = 0.20; forage, F = 1.66, p = 0.20; for size of home community, F = 0.002, p = 0.96; for degree sought,F = 2.13, p = 0.15; and for importance of environmental issues, F = 1.42; p = 0.24).The third ANCOVA examined the effect of the concomitant variables on whether participantsused one or both sides of the paper in the whole series of 30 post-manipulation lists. Here, too, noneof the concomitant variables had a significant effect (for sex, F = 1.76, p = 0.19; for age, F =0.74, p = 0.39; for size of home community, F = 0.01, p = 0.95; for degree sought, F = 2.09,p = 0.15; and for importance of environmental issues, F = 0.92; p = 0.34).These results rule out the possibility that resource use after the manipulations could be dueto sex, age, size of home community, and importance of environmental issues17.Hypotheses Concerning Groups with Less Premanipulation Task ExperienceThe first set of hypotheses concerns the groups with less task experience. Participants in theControl, Short Persuasion, and Short Behavior groups completed only five lists before viewing thevideotape in which the manipulations were delivered. According to the theory developed earlier in‘71t may seem surprising that participants’ evaluation of the importance of environmental issues intheir own lives would not have a significant effect on their resource use. Actually, it is not surprisingat all, for two reasons. First, it is well established that in the area of environmental issues, whatpeople say and what they do are often independent; second, it is also well established that globalattitudes are very poor predictors of specific behaviors. Therefore, even when people say that theenvironment is important in their lives, their behavior in any specific context may not reflect thatdegree of importance.173this dissertation research, it should be relatively easy to effect change in these groups, as theundesirable behavior has not been practised enough to become habit-like. Results of tests of thespecific hypotheses and certain underlying assumptions are detailed below.Assumption 1A. It was assumed that all participants in the low-practice groups (Control,Short Persuasion, Short Behavior) would initially do the task using one side of the paper only. Thisassumption was verified by counting the number of times each participant used both sides of the paperin the first series of lists (before the videotape containing the manipulations was shown). Allparticipants in the low-practice groups did all lists in the first series using only one side of the paper.Assumption 1 A is therefore confirmed.Assumption lB. It was assumed that in the low practice groups, the task itself would notinduce participants to switch from using one side of the paper to using both sides of the paper. Thisassumption was verified by counting the number of times participants in the Control group (whoreceived no manipulation encouraging them to switch from using one side to using both sides of thepaper) used both sides of the paper in the five lists completed before the videotape was shown, in theten lists immediately following the videotape, and in the final ten of the 30 lists made after thevideotape was shown. All participants in the Control group (n = 22) used only one side of the paperthroughout the study. Assumption lB is therefore confirmed.Hypothesis 1A. The statement of this hypothesis in theoretical terms is:If the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then a persuasive message that produces(or activates) a (more) positive attitude toward the desirable behaviorwill induce the desirable behavior.In the context of this study, the undesirable behavior is the use of one side of the paper; and thedesirable behavior is the use of both sides of the paper. Hypothesis 1A was tested two ways. First,participants in the Control and Short Persuasion groups were compared in terms of the mean numberof times they used both sides of the paper in the ten lists immediately following the videotape wascompared. As Table 10-8 shows, none of the participants in the Control group used both sides of174the paper after viewing the videotape; however, on average, participants in the Short Persuasiongroup used both sides of the paper .87 times in the ten lists immediately following the video (wherefive times would be the maximum possible). This represents a significant difference between the twogroups (t = 2.07,p = .022).Table 10-8Comparison of the Average Number of Times Participants in the Control and Short PersuasionGroups Used Both Sides of the Paper for Lists Immediately Following the VideotapeGroup Mean SD n t-value p (1-tail)Control 0.00 0.00 22 2.07 .022Short Persuasion 0.87 1.94 23Hypothesis 1A was also evaluated with the binomial test, a nonparametric test that was usedto indicate whether the proportion of participants in the Short Persuasion Group who switched fromusing one side of the paper to using both sides (4/23 = 0.17) was significantly greater than that inthe Control group (0/22 = 0.00). The binomial test confirmed that the proportions were significantlydifferent (p < .0001).Based on the results of the t-test and the binomial test, Hypothesis 1A is supported.Hypothesis lB. The statement of Hypothesis lB in theoretical terms is:If the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then a manipulation that both inducespositive attitudes toward the desirable behavior and reinforcespositive attitudes through limited practice of the desirable behaviorwill also induce the desirable behavior in the post-manipulationperiod.Hypothesis lB was tested by comparing the mean number of times participants in the Controland Short Behavior groups used both sides of the paper in the ten lists immediately following thevideotape. As Table 10-9 shows, none of the participants in the Control group used both sides ofthe paper after viewing the videotape; however, on average, participants in the Short Behavior groupused both sides of the paper 2.92 times in the ten lists immediately following the video (where five175times would be the maximum possible). This represents a significant difference between the twogroups (t = 5.67, p < 0.001).Table 10-9Comparison of the Average Number of Times Participants in the Control and Short BehaviorGroups Used Both Sides of the Paper for Lists Immediately Following the VideotapeGroup Mean SD n t-value p U-tail)Control 0.00 0.00 22 5.67 <.001Short Behavior 2.92 2.52 24Hypothesis lB was also evaluated with the binomial test, a nonparametric test that was usedto indicate whether the proportion of participants in the Short Behavior Group who switched fromusing one side of the paper to using both sides (14/23 = 0.61) was significantly greater than that inthe Control group (0/22 0.00). The binomial test confirmed that the proportions were significantlydifferent (p < .0001).Based on the results of the t-test and the binomial test, Hypothesis lB is supported.A stronger statement of Hypothesis lB would be:If the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then a manipulation that both inducespositive attitudes toward the desirable behavior and reinforcespositive attitudes through limited practice of the desirable behaviorwill be no more effective in inducing the desirable behavior in thepost-manipulation period than will a manipulation that (merely)induces positive attitudes toward the desirable behavior.The issue of whether the persuasive manipulation alone was as effective as the behavioralmanipulation was addressed with a t-test, as shown in Table 10-10.Table 10-10Comparison of the Average Number of Times Participants in the Short Persuasion and ShortBehavior Groups Used Both Sides of the Paper for Lists Immediately Following the VideotapeGroup Mean SD n t-value p (2-tail)Short Persuasion 0.87 1.94 23 3.13 .003Short Behavior 2.92 2.52 24176The results of the t-test indicate that, in the first ten lists after the videotape, participants in the ShortBehavior group used both sides of the paper significantly more often than did participants in the ShortPersuasion group (means of 2.92 times and 0.87 times, respectively; t = 3.13, p = .003).A chi-square test was also performed to determine whether participants in the ShortPersuasion and Short Behavior groups were equally likely to switch from using one side of the paperto using both sides. In the Short Persuasion group, 4 of 23 participants (17.4%) switched; in theShort Behavior group, 14 of 24 participants (58.3%) switched. This represents a significantdifference (x2 = 8.33, p <.004).Based on the results of the t-test and the chi-square test, the strong form of Hypothesis lBis rejected.Hypothesis 1C. The statement of this hypothesis in theoretical terms is:If the undesirable behavior has not been extensively repeated (i.e.,has not become habit-like), then as experience with the situationaccumulates, the behavior change induced by a manipulation thatboth induces positive attitudes toward the desirable behavior andreinforces positive attitudes through limited practice of the desirablebehavior will be sustained better than the behavior change inducedby a persuasive message that produces (or activates) a (more)positive attitude toward the desirable behavior.Hypothesis 1C was evaluated by comparing the number of times participants in the Short Behaviorand Short Persuasion groups used both sides of the paper in the first ten lists following the videotapewith the number of times they used both sides of the paper in the final ten (of 30) lists. In bothgroups, all participants maintained the same behavior for all lists following the videotape. Allparticipants in the Short Persuasion and Short Behavior groups who used both sides of the paper inthe first ten lists following the videotape continued to use both sides of the paper for all 30 lists.Hypothesis 1C is therefore rejected.177Hypotheses Concerning Groups with More Premanipulation Task ExperienceThe second set of hypotheses concerns the groups with more task experience. Participantsin the Long Persuasion and Long Behavior groups completed 30 lists before viewing the videotapein which the manipulations were delivered. According to the theory developed earlier in thisdissertation research, it should be relatively difficult to effect change in these groups, as theundesirable behavior has been practised enough to become habit-like. Results of tests of the specifichypotheses and certain underlying assumptions are detailed below.Note that there was no control group among the groups with more premanipulation taskexperience. Given the number of participants available, it was decided to sacrifice this control groupin order to increase numbers (and, therefore, power) in the treatment groups.Assumption 2A. As in the groups with less premanipulation task experience, it was assumedthat all participants would use only one side of the paper before seeing the videotape. Thisassumption was verified by counting the number of times each participant used both sides of the paperin the first series of lists (before the videotape containing the manipulations was shown). Allparticipants in the high-practice groups did all lists in the first series using only one side of the paper.Assumption 2A is therefore confirmed.Assumption 2B. It was assumed that, as in the low practice groups, the task itself wouldnot induce high practice participants to switch from using one side of the paper to using both sidesof the paper. Assumption 2B was not tested.Hypothesis 2A. The statement of this hypothesis in theoretical terms is:If the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then a persuasive message that produces apositive attitude toward the desirable behavior will not induce thedesirable behavior in the post-manipulation period.Hypothesis 2A was tested by examining the mean number of times participants in the LongPersuasion group used both sides of the paper in the first ten and the last ten (of 30) lists after thevideotape was shown. As Table 10-11 shows, on average, participants in the Long Persuasion group178used both sides of the paper 0.87 times (out of a maximum of five) in both the first ten lists and thelast ten lists after the videotape. A t-test indicates that this value is significantly greater than zero (t= 2.15, p = .022). This result does not support Hypothesis 2A.Table 10-11Comparison of the Average Number of Times Participants in the Long Persuasion Group UsedBoth Sides of the Paper for Lists Following the VideotapePost-manipulation Lists Mean SD nFirst ten 0.87 1.94 23Final ten 0.87 1.94 23Hypothesis 2A was also evaluated with the binomial test, a nonparametric test that was usedto indicate whether the proportion of participants in the Long Persuasion Group who switched fromusing one side of the paper to using both sides (5/23 = 0.22) was significantly greater than zero.The binomial test confirmed that the proportion was significantly different (p < .0001), again failingto support Hypothesis 2A.On the basis of the results of the t-test and the binomial test, Hypothesis 2A is rejected.Hypothesis 2B. The statement of this hypothesis in theoretical terms is:If the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then a manipulation that both induces positiveattitudes toward the desirable behavior and reinforces those positiveattitudes through limited practice of the desirable behavior willinduce the desirable behavior in the post-manipulation period.Hypothesis 2B was tested by examining the mean number of times participants in the Long Behaviorgroup used both sides of the paper in the first ten (of 30) lists after the videotape was shown. AsTable 10-12 shows, on average, participants in the Long Persuasion group used both sides of thepaper 0.72 times (out of a maximum of five) in the first ten lists after the videotape. A t-testindicates that this value is significantly greater than zero (t = 2.09, p = .024). This result supportsHypothesis 2B.179Table 10-12Average Number of Times Participants in the Long Behavior Group Used Both Sides of thePaper for Lists Immediately Following the VideotapeGroup Mean SD n t-value p (1-tail)Long Behavior 0.72 1.72 25 2.09 .024Hypothesis 2B was also evaluated with the binomial test, a nonparametric test that was usedto indicate whether the proportion of participants in the Long Behavior Group who switched fromusing one side of the paper to using both sides (4/25 0.16) was significantly greater than zero.The binomial test confirmed that the proportion was significantly different (p < .0001), providingadditional support for Hypothesis 2B.Based on the results of the t-test and the binomial test, Hypothesis 2B is supported.Hypothesis 2C. The statement of this hypothesis in theoretical terms is:If the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like), then as experience with the situation accumulates,behavior change induced by a manipulation that both induces positiveattitudes toward the desirable behavior and reinforces those attitudesthrough limited practice of the desirable behavior will decay.Hypothesis 2C was evaluated by comparing, for participants in the Long Behavior group, the numberof times both sides of the paper was used in the first ten lists after the videotape with the number oftimes both sides of the paper was used in the final ten (of 30) lists after the videotape. As Table 10-13 shows, contrary to Hypothesis 2C, use of both sides of the paper actually increased slightly,although the increase was not statistically insignificant (t 1.00, p = .327).Table 10-13Comparison of the Average Number of Times Participants in the Long Behavior Group UsedBoth Sides of the Paper for Lists Following the VideotapePost-manipulation Lists Mean SD nFirst ten 0.72 1.72 25Final ten 0.80 1.87 25Hypothesis 2C is therefore rejected.180Hypothesis 2D. The statement of this hypothesis in theoretical terms is:If the undesirable behavior has been extensively repeated (i.e., hasbecome habit-like) then as experience with the situation accumulates,behavior change induced by a manipulation that both induces positiveattitudes toward the desirable behavior and reinforces those positiveattitudes through Limited practice of the desirable behavior will begreater than behavior change (if any) following a manipulation thatproduces a positive attitude toward the desirable behavior.Hypothesis 2D was evaluated by comparing the number of times participants in the Long Persuasionand Long Behavior groups used both sides of the paper in the last ten lists (of 30) following thevideotape. As Table 10-14 shows, contrary to the prediction in Hypothesis 2D, participants in theLong Persuasion group used both sides of the paper more frequently than did participants in the LongBehavior group, although the difference is not statistically significant.Table 10-14Comparison of the Average Number of Times Participants in the Long Persuasion and LongBehavior Groups Used Both Sides of the Paper for the Final Ten Lists Following the VideotapeGroup Mean SD n t-value p (2-tail)Long Persuasion 0.87 1.94 23 0.13 .900Long Behavior 0.80 1.87 25Hypothesis 2D is therefore rejected.Hypotheses Concerning the Relative Effectiveness of the Manipulations Across Different Levelsof Premanipulation Task ExperienceThe third set of hypotheses concerns the relative effectiveness of persuasion alone versuspersuasion with behavioral reinforcement depending on whether the undesirable behavior had beenextensively practised.181Hypothesis 3A. The statement of this hypothesis in theoretical terms is:A manipulation that merely induces (or activates) (more) positiveattitudes toward the desirable behavior will be less effective if theundesirable behavior has been extensively repeated (i.e., has becomehabit-like) than if the undesirable behavior has not been extensivelyrepeated.Hypothesis 3A was tested by comparing the number of times participants in the Short Persuasion andLong Persuasion groups used both sides of the paper in the ten lists immediately following thevideotape. As Table 10-15 shows, contrary to Hypothesis 3A, the persuasive manipulation wasequally effective when the undesirable behavior had been extensively repeated and when it had notbeen extensively repeated.Table 10-15Comparison of the Number of Times Participants in the Short Persuasion and Long PersuasionGroups Used Both Sides of the Paper in the Lists Immediately Following the VideotapeGroup Mean SD n t-value p (2-tail)Short Persuasion 0.87 1.94 23 0.00 1.00Long Persuasion 0.87 1.94 23Hypothesis 3A is therefore rejected.Hypothesis 3B. The statement of this hypothesis in theoretical terms is:A manipulation that both induces (or activates) (more) positiveattitudes toward the desirable behavior and reinforces that behaviorwith limited practice will be equally effective whether or not theundesirable behavior has been extensively repeated (i.e., has becomehabit-like).Hypothesis 3B was tested by comparing the number of times participants in the Short Behavior andLong Behavior groups used both sides of the paper in the ten lists immediately following thevideotape. As Table 10-16 shows, contrary to Hypothesis 3B, the behavioral manipulation was moreeffective when the undesirable behavior had not been extensively repeated than when it had beenextensively repeated.182Table 10-16Comparison of the Number of Times Participants in the Short Behavior and Long BehaviorGroups Used Both Sides of the Paper in the Lists Immediately Following the VideotapeGroup Mean SD n t-value p (2-tail)Short Behavior 2.92 2.52 24 3.55 .001Long Behavior 0.72 1.72 25Hypothesis 3B is therefore rejected.Summary of the Experimental ResultsThere were several unexpected and interesting findings from this experiment. First, whenthe undesirable behavior had not been extensively practised (i.e., had not become habit-like), amanipulation that both induced positive attitudes toward the desirable behavior and reinforced thosepositive attitudes through limited practice of the desirable behavior was more effective in inducingthe desirable behavior than was a manipulation that merely induced positive attitudes. In otherwords, in the volitional group, the persuasion + behavior manipulation was more effective than thepersuasive manipulation. Furthermore, based on analyses associated with the manipulation check,the difference in effectiveness cannot be accounted for by differences in attitude toward the desirablebehavior, as there is no significant difference between the groups on that variable.The second unexpected and interesting finding was that when the undesirable behavior hadbeen extensively practised (i.e., had become habit-like), there was no advantage to the behavioralreinforcement. In other words, in the “habit-likes group, the behavioral manipulation was no moreeffective than was the persuasive manipulation. In fact, reinforcing the persuasive manipulation withlimited practice of the desirable behavior was slightly less effective than the persuasive manipulationalone, although the difference was not statistically significant.The third unexpected and interesting finding was that merely inducing a positive attitudetoward the desirable behavior was equally effective whether the undesirable behavior had beenextensively practised or not; however, the same was not true when the positive attitudes were183reinforced with limited practice. In the latter situation, the behavioral manipulation was significantlymore effective in changing the behavior of participants who had had less premanipulation experiencewith the undesirable behavior. In other words, the persuasive manipulation worked equally well inboth the “habit-like” and “volitional” groups, but the behavioral manipulation worked better in the“non-habit” group than it did in the “habit-like” group.The fourth unexpected and interesting finding is that, at the point when participants in theShort Behavior group had done the same total number of lists as the Long Behavior group had doneby the end of the videotape (i.e., 45 lists), they did not, unlike the Long Behavior participants, switchback to using one side of the paper.As in many exploratory studies, this experiment raises many new and interesting questions.The following chapter will discuss some of those questions, and point out additional factors andvariables which will need to be explored in future research.CHAPTER 11STUDY 2: DISCUSSIONThe previous two chapters described the design, analysis, and results of an experimentintended to test some of the theory developed in Chapter 8, which concerns the way that complex,habit-like behaviors change. The theory proposed that such change proceeds in three steps: deautomation of the old habit-like behavior, volitional change of that behavior, and consolidation of thenew behavior. Based on this theory, it was predicted that persuasion-based approaches to behaviorchange would be effective in changing behavior that had not yet been practised enough to becomehabit-like; but for behaviors that had been practised extensively enough to be considered habitual orhabit-like, persuasion alone would not be enough — behavioral reinforcement would also benecessary.To test these predictions, a laboratory experiment was run. Participants were asked to makea series of lists, each on a different page of a four-by-six-inch pad of paper. Some of them did fivelists (little pre-manipulation practice), and some did 30 lists (extensive pre-manipulation practice).At this stage, all the participants used only one side of the paper. Then the manipulations weredelivered. Some participants received a persuasive message suggesting that they use both sides ofthe paper, and some received the same persuasive message reinforced with limited practice of thedesired behavior. The dependent variable was the number of times participants used both sides of184185the paper in lists made after the manipulation was delivered. As Chapter 10 indicated, results of theexperiment were mixed, with several unexpected and interesting findings.This chapter discusses theoretical and methodological issues arising from Study 2. It beginswith a brief discussion of the experimental procedure itself. The organization of the remainder ofthe chapter parallels that of Chapter 10, examining first the results obtained in the Little PreManipulation Practice groups, then those for the Extensive Pre-Manipulation Practice groups, andfinally those comparing the effectiveness of the manipulations across the two levels of premanipulation practice.The Experimental ProcedureAs the review of the demarketing literature which appears in Chapter 2 indicates, there hasbeen little empirical research done in this area. Furthermore, there is almost no experimentalresearch. This gap may be due to the difficulty of designing an experimental procedure that capturesthe critical aspects of demarketing problems: people have become accustomed to using the resourcewastefully; the resource is merely a means to an end; changing behavior to use less of the resourceis costly (though not necessarily in financial terms); conserving the resource has uncertain returns thatwill accrue in the future, probably to people other than those making the effort; and there is noobvious reason (such as a shortage of the resource) to conserve. (See Chapter 9 for a more completediscussion of these criteria.)The procedure used in this experiment meets these criteria and therefore opens the door tofurther experimental research in demarketing. There were limitations, as the discussion belowindicates, but they were of an operational nature. The basic procedure seems to have been successfulin simulating the key characteristics of demarketing problems.186Little Pre-Manipulation Practice GroupsParticipants in the Little Pre-Manipulation groups completed only five lists before receivingthe manipulation. Such minimal practice is very unlikely to lead to habitual or habit-like use of oneside of the paper for the list-making task. It was expected that, since the use of one side of the paperfor this task was not habitual, a persuasive message alone should be enough to convince participantsto change their behavior. A persuasive message reinforced with limited practice of the desiredbehavior (using both sides of the paper) was also expected to be effective, but no more effective thanthe persuasive message alone. In other words, when the behavior is not habitual or habit-like, apersuasive message alone should lead to behavior change; behavioral reinforcement would beredundant.As expected, the results showed that the persuasive message alone did lead to behaviorchange. Also as expected, a persuasive message reinforced with limited behavioral practice waseffective. However, contrary to expectations, persuasion with behavioral reinforcement was muchmore effective than persuasion alone: 56% of participants who received a persuasive message andhad an opportunity to practice using both sides of the paper switched, while only 17% of those whoreceived the persuasive message alone switched. In this experiment, it appears that behavioralreinforcement was anything but redundant. Another unexpected result was that there was no decayin the conservation behavior of either the participants exposed to the persuasive message alone orthose exposed to both a persuasive message and behavioral reinforcement. The unexpected resultsraise both methodological and theoretical issues.Methodological Issues. On the methodological side, one explanation for the strong effectof behavioral reinforcement is that, because the number of lists done in the behavioral reinforcementportion of the experiment was three times the number done in the pre-manipulation practice portion(15 lists for behavioral reinforcement versus five lists for pre-manipulation practice), the behavioralreinforcement may have swamped the effect of the pre-manipulation practice. It could be argued that187participants with little pre-manipulation practice in the Persuasion + Behavior group got three timesas much practice using two sides of the paper as using one side. It is possible that for a task assimple as the one in the experiment a level of practice between five and fifteen lists might be enoughto form a habit-like tendency to use two sides of the paper.The counterargument is that participants in the Persuasion + Behavior group did not reallyhave 15 opportunities to use both sides of the paper; they only had seven. Only on the evennumbered lists could a participant use the second side of the paper. Consequently, although therewas an imbalance between the pre-manipulation practice and the behavioral reinforcement, it was notas large as it might initially seem. Nevertheless, without further research the possibility that thatdifference may have been large enough to cause the unexpected result cannot be eliminated.Further research using this experimental procedure should balance the number of premanipulation lists (which, based on the experience in this experiment, people invariably do one-sided)with the number of two-sided lists done in the manipulation. For instance, if five lists are done inthe pre-manipulation period, then ten lists should be done for behavioral reinforcement (providingfive opportunities to practice using the second side of the paper).A second possible methodological explanation for the unexpected importance of behavioralreinforcement in the Little Pre-manipulation Practice group concerns the strength of the persuasivemessage, particularly relative to the strength of the behavioral reinforcement. The persuasivemessage delivered in the videotape was:We are finding that a lot of paper is being wasted in this study, sincewe have to have each list on a separate page and most people useonly one side of the paper. If you could use ffi sides of the paper,that would waste less paper and save a few trees. Whether you useone side or two makes no difference to the study, but it would begood for the environment.188If persuasive messages can be thought of as being on a continuum from “soft sell” to “hard sell,” thisone would clearly be closer to the soft sell end18. A soft sell message was preferable to a hard sellmessage for two reasons. First, a hard sell message might be perceived by participants as aninstruction. Second, if the message were too attention-getting, it might raise suspicions inparticipants’ minds concerning the true purpose of the study. If either of these eventualities occurred,changes in behavior unrelated to participant attitudes toward conservation would likely occur.Nevertheless, even given that a soft sell message was desirable, the particular message usedin this experiment cannot be called persuasive in the sense of making any real attempt to changeattitudes toward using both sides of the paper. Rather, its purpose was to activate existing attitudes.In contrast, the behavioral reinforcement was directed, focused, and relatively hard sell.However, it cannot be said that it was so hard sell as to be mistaken for instructions or to raisesuspicions, since only 37% of all participants exposed to it (i.e., participants in the Persuasion +Behavior condition for both the Little and Extensive Pre-Manipulation Practice groups) changed theirbehavior.Future research should seek a better balance between the strength of the persuasive messageand the strength of the behavioral reinforcement. One way to strengthen the persuasive messagewithout drawing too much attention to it or having it be perceived as instructions might be to alterthe experimental task and cover story slightly. The study could be presented as an investigation ofthe memorability of persuasive messages. The same network model of memory could be explainedand the basic procedure would stay the same, except that participants would view a series of ads‘8Although closer to the soft sell end of the persuasive continuum than to the hard sell end, themessage used in the experiment was not the softest sell; an even milder version was pretested. Ina category called “Things you can do to help the environment” the following explanation was given:This is a big category. There are so many things we can do. In thiscategory I would list things like recycling soft drink cans, recyclingnewspapers, recycling glass, . . . , and even using two sides of the paper inthis experiment.”It changed no one’s behavior in the pretest.189(instead of making a series of lists), after each of which (or after having seen, say, a batch of five),they would write down on a piece of paper all that they remember from the ad. The ads before themanipulations would have no connections with each other; the ads after would. The persuasivemessage to use both sides of the paper would be delivered during the explanation of the networkmodel, just as it was in the present experiment, but it could be delivered in a much more openlypersuasive manner, as it would be just another one of many persuasive messages to which participantswere being exposed.The explanation for the lack of decay among participants who switched to using two sides ofthe paper may also be due to a methodological flaw: the number of post-manipulation lists may havebeen too few, and too close in time to the manipulations. An option for future research is to do 30lists after the manipulations have been delivered, as was done in this experiment, then add anotherdistractor videotape, and then do another series of lists.Theoretical Issue. The unexpected findings also raise a theoretical issue. Why wouldPersuasion + Behavior be so much more effective than Persuasion alone when the wasteful behaviorhad been practised so little that it is unlikely to have become habitual or habit-like? One possibleexplanation relates to the role of involvement.Zaichkowsky (1985) defines involvement as “a person’s perceived relevance of the objectbased on inherent needs, values, and interests” (p. 342). People whose involvement with an objectis high tend to process more information about it than do people whose involvement with an objectis low. Petty, Cacioppo, and Schumann (1983) differentiate between “central” and “peripheral”routes to attitude change. The former involves relatively deep processing of relevant information,while the latter involves relatively shallow processing:[TJhe central route views attitude change as resulting from a person’sdiligent consideration of information that s/he feels is central to thetrue merits of a particular attitudinal position. . . Attitude changesinduced via the central route are postulated to be relatively enduringand predictive of behavior.190Attitude changes that occur via the peripheral route do notoccur because an individual has personally considered the pros andcons of the issue, but because the attitude issue or object isassociated with positive or negative cues — or because the personmakes a simple inference about the merit of the advocated positionbased on various simple cues in the persuasion context. . . Attitudechanges induced under the peripheral route are postulated to berelatively temporary and unpredictive of behavior (Petty, Cacioppo,and Schumann 1983: 135-136).In this experiment, where the paper is merely a means to an end, and there is plenty of extrapaper in full view (so participants anticipate no impending shortage), involvement with the resourcewould tend to be low. This is consistent with what would normally occur in a typical clemarketingsituation: the resource is taken for granted.Given that involvement with the resource is low, the peripheral route would be favored.Consequently, in the Little Pre-Manipulation Practice groups, the persuasive message alone wouldmerely act as a cue to attitudes toward conservation. Most participants would continue to use oneside of the paper because they were not involved enough to bother assessing the pros and cons ofchanging their behavior. However, when the persuasive message was reinforced with limitedbehavioral practice, participants gained experience with both behavioral options: their initialexperience was using one side of the paper, while their later experience was using both sides of thepaper. Once they tried both options (i.e., weighed the pros and cons), more than half switched tothe desired behavior.This raises an interesting question. Compelling participants to try the two-sided optionessentially forced them to consider (or at least experience) the pros and cons of switching. In otherwords, they were emulating the deeper, central route to persuasion. The question is: did thisincrease participants’ level of involvement with the resource, or could the emulation of the deeper,central route to persuasion constitute some sort of increased pseudo-involvement? Unfortunately,involvement was not measured at any point in this experiment, so these data cannot answer that191question. There is no way to tell from Study 2 data whether involvement with the resource increasedor stayed the same when participants used both sides of the paper.The data from this experiment suggest that, in situations where people have not developedhabits or habit-like behaviors, demarketing programs that can get people to act as fthey had a higherlevel of involvement (it is unknown at this point whether their real level of involvement actuallyincreases) will be more effective than demarketing programs that do not increase people’s level ofinvolvement. Clearly, more theory concerning the role of involvement and, if it exists, pseudo-involvement, needs to be developed. That is a task for future research.Extensive Pre-Manipulation Practice GroupsParticipants in the Extensive Pre-Manipulation groups completed 30 lists before receiving themanipulation. Based on pretests, that level of practice was assumed to lead to habitual or habit-likeuse of one side of the paper for the list-making task. It was expected that, since the use of one sideof the paper for this task was habitual or habit-like, a persuasive message alone should not be enoughto convince participants to change their behavior. A persuasive message reinforced with limitedpractice of the desired behavior (using both sides of the paper) was expected to be effective. In otherwords, when the behavior is habitual or habit-like, a persuasive message alone should not lead tobehavior change; behavioral reinforcement would also be necessary.As expected, a persuasive message reinforced with limited behavioral practice was effective.But, contrary to expectations, the persuasive message alone did lead to behavior change. In addition,also contrary to expectations, persuasion with behavioral reinforcement was no more effective thanpersuasion alone: 22% of participants who received a persuasive message alone switched, while 17%of those who received the persuasive message reinforced by limited practice of the desired behaviorswitched (the difference was not statistically significant). In this experiment, it appears thatbehavioral reinforcement added nothing. Another unexpected result was that there was no decay in192the conservation behavior of either the participants exposed to the persuasive message alone or thoseexposed to both a persuasive message and behavioral reinforcement. The unexpected results raiseboth methodological and theoretical issues.Methodological Issues. On the methodological side, the explanation for the lack of decayamong participants who switched to using two sides of the paper is the same as it was for the LittlePre-manipulation Practice groups: the number of post-manipulation lists may have been too few, andtoo close in time to the manipulations. Again, an option for future research is to do 30 lists after themanipulations have been delivered, as was done in this experiment, then add another distractorvideotape, and then do another series of lists.The more serious question concerns why the behavioral reinforcement added nothing. Onepossibility is that the behavioral reinforcement simply was not strong enough. Participants in theExtensive Pre-Manipulation Practice groups had done 30 lists using one side of the paper. Those inthe Persuasion + Behavior group made 15 lists during the behavioral reinforcement portion of theexperiment. This represents only seven opportunities to use the second side of the paper. It may bethat the amount of behavioral reinforcement needed to overcome a habitual or habit-like behaviorwould be in the same range as the original amount of pre-manipulation practice. The results for theLittle Pre-Manipulation groups are consistent with this suggestion. Efforts should be made in futureresearch to determine what relative levels of behavioral reinforcement are necessary to overcome theeffects of existing practice.Theoretical Issue. The unexpected findings may also have a more theoretical explanation.It may be that, as people who habitually waste a resource gain experience with using less of it, theybecome more aware of the costs (e.g., inconvenience) associated with changing their behavior. Inother words, the behavioral reinforcement actually reminds them of all the reasons they do not wantto switch from using one side of the paper to using both sides (and in this experiment, thedisadvantages of using both sides of the paper would have been very evident to participants, because193they were under time pressure, and using both sides of the paper slowed them down). This wouldbe less likely to occur among people whose behavior is not habitual or habit-like simply because theirsystem for doing the task is (by definition) more flexible, so it is easier for them to accommodatechanges.In Chapter 8, it was proposed that changing habits or habit-like behaviors proceeds in threesteps: de-automation, volitional behavior change, and consolidation. It was pointed out there thatde-automation requires, first, that people know precisely what behavior(s) to change, second, thatthey can reliably anticipate the performance of the behavior, and third, that they have enoughmotivation to invest the cognitive energy necessary to bring the habit under conscious control. Atthis preliminary stage in testing the theory, choices were made concerning which variables tomanipulate, and which ones not to manipulate. The criterion used was the likely importance of thevariable in explaining and predicting behavior. In the absence of previous empirical tests of thetheory, the decision was made not to manipulate motivation. Based on the results in the ExtensivePre-Manipulation Practice groups, it would seem worthwhile for future research to investigate moreclosely the role of motivation.Relative Effectiveness of the Manipulations Across Pre-Manipulation LevelsThere were also unexpected results when the manipulations (Persuasion alone or Persuasion+ Behavior) were compared across different levels of pre-manipulation practice. Specifically, it wasexpected that Persuasion alone would be more effective for people who had had Little PreManipulation Practice of the undesirable behavior than for people who had had extensive PreManipulation Practice, because persuasion alone should not be powerful enough to overcome habitualor habit-like tendencies. In this experiment, however, Persuasion was equally effective for both Littleand Extensive Pre-Manipulation Practice. It was also expected that Persuasion + Behavior wouldbe equally effective for both levels of Pre-Manipulation Practice. Unexpectedly, it was more194effective for the Little Pre-Manipulation Practice group than for the Extensive Pre-Manipulationgroup.The first unexpected result, that Persuasion alone was equally effective for both Little andExtensive Pre-Manipulation Practice, might be explained as a methodological weakness. If, as seemslikely, the persuasive message used in the manipulation did not change attitudes, but merely activateexisting attitudes, then the only people to switch would be those who had existing favorable attitudestoward conserving paper by using both sides. The fact that the same percentage of each groupswitched tends to support this explanation: if attitudes toward using both sides of the paper arerandomly distributed through the population, then a random assignment of participants should yieldthe same proportions of participants with existing positive attitudes toward using both sides of thepaper. As noted earlier, attention should be paid in future research to ensuring that the persuasivemessage is actually persuasive, not just something that activates existing favorable attitudes, if any.The second unexpected result, that Persuasion + Behavior was more effective for the LittlePre-Manipulation Practice group than for the Extensive Pre-Manipulation group, is also consistentwith earlier explanations. Specifically, in the Little Pre-Manipulation Practice group, the behavioralreinforcement may have caused participants to increase involvement, or pseudo-involvement, withthe decision to use one or both sides of the paper, which led a sizeable number of them to changetheir behavior. In contrast, in the Extensive Pre-Manipulation Practice group, the behavioralreinforcement may have made salient the costs (in terms of inconvenience, loss of speed, etc.),thereby actually discouraging switching — once again highlighting the importance of motivation inthe de-automation process.SummaryChapter 11 has discussed some of the theoretical and methodological issues arising fromStudy 2. The chapter began with an evaluation of the suitability of the experimental procedure as195a method for studying typical demarketing problems. The conclusion was that, while there weresome implementation problems in Study 2, they are surmountable. The experimental procedure wasjudged to have been successful in capturing the essential characteristics of demarketing problems.This chapter also outlined some of the methodological problems encountered in Study 2, andmade suggestions for change. One recommendation was that future research using this experimentalprocedure should balance the number of pre-manipulation lists (which, based on the experience in thisexperiment, people invariably do one-sided) with the number of two-sided lists done in themanipulation.A second recommendation concerning methodology was that future research should seek abetter balance between the strength of the persuasive message and the strength of the behavioralreinforcement. One way to strengthen the persuasive message without drawing too much attentionto it or having it be perceived as instructions might be to alter the experimental task and cover storyslightly such that participants would view a series of persuasive messages, so that the persuasivemessage of interest could be delivered in a much more openly persuasive way, without drawing undueattention.A third recommendation was that, in order to provide an opportunity for decay to occur,another distractor videotape (or other distracting activity) should be added, which would be followedby another series of lists.A fourth recommendation of a methodological nature was further testing be done to determinewhat relative levels of behavioral reinforcement are necessary to overcome the effects of existingpractice. In Study 2, the amount of behavioral reinforcement may have been inadequate.Also discussed in this chapter were two issues of a more conceptual or theoretical nature.First, the data from Study 2 suggest that, in situations where people have not developed habits orhabit-like behaviors, demarketing programs that can get people to act as if they had a higher levelof involvement will be more effective than demarketing programs that do not. It is not clear that196engaging in two-sided list-making actually increased involvement; perhaps some “pseudo-involvement” (in which people act as if they are involved, but notice no increase in personalrelevance) is acting. Investigating the role of involvement and/or pseudo-involvement is a task forfuture research.The second theoretical issue concerns the role of motivation. Study 2 is a preliminaryempirical exploration of some of the theory developed in Chapter 8 to explain how habitual or habit-like behavior changes. Although motivation plays a prominent role in the theory, this experimentdid not attempt to manipulate it. Future research should.In summary, then, Study 2 shows that habit-like behavior does not respond to change attemptsin the same way that non-habit-like behavior does, which is one of the key points of this dissertation.It also raises some new and interesting questions. While some of the findings from this experimentdo not coincide with expectations, there are explanations for each unexpected finding that are notinconsistent with the theory presented earlier in this dissertation. It would seem worthwhile to refineboth the method and the theory, and to pursue further this research area.CHAPTER 12CONCLUSIONSThis chapter summarizes the dissertation research, discusses its implications and contributions,as well as its limitations, and outlines future research directions.Summary of the Dissertation ResearchThis research has explored, from a marketing point of view, some issues concerning thequestion of how to reduce discretionary consumption. This is an issue that has received relativelylittle research attention, probably due to the fact that countries of the developed world, where mostresearch has been conducted, have generally had economies in a state of excess supply. However,as the United Nations’ World Commission on Environment and Development pointed out, globalpressure on resources is becoming acute. The Commission recommended that a priority be put onsustainable development, which would require reductions in discretionary consumption of resources,particularly in the developed world, where per capita consumption is many times higher than it is inthe developing world.A review of the marketing literature revealed that, while the usual emphasis has been onissues related to increasing demand, some research has been done on decreasing demand. The termdemarketing was coined in the early 1970s, in recognition of the fact that marketing is about more197131The convergence between the Revolving Door Model and the habit reversal procedure canbe taken as an indicat