UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Trait concept formation and change as a function of the validity of behavioral information Farthing, Gerald Robert 1975

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

Item Metadata

Download

Media
831-UBC_1975_A1 F37_7.pdf [ 9.21MB ]
Metadata
JSON: 831-1.0100054.json
JSON-LD: 831-1.0100054-ld.json
RDF/XML (Pretty): 831-1.0100054-rdf.xml
RDF/JSON: 831-1.0100054-rdf.json
Turtle: 831-1.0100054-turtle.txt
N-Triples: 831-1.0100054-rdf-ntriples.txt
Original Record: 831-1.0100054-source.json
Full Text
831-1.0100054-fulltext.txt
Citation
831-1.0100054.ris

Full Text

TRAIT CONCEPT FORMATION AND CHANGE AS A FUNCTION OF THE VALIDITY OF BEHAVIORAL INFORMATION by GERALD ROBERT FARTHING B.A. ( D i s t . ) , U n i v e r s i t y of Saskatchewan, 1969 High Honours, University of Saskatchewan, 1970 M.A., Univ e r s i t y of Saskatchewan, 1971 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY i n the Department of Psychology We accept t h i s thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May, 1975 In p r e s e n t i n g t h i s t h e s i s in p a r t i a l f u l f i l m e n t o f the r e q u i r e m e n t s f o r an advanced degree at the U n i v e r s i t y o f B r i t i s h Co lumb ia , I a g ree tha t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s tudy . I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y purposes may be g r a n t e d by the Head o f my Department o r by h i s r e p r e s e n t a t i v e s . It i s u n d e r s t o o d tha t c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d w i thou t my w r i t t e n p e r m i s s i o n . Department o f Psychology  The U n i v e r s i t y o f B r i t i s h Co lumbia Vancouver 8, Canada Date July 24, 1975 6 i Abstract A search of the l i t e r a t u r e revealed that there are very few studies which have investigated two to p i c s : (a) t r a i t concept formation and change; and (b) t r a i t a t t r i b u t i o n as a function of the amount and type of behavioral information which i s used as s t i m u l i . The purpose of the present i n v e s t i g a t i o n i s to study some parameters of these processes. F i r s t , two aspects of t r a i t concept formation are studied: (a) the e f f e c t of d i f f e r e n t i a l l y - v a l i d behavioral information on subjects' a b i l i t y to form t r a i t concepts; and (b) the resistance to change of t r a i t concepts which are formed by reading either high-or low-valid behavioral information. Second, two aspects of t r a i t concept change are investigated: (a) the influence of d i f f e r e n t i a l l y - v a l i d behavioral information on t r a i t concept adjustment; and (b) the influence of two w i t h i n - t r a i t concept manipulations on t r a i t concept adjustment. One manipulation was the r e v e r s a l of one t r a i t dimension and the other was the replacement of a second t r a i t dimension with an e n t i r e l y new t r a i t dimension. Third, two aspects of t r a i t a t t r i b u t i o n are studied: (a) the e f f e c t of the v a l i d i t y of the behavioral information which was presented i n the t r a i t concept formation and the t r a i t concept adjustment parts of the study; and (b) the e f f e c t of the amount and the consistency of the behavioral information. Each of 96 introductory psychology students formed one of several conjunctive t r a i t concepts by reading either high- or low-valid behavioral information. Then subjects adjusted t h e i r t r a i t concept'to accommodate new and contradictory behavioral information of either high-.or l o w - v a l i d i t y . i i F i n a l l y , a f t e r they had s u c c e s s f u l l y adjusted t h e i r concepts, subjects gave t h e i r o v e r a l l t r a i t impressions. The behavioral information was obtained i n three stages. In the f i r s t stage, 185 introductory psychology students generated one behavioral statement for each of eight t r a i t terms which were used as s t i m u l i . In the second stage, two major ratings were obtained from independent samples of introductory psychology students: (a) the p r o b a b i l i t y of the t r a i t given that the behavior was performed (113 introductory psychology students gave t h i s r a t i n g ) ; and (b) the p r o b a b i l i t y of the t r a i t given that the behavior was not performed (63 introductory psychology students gave this r a t i n g ) . For each behavioral item, a v a l i d i t y index (Fishbein and A j z e n , 1974) was obtained by subtracting the second r a t i n g from the f i r s t one. On the basis of the v a l i d i t y i n d i c e s , two pools of behavioral statements (the high- or low-valid pools) were obtained and were used i n the experimental part of the study. The p r i n c i p a l r e s u l t s from analyses of variance are as follows: (a) t r a i t concept formation did not vary s i g n i f i c a n t l y as a function of d i f f e r e n t i a l l y v a l i d behavioral information; (b) t r a i t concepts which were formed by reading h i g h - v a l i d behavioral information were found to be more r e s i s t a n t to change than are those which were formed by reading low-valid behavioral information; (c) t r a i t concept adjustment varied as a function of d i f f e r e n t i a l l y - v a l i d behavioral information but not as expected; (d) the reversed component of the t r a i t concept was harder to adjust than was either the unchanged or the new component; (e) t r a i t a t t r i b u t i o n on the reversed component of the adjusted t r a i t concept did not vary s i g n i f i c a n t l y as a function of the v a l i d i t y of the behavioral information; and (f) t r a i t a t t r i b u t i o n r e f l e c t e d i i i the amount and the consistency of the behavioral information that subjects received throughout the e n t i r e study. The g e n e r a l i z a b i l i t y of some of these findings i s l i m i t e d because the contextual arrangement of the t r a i t dimensions was found to influence both t r a i t concept formation and change and t r a i t a t t r i b u t i o n . A discussion of the r e s u l t s centered on: (a) the c u r v i l i n e a r hypothesis of Jones and Goethals ( 1972); (b) some content- and context-related e f f e c t s on impression formation; (c) some aspects of t r a i t a t t r i b u t i o n and i m p l i c i t p ersonality theory; (d) the summary view of t r a i t s (Wiggins, 1974); and (e) the Fishbein and Azjen (1974) approach to behavioral v a l i d i t y . i v Table of Contents Page Chapter 1 The Problem 1 Models of t r a i t concept usage by laymen . . . . . . . 4 Goals of the present i n v e s t i g a t i o n 7 D e f i n i t i o n s .' 9 Related l i t e r a t u r e and expectations 11 1. The f i r s t goal 12 2. The second goal 14 3. The t h i r d goal 26 The t r a i t dimensions 29 Method of obtaining the v a l i d i t y of the behavioral information 32 Hypotheses 36 Chapter 2 Method 40 Phase 1 .41 Stage 1: Generation of behavioral statements., . 41 Stage 2: The r a t i n g tasks 46 Stage 3: The elimination of behavioral statements 54 Phase 2: The experiment 56 Method 65 Dependent measures 74 Chpater 3 Results 76 Learning data: A c q u i s i t i o n 76 Behavior set 1: Design 1 76 Behavior set 1: Design 2 82 Behavior set 2: Design 1 91 Behavior set 2: Design 2 . . . . . . . . . . . 92 Learning data: Reversal 99 Behavior set 3: Design 3 100 Behavior set 3: Design 4 106 Behavior set 4: Design 3 . 123 Behavior set 4: Design 4 126 •Attribution;.data 137.; .... • ..J: : -j. Chapter 4 Discussion 160 Prospect • • • References 192 Appendix A • 199 v i L i s t of Tables Page Table 1 Norman's personality factors with abbreviated descriptions of t h e i r r a t i n g scales 31 Table 2 "Factor score" i n t e r c o r r e l a t i o n s based on the peer nomination r a t i n g scale data from two samples 33 Table 3 C l a s s i f i c a t i o n of responses to the behavior generation task 47 Table 4 Comparisons of the v a l i d i t y indices f o r the items i n the f i n a l item pool 57 Table 5 Combinations of t r a i t dimensions f o r each p r o f i l e 60 Table 6 Composition of the characters 62 Table 7 The dimensions i n each condition f o r each character 64 Table 8 Analysis of variance: Design 1 77 Table 9 Summary of the analysis of variance on t r i a l s to c r i t e r i o n i n behavior set 1 78 Table 10 Mean t r i a l s to c r i t e r i o n f o r the characters i n behavior set 1 80 Table 11 Analysis of variance: Design 2 83 Table 12 Dimensions treated as nested within p r o f i l e s .. 85 Table 13 Summary of the analysis of variance on errors to c r i t e r i o n i n behavior set 1 86 Table 14 Mean errors to c r i t e r i o n f o r the characters i n behavior set 1 87 v i i Table 15 Mean errors to c r i t e r i o n f o r the dimensions i n behavior set 1 89 Table 16 Summary of the analysis of variance on t r i a l s to c r i t e r i o n i n behavior set 2 . — 93 Table 17 Mean t r i a l s to c r i t e r i o n f o r the characters i n behavior set 2 94 Table 18 Summary of the analysis of variance on errors to c r i t e r i o n i n behavior set 2 95 Table 19 Mean errors to c r i t e r i o n f o r the characters i n behavior set 2 96 Table 20 Analysis of variance: Design 3 • ••• 101 Table 21 Summary of the analysis of variance on t r i a l s to c r i t e r i o n i n behavior set 3 102 Table 22 Mean t r i a l s to c r i t e r i o n f o r the characters i n behavior set 3 104 Table 23 Analysis of variance: Design 4 107 Table 24 Summary of the analysis of variance on errors within the f i r s t t r i a l i n behavior set 3 109 Table 25 Mean errors to c r i t e r i o n i n the f i r s t t r i a l f o r characters i n behavior set 3 112 Table 26 Summary of the analysis of variance on errors within f i x e d t r i a l s (5) i n behavior set 3 119 Table 27 Mean errors within a fi x e d number of t r i a l s f o r the characters i n behavior set 3 120 Table 28 Summary of the analysis of variance on t r i a l s to c r i t e r i o n i n behavior set 4 125 v i i i Table 29 Mean t r i a l s to c r i t e r i o n f o r the characters i n behavior set 4 127 Table 30 Summary of the analysis of variance on errors to c r i t e r i o n i n behavior set 4 128 Table 31 Summary of the analysis of the l a t i n square r e s i d u a l f o r each p r o f i l e 136 Table 32 Analysis of variance: Design 5 139 Table 33 Summary of the analysis of variance on a t t r i b u t i o n data 140 Table 34 Mean scale score of each character 144 Table 35 Summary of the analysis of the dimension v a r i a b l e f o r each p r o f i l e 153 i x L i s t of Figures Page Figure 1 Mean number of errors i n Behavior Set 1 as a function of the characters (CH) and the dimensions 90 Figure 2 Mean number of errors i n Behavior Set 2 as a function of the characters (CH) and the dimensions 99 Figure 3 Mean number of t r i a l s i n Behavior Set 3 as a function of the p r o f i l e s and r e v e r s a l groups.. 105 Figure 4 Mean number of errors i n the f i r s t r e v e r s a l t r i a l of Behavior Set 3 as a function of the a c q u i s i t i o n and the rever s a l groups 110 Figure 5 Mean number of errors i n Behavior Set 3 as a function of the conditions arid the rev e r s a l groups 113 Figure 6 Mean number of errors i n Behavior Set 3 as a function of the conditions, the p r o f i l e groups and the r e v e r s a l groups 115 Figure 7 Mean number of errors within f i x e d t r i a l s i n Behavior Set 3 as a function of the characters and the re v e r s a l groups 122 Figure 8 Mean number of errors i n Behavior Set 4 as a function of the conditions and the a c q u i s i t i o n X Figure 9 Mean number of errors i n Behavior Set 4 as a function of the conditions, the a c q u i s i t i o n groups and the re v e r s a l groups , . 132 Figure 10 Mean number of errors i n Behavior Set 4 as a function of the conditions and the characters. 134 Figure 11 Mean scale score as a function of the conditions and the poles 146 Figure 12 Mean scale score as a function of the conditions and the a c q u i s i t i o n groups 147 Figure 13 Mean scale score as a function of the conditions and the p r o f i l e groups 149 Figure 14 Mean scale score as a function of the conditions and the characters 151 Figure 15 Mean scale score as a function of the poles and the characters 155 Figure 16 Mean scale score as a function of the poles, the conditions and the re v e r s a l groups 156 x i Acknowledgments Many people have made various contributions to the completion of t h i s d i s s e r t a t i o n . Dr. Jerry S. Wiggins, my p r i n c i p l e adviser, has broadened my store of knowledge, has guided the formulation of my t h e o r e t i c a l and research ideas and has offered numerous suggestions to perfect my research as I passed through each stage of i t . I thank him for h i s i n s i g h t f u l guidance. Dr. Jim Johnson has provided invaluable input into the experimental design and data a n a l y s i s . I thank him for passing on to me h i s enthusiasm for experimental design. Dr. Demetrios Papagoergis has made contributions to the planning and execution of my research ideas. I thank him for h i s suggestions concerning the type of experimental co n t r o l . F i n a l l y , Dr. Sue Butt-Finn has offered several suggestions which have aided the communication of my methodology. In a s l i g h t l y d i f f e r e n t capacity, Jim Bjering and Malcolm Greig, who are s t a t i s t i c a l analysts at the Univ e r s i t y of B r i t i s h Columbia Computing Center, have guided the computer analysis of my data. I thank them for t h e i r enthusiasm f o r the new and the complex. Two very able t y p i s t s , Diana Turner and Sandy Gallacher, have prepared the f i n a l copy of the manuscript. I thank them for t h e i r patience and conscientiousness. I would l i k e to give s p e c i a l thanks to my wife, Grace, who has given constant support, who has made many s a c r i f i c e s without complaint and who has performed many c l e r i c a l tasks. She has so ably created the i n t e l l e c t u a l , emotional and material environment without which my d i s s e r t a t i o n would not have been completed. In addition, I would l i k e to thank my ch i l d r e n , J e f f r e y , Daniel and Andrea, who have provided many hours of enjoyment during periods which were spent away from my research. F i n a l l y , I would l i k e to thank my parents, Mr. and Mrs. R.W. Farthing and my wife's parents, Mr. and Mrs. G.P. Mario for t h e i r constant and multifaceted support. I appreciated f i n a n c i a l support from a Canada Council Doctoral Fellowship (1972, 1973, 1974), from a Saskatchewan Government Bursary and from a U n i v e r s i t y of B r i t i s h Columbia Teaching Assistantship. In addition I received f i n a n c i a l support from both my parents and my wife parents. G.R. Farthing June 18, 1975 1 Chapter 1 The Problem A question which i s being r a i s e d with increasing frequency i n the personality and s o c i a l psychological l i t e r a t u r e i s : how does the layman use t r a i t concepts (e.g., Mischel, 1973; Schneider, 1973)? As yet, we know l i t t l e about t h i s subject area, but "the function of t r a i t concepts f o r the layman deserves serious attention and hopefully w i l l inform.us about the psychological uses and abuses of t r a i t categorizations" (Mischel, 1973, p. 263). To a large extent, the impetus for t h i s area of inquir y has come from the controversy which was begun by Mischel (1968, 1971, 1973) and which concerned the u t i l i t y of the global d i s p o s i t i o n a l approach to a psychology of personality. This approach i s based on the fundamental assumption that "personality d i s p o s i t i o n s or t r a i t s are r e l a t i v e l y stable, highly consistent a t t r i b u t e s that exert widely generalized causal e f f e c t s on behavior" (Mischel, 1973, p. 253). Mischel (1968)' presented three basic ideas. F i r s t , he argued that we overemphasize i n d i v i d u a l differences at the t r a i t l e v e l and s l i g h t the impact of s i t u a t i o n a l v a r i a b l e s . In support of h i s claim, he reviewed the evidence f o r the existence of several dimensions of behavior which are usually presumed to be manifestations of a t r a i t (e.g., honesty, dependency, att i t u d e s toward authority, r i g i d i t y or intolerance f o r ambiguity, p e r s u a s a b i l i t y ) . He concluded that there i s l i t t l e evidence that t r a i t s e x i s t anywhere but i n the cognitive structure of observers. 2 Secondly, Mischel (1968) reviewed many studies which attempted to f i n d behavioral g e n e r a l i t y across the settings which are usually presumed to be associated with a given t r a i t . A l l of the t r a i t s he reviewed except the a b i l i t y and a b i l i t y - r e l a t e d t r a i t s were found to have low generality across s i t u a t i o n s . He concluded that there i s very l i t t l e evidence to support the notion of cross-s i t u a t i o n a l s t a b i l i t y of s o c i a l behavior. T h i r d l y , Mischel (1968) examined the attempts to predict behavior from paper and p e n c i l t r a i t measurements. He claimed that a " c o r r e l a t i o n between .20 and .30 i s found p e r s i s t e n t l y when v i r t u a l l y any personality dimension i n f e r r e d from a questionnaire i s r e l a t e d to almost any conceivable external c r i t e r i o n i n v o l v i n g responses sampled i n a d i f f e r e n t medium - that i s , not by another questionnaire" (p. 78). Consequently, he concluded that the glo b a l d i s p o s i t i o n a l approach to personality has very l i t t l e p r e d i c t i v e u t i l i t y . Mischel (1973) equated the fundamental assumption of the global d i s p o s i t i o n a l approach to per s o n a l i t y with what he c a l l e d the psychodynamic approach to consistency. According to the psychodynamic approach, diverse behavioral patterns are thought to serve the same underlying dynamic d i s p o s i t i o n and are, therefore, taken as signs of that d i s p o s i t i o n even though a contradiction may exist at the behavioral l e v e l . Because he found a lack of cross-s i t u a t i o n a l s t a b i l i t y to s o c i a l behavior, Mischel c r i t i c i z e d the 3 pra c t i c e of " i n f e r r i n g d i s p o s i t i o n s from behavioral signs as the basis f o r t r y i n g to explain the phenomenon of pe r s o n a l i t y " (p. 262). From h i s point of view, the behavioral signs are not l i k e l y to be consistent with each other and, as such, are not l i k e l y to be governed by the same d i s p o s i t i o n . The responses to Mischel's basic c r i t i q u e have been varied' and widespread (e.g., A v e r i l l , 1973; Bowers, 1973; Gormly & Edelberg, 1974; Wachtel, 1973). Most of these r e p l i e s have e i t h e r disputed Mischel's evidence, pointed out inadequacies i n h i s reasoning or presented data which demonstrated a consistent r e l a t i o n s h i p between t r a i t s and behavior. These arguments have broadened the issues and have stimulated new ideas. A review of the controversy concerning the t r a i t approach to personality i s not c e n t r a l to t h i s i n v e s t i g a t i o n . However, a few new ideas which have c l a r i f i e d some of the issues are pertinent to research oh t r a i t concept usage by laymen. Wiggins (1974) attempted to c l a r i f y some of the issues. He argued that the controversy i s based on an unfortunate confusion of one p a r t i c u l a r theory of t r a i t s with the whole of the psychometric viewpoint. One r e s u l t of t h i s confusion "has been that the psycho-metric viewpoint has been judged g u i l t y i n v i r t u e of i t s a s s o c i a t i o n with c e r t a i n p ersonality theories"(Wiggins, 1974). Wiggins proposed an a l t e r n a t i v e view of t r a i t s . He t r i e d to "illu m i n a t e the nature of the t r a i t concept by considering ways i n 4 which t r a i t terms are employed i n everyday discourse." He came to two important conclusions. F i r s t , he concluded that "when a t r a i t q u a l i t y i s a t t r i b u t e d to an a c t i o n " i t means that "the a c t i o n belongs to a class of actions that are l i k e l y to lead to a p a r t i c u l a r outcome." Secondly, with respect to t r a i t s as a t t r i b u t e s of people, Wiggins rejected the "causal d i s p o s i t i o n a l " view i n favor of a summary view (Hampshire, 1953). According to the summary view, rather than being i n f e r r e d " i n t e r n a l " d i s p o s i t i o n s which cause c e r t a i n behaviors, t r a i t s are c a t e g o r i c a l summaries of a person's past conduct to date. This a l t e r n a t i v e view has important implica-tions f o r an analysis of t r a i t concept usage by laymen. Models of T r a i t Concept Usage by Laymen At the same time as he c r i t i c i z e d the global d i s p o s i t i o n a l approach to personality, Mischel (1971, 1973) recognized that t r a i t concepts are used ubiquitously by people i n many l i f e s i t u a t i o n s and therefore must serve some useful purpose. He recommended that researchers begin to investigate the functions t r a i t concepts serve f o r laymen. Mischel (1968) claimed that observers, even pr o f e s s i o n a l psychological observers, are prone to think about the p e r s o n a l i t i e s of other people i n terms of a c o l l e c t i o n of broad d i s p o s i t i o n s or t r a i t s , despite the scant empirical evidence f o r t h e i r existence. He argued that t h i s tendency r e s u l t s from d e f i c i t s and biases i n the information a v a i l a b l e to the observer and from a wide v a r i e t y of biases i n the processing 5 of information at the perceptual, cognitive and l i n g u i s t i c l e v e l s . Mischel thought that t h i s conceptual tendency very c l o s e l y resembled the psychodynamic view of t r a i t s and therefore he proposed . a psychodynamic model of t r a i t concept usage by laymen. This model i s most c l e a r l y presented i n a paper by Hayden and Mischel (1973). They stated that laymen " a t t r i b u t e diverse seemingly discrepant behaviors to the same underlying motive (or other causal d i s p o s i t i o n ) . Perhaps t h i s t a c t i c i s seen most p l a i n l y i n the t r a d i t i o n a l ' i n d i r e c t sign' paradigm f o r studying p e r s o n a l i t y . In t h i s approach, diverse overt behaviors maybe/construed as i n d i r e c t sympto-matic manifestations of the same basic underlying genotypic motivational d i s p o s i t i o n s " (p. 3). Wiggins (1974) d i d not think that laymen follow a psychodynamic model when they use t r a i t concepts. His summary model, which suggested that t r a i t concepts are used to summarize a target person's past conduct to date i n terms of categories which code the l i k e l y e f f e c t s of a person's behavior i n c e r t a i n s i t u a t i o n s , does not ascribe any explanatory function to t r a i t concepts. When t h i s model i s applied to t r a i t concept usage by laymen, i t follows that the layman's chief purpose i s not to explain the behavior of other people but rather to describe such behavior by coding i t i n terms of categories he has learned through the language he has been taught and through other c u l t u r a l experiences. Once a person has c l a s s i f i e d a l l of the a v a i l a b l e information about a target person, he can step back, as i t were, and summarize that person's behavior i n "more than" or " l e s s than" terms. The end point of t h i s summarization process i s the a t t r i b u t i o n of a t r a i t concept. 6 There are three important implications of the summary model. The f i r s t i s that observations or reports v of behavioral performances are c r u c i a l to the a t t r i b u t i o n of t r a i t concepts. The t r a i t concept i s the end point of the categorization of actions. Secondly, the summary model implies a p a r t i c u l a r t r a i t concept formation process. The layman begins by sorting the behavioral performances into the categories which he has learned. Then, i n a continuous monitoring fashion, the layman engages i n the i n t e l l e c t u a l task of "weighing the evidence" as i t occurs and of a t t r i b u t i n g a t r a i t concept according to that evidence. F i n a l l y , the summary model implies that as ongoing behavioral performances are s c r u t i n i z e d , a d d i t i o n a l evidence may cause an immediate r e t r a c t i o n of the a t t r i b u t e d t r a i t concept i n favor of another t r a i t concept which more adequately represents the current end point of the summarization process. Note that the emphases of the summary model are d i f f e r e n t from those of the psychodynamic model. F i r s t , the psychodynamic model stresses the explanatory function of t r a i t s whereas the summary model stresses the d e s c r i p t i v e aspect. The summary model does not "comment on c a u s a l i t y . " A second difference i s that the psychodynamic model stresses the* r i g i d nature of t r a i t concepts whereas the summary model implies that t r a i t concepts are amenable to change. This second diffe r e n c e i s c l e a r l y shown i n the predictions of the two contrasting models concerning the e f f e c t of contradictory information on i n i t i a l t r a i t concepts. Hayden and Mischel (1973) maintained that laymen hold on tenaciously to i n i t i a l t r a i t concepts 7 even when presented with contradictory evidence. On the other hand, Wiggins doubted that laymen hold on tenaciously to t h e i r i n i t i a l t r a i t concepts. In h i s view, contradictory evidence merely changes the balance of the summary of past behavior to date. I f there i s enough evidence which contrasts the old concept, a new t r a i t concept w i l l r e a d i l y be chosen to replace i t . The purpose of t h i s i n v e s t i g a t i o n i s to study a subset of t r a i t concept usage by laymen, namely, t r a i t concept formation and change. Both of the general models of t r a i t concept usage are pertinent to t h i s goal. However, because i t s p e c i f i e s p o s s i b l e ways i n which t r a i t concepts might be learned, the summary model i s the more useful model f o r generating predictions about t r a i t concept formation and change. Goals of The Present Investigation There are several goals to t h i s i n v e s t i g a t i o n . The f i r s t goal i s to study one parameter of t r a i t concept formation, namely, the v a l i d i t y of the behavioral information which i s used as the basis f o r the formation of an i n i t i a l t r a i t concept. The i n t e n t i o n i s to see i f d i f f e r e n t i a l l y v a l i d behavioral information makes a di f f e r e n c e i n the speed with which subjects form an i n i t i a l concept. The second goal i s to study three parameters of t r a i t concept change. In t h i s i n v e s t i g a t i o n , t r a i t concept change i s i n i t i a t e d by g iving subjects new behavioral information which d i f f e r s systematically from the information upon which t h e i r i n i t i a l t r a i t concept was formed. The f i r s t parameter i s the resistance to change of i n i t i a l t r a i t concepts which have been formed by reading behavioral information of varying v a l i d i t y . The i n t e n t i o n i s to see i f t r a i t '. concepts which are formed by reading behavioral information of one l e v e l of v a l i d i t y are easier to change than t r a i t concepts which are formed by reading information of a d i f f e r e n t l e v e l of v a l i d i t y . The second parameter of t r a i t concept change i s the v a l i d i t y of the behavioral information according to which a change i n the i n i t i a l t r a i t concept takes place. The i n t e n t i o n here i s to see i f the rate of t r a i t concept change d i f f e r s as a function of the type of information which i s being read by subjects and which requires a new basis f o r i t s conceptualization. The t h i r d parameter of t r a i t concept change i s the type of change required within each t r a i t concept. Two types of changes are required by the new behavioral information: (a) a re v e r s a l of one of the dimensions which make up that t r a i t concept; and (b) the addi t i o n of a new dimension to that i n i t i a l t r a i t concept. The f i r s t type i s required because new behavioral information which contradicts the o l d information on one p a r t i c u l a r dimension i s presented. The second type of change i s required because behavioral information which suggests a second dimension of the i n i t i a l t r a i t concept i s replaced by behavioral information which suggests an e n t i r e l y new dimension. No change i s required on the t h i r d dimension of the i n i t i a l t r a i t concept because subjects merely read new-behavioral-information which suggests the same t r a i t dimension that was 9 suggested during t r a i t concept formation. The i n t e n t i o n i n making these changes i s to study the e f f e c t of behavioral information both of a completely new and of a contradictory type ori t r a i t concept change. The t h i r d goal has two aspects. The f i r s t i s to study the e f f e c t of the v a l i d i t y of behavioral information on the a t t r i b u t i o n of an o v e r a l l t r a i t concept to the i n d i v i d u a l s about whom behavioral information was read. The second aspect i s to study t r a i t a t t r i b u t i o n as a function of the type of change which i s required by the new and contradictory behavioral information. D e f i n i t i o n s A t r a i t concept i s an idea about the per s o n a l i t y a t t r i b u t e s of another person. In t h i s study, t r a i t concepts of two f i c t i t i o u s people whose personality a t t r i b u t e s have been preset according to d i f f e r e n t t r a i t concept patterns are formed. The patterns are described i n the next chapter. The t r a i t concepts are conjunctive i n nature because they consist of three separate dimensions each of which refe r s to a d i f f e r e n t p ersonality a t t r i b u t e . The term, i n i t i a l t r a i t concept, refers to the f i r s t idea formed about the personality a t t r i b u t e s of another person. In t h i s study, the i n i t i a l t r a i t concept has three dimensions andfthe t r a i t concept formation stage of the study i s considered to be complete when subjects c o r r e c t l y d i f f e r e n t i a t e the behavior of the two f i c t i t i o u s people. Note 10 that the i n i t i a l t r a i t concept i s i n f e r r e d from the subject's a b i l i t y to i d e n t i f y which f i c t i t i o u s person performed each of the behaviors i n the f i r s t stage of the study. T r a i t concept formation r e f e r s to the process of coming to have an idea about the personality a t t r i b u t e s of another person. In the present i n v e s t i g a t i o n , t h i s process takes place when subjects read successive behavioral information and pick up something (a t r a i t concept) which helps them to d i f f e r e n t i a t e the actions of the f i c t i t i o u s people. The process of forming t h i s idea i s c a l l e d t r a i t concept formation. T r a i t concept change or adjustment ref e r s to the adjustment of an i n i t i a l t r a i t concept to accommodate new information which may be independent of or contradictory to that i n i t i a l t r a i t concept. In everyday l i f e , we encounter many instances where an i n d i v i d u a l about whom we hold a t r a i t concept acts i n ways which are not consistent with that t r a i t concept. I f people continue to act i n ways which are inconsistent with our t r a i t concept, we tend to adjust our t r a i t concept to incorporate the new information. This process i s c a l l e d t r a i t concept change or adjustment. In t h i s study, t r a i t concept change or adjustment r e f e r s to the adjustment of an i n i t i a l t r a i t concept which takes place when subjects read new and contra-d i c t o r y information about the f i c t i t i o u s people. V a l i d i t y i s a psychometric concept which, i n the present i n v e s t i g a t i o n , i s used to mean the extent to which a p a r t i c u l a r behavior suggests the t r a i t term from which that behavior had 11 o r i g i n a l l y been generated. When one i s presented with a t r a i t term, one can think of several actions i n several s i t u a t i o n s which can be classed as suggesting that p a r t i c u l a r t r a i t term. However, some of these actions more c l e a r l y suggest that t r a i t term than other actions do. Hence, these actions d i f f e r i n v a l i d i t y . For example, one action may suggest several t r a i t terms and therefore the r e l a t i o n -ship of that action to the t r a i t term from which i t had been generated i s obscure. As a suggestor of t h i s t r a i t term, the action i n question would be low i n v a l i d i t y . The method of e s t a b l i s h i n g the v a l i d i t y of the behavioral information which i s used i n t h i s i n -v e s t i g a t i o n i s described presently. Related L i t e r a t u r e and Expectations Research which has investigated how laymen form t r a i t concepts and then adjust those concepts to accommodate new and/or contradictory information i s almost non-existent'. Hence, there i s a lack of empirical data and t h e o r e t i c a l propositions which would guide research i n t h i s area. This lack i s probably due most to the d i f f i c u l t y of studying concept formation i n general and to the very recent focus upon the study of the functions t r a i t concepts serve for laymen. Secondly, no one has organized the important v a r i a b l e s of t r a i t concept usage into a coherent conceptual system and therefore i t i s d i f f i c u l t to decide what research areas are relevant to the goals of any p a r t i c u l a r i n v e s t i g a t i o n . Most of the l i t e r a t u r e which i s relevant to t h i s study appears i n the subject area of 12 impression formation and i s discussed i n terms of the goals of t h i s study to which i t r e l a t e s . 1. The F i r s t Goal In an extensive search of the l i t e r a t u r e , I did not f i n d a sin g l e study which investigated t r a i t concept formation as a function of d i f f e r e n t i a l l y v a l i d behavioral information. There i s a big l i t e r a t u r e on impression formation (Cook, 1971; Hastorf, Schneider & Polefka, 1970) which seems to be very c l o s e l y r e l a t e d to t r a i t concept formation, but the method of studying the formation process i n the present i n v e s t i g a t i o n d i f f e r s from that used i n most impression formation studies. In most of the impression formation l i t e r a t u r e , subjects are given a systematically constructed adjective l i s t and are asked to give t h e i r impression (either by w r i t i n g a free d e s c r i p t i o n or by responding to an adjective check l i s t ) of a stimulus person who had been described by these a d j e c t i v e s . The purpose of t h i s procedure i s to determine how perceivers combine inconsistent t r a i t s t i m u l i to produce an o v e r a l l impression. Asch (1946) f e l t that the study of how inconsistent information i s resolved would reveal something about impression formation i n general. Similar to Asch (1946), the intent of. t h i s i n v e s t i g a t i o n i s to reveal some important processes of t r a i t concept formation but the method of i l l u m i n a t i n g these processes i s d i f f e r e n t . The approach here i s to have subjects read behavioral information and to discriminate 13 between two people who have performed d i f f e r e n t behaviors. The . formation of a t r a i t concept occurs through the process of i n t e g r a t i n g the behavioral information about each f i c t i t i o u s person and of discriminating the behavior of one person from the other one. Unlike the impression formation researchers, I do not begin by giving subjects t r a i t information. Subjects must form t h e i r own t r a i t impression. In s p i t e of t h i s d i f f e r e n c e of approach, the t h e o r e t i c a l ideas which arose i n impression formation research seem applicable to the processes which are studies here. In t h e i r discussion of the impression formation l i t e r a t u r e , Jones and Goethals (1972) stated that "Cognitive categories are more r e a d i l y formed when the e n t i t y being judged i s stable and i t s manifestations are assumed to be equally d i s t r i b u t e d over time. Unstable e n t i t i e s with capricious, c y c l i c a l or episodic manifestations should contribute to the suspension of conclusive a t t r i b u t i o n a l judgments and to the avoidance of the kind of premature closure on which a s s i m i l a t i o n apparently depends" (p. 43). When these comments are applied to t r a i t concept formation, c e r t a i n predictions emerge. In p a r t i c u l a r , when t r a i t concepts are formed by reading h i g h - v a l i d behavioral information they should be easy to form because t h i s information most c l e a r l y suggests the t r a i t term from which i t had been generated. Hence, t h i s information should cause subjects to experience more s t a b i l i t y and evenness to the personality of i n d i v i d u a l s who perform these actions and therefore subjects should r e a d i l y form a t r a i t concept which summarizes the behavior. 14 In contrast, subjects who read low-valid behavioral informa-t i o n which does not so c l e a r l y suggest the appropriate t r a i t term should suspend a t t r i b u t i o n a l judgments. Consequently, they should have more d i f f i c u l t y formulating categories which w i l l help them d i f f e r e n t i a t e people who have performed the behaviors. 2. The Second Goal The second goal of t h i s study has three parts: (a) the f i r s t part concerns the strength of the i n i t i a l t r a i t concept; (b) the second part concerns the v a l i d i t y of subsequent information; and (c) the t h i r d part concerns the type of change within each t r a i t concept. L i t e r a t u r e which r e l a t e s to each part i s discussed i n turn. Strength of the i n i t i a l t r a i t concept. One study which r e l a t e s to the resistance to change of i n i t i a l t r a i t concepts was found. Hayden and Mischel (1973) designed two experiments which showed the tendency of laymen to maintain t r a i t consistency i n the r e s o l u t i o n of behavioral inconsistency. In the f i r s t experiment, they had subjects read paragraph descriptions of a stimulus person's behavior i n order to form an i n i t i a l t r a i t impression. Then they gave the subjects cartoons which portrayed that person behaving i n a way which was consistent, inconsistent, or neutral with regard to the behaviors which appeared i n the paragraphs. F i n a l l y , they asked the subjects to judge the extent to which various t r a i t s motivated the cartoon behaviors. The experimenters found that subjects tended to i n f e r more motivation from the t r a i t s which were consistent with t h e i r i n i t i a l 15 Impression, and l e s s motivation from t r a i t s which were inconsistent with t h e i r i n i t i a l impression. The authors concluded that an observer's i n i t i a l impression of other people biased the way i n which he interpreted the motivations which underly t h e i r subsequent behavior, i n the d i r e c t i o n of t r a i t consistency. In the second experiment which was c a r r i e d out concurrently with the f i r s t experiment, the experimenters asked the subjects to a t t r i b u t e causes f o r the behavioral performances which appeared i n the cartoons. The authors found that when the cartoon behaviors were consistent with the i n i t i a l impression of the stimulus person, they were a t t r i b u t e d to the di s p o s i t i o n s of the stimulus person, but when they were inconsistent with the i n i t i a l impression, the cartoon behaviors were a t t r i b u t e d to s u p e r f i c i a l and transient causes. These findings suggest a strong influence of the i n i t i a l t r a i t concept i n determining the mode of i n t e g r a t i o n of a d d i t i o n a l behaviors. Given t h i s strong determining tendency, i t appears that i n i t i a l t r a i t concepts are strongly r e s i s t a n t to change. To some extent, the r e s o l u t i o n of inconsistency i n impression formation research can be construed as t e s t i n g the resistance to change of i n i t i a l t r a i t concepts. In these studies, subjects are usually given a l i s t of adjectives the second h a l f of which contains adjectives which are inconsistent with those i n the f i r s t h a l f . The manner i n which subjects resolve t h i s inconsistency i s compared to the solutions of other subjects who have been given the adjectives 16 i n the reverse order. To the extent that the f i r s t part of the adjective l i s t causes an i n i t i a l impression, the second part can be seen as t e s t i n g the resistance to change of that i n i t i a l impression. Several impression formation studies have suggested possible parameters that a f f e c t the r e s o l u t i o n of inconsistency. Some va r i a b l e s which have been shown to be important i n determining a primacy e f f e c t are: c r e d i b i l i t y of the source (Rosenbaum & Levin, 1969), response d i s p o s i t i o n (Kaplan, 1971, 1972, 1973), discounting i n s t r u c t i o n s (Anderson & Jacobson, 1965), cognitive tuning set (Cohen, 1961), f a v o r a b i l i t y of the adjectives (Briscoe, Woodyard & Shaw, 1967), amount of information (Schmidt, 1969 Wyer, 1970, 1974), and the order of information (e.g., Anderson, 1965; Anderson & Norman, 1964). However, there i s no general agreement about the importance of these v a r i a b l e s . From his review of studies which have var i e d stimulus consistency i n one way or another, Hendrick (1972) concluded that "the evidence that consistency-inconsistency i s an important v a r i a b l e i s rather meagre and the e f f e c t s which do e x i s t are s l i g h t and subtle." Jones and Goethals (1972) used the analysis of the r e s u l t s from the order-effect paradigm to suggest some of the processes which underly impression formation. They discussed three processes that are l i k e l y to r e s u l t i n a primacy e f f e c t . The f i r s t process i s an attention decrement. I t i s thought that subjects pay l e s s attention to the a d j e c t i v a l information i n the l a t t e r part of the 17 l i s t because they are fatigued or because they are protecting t h e i r f i r s t impression against the disturbing implications of incongruous information. The second process i s discounting. In t h i s process, incon-gruent information i s l i k e l y to be discounted e s p e c i a l l y when factors i n the p a r t i c u l a r context "permit judgments of d i f f e r e n t i a l r e l i a b i l i t y or v a l i d i t y " (p. 43). T r a i t s of l e s s e r v a l i d i t y are discounted i n an o v e r a l l i n t e g r a t i v e attempt. The t h i r d process i s a s s i m i l a t i o n . For t h i s process to occur, Jones and Goethals (1972) assumed that "... information creates or finds i t s way into p r e - e x i s t i n g categories. These categories are l i k e hypotheses about the nature of r e a l i t y being confronted. Once a c a t e g o r i c a l decision i s made, subsequent evidence i s d i s t o r t e d to f i t the category or to confirm the hypothesis - as long as i t i s not too discrepant from the category's t y p i c a l instance" (p. 43). This process d i f f e r s from the discounting or att e n t i o n decrement in t e r p r e t a t i o n s because i t implies that "the value, i n t e n s i t y , or frequency of l a t e r u nits are shaped by the categories (expectancies) established by early information" (Jones & Goethals, 1972, p. 43). In t h i s sense, a lower weight i s always given.to more recent events because they are viewed i n terms of the e a r l i e r a s s i m i l a t i v e product. Jones and Goethals (1972) made two a d d i t i o n a l points. F i r s t , they stated that "any factor that contributes to the speed with which a category can be formed and to the degree of confidence with which i t i s held should increase the tendency toward a s s i m i l a t i o n " 18 (p. 43). In other words, a r e a d i l y formed category should press toward the a s s i m i l a t i o n of subsequent information. Secondly, they stated that " a s s i m i l a t i o n probably depends i n a c r i t i c a l way on the discrepancy between items of unfolding information" (p. 43). They recognized that there i s a point of incongruity beyond which people do not press f o r a s s i m i l a t i o n . As yet, we have l i t t l e information about factors which determine t h i s discrepancy point. With regard to the l e v e l of incongruity, Jones and Goethals (1972) proposed a c u r v i l i n e a r hypothesis which r e l a t e d the d i s -crepancy magnitude to primacy e f f e c t s : " I f l a t e r manifestations are only s l i g h t l y d i f f e r e n t from e a r l i e r manifestations i n t h e i r e n t i t y i m p l i c a t i o n , they should be r e a d i l y assimilated i n value. The l a t t e r manifestations should, i n e f f e c t , be seen as more l i k e the e a r l i e r manifestations than i n fact - i s the case, giving r i s e to a primacy e f f e c t . With s l i g h t l y greater discrepancies, however, we might expect contrast to occur. Later items should exert a disproportionate, influence on the category ( e n t i t y , expectancy ?) being formed and should bring about recency. But beyond some point of moderate discrepancy, the processes of i n c r e d u l i t y , d e n i a l and discounting should give r i s e to primacy again" (p. 43). It i s not c l e a r to what extent the variables which have been found to a f f e c t the r e s o l u t i o n of inconsistent t r a i t information determine the i n t e g r a t i o n of behavioral i n c o n s i s t e n c i e s . In an extensive search of the impression formation l i t e r a t u r e , I found only one study which used behavioral information. Richey, McClelland and Shimkunas (1967) gave u n i v e r s i t y students written behavioral information and asked them to rate the character of a stranger. Half of the subjects received the information i n a 19 positive-negative order and the other h a l f i n a.negative-positive order. The experimenters found that i n i t i a l ratings which were based on s i n g l e univalent paragraphs were s i g n i f i c a n t l y a l t e r e d i n both groups by subsequent incompatible information, but the change was not equally permanent for both orders of presentation. O r i g i n a l l y p o s i t i v e impressions were found to be l a s t i n g l y changed by negative information, but o r i g i n a l l y negative impressions which had been revised upward became s i g n i f i c a n t l y more negative a f t e r nine days. These r e s u l t s stress the importance of the order i n which the behavioral information i s presented. As f a r as I have been able to determine, there i s no evidence which suggests that c r e d i b i l i t y of the source, response d i s p o s i t i o n , discounting i n s t r u c t i o n s , f a v o r a b i l i t y of the information or the amount of information a f f e c t s the r e s o l u t i o n of inconsistencies i n behavioral information. The processes which were described by Jones and Goethals(1972) with regard to the primacy e f f e c t appear to r e l a t e to the resistance to change of the t r a i t concepts i n the present i n v e s t i g a t i o n . On the basis of the primacy processes, one can predict which t r a i t concepts are more r e s i s t a n t to change. In p a r t i c u l a r , t r a i t concepts which were formed on the basis of h i g h - v a l i d behavioral information should be hard to adjust ( i . e . , more r e s i s t a n t to change) for the following reasons. 1. The i n i t i a l impression should be strong and therefore more r e a d i l y protected against the disturbing implications of incongruous information. 20 2. Because the evidence f o r the i n i t i a l t r a i t concept i s very r e l i a b l e (more v a l i d ) , subjects are more l i k e l y to discount subsequent information. 3. High-valid behavioral information should contribute to the speed with which a category i s formed and to the degree of confidence with which i t i s held. Therefore, there should be a greater press toward the a s s i m i l a t i o n of subsequent material into the e x i s t i n g t r a i t concept because the value of the subsequent information should be shaped by the i n i t i a l t r a i t categories. These conditions or processes should be less applicable to t r a i t concepts which have been formed by reading low-valid behavioral informa-t i o n . Therefore, when presented with new and contradictory information, subjects should more r e a d i l y change t h e i r t r a i t concept i n the d i r e c t i o n of that new information. V a l i d i t y of subsequent information. No in v e s t i g a t i o n s which studied the second parameter of t r a i t concept change were found. However, once again, we can look to the impression formation l i t e r a -ture to f i n d clues which would suggest how t h i s parameter may influence t r a i t concept adjustment. In the impression formation l i t e r a t u r e , some conditions which favor a recency e f f e c t i n the i n t e g r a t i o n of incongruent t r a i t information have been found. Jones and Goethals (1972) have reviewed the major findings and have suggested three processes which favor recency e f f e c t s . The f i r s t process i s r e c a l l readiness which means that immediate past events are better remembered than more remote 21 events. Jones and Goethals maintained that r e c a l l readiness depends upon how remote the early events are. The further away the early events, the more the tendency w i l l be toward recency. The second process i s judgmental contrast. According to Jones and Goethals (1972), "e a r l y information creates anchoring expectancies i n terms of which l a t e r expectancies are judged" (p. 43). However, i f the l a t e r expectancies are not met, the anchor i s e i t h e r defended or changed to f i t the new information. The more the subsequent information constrasts with the early information, the more w i l l be the tendency toward recency ( i . e . , to adjust the impression to f i t the more recent information). However, i f the new information i s too discrepant i t tends to be denied and the impression remains as i t was i n i t i a l l y ( c f . the c u r v i l i n e a r hypothesis). The f i n a l group of processes which promote recency e f f e c t s concern some content-and context-related hypotheses. F i r s t , " i f the e n t i t y i s known to be capable of progressive changes or develop-ment, i t s l a t e r manifestations are obviously more s i g n i f i c a n t than e a r l i e r manifestations f o r in t e r a c t i o n s with the e n t i t y " (Jones & Goethals, 1972, p. 44). Secondly, a c e r t a i n context and/or content of the judgments may i n i t i a t e cognitive processes that d i s c r e d i t early information f o r d i f f e r e n t reasons. Jones and Goethals (1972) gave the example of an i n d i v i d u a l who seemed shy at f i r s t , but more confident l a t e r . In t h i s circumstance, the early 22 information i s l i k e l y to be discounted. F i n a l l y , the content may be ordered i n such a way that i t suggests a gradual progression to the important l a t e r information. ."Later data may explain or subsume e a r l i e r evidence" (Jones & Goethals, 1972, p. 44). These comments can be applied to t r a i t concepts which have been formed on the basis of behavioral information. On the basis of the processes outlined by Jones and Goethals, one can make some predictions concerning the v a l i d i t y of the subsequent behavioral information. In p a r t i c u l a r , h i g h - v a l i d behavioral information should r e s u l t i n a rapid adjustment of the i n i t i a l t r a i t concept f o r the following reasons. 1. The information w i l l be more r e l i a b l e and therefore there w i l l be les s of a tendency to ignore i t or to discount i t . 2. With reference to the c u r v i l i n e a r hypothesis of Jones and Goethals (1972), h i g h - v a l i d behavioral information w i l l suggest a l e v e l of incongruity beyond which subjects w i l l press f o r the as s i m i l a t i o n of that information into the i n i t i a l t r a i t concept. The contrast w i l l be obvious and subjects w i l l more r e a d i l y reassess t h e i r anchoring information and change t h e i r anchor to accommodate the new information. • These processes are not l i k e l y to come into e f f e c t when low-valid information is. the basis f o r t r a i t concept, change. This low-valid information should be r e a d i l y assimilated into the i n i t i a l concept. Type of change within each t r a i t concept. A search of the human conceptual behavior l i t e r a t u r e and the impression formation 23 l i t e r a t u r e h a s . f a i l e d to uncover a s i n g l e study which has investigated the e f f e c t on an i n i t i a l t r a i t concept of making two types of changes: (a) reversing the behavioral information on one dimension of that t r a i t concept; and (b) replacing the behavioral information on a second dimension with behavioral information which suggests an e n t i r e l y new dimension. Hence, there i s l i t t l e to guide research i n t h i s area. Most of the s h i f t s which have been studied i n the human conceptual behavior l i t e r a t u r e are d i f f e r e n t from the s h i f t s which are used i n t h i s i n v e s t i g a t i o n ( c f . Bourne, 1966; Bourne, Ekstrand, & Dominowski, 1971; Slamecka, 1968). Bourne and M i l l e r (1973) noted three main forms of s h i f t s . F i r s t , the intradimensional r e v e r s a l s h i f t (IDR) imposes a change i n a l l i n i t i a l l y learned stimulus category assignments. For example, with color and shape as the main features of the s t i m u l i , subjects f i r s t l earn to place red s t i m u l i i n Category A and green s t i m u l i i n Category B. The IDR s h i f t requires that subjects place red s t i m u l i i n Category B and green s t i m u l i i n Category A. This s h i f t i s used i n the present i n v e s t i g a t i o n when subjects f i r s t l earn the a t t r i b u t e s of two f i c t i -t ious people on one t r a i t dimension during the a c q u i s i t i o n part of the study and then learn the reversed a t t r i b u t e s of those f i c t i t i o u s people during the r e v e r s a l part of the study. The second s h i f t i s an extradimensional s h i f t (ED). This s h i f t imposes a change of relevant stimulus dimensions. For example, with 24 color and shape as the main dimensions of the s t i m u l i , subjects f i r s t l e a r n to place squares i n Category A and t r i a n g l e s i n Category B. An ED s h i f t requires subjects to place reds i n Category A and greens i n Category B. When learning on t h i s s h i f t i s compared with learning on the IDR s h i f t , researchers have found that adult subjects perform better on the IDR s h i f t . The ED s h i f t i s not used i n the present i n v e s t i g a t i o n . Here, a s h i f t i s made from one t r a i t dimension to a new t r a i t dimension, but both dimensions are t r a i t dimensions. In both the pre- and po s t - s h i f t parts of the study, the relevant s t i m u l i are behavioral data, a l b e i t d i f f e r e n t behavioral data. The change i n the t r a i t concept i s merely a change from one t r a i t dimension to another t r a i t dimension and not a change to some other kind of dimension as the ED s h i f t requires. The t h i r d type of s h i f t i s an intradimensional nonreversal s h i f t (IDN). In t h i s s h i f t , the same dimension i s relevant before and a f t e r the s h i f t , but d i f f e r e n t values are used to represent i t . For example, with color and shape as the main dimensions of the s t i m u l i , subjects f i r s t l earn to place yellow s t i m u l i i n Category A and blue s t i m u l i i n Category B. An IDN s h i f t requires subjects to place reds i n Category A and greens i n Category B. The same dimension (color) i s relevant, but i t has d i f f e r e n t values. A common pr a c t i c e has been to compare the learning of the IDR 25 and ED s h i f t s to the learning of the IDN s h i f t . Research has shown that f o r adult subjects, the IDR and IDN s h i f t s are s i g n i f i c a n t l y easier to learn than the ED s h i f t (Issacs & Duncan, 1962; Johnson, 1966). Moreover, the IDN s h i f t has been found to be s i g n i f i c a n t l y e a sier to l e a r n than the IDR s h i f t . This f i n d i n g i s expected because the IDN s h i f t i s a p o s i t i v e t r a n s f e r paradigm while the IDR s h i f t i s a negative t r a n s f e r paradigm. D i f f e r e n t researchers have d i f f e r e n t ideas about what kind of t r a n s f e r takes place i n the ED s h i f t (Bourne, Ekstrand, & Dominowski, 1971; Bourne & M i l l e r , 1973). In the present i n v e s t i g a t i o n where the same stimulus property i s relevant (upper or lower pole p o s i t i o n along the dimensions which describe the a t t r i b u t e s of the f i c t i t i o u s people) , a s h i f t which very c l o s e l y resembles the IDN s h i f t i s used. A f t e r the s h i f t i n t h i s i n v e s t i g a t i o n , the pole positions remain the same i n r e l a t i o n to one another but they take on new values ( i . e . , the pole pos i t i o n s are described by.newJbehaviourai"statements). Given that i t i s a p o s i t i v e t r a n s f e r paradigm, t h i s s h i f t i s expected to be the easiest component of the adjusted t r a i t concept to learn. Secondly because i t i s a negative t r a n s f e r paradigm, i t i s expected that the r e v e r s a l s h i f t w i l l be the most d i f f i c u l t to learn. F i n a l l y , i t is. expected that subjects w i l l carry some of the negative trans f e r on the reversed dimension over to the unchanged dimension. This tendency should make the unchanged dimension harder to learn than the new dimension ( i n the IDN s h i f t ) . However, the negative transfer should not l a s t as long as i t w i l l on the reversed 26 dimension because subjects w i l l soon discover that t h e i r previous mode of response on t h i s dimension i s s t i l l being reinforced. 3. The Third Goal The t h i r d goal of t h i s i n v e s t i g a t i o n was to study t r a i t , a t t r i b u t i o n as a function of d i f f e r e n t i a l l y v a l i d behavioral information during both the t r a i t concept formation and the t r a i t concept adjust-ment stage of the study and as a function of the type of change required within each t r a i t concept. In my search of the l i t e r a t u r e , I have not found any study which investigated the e f f e c t of d i f f e r e n t i a l l y v a l i d behavioral information on t r a i t a t t r i b u t i o n . Nor have I found a study which investigated the e f f e c t of systematic changes within a t r a i t concept on the o v e r a l l t r a i t impression. Consequently, there i s very l i t t l e to guide research i n t h i s area. To some extent, the study of how subjects resolve inconsistent t r a i t information can be construed as the study of t r a i t a t t r i b u t i o n following some kind of r e s o l u t i o n of inconsistent information. The t h e o r e t i c a l propositions concerning the processes which favour primacy and recency e f f e c t s (Jones & Goethals 1972) suggest how t h i s t r a i t a t t r i b u t i o n takes place. I f the subsequent information i s ignored, . discounted or assimilated, then t r a i t a t t r i b u t i o n should r e f l e c t the i n i t i a l impression. On the other hand, i f the contrast i s too great to be ignored, discounted or assimilated, t h e n . t r a i t a t t r i b u t i o n should r e f l e c t the d i r e c t i o n of the new information. 27 With reference to t h i s i n v e s t i g a t i o n , t r a i t a t t r i b u t i o n should be d i f f e r e n t i a l l y a ffected by the type of behavioral information subjects read while adjusting t h e i r t r a i t concept. In general, the more v a l i d the behavioral information, the more t r a i t a t t r i b u -t i o n should be affected by that information. This d i f f e r e n c e may be seen only on the reversed dimension of the t r a i t concept. Here, d i f f e r e n t predictions are advanced, which correspond to the type of information which was received during both the i n i t i a l concept formation and the concept adjustment parts of the study. F i r s t , i f subjects formed t h e i r i n i t i a l t r a i t concept by reading ei t h e r high- or low-valid behavioral information, but adjusted that concept on the basis of hi g h - v a l i d behavioral information, we would expect t r a i t a t t r i b u t i o n to r e f l e c t a recency e f f e c t because the new and contradictory information i s too v a l i d to be ignored, discounted or assimilated. In contrast, i f subjects formed t h e i r i n i t i a l t r a i t impression on the basis of ei t h e r high-or low-valid behavioral information, but adjusted that concept on the basis of low-valid behavioral information, we would expect t r a i t a t t r i b u t i o n to r e f l e c t a primacy e f f e c t because the new and contradictory information w i l l tend to be ignored, discounted or assimilated into that i n i t i a l t r a i t concept. It i s i n t e r e s t i n g to study t r a i t a t t r i b u t i o n as a function of the o v e r a l l amount of information subjects received about each of the t r a i t dimensions upon which t h e i r o v e r a l l impression was based. On 28 every dimension, subjects read information about two f i c t i t i o u s people whose per s o n a l i t y a t t r i b u t e s are opposite to each other. However, the amount of information and the place where the information was obtained d i f f e r s f o r each dimension. F i r s t , on the reversed dimension, subjects w i l l have read four behaviors which were suggestive of one pole of that dimension during the t r a i t concept formation part of the study, but i n the adjustment part, they w i l l have read four behaviors which were suggestive of the opposite pole of that same dimension for each f i c t i t i o u s person. Hence, subjects have an equal amount of informa-t i o n which i s suggestive of each pole. From a summary view of t r a i t concepts, I predict that the subjects' impressions along t h i s dimension w i l l be very near the middle of the two poles of that dimension. Secondly, on the unchanged dimension, subjects w i l l have read eight behaviors which are suggestive of the same pole of that dimension ( i . e . , four behaviors i n the t r a i t concept formation part and four i n the t r a i t concept adjustment part of the study). Again, from a summary view of the t r a i t s , I p r e d i c t that subjects w i l l i n d i c a t e a strong impression by a t t r i b u t i n g a t r a i t concept near the extreme poles which were suggested by the behaviors they read. T h i r d l y , on both the replaced dimension and the new dimension, subjects w i l l have read four behaviors which suggested each t r a i t dimension, but the information w i l l have come from d i f f e r e n t places i n the study. In the replaced dimension, the information w i l l have 29 come only i n the t r a i t concept formation part while, on the new dimension, the information w i l l have come only i n the adjustment part of the study. On these two dimensions, subjects are expected to form moderate impressions and to i n d i c a t e those impressions by a t t r i b u t i n g a moderate t r a i t concept on both dimensions i n the d i r e c -t i o n which.was suggested by the behaviors they read. However, some of the information about the replaced dimension may have been forgotten. Therefore, i t i s expected that the impression on t h i s dimension may not be as extreme as i t w i l l be f o r the new dimension. The T r a i t Dimensions In order to carry out an adequate test of the experimental hypotheses, I needed to have t r a i t dimensions which meet two q u a l i f i c a t i o n s . F i r s t , they must be as independent of each other as possible and second, y they must be f a m i l i a r to laymen. Four t r a i t dimensions which f u l f i l l both of these requirements have been obtained through peer r a t i n g research. Wiggins (1973) has discussed the basic procedures i n the study of t r a i t a t t r i b u t i o n as i t unfolded i n peer r a t i n g research and has reviewed the major attempts to f i n d the basic s t r u c t u r a l dimensions which account f o r most of the variance i n peer r a t i n g research. The studies which are of most i n t e r e s t here are the Norman studies. Norman (1963) selected f i v e r a t i n g dimensions which adequately represented each of f i v e f a c t o r s , f i r s t uncovered by Tupes and 30 C h r i s t a l (1961): surgency (extraversion), agreeableness, dependability (conscientiousness), emotional s t a b i l i t y and c u l t u r e . He c a r r i e d out a s e r i e s of studies which were designed to test the f a c t o r i a l consistency of these f i v e f a c t o r s . The data for these studies were obtained p r i m a r i l y from groups of u n i v e r s i t y students, which var i e d i n the length and intimacy of t h e i r members' previous associations. Norman's r e s u l t s yielded c l e a r and consistent evidence for the existence of f i v e p ersonality factors which were r e l a t i v e l y orthogonal (independent) and which were e a s i l y interpreted. These personality factors along with a d e s c r i p t i o n of t h e i r peer nomination r a t i n g scales are shown i n Table 1. In more recent studies, these dimensions have reoccurred with s u r p r i s i n g r e g u l a r i t y and appear to be highly r e p l i c a b l e across s i t u a t i o n s and samples of people (Norman, 1969; Norman & Goldberg, 1966; P a s s i n i & Norman, 1966). Therefore, they are the dimensions of choice for a study of t r a i t concept formation. The f i v e personality factors were described as being r e l a t i v e l y orthogonal. The i m p l i c a t i o n i s that these factors are not completely independent of one another. Norman (1963) presented some data which in d i c a t e the degree to which these factors were independent i n two samples of subjects: f r a t e r n i t y groups and residence h a l l groups. For each of the two groups, Norman obtained " f a c t o r scores" by "summing each i n d i v i d u a l ' s r a t i n g scale scores for the four a p r i o r i s a l i e n t s on each of the f i v e f a c t o r s " (p. 580). Then within each sample, he correlated each p a i r of factors across the i n d i v i d u a l s who made up each sample. The c o r r e l a t i o n matrix for each sample i s given i n 31 Table 1 Norman's Personality Factors with Abbreviated Descriptions of t h e i r Peer Rating Scales Pole A Pole B Factor : — Scale Labels Scale Labels Extraversion - t a l k a t i v e - frank, open - adventurous - sociable Introversion - s i l e n t • - secretive - cautious - r e c l u s i v e II III IV Agreeableness - goodnatured - not jealous - mild, gentle - cooperative Conscientiousness - fussy, t i d y - responsible - scrupulous - perservering Emotional S t a b i l i t y - poised - calm - composed - not hypochondriacal Culturedness - a r t i s t i c a l l y s e n s i t i v e - i n t e l l e c t u a l - polished, refine d - imaginative Disagreeableness - i r r i t a b l e - jealous - headstrong - n e g a t i v i s t i c Unconscienfeiousness - careless - undependable - unscrupulous - q u i t t i n g , f i c k l e Emotional I n s t a b i l i t y - nervous, tense - anxious - e x c i t a b l e - hypochondriacal Unculturedness - a r t i s t i c a l l y i n s e n s i t i v e - u n r e f l e c t i v e , narrow - crude, boorish - simple, d i r e c t Source: Adapted from Norman, 1963, p. 577. 32 Table 2. Note that the c o r r e l a t i o n s between Factors II and I I I , II and IV, and III and V are r e l a t i v e l y high. According to Norman (1963), these r e s u l t s showed that "some degree of o b l i q u i t y may exist among c e r t a i n p a i r s of the dimensions of personality f o r which these factor measures have been developed" (p. 581). In t h i s i n v e s t i g a t i o n , only the f i r s t four f actors which were found by Norman (1963) are used. These factors have the cl e a r e s t structure. That i s , the scales show high loadings on factors they are supposed to mark and low loadings on the independent factors they are not supposed to mark (see Norman, 1963, p. 579, Table 2). Method of Obtaining the V a l i d i t y of the Behavioral Information Up to t h i s point, I have been t a l k i n g about behavioral information of d i f f e r e n t i a l v a l i d i t y without i n d i c a t i n g how that v a l i d i t y i s established. Here I review the area of research from which the method of obtaining v a l i d i t y indices was borrowed. Argyle and L i t t l e (1972) have ra i s e d the question: "Do t r a i t s apply to s o c i a l behavior?" In a re l a t e d research area, Fishbein and Ajzen (1972, 1974) have ra i s e d a s i m i l a r question with regard to the attitude-behavior r e l a t i o n s h i p . These authors reviewed the research which has t r i e d to f i n d such a r e l a t i o n s h i p and they noted the inconsistency of the evidence. Although several reasons can be presented to explain the low r e l a t i o n s h i p , Fishbein and Ajzen 33 Table 2 a "Factor Score" I n t e r c o r r e l a t i o n s Based on the Peer Nomination Rating Scale Data From Two Samples Factor Factor I II I II IV V FG Sample b I II .08 III -.27 .48 IV .20 .55 .10 V .02 .37 .63 .18 RH Sample 0 I II -.02 III -.49 .49 IV .22 .44 .05 V -.02 .31 .59 -19 3 A f t e r Norman, 1963, p. 580. k F r a t e r n i t y group Residence H a l l group 34 (1974) suggested a s h i f t from blaming d e f i n i t i o n s and measures of attitudes to an examination of the inadequacies of the behavioral c r i t e r i a . Fishbein and Ajzen reasoned that "although at t i t u d e s w i l l p redict multiple-act c r i t e r i a , there i s no reason to assume a re l a t i o n s h i p between a t t i t u d e and single-act c r i t e r i a " (p. 61). For these authors, the problem was to ensure that the behavioral items one sele c t s constitute " v a l i d behavioral c r i t e r i a " : that i s , some assurance must be given that the behaviors are re l a t e d to the a t t i t u d e under consideration i n some systematic manner. To get t h i s assurance, the researchers subjected the behavioral c r i t e r i a to the same kinds of psychometric considerations to which attitudes are usually subjected. As a s o l u t i o n to t h i s v a l i d i t y problem, Fishbein and Ajzen f i r s t considered the a p p l i c a t i o n of standard s c a l i n g procedures (Guttman, L i k e r t , Thurstone). They were quickly dissuaded because of t h e o r e t i c a l considerations such as the d i f f e r e n t d e f i n i t i o n s of v a l i d i t y assumed by d i f f e r e n t s c a l i n g procedures, and by p r a c t i c a l l i m i t a t i o n s of the methods i n circumventing these t h e o r e t i c a l d i f f i c u l t i e s . Instead, the authors devised an a l t e r n a t i v e procedure which i s based on the c a l c u l a t i o n of l i n e a r i t y and v a l i d i t y i n d i c e s . To c a l c u l a t e these i n d i c e s , several item properties must be obtained. These include: (a) the p r o b a b i l i t y of an a t t i t u d e given that the behavior i s performed, p(A|B); 35 (b) the p r o b a b i l i t y of an a t t i t u d e given the behavior was not performed, p(A|B); (c) the base rate of the behavior i n the s i t u a t i o n , p(B); (d) the p r o b a b i l i t y of someone who has the a t t i t u d e performing the behavior, p(B|A); and (e) the p r o b a b i l i t y of someone who has the a t t i t u d e not performing the behavior, p(B|A). The l i n e a r i t y index i s obtained by subtracting (e) from (d). The v a l i d i t y index i s obtained by subtracting (b) from (a), and (c) i s used to c a l c u l a t e (a) and (b) through the a p p l i c a t i o n of Bayes'theorem. In t h e i r study, Fishbein and Ajzen (1974) demonstrated that the l i n e a r i t y index i s p r e d i c t i v e of the attitude-behavior r e l a t i o n -ship and also that v a l i d i t y i s a necessary but not a s u f f i c i e n t condition for l i n e a r i t y . The l a t t e r i s based on the f i n d i n g that the performance or non-performance of " v a l i d " behaviors need not ne c e s s a r i l y be correlated with ( i . e . , l i n e a r l y r e l a t e d to) a person's a t t i t u d e . These findings should have at l e a s t t h e o r e t i c a l relevance to t r a i t constructs because the t r a i t - b e h a v i o r r e l a t i o n s h i p i s analogous to the attitude-behavior r e l a t i o n s h i p . As mentioned at the beginning of t h i s chapter, an e a r l i e r review (Mischel, 1968) f a i l e d to f i n d s u f f i c i e n t evidence for a systematic r e l a t i o n s h i p between per s o n a l i t y t r a i t s and behavior. More recently, Jaccard (1974) has extended Fishbein and Ajzen's method to the evaluation of t r a i t - b e h a v i o r r e l a t i o n s h i p s . He reasoned that " i f a 36 personality t r a i t i s assumed to mediate the o v e r a l l degree of (e.g.) dominance exhibited i n an i n d i v i d u a l ' s behavior but not any s p e c i f i c behavior per se, then a given personality measure should be highly related to a multiple-act c r i t e r i o n , but not n e c e s s a r i l y to a single-ac t c r i t e r i o n " (Jaccard, 1974, p. 361). In an attempt to f i n d multiple behaviors which were l i n e a r l y related to dominance, Jaccard had subjects generate and then rate (for the appropriate p r o b a b i l i t i e s ) behaviors which were performed by people who were considered to be dominant. From t h i s information, he calculated l i n e a r i t y and v a l i d i t y i n d i c e s . He f i r s t showed that there i s a s i g n i f i c a n t c o r r e l a t i o n between the l i n e a r i t y index and his traitr-behavior c o r r e l a t i o n s f o r a l l measures of dominance. Secondly, he showed that the v a l i d i t y index was a useful t o o l i n est a b l i s h i n g t r a i t - b e h a v i o r r e l a t i o n s h i p s . In sum, he found that the magnitude of v a l i d i t y c o e f f i c i e n t s i s improved considerably when one expresses the dominance-behavior r e l a t i o n s h i p as a c o r r e l a t i o n between a measure of dominance and a multiple-act c r i t e r i o n . In the present i n v e s t i g a t i o n , the v a l i d i t y index i s used to obtain behavioral information which i s d i f f e r e n t i a l l y suggestive of each pole of the f i r s t four personality factors which were found by Norman (1963). These behaviors are then used i n a t r a i t concept formation and change paradigm to study some parameters of these processes. Hypotheses Although the predictions i n t h i s study were discussed along with 37 l i t e r a t u r e which r e l a t e d to them, the s p e c i f i c hypotheses are outlined here. Hypothesis 1. T r a i t concepts about f i c t i t i o u s people are easier to form when t h e i r basis i s h i g h - v a l i d behavioral information than when t h e i r basis i s low-valid behavioral information. Hypothesis 2. There are two parts to Hypothesis 2: (a) i n i t i a l t r a i t concepts about f i c t i t i o u s people are easier to adjust ( i . e . , are les s r e s i s t a n t to change) when they were formed by reading low-valid behavioral information than" when they were formed by reading h i g h - v a l i d behavioral information; and (b) t r a i t concept adjustment i s easier when i t occurs while reading h i g h - v a l i d behavioral information than when i t occurs while reading low-valid behavioral information. Hypothesis 3. There are two parts to hypothesis 3: (a) when t r a i t concepts are adjusted to accommodate new information, the dimension which was reversed i s harder to learn than are both the unchanged and the new (control) dimensions; and (b) the unchanged.dimension i s easier to learn than i s the new dimension.. Hypothesis 4. Subjects' impressions of the f i c t i t i o u s people are i n accordance with the preset personality a t t r i b u t e s of those f i c t i t i o u s people. This hypothesis s p e c i f i e s the d i r e c t i o n that t r a i t a t t r i b u t i o n must take i n order f o r the experiment to be i n t e r n a l l y v a l i d . Hypothesis 5. Subjects' impressions of the f i c t i t i o u s people vary as a function of each condition i n the following manner: (a) 38 the impression on the reversed dimension (Condition 1) i s near the midpoint between the two poles of whatever t r a i t dimension happens to be i n that condition; (b) the impression on the un-changed dimension (Condition 2) i s the most extreme i n the d i r e c t i o n of the appropriate pole of whatever t r a i t dimension happens to be i n that condition; and (c) the impressions on the replaced and the new dimensions (Conditions 3 and 4 respectively) are toward the appropriate poles but not so extreme as the impression on the unchanged dimension. Hypothesis 6. T r a i t a t t r i b u t i o n v a r i e s as a function of the type of behavioral information subjects read, i n the following manner: (a) i f subjects adjusted t h e i r t r a i t concept by reading h i g h - v a l i d behavioral information, t r a i t a t t r i b u t i o n on the reversed dimension w i l l r e f l e c t a recency e f f e c t such that the impression on that dimension w i l l be i n the d i r e c t i o n of the pole p o s i t i o n s i n the p o s t - s h i f t part of the study; and (b) i f subjects adjusted t h e i r t r a i t concept by reading low-valid behavioral information, t r a i t a t t r i b u t i o n on the reversed dimension w i l l r e f l e c t a primacy e f f e c t such that the impression on t h i s dimension w i l l be i n the d i r e c t i o n which was suggested by the pole positions i n the pre-s h i f t part of the study. In the present i n v e s t i g a t i o n , the g e n e r a l i z a b i l i t y of any findings concerning the e f f e c t of the v a l i d i t y of behavioral information on t r a i t concept formation and change was tested by using several t r a i t concepts. The type of t r a i t concepts which were used was 39 determined in the following manner. First content groups of t r a i t concepts were obtained by arranging a l l possible combinations of the f i r s t four orthogonal personality factors (Norman 1963) which were taken three at a time. Secondly, within each content group, specific contextual arrangements of the attributes of two f i c t i t i o u s people were designed. Each contextual arrangement was a t r a i t concept. The t r a i t concepts which were used are specified in the next chapter. 4 0 Chapter 2 Method In t h i s study, there were two phases, each re q u i r i n g a s p e c i f i c method. The f i r s t phase was designed to obtain behavioral statements which were then used i n the second (experimental) phase. There were three parts to the f i r s t phase. F i r s t , behavioral s t a t e -ments were obtained by giving subjects a d e f i n i t i o n of various t r a i t terms and then asking them to record a s p e c i f i c a ction which was performed by someone they knew who was described by the t r a i t term i n question. In the second part of t h i s phase, two types of judgments from independent groups were obtained i n order that the v a l i d i t y of each behavioral statement could be assessed. In the t h i r d part, the behavioral statements were divided into four sets for use i n the experimental phase. There were three parts to the experimental phase: the a c q u i s i t i o n part, the r e v e r s a l part (sometimes re f e r r e d to as the adjustment p a r t ) , and the t r a i t a t t r i b u t i o n part. In the a c q u i s i t i o n part, subjects formed an i n i t i a l t r a i t concept of two f i c t i t i o u s people by reading statements which described the behavior of these people. In the second part, subjects were required to adjust t h e i r i n i t i a l t r a i t concept by reading d i f f e r e n t behavioral statements. An adjustment was required because one aspect of the i n i t i a l t r a i t concept was reversed, another aspect was l e f t unchanged and a t h i r d 41 was completely new. F i n a l l y , the t h i r d part was concerned with post-experimental t r a i t a t t r i b u t i o n . In t h i s part, subjects were asked to rate t h e i r o v e r a l l impression of the f i c t i t i o u s people by responding to a seven-place scale. A l l of the subjects who p a r t i c i p a t e d i n t h i s study were enrolled i n introductory psychology courses at The Uni v e r s i t y of B r i t i s h Columbia. Most were i n t h e i r f i r s t year of u n i v e r s i t y and the remainder were i n t h e i r second year. Secondly, a l l subjects were volunteers and were not paid f o r t h e i r p a r t i c i p a t i o n . F i n a l l y , a l l subjects were given a d e s c r i p t i o n of the task they would perform p r i o r to t h e i r p a r t i c i p a t i o n . Phase 1 There were three stages to phase 1: (a) the behavior generation stage; (b) the behavior r a t i n g stage; and (c) the behavior e l i m i n a t i o n stage. Each stage i s described i n turn. Stage 1: Generation of Behavior Statements Subjects. In t h i s stage, 185 subjects (82 female and 103 male subjects) were used. Information from the male subjects was not used i n the present study.-'- The ages of the female subjects ranged from 17 years to 25 years with a mean age of 18.4 years. Materials. Nine-page booklets which had a face sheet and eight randomly ordered pages were constructed. The face sheet 1 Even though they were c o l l e c t e d i n each r a t i n g procedure, the data from male subjects were not used because male subjects were not used i n the experimental part of the study. 42 contained three sections. The f i r s t section requested some demographic information (age. sex, and year of u n i v e r s i t y ) . The second section described the task. In essence, subjects were t o l d that the experimenter was interested i n t h e i r experiences with people who may be classed according to c e r t a i n t r a i t dimensions. The f i n a l section presented the i n s t r u c t i o n s . There were four parts to the i n s t r u c t i o n s . F i r s t , subjects were asked to read the t r a i t d e s c r i p t i o n which appeared at the top of each page. Secondly, they were t o l d to think of someone they knew quite w e l l who " f i t " the p a r t i c u l a r d e s c r i p t i o n they j u s t read. This person had to be the same sex and approximately the same age as the subjects were. T h i r d l y , subjects were asked to r e c a l l a s i t u a t i o n i n which that person had acted i n a way which was con-s i s t e n t with the d e s c r i p t i o n i n question and to describe, i n a sentence, the action and the s i t u a t i o n . F i n a l l y , subjects were asked not to include themselves as an example of the person to whom any of the t r a i t terms applied and were asked to work i n d i v i d u a l l y . At the top of each randomly ordered page, one of the following t r a i t d e s c r i p t i v e sentences was printed: (a) Extraverted people are t a l k a t i v e , frank, open, adventurous and sociable; (b) Introverted people are s i l e n t , s e c r e t i v e , cautious and re c l u s i v e ; (c) Agreeable people are good-natured, mild, gentle, cooperative and not prone to jealousy; 43 (d) Disagreeable people are i r r i t a b l e , headstrong, n e g a t i v i s t i c and prone to jealousy; (e) Conscientious people are fussy, t i d y , responsible, scrupulous and persevering; (f) Unconscientious people are careless, undependable, unscrupulous, q u i t t i n g and f i c k l e ; (g) Emotionally stable people.are, poised, cal.m, and composed; or (h) Emotionally unstable people are nervous, tense, anxious and excitable. These sentences were constructed using Norman's (1963) e x p l i c a t i v e adjectives which described four of the f i v e orthogonal t r a i t dimen-sions he found i n peer r a t i n g studies (see Table 1). Immediately following each sentence, a b r i e f version of the i n s t r u c t i o n s was printed. On each page, the i n s t r u c t i o n version was modified to., " f i t the p a r t i c u l a r t r a i t term which was described on the top of that page. Procedure. The data were c o l l e c t e d during one c l a s s period. At the beginning of the class period, one booklet was given to each subject. Subjects were asked not to page through the booklet u n t i l they were t o l d to do so. As soon as they were seated, subjects were asked to f i l l i n the demographic section. Then they were asked to read the remainder of the face sheet s i l e n t l y while the experimenter read i t aloud. When he was f i n i s h e d , the experimenter asked f o r questions and then t o l d subjects to begin. F i n a l l y , subjects were asked to bring t h e i r com-pleted booklet to the front of the room and to leave q u i e t l y . 44 On the basis of a p i l o t study, I expected to receive several types of statements. F i r s t , I expected that subjects would not express themselves i n one sentence or even i n complete sentences. Secondly, I expected that some subjects would not describe a s i t u a t i o n but would merely describe with d i f f e r e n t t r a i t adjectives the person about whom they were thinking. In order to obtain acceptable statements, I developed a procedure f o r evaluating the statements. This procedure consisted of three steps. The f i r s t step was a general evaluation i n which a l l statements or paragraphs were c l a s s i f i e d into one of three categories. The f i r s t category included errors i n responding to the task and omissions. A response was considered an error i f (a) a person of the opposite sex was described, (b) no s i t u a t i o n was given or (c) a s i t u a t i o n but not an action was given. The second category consisted of statements or basic ideas which were very s i m i l a r to previous statements. F i l i a l l y , the t h i r d category consisted of statements or paragraphs which were regarded as acceptable. The second step was a s i m p l i f y i n g task. Most of the p i l o t subjects expressed themselves i n uneconomical ways and used slang and other i d i o s y n c r a t i c forms of expression. To obtain the most economical, yet grammatically correct, form of expression, I developed a simple e d i t i n g procedure. F i r s t , where possible, the d e s c r i p t i o n of the s i t u a t i o n was placed i n an adjective clause at the beginning of the sentence. Secondly, the 45 subject of the action and the action i t s e l f were placed immediately following the adjective clause. T h i r d l y , i n so f a r as an alternate word could be found, slang words and i d i o s y n c r a t i c forms of expression were replaced. An example of a corrected d e s c r i p t i o n i s : "When she was cleaning her apartment, she pulled out every piece of fur n i t u r e and vacuumed the carpet which was beneath i t . " The t h i r d step was designed to reduce the t o t a l number of statements to 288 ( i . e . , 36 f o r each of the 8 t r a i t d e s c r i p t i o n s ) . More items than 288 would have made the r a t i n g task monstrous. The following guidelines for eliminating statements were developed: (a) omit statements which were e x t r a o r d i n a r i l y humorous; (b) omit statements which were very generalized (e.g., one p i l o t item read, "Whenever she works, she works very hard."); (c) omit statements with high moral overtones ( i . e . , statements which r e l a t e to f o r n i c a t i o n , pre-marital pregnancy, extra-marital a f f a i r s , abortion, e t c . ) ; (d) omit statements which suggested criminal records; (e) omit statements which suggested ^ brutally''- aggressive acts; and (f) omit statements which were, i n the opinion of the experimenter, strongly r e l a t e d to a dimension other than the one from which the statement had been generated. Results. In Table 3 are the number of responses i n each group for each of the eight t r a i t terms a f t e r the general evaluation. The number of items i n each category does not d i f f e r considerably 46 across the eight t r a i t terms. When the guidelines f o r eliminating statements were applied, several items were eliminated. S p e c i f i c a l l y , 8 statements were eliminated from the extraversion pool, 9 from the i n t r o v e r s i o n pool, 12 from the agreeable pool, 16 from the disagreeable pool, 12 from the conscientious pool, 5 from the unconscientious pool, 8 from the emotional s t a b i l i t y pool, and 9 from the emotional i n -s t a b i l i t y pool. There were 36 statements remaining f o r each t r a i t term except f o r the unconscientious t r a i t term. Two items were supplied f o r t h i s group. These items were: (a) Af t e r she spent some money, she forgot to record i t i n her cheque book and over-drew her account; and (b) When she had an appointment with her doctor, she ar r i v e d 20 minutes l a t e . Stage 2: The Rating Tasks In order to obtain a v a l i d i t y index f o r each item, I needed to obtain two types of judgments (Fishbein and Ajz ;en, 1974). The f i r s t was the p r o b a b i l i t y of the t r a i t given that the behavior was performed, p(T|B), and the second was the p r o b a b i l i t y of the t r a i t given that the same behavior was not performed, p(T|B). The difference between these two judgments i s the v a l i d i t y index. One a d d i t i o n a l item property, the base rate, was obtained i n a t h i r d judgment task. On the basis of t h i s r a t i n g , items which occur very r a r e l y or very frequently can be eliminated from the item pool. Table 3 C l a s s i f i c a t i o n of Responses to the Behavior Generation Task Category a Group I ' II III Extraversion 17 21 44 Introversion 25 12 45 Agreeable 21 13 48 Disagreeable 22 8 52 Conscientious 27 17 48 Unconscientious 30 13 39 Emotional S t a b i l i t y 30 8 44 Emotional I n s t a b i l i t y 29 8 45 The category groups are defined as follows: I i s errors and omissions; II i s duplicate ideas; and III i s acceptable statements. 48 The materials and the procedure f o r the r a t i n g tasks are very s i m i l a r . However, for c l a r i t y , each r a t i n g task i s described separately. Subjects f o r the f i r s t r a t i n g task. The subjects f o r t h i s task were 113 introductory psychology students (53 females and 60 males). Only the data from the female subjects were used i n t h i s study. The ages of the female subjects ranged from 17 to 26 years with a mean age of 18.6 years. Materials for the f i r s t r a t i n g task. Booklets which contained one face sheet and eight sections were constructed. The face sheet had three parts. The f i r s t part requested demographic information (age, sex, and year of u n i v e r s i t y ) . The second part gave a b r i e f introduction to the task. In short, subjects were t o l d that the object of the study was to learn how people c l a s s i f y other people who have performed c e r t a i n actions i n c e r t a i n s i t u a t i o n s . F i n a l l y the t h i r d part presented the i n s t r u c t i o n s . There were three parts to the i n s t r u c t i o n s . The f i r s t part described the booklet i n d e t a i l . In p a r t i c u l a r the p o s i t i o n of the t r a i t d e s c r i p t i v e sentences, the arrangement of the behavioral items and the type of scale f o r each item were described. Secondly, subjects were instructed to read the t r a i t d e s c r i p t i o n which appeared at the beginning of each section. Then they were asked to read each behavioral statement which followed the t r a i t d e s c r i p t i o n separately and to estimate how l i k e l y (probable) i t i s that the t r a i t d e s c r i p t i o n described a female college .student who 4 9 was approximately the same age as they were and who has performed the p a r t i c u l a r behavior i n question. F i n a l l y , subjects were asked to give t h e i r estimate by marking a l i n e perpendicular to the scale which followed each behavioral statement. The t h i r d part of the i n s t r u c t i o n s s p e c i f i e d the performance that was expected. Subjects were t o l d that they were not l i k e l y to have the same booklet as t h e i r neighbor and therefore that they should work i n d i v i d u a l l y . In addition, they were t o l d to work as quickly as possible and not to spend too much time with any one item. F i n a l l y , subjects were t o l d to keep i n mind that the only information they had about each college student was that she had performed the p a r t i c u l a r behavior i n question. The remaining pages of the booklet were divided i n t o eight sections, one for each of the eight t r a i t d e s c r i p t i v e sentences. These sentences were the same as those which were used f o r the behavior generation procedure, and were printed at the top of the f i r s t page i n each section. Following t h i s sentence was a b r i e f i n s t r u c t i o n statement which was adapted from the general i n s t r u c t i o n statement on the face sheet to f i t the p a r t i c u l a r t r a i t d e s c r i p t i o n sentence i n question. F i n a l l y , on the remainder of the f i r s t page and on the second page were 12 behavioural statements each of which was followed by a ra t i n g scale. The scale was four inches i n length and ranged from 0 (unlikely) to 100 ( l i k e l y ) with 50 as the midpoint. As an example, the extraversion section i s described. A sentence which described an extraverted person appeared at the top 50 of the f i r s t page of t h i s section. This sentence was followed by the open question: "How l i k e l y i s i t that the t r a i t term, extraverted, describes a female college student who:" which i n turn was followed by 12 behavioral statements with t h e i r corresponding scales. Three groups of booklets were constructed. The groups d i f f e r e d only i n terms of the behavior statements which appeared i n each section. Each group of booklets had a d i f f e r e n t set of 12 behaviors for each section, but together the three booklets contained the 36 statements which were obtained i n the f i r s t stage f o r each t r a i t term. The purpose of the three booklets was to gauge the number of ratings each subject would make i n a 50 minute period. I t was decided not to have any subject rate more than 96 statements (12 fo r each of the 8 t r a i t terms) i n one class period. The behavioral statements were randomly assigned to booklets and were randomly l i s t e d below the appropriate t r a i t d e s c r i p t i v e sentence. In each booklet, the sections were randomly ordered. F i n a l l y , the booklets were stacked such that a booklet from each group appeared a l t e r n a t e l y throughout the stack. Procedure f o r the f i r s t r a t i n g task. The data were c o l l e c t e d during one class period. At the beginning of the period, subjects were asked to come to the front of the room and to pick up the booklet which appeared on the top of the stack. Female subjects came to the r i g h t side of the table and males came to the l e f t side. Subjects were asked not to page through the booklet u n t i l 51 instructed to do so. When they were seated, subjects were asked to f i l l i n the demographic section of the face sheet. Then they were t o l d to read the rest of the face sheet s i l e n t l y while the experimenter read i t aloud. When he was f i n i s h e d , the experimenter paused f o r questions and then t o l d the subjects to begin. F i n a l l y , subjects were asked to bring t h e i r completed booklet to the front of the room and to leave q u i e t l y . The booklets were scored by measuring the distance of the perpendicular l i n e s from the zero-point of the scale. The measuring u n i t , tenths of an inch, was chosen because i t was easy to convert into a p r o b a b i l i t y measure. The p(T|B) was determined f o r each item by computing a mean score f o r that item. The number of observations which entered into the c a l c u l a t i o n of the mean d i f f e r e d because, most subjects did not respond to every item. For most-items, there were 17 observations. Subjects f o r the second r a t i n g task. The subjects f o r t h i s r a t i n g task were 63 introductory psychology students (30 female and 33 males). The ages of the female subjects ranged from 17 to 24 years with a mean age of 18.3 years. The ratings by one subject were not used because<she f a i l e d to follow the in s t r u c t i o n s f o r t h i s very d i f f i c u l t task. -Materials f o r the second r a t i n g task. Booklets which were almost i d e n t i c a l to those used i n the f i r s t r a t i n g task were constructed. The only d i f f e r e n c e between these booklets and the 52 ones f o r the previous task was the i n s t r u c t i o n s which were converted to obtain the second p r o b a b i l i t y judgments. The s p e c i f i c changes were as follows. F i r s t , the second part of the face sheet t o l d subjects that the object of the study was to learn how people c l a s s i f y other people who have not performed c e r t a i n actions i n c e r t a i n s i t u a t i o n s . Secondly, the i n s t r u c t i o n s t o l d subjects to estimate how l i k e l y (probable) i t i s that the t r a i t d e s c r i p t i o n described a female college student who did not perform the p a r t i c u l a r behavior i n question. T h i r d l y , the i n s t r u c t i o n s asked subjects to keep i n mind that the only information they had about each college student was that she had not performed the p a r t i c u l a r behavior i n question. F i n a l l y , i n each section, the adapted open question read: "... describes a female college student who did not:" which was following by the appropriate behavior statements and t h e i r scales. S l i g h t modifications to the verb i n each behavior statement were required to make the statements grammatically correct. The booklets were constructed and stacked according to the randomization procedures which were used i n the f i r s t r a t i n g task. Procedure f o r the second r a t i n g task. The procedures f o r data c o l l e c t i o n and data scoring were exactly the same as they were f o r the f i r s t r a t i n g task. The average number of responses to each item was 10. Subjects for the t h i r d r a t i n g task. The subjects f o r t h i s r a t i n g task were 54 introductory psychology students (28 female and 26 53 males). The age range of the female subjects was 17 to 27 years with a mean of 18.7 years. Materials f o r the t h i r d r a t i n g task. Booklets which contained a face sheet and f i v e pages of behavioral statements were constructed. There were two parts to the face sheet. The f i r s t part requested demographic information (age, sex and year of u n i v e r s i t y ) . The second part presented the i n s t r u c t i o n s . F i r s t , the i n s t r u c t i o n s t o l d subjects to read each behavior statement i n turn. Secondly, subjects were asked to think of 100 t y p i c a l female college students who were approximately the same age as they were and to indic a t e how many have probably performed the p a r t i c u l a r behavior i n question. Subjects were asked to place t h e i r estimate i n the space which was provided./ The remaining pages of the booklet contained randomly ordered behavioral statements which were used i n the f i r s t two r a t i n g tasks. A f t e r each statement, a five-^space blank appeared. Three d i f f e r e n t groups of booklets were constructed by randomly assigning one of three behaviors i n each t r a i t term pool to respective booklets. The booklets were stacked i n the same manner as the booklets f o r the previous r a t i n g procedures were stacked. Procedure f o r the t h i r d r a t i n g task. The procedure f o r t h i s task was exactly the same as i t was f o r the two previous tasks. The scoring procedure was not so complex. For each item, a mean score was computed by averaging the responses of subjects to that 54 p a r t i c u l a r item. Because not a l l subjects responded to a l l items, there were a d i f f e r e n t number of observations for most items. Most items had nine observations. Stage 3: The E l i m i n a t i o n of Behavior Statements For each item i n the r a t i n g tasks, a v a l i d i t y index was c a l -culated by subtracting the second r a t i n g from the f i r s t one. For use i n the experimental phase, I needed 16 items (8 items i n each of 2 v a l i d i t y groups) f o r each t r a i t term. To obtain these items, I followed two steps. The f i r s t step was concerned with three rules f o r i n i t i a l i n c l u s i o n and the second, step was 'the choice of the f i n a l items. Three decision rules f o r the i n i t i a l i n c l u s i o n of behavior statements were developed. The f i r s t r u l e r e l a t e d to the f i r s t r a t i n g . I t stated that, i n order f o r an item to enter the second stage of evaluation, i t had to have a p(T|B) r a t i n g equal to or above ..5500y The r a t i o n a l e f o r t h i s rule was that, i n order to carry out an adequate test of the experimental hypotheses, I needed items which had at least a moderate p r o b a b i l i t y r e l a t i o n s h i p to the p a r t i c u l a r t r a i t term from which they had been generated. An item which was judged to have, l e s s than a 50% r e l a t i o n s h i p to a t r a i t term was considered to be inadequate. To ensure at least a moderate r e l a t i o n -ship, the lower cut-off point was set at .5500. The second r u l e r e l a t e d to the second r a t i n g . In order f o r i t to be placed i n the f i n a l item pool, an item had to have a 55 p(T|B) ra t i n g of l e s s than .4500V A r a t i n g higher than t h i s cu-off point was considered troublesome. C l e a r l y , any ra t i n g above t h i s l e v e l would suggest that both the performance of the behavior (as suggested by the behavior generator) and the non-performance of that same behavior (as indicated by the raters i n the second r a t i n g task) are rel a t e d to t h e . a t t r i b u t i o n of the t r a i t term i n question. The t h i r d rule was to eliminate any behavior which had a base rate less than .05 or more than 195.' It was- thought that any behavior with a base rate lower than .05 would have s p e c i a l s i g n i f i c a n c e because of i t s rare occurrence and that any ra t i n g with a base rate higher than iQS'would not give s u f f i c i e n t information to be us e f u l . The second step was the s e l e c t i o n of a f i n a l set of items from the pool of items which was obtained from the f i r s t step. For each t r a i t term group, I obtained the f i n a l set of items by plac i n g the eight items with the highest v a l i d i t y index i n the h i g h - v a l i d group and the eight items with the lowest v a l i d i t y index i n the low-valid group. Results. Because they did not meet the s p e c i f i c t i o n s of the f i r s t r u l e , 43 items were eliminated from fur t h e r consideration. Approximately f i v e items were eliminated from each of the t r a i t term pools. In ad d i t i o n . 42 items were eliminated because they f a i l e d the second c r i t e r i o n . Again, these items were evenly d i s -t r i b u t e d across the eight t r a i t term pools. One f i n a l item was eliminated because i t s base rate was below .05. 56 The f i n a l item pool i s l i s t e d i n Appendix A. The items are grouped according to t h e i r v a l i d i t y and according to the t r a i t term from which they were generated. In Table 4 are comparisons of the average v a l i d i t y index f o r each of the groups i n the f i n a l item pool. A few differences are noteworthy. F i r s t , i n the h i g h - v a l i d group, the v a l i d i t y indices f o r the agreeable and the conscientious pool are 10 points higher than t h e i r nearest competitor. In contrast, the v a l i d i t y index f o r the emotional i n s t a b i l i t y pool i s comparatively low. The remaining indices are f a i r l y close to one another. Secondly, i n the low-valid group, the v a l i d i t y index f o r emotional i n s t a b i l i t y i s comparatively low. The remaining indices are f a i r l y close to one another. Phase 2: The Experiment There were three.parts to the experimental phase of the present study. In the f i r s t part, the a c q u i s i t i o n part, subjects formed one of several t r a i t concepts about two f i c t i t i o u s people by reading behavioral information of eit h e r high or low v a l i d i t y . The second part of the experiment was the re v e r s a l part (sometimes c a l l e d the adjustment p a r t ) . In t h i s part, subjects adjusted t h e i r t r a i t concepts to accommodate new and contradictory information of e i t h e r high or low v a l i d i t y . The t h i r d part which immediately followed the second part was c a l l e d the t r a i t a t t r i b u t i o n part. Here, subjects gave t h e i r o v e r a l l impression of the two f i c t i t i o u s people by responding to a seven-place scale. 57 Table 4 Comparisons of the V a l i d i t y Indices f o r the Items i n the F i n a l Item Pool V a l i d i t y Index Item Groups Lowest Highest Mean High-valid Extraverted .4805 .6372 .5600 Introverted .4111 :6691 .4791 Agreeable .5697 . 7719 .6666 Disagreeable .4947 : 5,859 .5416 Conscientious .6000 .7700 .6691 Unconscientious .4766 16193 .5162 Emotionally Stable .5050 .6454 .5598 Emotionally Unstable .3150 :6906 .4132 i l i d Extraverted .2850 .4157 .3600 Introverted .2444 .3541 .3188 Agreeable .3271 -.4272 .3708 Disagreeable .2522 .4384 .3878 Conscientious .2746 .4125 .3263 Unconscientious .2675 .3841 .3326 Emotionally Stable .2913 .4005 .3406 Emotionally Unstable .1736 .2925 .2360 5 8 The design of the experiment and the experimental materials are very complex. To help s i m p l i f y the presentation of the methodology, i t i s useful to describe the experimental conditions at the beginning. For the experimental phase taken as a u n i t , there are four between-subject and three within-subject c l a s s i f i c a t i o n s . Not a l l of the c l a s s i f i c a t i o n s are used i n every data analysis but i t i s useful to know a l l c l a s s i f i c a t i o n s at the outset. F i r s t , the between-subject c l a s s i f i c a t i o n s are defined. The f i r s t c l a s s i f i c a t i o n was a c q u i s i t i o n groups. There are two a c q u i s i -t i o n groups: the h i g h - v a l i d a c q u i s i t i o n group and the low-valid  a c q u i s i t i o n group. The former group designates subjects who formed an i n i t i a l t r a i t concept of the f i c t i t i o u s people by reading high-v a l i d behaviors. The l a t t e r designates subjects who formed an i n i t i a l t r a i t concept by reading low-valid behaviors. • There are two r e v e r s a l groups: the h i g h - v a l i d r e v e r s a l group and the low-valid r e v e r s a l group. The former group consists of subjects who adjusted t h e i r i n i t i a l t r a i t concept to f i t a set of new h i g h - v a l i d behaviors. The l a t t e r group consists of subjects who adjusted t h e i r i n i t i a l t r a i t concept t o ' f i t a new set of low-valid behaviors. The r e v e r s a l groups are f a c t o r i a l with the a c q u i s i t i o n groups. A p r o f i l e i s defined as a p a r t i c u l a r combination of t r a i t dimensions which formed the content upon which t r a i t concept formation i s based. There are four possible combinations of four 59 t r a i t dimensions which were taken three at a time. Each combination represents a p r o f i l e group. Table 5 shows the t r a i t dimensions which define each of the four p r o f i l e s . A character i s the predetermined personality a t t r i b u t e s of both f i c t i t i o u s people whose behaviors were presented together. In the present study, the personality a t t r i b u t e s of the f i c t i t i o u s people are preset by giving each person opposite pos i t i o n s along each of the three dimensions which make up each p r o f i l e . The combination of positions along each dimension for the two people taken together i s c a l l e d a character. In the present study, the number of d i f f e r e n t characters was determined by applying the following r e s t r i c t i o n s to the combination of p o s i t i o n s along the t r a i t dimensions i n each p r o f i l e . The f i r s t r e s t r i c t i o n was that there can be only two possible p o s i t i o n s i n each t r a i t dimension: an upper p o s i t i o n and a lower p o s i t i o n . The second r e s t r i c t i o n was that there can be only two forms the combina-t i o n of positions along each dimension may take. The two forms followed the r u l e that the p o s i t i o n along two dimensions must d i f f e r from the p o s i t i o n along the t h i r d dimension i n each combina-t i o n . These forms were: (a) any two of the dimensions must have an upper p o s i t i o n while the t h i r d must have a lower p o s i t i o n ; or (b) any two of the dimensions must have a lower p o s i t i o n while the t h i r d must have an upper p o s i t i o n . Table 5 Combinations of T r a i t Dimensions f o r Each P r o f i l e Dimensions F i r s t Second Third P r o f i l e 1 E - I a A-Db C-UC P r o f i l e 2 E-I A-D ES-EI P r o f i l e 3 E-I C-U ES-EI P r o f i l e 4 A-D C-U ES-EI Extraversion-Introversion Agreeable-Disagreeable Conscient ious-Unconscient ious Emotional Stability-Emotional I n s t a b i l i t y 61 In each p r o f i l e there are three ways i n which these r e s t r i c t i o n s are met. Table 6 presents the range of characters f o r each p r o f i l e . Note that within each p r o f i l e , each dimension gets a chance to be on the pole opposite to two other dimensions. Secondly, note that each character has two parts: (a) the combination of dimension positions which defines Jane; and (b) the combination which defines S a l l y . F i n a l l y , note that i n each character, Jane i s opposite to S a l l y on each dimension and that the character groups are nested within the p r o f i l e groups .' As a c o n t r o l , one a d d i t i o n a l between subject c l a s s i f i c a t i o n was formed but was not used i n any of the analyses. Two groups were formed within each character to counterbalance the names, Jane and S a l l y , with the t r a i t a t t r i b u t e s which appeared i n each character. This procedure guarded against possible e f f e c t s due to the asso c i a t i o n of a p a r t i c u l a r name with a p a r t i c u l a r group of personality a t t r i b u t e s . The f i r s t within-subject c l a s s i f i c a t i o n appeared i n a l l three parts of the experimental phase. It i s c a l l e d the dimension v a r i a b l e and refers to the three t r a i t dimenions which appear i n each p r o f i l e . We w i l l see i n the next chapter that because each p r o f i l e contains a d i f f e r e n t set of three dimensions, the dimensions were treated as nested within p r o f i l e s . The second within-subject c l a s s i f i c a t i o n appeared only i n the revers a l and the t r a i t a t t r i b u t i o n parts of the experiment. However, i t was d i f f e r e n t i n each part. This c l a s s i f i c a t i o n i s the condition 62 Table 6 Composition of the Characters A t t r i b u t e P r o f i l e Character F i c t i t i o u s Person : — F i r s t Second Third 1 1 Jane E A U S a l l y I D C 1 2 Jane C E D S a l l y u I A 1 3 Jane A C I S a l l y D u E 2 4 Jane E A EU S a l l y I D ES 2 5 Jane ES E D S a l l y EU I A 2 6 Jane A ES I S a l l y D EU E 3 •7 Jane E C EU S a l l y I u ES 3 8 Jane ES E U S a l l y EU I C 3 9 Jane C ES I S a l l y U EI E 4 10 Jane A C EU S a l l y D u ES 4 11 Jane ES A U S a l l y EU D C 4 12 Jane C EI D S a l l y u EU A 3 The l e t t e r s represent the following a t t r i b u t e s : E i s extraverted; I i s introverted; A i s agreeable; D i s disagreeable; C i s conscient U i s unconscientious; ES i s emotional stable; and EU i s emotionally unstable 63 v a r i a b l e . The d i v i s i o n s of t h i s v a r i a b l e r e f l e c t e d the changes i n the t h i r d and fourth set of behaviors which required t r a i t concept adjustment. There are four conditions. The f i r s t condition, the reversed condition, i s the dimension ( i n each character) on which the posit i o n s of Jane and S a l l y were reversed. The second condition, the unchanged condition, i s the dimension ( i n each character) on which the posit i o n s of Jane and S a l l y remained the same as they were during the a c q u i s i t i o n part of the experiment. The t h i r d condition was changed i n the r e v e r s a l and a t t r i b u t i o n parts of the experiment. In the r e v e r s a l part, Condition 3 i s the new dimension or the con t r o l condition. I t i s the dimension which had not appeared during the a c q u i s i t i o n part of the experiment and therefore can be used as a frame of reference f o r gauging t r a i t concept change on the other two dimensions. In the t r a i t a t t r i b u t i o n part of the experiment, t h i s control condition i s c a l l e d the fourth condition. Here the t h i r d condition i s c a l l e d the replaced condition because i t appeared during the a c q u i s i t i o n part but not i n the rev e r s a l part of the experiment. Table 7 outlines the dimensions which appeared i n each condition for each character. Note that each dimension appears i n each condition when a l l of the characters are taken into account. The f i n a l within-subject c l a s s i f i c a t i o n was c a l l e d the pole  v a r i a b l e . This v a r i a b l e r e f e r s to the two ends of each t r a i t dimension. There are two poles: the upper pole and the'lower pole. The upper 64 Table 7 The Dimensions i n Each Condition f o r Each character Condition Character 1 E - I d A-D C-U ES-EI P l e 2 C-U E-I A-D ES-EI 3 A-D C-U E-I ES-EI 4 E-I A-D ES-EI C-U P2 5 ES-EI E-I A-D C-U 6 A-D ES-EI E-I C-U 7 E-I C-U ES-EI A-D P3 8 ES-EI E-I C-U A-D 9 C-U ES-EI E-I A-D 10 A-D C-U ES-EI E-I P4 11 ES-EI A-D C-U E-I 12 C-U ES-EI A-D E-I The f i r s t condition i s the reversed dimension; the second i s the unchanged dimension; the t h i r d i s the replaced dimension; and the k fourth i s the new (control) dimension. This condition does not appear i n the r e v e r s a l part. ^ In the reve r s a l part, t h i s i s the t h i r d condition. E-I i s extraversion-introversion; A-D i s agreeable-disagreeable; C-U i s conscientious-unconscientious; and ES-EI i s emotional s t a b i l i t y -emotional i n s t a b i l i t y . e P i s p r o f i l e . 65 pole categorizes a p o s i t i o n on one or more of the following: extraversion, agreeableness, conscientiousness and emotional s t a b i l i t y . In contrast, the lower pole categorizes one or more of the following: i n t r o v e r s i o n , unconscientiousness, disagreeable, and emotional i n s t a b i l i t y . The names of the poles r e f l e c t ' the scores which were assigned to the d i v i s i o n s on the seven-place scale. Recall that a score of 1 was assigned to the space nearest the lower pole and 7 was nearest the upper pole. Method Subjects. The subjects f o r the experimental phase were 96 female introductory psychology students. A l l subjects were drawn from laii introductory psychology subject pool and were s o l i c i t e d by a telephone conversation. Their ages ranged from 17 to 24 years with a mean age of 18.4 years. Two subjects did not respond to the task c o r r e c t l y and therefore were replaced. Experimental materials. The experimental materials were 96 decks of index cards, 96 corresponding data contingency sheets, 96 question sheets, 96 t r a i t a t t r i b u t i o n scales, four sample index cards and one clock. The materials are described i n turn. The procedure for assembling the 96 decks of index cards was very complex. Five steps were followed. F i r s t , 72 copies of each item i n Appendix A were made using a S p i r i t Duplicator. As many items as possible were typed i n a 2 x 4 inch space on the S p i r i t Masters. A f t e r the S p i r i t Masters were run on the duplicator, the items were separated by a paper cutter and were pasted on 3 x 5 66 p l a i n index cards such that one item appeared on each card. The cards were sorted into the same groups that appear i n Appendix A. The second step was the separation of the 16 groups (8 t r a i t term groups i n each of 2 v a l i d i t y groups) into four behavior sets. These sets were obtained as follows. F i r s t , from each of the 16 groups, I randomly selected four of the eight index cards. Of these, the f i r s t three cards which were selected formed Behavior Set 1 and the remaining index card ;'in each group formed Behavior Set 2. Secondly, of the four index cards which remained i n each of the 16 groups, three were chosen, to form Behavior Set 3. The fourth card i n each group formed Behavior Set 4. Note that there are 16 groups of index cards i n each behavior set (8 t r a i t term groups i n each of 2 v a l i d i t y groups). Behavior Sets 1 and 2 were to be used i n the a c q u i s i t i o n part of the experiment and Behavior Sets 3 and 4 were to be used i n the r e v e r s a l part of the experiment. The t h i r d step had two parts. The f i r s t part concerned Behavior Sets 1 and 2 and the second part applied to Behavior Sets 3 and 4. To f a c i l i t a t e the completion of t h i s step, I divided each behavior set into two groups according to the v a l i d i t y of the items on the index cards. Remember that i n each of the 16 groups i n Behavior Sets 1 and 3 there were three index cards each of which had 72 copies. In Behavior Sets 2 and 4, there was only one index card i n each of the 16 groups and that card was duplicated 72 times. Since i t was the same f o r both v a l i d i t y groups, the procedure i n step 67 three i s described only f o r the high v a l i d i t y group. F i r s t , the procedure f o r Behavior Sets 1 and 2 i s described. There were four parts to t h i s procedure. F i r s t , index cards which suggested the a t t r i b u t e s of Jane and S a l l y f o r each character i n Table 6 were chosen and stacked into 12 groups (one f o r each character). Secondly, the name, Jane, was stamped on the back of the index cards which suggested Jane's a t t r i b u t e s and the name, S a l l y , was stamped oh those which suggested her a t t r i b u t e s . T h i r d l y , because the names, Jane and S a l l y , were counterbalanced, 12 a d d i t i o n a l groups of index cards were stamped such that the a t t r i b u t e s of Jane and S a l l y were completely reversed. F i n a l l y , a duplicate copy of each of the 24 groups (2 counterbalanced orders f or each of 12 characters) was made. The same procedure was used to obtain 48 groups of low-valid index cards. The procedure f o r Behavior Sets 3 and 4 was more complex. Because they were to be used i n the rever s a l part of the experiment, these behavior sets had to r e f l e c t the changes i n the a t t r i b u t e s of both Jane and S a l l y . These changes are s p e c i f i e d by combining the information i n Tables 6 and 7. To si m p l i f y t h i s information, I designed a chart which recorded the adjustments to the a t t r i b u t e s f o r each character. There were four parts to the procedure f or Behavior Sets 3 and 4. F i r s t , index cards which suggested the adjusted a t t r i b u t e s of Jane and S a l l y f o r each character were chosen and stacked into 12 groups (one group f or each character). Secondly, the names, Jane 68 or S a l l y , were stamped on the back of the index cards which suggested the a t t r i b u t e s of each f i c t i t i o u s person r e s p e c t i v e l y . T h i r d l y , a counterbalanced order was stamped exactly opposite to the f i r s t order. F i n a l l y , a duplicate copy was made of each of the 24 groups. The same procedure was followed to obtain 48 groups of low-valid index cards. When the t h i r d step was completed, each behavior set had 96 components. For each behavior set, the components were 2 groups of hig h - v a l i d index cards which were separated into 2 counterbalanced person a t t r i b u t e arrangements ( S a l l y vs. Jane) and 2 groups of low-v a l i d index cards which were seperated i n the same manner. The fourth step was the construction of the decks by combining index cards from each of the four behavior sets. There were four d i v i s i o n s i n each deck: the f i r s t d i v i s i o n consisted of index cards from Behavior Set 1: the second d i v i s i o n consisted of index cards from Behavior Set 2; the t h i r d d i v i s i o n consisted of those from Behavior Set 3; and, f i n a l l y , the fourth d i v i s i o n consisted of those from Behavior Set 4. Each deck was made up by combining the four successive d i v i s i o n s of index cards which suggested the same order of the same character. There were four groups of decks with 24 decks i n each group. These 24 decks were two counterbalanced orders of each of the 12 characters. The f i r s t group had hi g h - v a l i d index cards i n each of the four d i v i s i o n s . The second group had high - v a l i d index cards i n the f i r s t two d i v i s i o n s (Behavior Sets 1 and 2) but low-valid index 69 cards i n the f i n a l two d i v i s i o n s (Behavior Sets 3 and 4). The t h i r d group had low-valid index cards i n the f i r s t two d i v i s i o n s , but hi g h - v a l i d index cards i n the f i n a l two d i v i s i o n s . F i n a l l y , the fourth group had low-valid index cards i n a l l of the d i v i s i o n s . In a l l , there were 96 decks, one f o r each subject. One f i n a l step remained to be completed. In each of the four d i v i s i o n s i n each deck, the index cards were randomly ordered with the following r e s t r i c t i o n s which apply to the sequence of the names, Jane and S a l l y . F i r s t either name could not appear i n the deck more than three times i n a row. Secondly, f o r each name a sequence of three names which were the same (either Jane or Sal l y ) could not appear more than once throughout any d i v i s i o n of the deck. T h i r d l y , a pattern of two or more names could not be repeated more than once i n a row. For example, the pattern, Jane S a l l y Jane S a l l y Jane S a l l y , was not allowed because the pattern of two names was repeated twice. Fourthly, the f i r s t card i n the t h i r d d i v i s i o n of each deck was always a randomly selected card from the reversed dimension. For a l l subjects, t h i s r e s t r i c t i o n ensured that the o r i g i n a l l y correct response was negatively reinforced at the same time. F i n a l l y , the d i v i s i o n s i n each deck were separated by an orange-colored index card which f a c i l i t a t e d the use of the d i v i s i o n s one at a time. The construction of 96 data contingency sheets was not as d i f f i c u l t . These sheets were made by l i s t i n g the appropriate order of names i n each d i v i s i o n of each deck along the l e f t margin of the page and by making columns to record the errors i n each t r i a l . For 70 each d i v i s i o n , one t r i a l consisted of giving a response to each of the index cards which appeared i n that d i v i s i o n . A h o r i z o n t a l l i n e was placed i n the appropriate space i f the subject made an error and a v e r t i c a l l i n e was made i f the subject was correct. A l l v e r t i c a l l i n e s i n one t r i a l indicated that the c r i t e r i o n had been reached. Each deck and i t s corresponding data contingency sheet were given a number from one to 96. Then, according to a table of random numbers, the index cards and data contingency sheets were randomly ordered. The cards and sheets were renumbered and stacked consecutively. In the order that they came to the experiment, subjects were given successive decks beginning at the top of the p i l e . The question sheets were made by typing two questions on a S p i r i t Master and then making 96 copies on the S p i r i t Duplicator. The two questions were: (a) How did you figu r e out who performed each action; and (b) Did you notice any changes i n eit h e r S a l l y ' s or Jane's behavior from an ea r l y to a l a t e r part of the experiment? The f i r s t question was designed to get some idea of how subjects were responding to the task. I f they . t r i e d to memorize or to use inappropriate devices (associating c e r t a i n l e t t e r s to c e r t a i n names), subjects were eliminated. One subject was eliminated because she t r i e d to associate the l e t t e r t with S a l l y . The second question was designed to determine the type of s h i f t subjects noticed i n the decks. 71 The 96 scales were made i n the following manner. Each of the four t r a i t dimensions were separated h o r i z o n t a l l y by seven evenly-spaced blanks and were placed on separate pages. There were two sets of these four pages; one was to be used f or Jane and the other fo r S a l l y . Within each scale, the order i n which the t r a i t dimensions appeared was randomized. Secondly, which scale, Jane's or S a l l y ' s , came f i r s t was randomly determined. Four sample index cards were made to give subjects p r a c t i c e at the task. Two of the cards had the name, Mary, stamped on the back and two had the name, Sue, on them. The items which were used on the sample index cards were taken from the items which were l e f t over when the eight highest items and the eight lowest items were picked from the following t r a i t term pools: extraversion, agreeable-ness, conscientiousness and emotional s t a b i l i t y . The items which were associated with Mary were: (a) When a customer entered the store, she remembered h i s name and the subject of t h e i r previous conversation; and (b) When harrassed by a customer over a bank error, she t r i e d to explain the error, but having f a i l e d she immediately r e f e r r e d that customer to the manager. The items which were associated with Sue were: (a) Even though she was troubled with her own problems, she responded to another's request for help with regard to the l a t t e r ' s personal problems; and 72 (b) When asked to borrow her assignment, she r e p l i e d that her fr i e n d should t r y to do as much as possible by h e r s e l f before copying someone else's work. F i n a l l y , an ordinary desk clock was used to keep track of the exposure time of each index card., Procedure. The subjects were tested i n d i v i d u a l l y . As soon as they were seated comfortably, subjects were shown a deck of index cards and were t o l d that, on each index card, statements which described c e r t a i n actions i n s p e c i f i c s i t u a t i o n s were printed. They were t o l d that by design, each action had been performed by one of two people, S a l l y or Jane, and that t h e i r task was to f i g u r e out who performed each action. They were t o l d to do t h i s i n the following manner. F i r s t , they were to read the statement which appeared on the front of the index card. Then they were to give t h e i r best guess as to who performed the action which was printed on that card. F i n a l l y , they were to turn the card over to view the correct response. In addition, they were t o l d to give a response within a 7-second duration. I f 7 seconds passed before they gave a response, the experimenter would turn the card over; they could view the correct response and then move on to the next card. F i n a l l y , the subjects were t o l d that the deck of index cards was assembled into four d i v i s i o n s and that they would go through each d i v i s i o n i n order u n t i l t h e i r response matched the name on the back of each card i n the d i v i s i o n . 73 To make sure that the subjects understood the procedure, the experimenter gave them the sample cards. A f t e r the subjects went through the sample cards, the experimenter asked for questions. Then he gave the subjects the f i r s t d i v i s i o n of the deck. The subjects were reminded that the actions i n the d i v i s i o n had been performed by ei t h e r S a l l y or Jane and that t h e i r task was to fi g u r e out who performed each action. F i n a l l y , subjects were encouraged by the experimenter who commented that the task w i l l be d i f f i c u l t at f i r s t but that i t should get easier as they progressed. As soon as the subjects attended to each index card, the experimenter glanced at the second-hand on the clock. If the subjects took longer than 7 seconds, the experimenter turned the card over, recorded an error and began to time the next response. Only r a r e l y did the experimenter have to turn a card over. A f t e r the subject had gone through the d i v i s i o n of the deck, the experimenter turned the d i v i s i o n of the deck face upward and asked the subject to go through the deck again. When the c r i t e r i o n of a correct response to each card i n the e n t i r e t r i a l had been reached, the experimenter announced that he would l i k e the subject to view some new action statements and gave the subject the second d i v i s i o n . This procedure was repeated with the t h i r d and fourth d i v i s i o n s . When the c r i t e r i o n had been reached i n the fourth d i v i s i o n , the subjects were asked the two questions which were printed on the question sheet. Subjects were asked to explain t h e i r vague responses and were encouraged to remember as much as possible. Secondly, the 74 subjects were asked to give t h e i r o v e r a l l impression of S a l l y and Jane by responding to a seven-place scale f or each f i c t i t i o u s person. When subjects had f i n i s h e d , they were given an opportunity to ask any questions that they wished. Just before they l e f t , they were thanked and were asked not to discuss the experiment with anyone for a month. From beginning to end, the experiment took from 20 to 50 minutes with most people f i n i s h i n g i n 40 minutes. The performance of two subjects was rejected because they t r i e d to memorize the sequence of Jane vs. S a l l y , and data from new subjects were c o l l e c t e d for the appropriate decks. Dependent Measures The learning and the a t t r i b u t i o n data were scored according to several dependent measures. The dependent measures are described according to the part of the experiment to which they apply. F i r s t , the measures for the a c q u i s i t i o n part of the experiment were t r i a l s to c r i t e r i o n and errors to c r i t e r i o n i n each of Behavior Sets 1 and 2. These dependent measures are somewhat redundant but both were used because d i f f e r e n t information was obtained with each measure. Secondly, the dependent measures f o r the r e v e r s a l part of the experiment were d i f f e r e n t i n Behavior Sets 3 and 4. For Behavior Set 3, the dependent measures were t r i a l s to c r i t e r i o n , errors within a f i x e d number of t r i a l s , and errors on the f i r s t r e v e r s a l t r i a l . The errors i n the f i r s t t r i a l gave some i n d i c a t i o n of the i n i t i a l e f f e c t of the r e v e r s a l part of the experi-75 ment. Secondly, the errors In each condition within a f i x e d number of t r i a l s guarded against biases due to p o s t - c r i t e r i a l drop i n performance. F i n a l l y , the t r i a l s to c r i t e r i o n measure gave some i n d i c a t i o n of the o v e r a l l e f f e c t of the r e v e r s a l procedure. The a t t r i b u t i o n data from the scales were scored by giving the place nearest the lower pole of each dimension a score of 1 and the place nearest the upper pole a score of 7. The middle spaces were numbered o r d i n a l l y . For the purpose of analysis, the impressions of Jane and S a l l y were combined and scored according to the preset a t t r i b u t e s of these f i c t i t i o u s people. That i s , i f the reversed dimension was agreeableness-disagreeableness, and i f S a l l y ' s a t t r i b u t e was preset,, to be agreeable, the score f o r S a l l y (e.g., 6) would be placed i n the upper pole of that dimension (agreeableness). On. the other hand, i f S a l l y ' s a t t r i b u t e was preset to be disagreeable, the score for S a l l y (6) would be placed i n the lower pole (disagreeable). In t h i s way one can determine whether subjects' impressions were i n the d i r e c t i o n which was suggested by the behaviors they read. 76 Chapter 3 Results Three sets of data were analyzed separately: (1) learning data: a c q u i s i t i o n part of the experiment; (2) learning data: r e v e r s a l part of the experiment; and (3) the a t t r i b u t i o n data. A l l of the analyses except those on the 'latinized' v a r i a b l e were done using the 2 computer package: UBC Anova. Learning Data: A c q u i s i t i o n The learning data i n the a c q u i s i t i o n part of the experiment were subdivided into two groups: data from Behavior Set 1 and data from Behavior Set 2. There were two dependent measures f o r each behavior set: t r i a l s to c r i t e r i o n and errors to c r i t e r i o n . A separate design was used to analyze each measure. Behavior Set 1: Design 1 A 2 ( a c q u i s i t i o n groups) x 4 ( p r o f i l e s ) x 3 (characters) between-subjects analysis of variance was performed on the number of t r i a l s to c r i t e r i o n i n the f i r s t behavior set. Table 8 presents the design of t h i s analysis of variance. Notice that the t h i r d f a c t o r i s nested within the second f a c t o r . In Table 9 i s a summary of the analysis of variance on t r i a l s to c r i t e r i o n i n Behavior Set 1. In opposition to the f i r s t hypothesis, the h i g h - v a l i d a c q u i s i t i o n group required more t r i a l s to reach Malcolm Greig, University of B r i t i s h Columbia adapted UBC Anova from a Brigham Young Uni v e r s i t y Documentation i n October, 1974. 77 Table 8 Analysis of Variance: Design 1 Factors A l Bl B2 B3 B4 Cl C2 C3 C4 C5 C6 C7 C8 C9 CIO C l l C12 Dl-4 A l Bl B2 B3 B4 Cl C2 C3 C4 C5 C6 C7 C8 C9 cio C l l C12 93-96 The factors are as follows: A i s a c q u i s i t i o n groups; B i s p r o f i l e groups; C i s character groups; and D i s subjects. Note that C i s nested within B. 78 Table 9 Summary of the Analysis of Variance on T r i a l s to C r i t e r i o n i n Behavior Set 1 Source 3 D.F. S.S. M.S. F P Tot a l 95 2366.958 A 1 15.041 .15-041 0.649 0.428 B 3 68.375 22.791 0.983 0.406 C/B 8 488.833 61.104 2.637 0.013 AB 3 45.208 15.069 6.650 0.589 AC/B 8 81.500 10.187 0.439 0.893 SS/AC/B 72 1668.000 23.166 aA = a c q u i s i t i o n group, B = p r o f i l e , C = character, and SS = subjects c r i t e r i o n than did the low-valid a c q u i s i t i o n group, but t h i s d i f f e r e n c e does not reach an acceptable l e v e l of s i g n i f i c a n c e , F_ (1, 72) = .6490; p = .428.';. In order, the a c q u i s i t i o n group means are 9.00 and 8.20. The difference between the two groups i s very small. The performance of subjects i n P r o f i l e s 1 through 4 does not d i f f e r r e l i a b l y , F (3,72)= .983-'; p = .406 . The mean t r i a l s to c r i t e r i o n i n P r o f i l e s 1 to 4 are 8.833; 7.458, 8.333 and 9.792 respectively. There i s a s i g n i f i c a n t difference among the characters^C/B) F (8,72) = 2.637 ; p = .013'. Table 10 presents the character 3 means within each p r o f i l e . The Newman Keuls test was used to make a l l possible pairwise comparisons among the character means. Only one difference i s r e l i a b l e . In P r o f i l e 1, subjects i n Character 3 required s i g n i f i c a n t l y fewer t r i a l s to reach the c r i t e r i o n than d i d subjects i n Characters 1 and 2. Looking back at Table 6, we see the content of each character. In Character 1, Jane's a t t r i b u t e s are extraverted, agreeable and unconscientious whereas S a l l y ' s are j u s t the opposite. In Character 2, Jane's a t t r i b u t e s are conscientious, extraverted and disagreeable whereas S a l l y ' s are j u s t the opposite. F i n a l l y , i n Character 3, Jane's a t t r i b u t e s are agreeable, conscientious and introverted whereas S a l l y ' s are the opposite. Unless indicated otherwise, the l e v e l of s i g n i f i c a n c e f o r a l l Newman Keuls comparisons was set at p <.. 05. Table 10 Mean T r i a l s to C r i t e r i o n f o r the Characters i n Behavior Set 1 Factor a 2 Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 11.375 Character 2 10.750 Character 3 4.375 Character 4 8.500 Character 5 "7.250 Character 6 • 6.625 Character 7 11.-250 Character 8 81750 Character 9 5.875 Character 10 6.750 Character 11 11.875 Character 12 10.750 a The number of observations f o r each mean was 8. 8 1 There i s one feature which both Characters 1 and 2 share i n common, but which i s not found i n Character 3. This feature i s the contextual arrangement of the t r a i t dimensions, agreeableness-disagreeableness (A-D) and conscientiousness-unconscientiousness (C-U). In Characters 1 and 2, the a t t r i b u t e s of the f i c t i t i o u s people are such that an upper p o s i t i o n on e i t h e r A-D or C-U e x i s t s i n a contextual arrangement with the lower p o s i t i o n on the opposite dimension. That i s , the a t t r i b u t e s , agreeable and unconscientious, or disagreeable and conscientious appear together i n the f i c t i t i o u s people. In Character 3, which was easier to learn, the f i c t i t i o u s people have a t t r i b u t e s from the same pole of the dimensions A-D and C-U. That i s , agreeable and conscientious appear together as do disagreeable and unconscientious. I f we look back to Table 4, we f i n d that, for both the high-and low-valid groups, the agreeable group has comparatively high v a l i d i t y i n d i c e s . Secondly, the indices i n the conscientious groups are comparatively high i n the h i g h - v a l i d group. Perhaps the learning d i f f i c u l t y of Characters 1 and 2 can be a t t r i b u t e d to the high indices as well as to the contextual arrangement of the dimensions, A-D and C-U. We can never know with the present design. Two a d d i t i o n a l characters, Characters 11 and 12, resemble Characters 1 and 2 i n terms of the contextual arrangement of the dimensions, A-D and C-U. Both of these characters were comparatively hard to learn. This observation lends a d d i t i o n a l support to the impression that the contextual arrangement of A-D and C-U i s . J ..82 an important factor i n t r a i t concept formation. Behavior Set 1: Design 2 In order to assess possible e f f e c t s of the dimensions, a 2 ( a c q u i s i t i o n groups) x 4 ( p r o f i l e s ) x 3 (characters) x 3 (dimensions) mixed analysis of variance was performed on the number of errors to c r i t e r i o n . Table 11 presents t h i s design. Notice that the f i r s t within-subject factor (dimensions) and the t h i r d between-subject fa c t o r (characters) are nested within the second between-subject factor ( p r o f i l e s ) . It may seem strange that there are 12 dimensions i n Table 11 when only four dimensions were used i n the experiment. Recall that the four p r o f i l e s were defined by deriving a l l possible combinations of four dimensions which were taken three at a time. In each p r o f i l e , a d i f f e r e n t combination of the four t r a i t dimensions appeared. The context of the dimensions changed as they appeared i n d i f f e r e n t p r o f i l e s . This contextual diff e r e n c e made i t reasonable to treat the dimensions as i f they were nested within the p r o f i l e s even through the same dimension appeared i n three of the four p r o f i l e s . A second reason f o r nesting the dimension v a r i a b l e within the p r o f i l e v a r i a b l e was e f f i c i e n c y i n the data a n a l y s i s . Because each dimension appeared i n only three of four p r o f i l e s , the dimensions are not f a c t o r i a l with the p r o f i l e s . This created a problem f o r the computer an a l y s i s . The easiest way to computer analyze these data 83 Table 11 Analysis of Variance: Design 2 Factors A l A2 B4 C l D 1-4 C2 C3 C4 C5 C6 C7 C8 C9 CIO C l l C12 C l C2 C3 C4 C5 C6 C7 C8 C9 CIO C l l C12 D93-96 F l F2 F3 F4 F5 F6 F7 F8 F9 F10 F l l F12 F l F2 F3 F4 F5 F6 F7 F8 F9 F10 F l l F12 a The factors are as follows: A i s a c q u i s i t i o n groups; B i s p r o f i l e s ; C i s character; D i s subjects; and F i s dimensions. Note that both C and F are nested within B. 84 was to tr e a t the dimensions as i f they were nested within the p r o f i l e s . If the dimension e f f e c t was s i g n i f i c a n t , the plan was to evaluate that dimension e f f e c t by c a l c u l a t i n g the sum of squares f o r each dimension across the three p r o f i l e s i n which i t appeared and add those sums together. If we nest the dimensions within the p r o f i l e s , we have a t o t a l of 12 dimensions. Table 12 displays the numbers which were associated with the dimensions i n each p r o f i l e . In Table 13 i s a summary of t h i s analysis of variance. The h i g h - v a l i d a c q u i s i t i o n group made s l i g h t l y more errors than did the low-valid a c q u i s i t i o n group, but the differences are not s i g n i f i c a n t , F_ (1,72) = .271"'; p = .609.^. The mean errors of a c q u i s i t i o n groups 1 and 2 are 12.639 and 11.854 re s p e c t i v e l y . Secondly, the small, p r o f i l e differences do not reach an acceptable l e v e l of s i g n i f i c a n c e , F (3,72) = 1.55' ; p = .205 ':. The mean number of errors i n P r o f i l e s 1 through 4 are 11.792, 10,681, 11.542 and 14.972 re s p e c t i v e l y . As i n the previous a n a l y s i s , the main e f f e c t due to the characters i s s i g n i f i c a n t , F (8,72) = 2.501. ; p = .018'-/. In Table 14 are the character means for each p r o f i l e . The Newman Keuls t e s t (with the l e v e l of s i g n i f i c a n c e set at .05) was used to make a l l possible pairwise comparisons among charactert(G/B) "means..In,Profile-1, subjects who learned Character 3 made s i g n i f i c a n t l y fewer errors than subjects who learned either Character 2 or Character 1. The diff e r e n c e between Character 1 and 2 i s not s i g n i f i c a n t . Secondly, i n P r o f i l e 4, subjects who learned Character 11 made s i g n i f i c a n t l y more errors than subjects 85 Table 12 Dimensions Treated as Nested Within P r o f i l e s P r o f i l e Dimension Name P r o f i l e 1 Dimension 1 Extraversion-Introversion Dimension 2 Agreeableness-Disagreeableness Dimension 3 Conscientiousness-Unconscientiousness P r o f i l e 2 Dimension Dimension Dimension P r o f i l e 3 Dimension 7 Extraversion-Introversion Dimension 8 Conscientiousness-Unconscientiousness Dimension 9 Emot. Stability-Emot. I n s t a b i l i t y P r o f i l e 4 Dimension 10 Agreeableness-Disagreeableness Dimension llConscientiousness-Unconscientiousness Dimension 12 Emot. Stability-Emot. I n s t a b i l i t y 4 Extraversion-Introversion 5 Agreeableness-Disagreeableness 6 Emot. Stability-Emot. I n s t a b i l i t y Emot. i s an abbreviation f o r Emotional. 86 Table 13 Summary of the Analysis of Variance on Errors to C r i t e r i o n i n Behavior Set 1 Source 3 D.F. S.S. M.S. F P To t a l 287 25061.5 Between Subjects 95 16407.494 A 1 44.336 44.336 0.271 0.609 B 3 762.149 254.049 1.558 0.205 C/B 8 3262.888 407.861 2.501 0.018 AB 3 249.760 83.253 0.510 0.680 AC/B 8 348.111 43.513 0.266 0.973 SS/AC/B 72 11740.250 163.059 3.865 Within Subjects 192 8653.998 F/B 8 118.722 14.840 10.351 0.943 F/B X Subjects 184 8535.276, AF/B 8 409.777 51.222 1.214 0.294 CF/B 16 1637.611 102.350 2.426 0.002 ACF/B 16 413.888 25.868 0.613 0.869 F/B X SS/AC/B 144 6074.000 42.180 aA=acquisition group, B = p r o f i l e , C = character, S ;S = subj ects and F=d intensions. Table 14 Mean Errors to C r i t e r i o n f o r the Characters i n Behavior Set 1 P r o f i l e Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 Character 2 Character 3 14.667 15.417 5.292 Character 4 Character 5 Character 6 11.330 11.201 9.500 Character 7 Character 8 Character 9 14.792 10.750 9.083 Character 10 Character 11 Character 12 9.792 20.083 15.042 a The number of observations for each mean i s 8. 8 8 who learned character 10. No other differences are s i g n i f i c a n t . The s i m i l a r i t i e s and differences among Characters 1, 2, 3, and 11 were described i n the previous analysis and w i l l not be repeated here. In Character 10, the contextual arrangement of the dimensions, A-D and C-U, i s the same as i t i s i n Character 3. However, according to t h i s a n a l y s i s , Character 10 i s not comparatively easy to learn. No o v e r a l l e f f e c t due to dimensions (.F/B.)" i s found,' F_ (8,144)= .351 :; p = . 943':. In Table 15 are the dimension means. The differences among the means are very s l i g h t . One i n t e r a c t i o n , the character x dimension(@F^B) i n t e r a c t i o n , i s s i g n i f i c a n t , F (16,144) = 2.426 ; p = .002 . In Figure 1 are the mean dimensional errors f o r each character. A l l possible pairwise comparisons among dimension means i n each character were made using the Newman Keuls t e s t . One differ e n c e i s s i g n i f i c a n t . In Character 11, there were s i g n i f i c a n t l y more errors made on Dimension 11 than were made on Dimensions 10 and 12 which do not d i f f e r from each other. According to Table 12, Dimension 10 i s A-D; Dimension 11 i s C-U and Dimension 12 i s ES-EI. The contextual arrangement of these three dimensions (for Character 11) i s shown i n Table 6. For th i s character, Jane's a t t r i b u t e s were emotionally unstable, disagreeable, and conscientious. Once again, we see that the contextual arrangement of the dimensions, A-D and C-U are such that a p o s i t i o n on one i s the opposite of the p o s i t i o n on the other one with each f i c t i t i o u s person. One reason f o r t h i s f i n d i n g could be that the dimension Table 15 Mean Errors to C r i t e r i o n f o r the Dimensions i n Behavior Set 1 a P r o f i l e Dimension Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Dimension 1 12.333 Dimension 2 10.917 Dimension 3 12.125 Dimension 4 11.000 Dimension 5 11.083 Dimension 6 9.958 Dimension 7 11.458 Dimension 8 11.250 Dimension 9 .11.917 Dimension 10 14.042 Dimension 11 16.292 Dimension 12 14.583 a The number of observations i n each mean i s 24. 90 25 20 CO Pi Pi w 10 v-v = CH 1 = CH 7 A—A = CH 2 = CH 8 o-o = CH 3 = CH 9 Y — Y = CH 4 O-O = CH 10 • = CH 5 = CH 11 • -• = CH 6 <-< = CH 12 1 2 3 1 2 4 1 3 4 2 3 4 DIMENSION Figure 1. Mean number of errors i n Behavior Set 1 as a function of characters (CH) and dimensions. Note that the dimensions are graphed i n four groups - one for each p r o f i l e within which three dimensions are nested. Dimension 1 i s E-I; Dimension 2 i s A-D; Dimension 3 i s C-U; and Dimension 4 i s ES-EI. 91 A-D and ES-EI have s i m i l a r pole positions f o r each f i c t i t i o u s person and these positions are opposite to the p o s i t i o n on C-U. Inspection of Figure 1 and Table' 6 together leads to the observation that, f o r each character, the dimension whose pole p o s i t i o n was opposite to the pole positions of the remaining two dimensions for that character was the hardest to learn. This observa-t i o n i s i n t e r e s t i n g because the pole p o s i t i o n s along the dimensions seem to be confounded with s o c i a l d e s i r a b i l i t y ( i . e . , the upper poles are desirable and the lower poles are undesirable). Behavior Set 2: Design 1 When they had reached c r i t e r i o n on the f i r s t set of behaviors, subjects were given Behavior Set 2. This second set of behaviors was constructed using the same p r i n c i p l e s that were used to construct Behavior Set 1. There were two purposes behind t h i s procedure. F i r s t , i t was designed to get subjects f a m i l i a r with reading new behaviors before the re v e r s a l part of the experiment. A second purpose was to see i f subjects i n a l l conditions could e a s i l y apply the t r a i t concept they had formed to a new set of s i m i l a r i l y c a l i b r a t e d and contextually arranged behaviors. The dependent measure for the f i r s t analysis of Behavior Set 2 was t r i a l s to c r i t e r i o n . A 2 ( a c q u i s i t i o n groups) x 4 ( p r o f i l e s ) x 3 (characters) between-subjects analysis of variance with the t h i r d f a c t o r nested within the second factor was used to analyze the data. This design i s the same as the one presented i n Table 8. .92' Once they reached the c r i t e r i o n on Behavior Set 1, subjects i n a l l conditions e a s i l y applied t h e i r t r a i t concept to a new set of behaviors. In Table 16 i s a summary of t h i s analysis of. variance. None of the main e f f e c t s or i n t e r a c t i o n s reaches an acceptable l e v e l of s i g n i f i c a n c e . The mean number of t r i a l s f o r A c q u i s i t i o n Groups 1 and 2 are 1.50 and 1.583 re s p e c t i v e l y . The mean t r i a l s i n P r o f i l e s 1 through 4 are 1.708, 1.208. 1.500 and 1.750 r e s p e c t i v e l y . The character means are shown i n Table 17. Behavior set 2: Design 2 In order to assess the e f f e c t of the dimension v a r i a b l e , I performed a second analysis on Behavior Set 2. The design which was used f o r t h i s analysis was a 2 (acquisition groups) x 4 ( p r o f i l e s ) x 3 (characters) x 3 (dimensions) mixed analysis of variance. The within-subject factor (dimensions) and the t h i r d between-subject fa c t o r (characters) are nested within the second between-subject fa c t o r ( p r o f i l e s ) . This design was shown i n Table 11. The dependent measure f o r t h i s analysis was the t o t a l number of errors on each dimension. ' In Table 18 i s a summary of t h i s mixed analysis of variance. None of the main e f f e c t s Is s i g n i f i c a n t . The mean number of errors f o r A c q u i s i t i o n Groups 1 and 2 are .958 and .965 re s p e c t i v e l y . The mean number of errors i n P r o f i l e s 1 through 4 are 1.056, .972, .806 and 1.014 re s p e c t i v e l y . F i n a l l y , the character means are shown i n Table 19. 93 Table 16 Summary of the Analysis of Variance on T r i a l s to C r i t e r i o n i n Behavior Set 2 Source a D.F. S.S. M. S. F F. To t a l 95 117.833 A 1 0.166 0.166 0.138 0.710 B 3 4.416 1.472 1.225 0.306 C/B 8 12.166 1.520 1.265 0.274 AB 3 3.083 1.027 0.855 0.470 AC/B 8 11.500 1.437 1.196 0.313 SS/AC/B 72 86.500 1.201 a. . . . A = a c q u i s i t i o n group, B= p r o f i l e , C = character, and SS = subjects Table 17 Mean T r i a l s to C r i t e r i o n f o r the Characters i n Behavior Set 2 P r o f i l e Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 Character 2 Character 3 1.875 2.250 1.000 Character 4 Character 5 Character 6 1.500 .875 1.250 Character 7 Character 8 Character 9 1.750 1.625 1.125 Character 10 Character 11 Character 12 1.375 1.750 2.125 a The number of observations f o r each mean i s 8. 95 Table 18 Summary of the Analysis of Variance on Errors to C r i t e r i o n i n Behavior Set 2 Source a D.F. S.S. M.S. F P T o t a l 287 388.579 Between Subjects 95 185.243 A 1 0.003 0.003 0.001 0.918 B 3 2.593 0.864 0.429 0.736 C/B 8 20.297 2.534 1.257 0.278 AB 3 2.732 0.910 0.452 0.720 AC/B 8 14.555 1.819 0.902 0.519 SS/AC/B 72 145.083 2.015 Within-Subj ects 192 203.336 F/B 8 11.611 1.451 1.501 0.160 F/B X SS 184 191.725 AF/B 8 9.388 1.173 1.214 0.294 CF/B 16 28.472 1.779 1.841 0.031 ACF/B 16 14.694 0.918 0.950 0.514 FB X SS/AC/B 144 139.166 0.966 A = A c q u i s i t i o n group, B = p r o f i l e , C = character, SS = subjects and F = dimension. Table 19 Mean Errors to C r i t e r i o n f o r the Characters In Behavior Set 2 a P r o f i l e Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 1.333 Character 2 1.292 Character 3 .542 Character 4 1.208 Character 5 .708 Character 6 1.000 Character 7 .958 Character 8 .958 Character 9 .500 Character 10 .667 Character 11 1.208 Character 12 1.167 3 The number of observations for each mean i s 8. 97 One i n t e r a c t i o n , the character x dimension (CF/B) .interaction, i s s i g n i f i c a n t , _F (16,144) = 1.841;;; p = .031 \ Figure 2 presents t h i s i n t e r a c t i o n . When used to make a l l possible pairwise comparisons among dimension means f o r each character(c/B) , the Newman Keuls test y i e l d e d the following r e s u l t s . In Character 4, subjects made s i g n i f i c a n t l y more errors on Dimension 6 than they did on Dimension 4 and 5 which do not d i f f e r from each other. Secondly, i n Character 10, subjects made s i g n i f i c a n t l y more errors on Dimension 12 than they did on Dimensions 10 and 11 which do not d i f f e r from each other. Let us look at Characters 4 and 11 separately. There are two components to Character 4: (a) Jane's a t t r i b u t e s which are extraverted, agreeable and emotionally unstable; and (b) S a l l y ' s a t t r i b u t e s which are introverted, disagreeable and emotionally stable. In t h i s character, Dimension 6, which was the most d i f f i c u l t to learn, was ES-EI. Dimension 5 i s A-D and Dimension 4 i s E-I. In terms of the a t t r i b u t e s of the f i c t i t i o u s people, the pole p o s i t i o n of the dimension, ES-EI i s opposite to the pole positions on A-D and E-I which are s i m i l a r to one another. There are two components to Character 10: (a) Jane's a t t r i b u t e s which are agreeable, conscientious and emotionally unstable; and (b) S a l l y ' s a t t r i b u t e s which are disagreeable, unconscientious and emotionally stable. In t h i s character, Dimension 12 which was the most d i f f i c u l t to learn i s ES-EI; Dimension 10 i s A-D; and Dimension 11 i s C-U. In terms of the a t t r i b u t e s of the f i c t i t i o u s people, 98 CO Pi O Pi Pi w 2.5 2.0 1.5 v - v ~ CH 1 B — • = CH 7 A—A = CH 2 • -• = CH 8 O-O = CH 3 • 0 - • = CH 9 T - T = CH 4 o-o = CH 10 A—A = CH 5 >- • = CH 11 •-• = CH 6 <-< = CH 12 1 2 3 1 2 4 1 3 4 DIMENSION Figure 2. Mean number of errors i n Behavior Set 2 as a function of characters (CH) and dimensions. Note that, i n l i k e manner to Figure 1, the dimensions are graphed i n four groups of three i n each group. Dimension 1 i s E-I; Dimension 2 i s A-D; Dimension 3 i s C-U; and Dimension 4 i s ES-EI. 99 the pole p o s i t i o n of ES-EI i s opposite to the pole p o s i t i o n of the dimensions A-D and C-U which are s i m i l a r to one another. By inspecting Figure 2 and Table 6 together, we see that i n Behavior Set 2, the dimension whose pole p o s i t i o n i s opposite to the pole p o s i t i o n on the remaining two dimensions for each character was not always the most d i f f i c u l t dimension to learn. In f a c t , i t was the most d i f f i c u l t dimension to learn i n only h a l f of the characters. This suggests that a primary cause of the s i g n i f i c a n c e of the character x dimension (CF/B) i n t e r a c t i o n i s not the. contextual r e l a t i o n s h i p of negative to p o s i t i v e pole p o s i t i o n s on the dimensions. One a d d i t i o n a l observation i s reported. In Characters 4 and 11, the pole p o s i t i o n on the dimension, ES-EI, which was most d i f f i c u l t to l e a r n , i s opposite to the pole p o s i t i o n on the,dimension, A-D. Perhaps i n t e g r a t i n g d i f f e r e n t pole information on these dimensions i s a d i f f i c u l t task. Learning Data: Reversal When they reached the c r i t e r i o n on Behavior Set 2, subjects were given Behavior Set 3. As was pointed out i n the chapter on method, Behavior Set 3 d i f f e r e d from Behavior Set 2 on two accounts: one dimension was reversed, and another was replaced with a new dimension. These changes required that subjects adjust t h e i r i n i t i a l t r a i t concept to accommodate new and contradictory behavioral information. The analysis of the learning data i n Behavior Set 3 provides a test of hypotheses 2 (a and b) and 3 (a and b). 100 The learning data are subdivided into two groups: (a) data from Behavior Set 3, and (b) data from Behavior Set 4. There are three dependent measures for Behavior Set 3 and two for Behavior Set 4. For Behavior Set 3, the dependent measures are t r i a l s to c r i t e r i o n , errors within the f i r s t r e v e r s a l t r i a l and the t o t a l number of errors within a fi x e d number of t r i a l s ( i . e . , 5 t r i a l s ) . For Behavior Set 4, the dependent measures are t r i a l s to c r i t e r i o n and errors to c r i t e r i o n . Behavior Set 3: Design 3 To analyze the t o t a l number of t r i a l s to c r i t e r i o n i n the t h i r d behavior set, a 2 ( a c q u i s i t i o n groups) x 2 (reversal groups) x 4 ( p r o f i l e s ) x 3 (characters) between-subject analysis of variance was performed on the data. Table 20 presents t h i s design. Notice that the fourth factor i s nested within the t h i r d f a c t o r . In Table 21 i s a summary of the analysis of the t r i a l s to c r i t e r i o n i n Behavior Set 3. Subjects who were i n the hi g h - v a l i d a c q u i s i t i o n group took s i g n i f i c a n t l y longer to reach c r i t e r i o n than did subjects who were i n the low-valid a c q u i s i t i o n group, F_ (1,48) = 5.859; p = .018. In order, the mean t r i a l s to c r i t e r i o n for A c q u i s i t i o n Groups 1 and 2 are 5.521 and 4.229. This d i f f e r e n c e i s i n the predicted d i r e c t i o n . Subjects who were i n the h i g h - v a l i d r e v e r s a l group took more t r i a l s to reach c r i t e r i o n than d i d subjects who were i n the low v a l i d r e v e r s a l group, but t h i s d i f f e r e n c e does not reach an acceptable l e v e l of s i g n i f i c a n c e , F (1,48) = .298; p = .593. The t r i a l s to 101 Table 20 Analaysis of Variance: Design 3 Factors C l Dl D2 D3 F 1-2 B l C2 C3 D4 D5 D6 D7 D8 D9 C4 A l D10 D l l D12 C l Dl D2 D3 B2 C2 C3 D4 D5 D6 D7 D8 D9 C4 D10 D l i D12 A2 F95-96 The factors are as follows: A i s a c q u i s i t i o n groups;. B i s r e v e r s a l groups; C i s p r o f i l e s ; D i s characters and F- i s subjects. Note that D i s nested within C. A2 had the same arrangement as does A l . 102 Table 21 Summary of the Analysis of Variance on T r i a l s to C r i t e r i o n i n Behavior Set 3 Source 3 D.F. S.S. M.S. F P T o t a l 95 860.500 A 1 40.041 40.041 5.859 0.018 B 1 2.041 2.041 0.298 0.593 C 3 35.250 11.750 1.719 0.174 D/C 8 65.250 8.156 1.193 0.322 A B 1 0.166 0.166 0.024 0.850 A C 3 .12.375 4.125 0.603 0.619 B C 3 70.208 23.402 3.424 0.024 ADAC 8 87.083 10.885 1.593 0.151 B D / C 9 87.250 10.906 1.596 0.150 A B C 3 33.083 11.027 1.613 0.197 A B D / C 8 99.750 12.468 1.824 0.095 SS/ABD/C 48 328.000 6.833 a . . ._ . A = a c q u i s i t i o n group, B = re v e r s a l group, C = p r o f i l e , D = character and SS = subjects. 103 c r i t e r i o n f or Reversal Groups 1 and 2 are 5.021 and 4.729 r e s p e c t i v e l y . The small difference between the groups appears i n the d i r e c t i o n opposite to Hypothesis 2(b). The a c q u i s i t i o n group x re v e r s a l group i n t e r a c t i o n i s not s i g n i f i c a n t , F (1,48) = .024-; p = .850.. As expected, the mean number of t r i a l s i n P r o f i l e s 1 through 4 do not d i f f e r s i g n i f i c a n t l y from each other, F_ ( 3,48) = 1.719'-; p = .174-;. The mean t r i a l s to c r i t e r i o n i n P r o f i l e Groups 1 through 4 are 5.875, 4.250, 4.625 and 4.750 re s p e c t i v e l y . F i n a l l y , the main ef f e c t due to the character v a r i a b l e i s not s i g n i f i c a n t , _F (8,48) = 1.193' j p = .322". The mean t r i a l s to c r i t e r i o n f or the characters are shown i n Table 22. Character 1 i s comparatively more d i f f i c u l t to adjust and Character 5 i s comparatively easy to adjust, but neither character d i f f e r s s i g n i f i c a n t l y from any of the others within the p a r t i c u l a r p r o f i l e . One i n t e r a c t i o n , the p r o f i l e x r e v e r s a l group i n t e r a c t i o n , i s s i g n i f i c a n t , F (3,48) = .424.!; p = .024.. In Figure 3 are the comparisons of p r o f i l e means f o r each r e v e r s a l group. The Newman Keuls test was used to make a l l possible pairwise comparisons among p r o f i l e means. One di f f e r e n c e i s s i g n i f i c a n t . In P r o f i l e 1, subjects who were i n the h i g h - v a l i d r e v e r s a l group required s i g n i f i c a n t l y more t r i a l s to reach c r i t e r i o n than did subjects who were i n the low-valid r e v e r s a l group. This d i f f e r e n c e contradicts Hypothesis 2(b). The differences between the high-and low-valid r e v e r s a l groups for P r o f i l e s 2 to 4 are not s i g n i f i c a n t . 104 Table 22 Mean T r i a l s to C r i t e r i o n f o r the Characters i n Behavior Set 3 P r o f i l e Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 Character 2 Character 3 7.250 4.500 5.875 Character 4 Character 5 Character 6 5.250 3.250 5.250 Character 7 Character 8 Character 9 4.375 4.000 5.500 Character 10 Character 11 Character 12 3.875 5.125 5.250 a The number of observations for each mean i s 8. 105 Figure 3. Mean number of t r i a l s i n Behavior Set 3 as a function of the p r o f i l e s and re v e r s a l groups (HVR - h i g h - v a l i d r e v e r s a l group and LVR = low-valid r e v e r s a l group). 106 During the a c q u i s i t i o n part- of the experiment, P r o f i l e 1 consisted of the dimensions: E-I, A-D and C-U. Upon r e v e r s a l , each dimension i n the a c q u i s i t i o n part had a chance to drop out and to be replaced by the dimension, ES-EI. Perhaps when one i s reading hi g h - v a l i d behaviors, i t i s more d i f f i c u l t to integrate a s h i f t on any of the f i r s t three dimensions with information on the dimension, ES-EI. This may be because the pole p o s i t i o n of ES-EI i s always the same as the pole p o s i t i o n on the reversed dimension. Behavior Set 3: Design 4 There were two a d d i t i o n a l analyses which were performed on Behavior Set 3. The f i r s t analysis was performed on the errors within the f i r s t r e v e r s a l t r i a l and the second was performed on the number of errors within a f i x e d number of t r i a l s . Both dependent measures were analyzed with Design 4. This design was a 2 ( a c q u i s i t i o n groups) x 2 (reversal groups) x 4 ( p r o f i l e s ) x 3 (characters) x 3 (conditions) mixed analysis of variance and i s described i n Table 23. Notice that the fourth between-subject factor (characters) i s nested within the t h i r d between-subject f a c t o r ( p r o f i l e s ) . Secondly, notice that the dimension variable; i s "incompletely l a t i n i z e d " " w i t h i f f each p r o f i l e . However, across the four p r o f i l e s , each dimension has a chance to be i n each condition an equal number of times and therefore the l a t i n square becomes complete. Analysis of the f i r s t r e v e r s a l t r i a l . The r e s u l t s from the 107 Table 23 Analysis of Variance: Design 4 Factors Gl G2 G3 Dl F 1-2 HI H2 H4 C l D2 H3 HI H4 D3 H2 H3 H4 D4 HI H2 H3 C2 D5 H4 HI H3 D6 H2 H4 H3 D7 HI H3 H2 C3 D8 H4 HI H2 D9 H3 H4 H2 D10 H2 H3 HI C4 D l l H4 H3 HI D12 H3 H4 HI Dl HI H2 H4 C l D2 H3 HI H4 D3 H2 H3 H4 D4 HI H2 H'3 C2 D5 H4 HI H3 D6 H2 H4 H3 D7 HI H3 H2 C3 D8 H4 HI H2 D9 H3 H4 H2 D10 H2 H3 HI C4 D l l H4 H2 HI D12 H3 H4 HI A2 b F95-96 aThe factors are as follows: A i s a c q u i s i t i o n groups; B i s r e v e r s a l groups; C i s p r o f i l e s ; D i s characters; F i s subjects; G i s conditions and H i s dimensions. Note that D and H are nested within C and that ,H i s l a t i n i z e d within the G x D i n t e r a c t i o n . The variables i n A2 are the same as they were i n A l . 108 analysis of the learning data i n the f i r s t r e v e r s a l t r i a l are reported f i r s t . In Table 24 i s a summary of the analysis of variance on the number of errors within the f i r s t r e v e r s a l t r i a l i n Behavior Set 3. Subjects who were i n the h i g h - v a l i d a c q u i s i t i o n group made s l i g h t l y more errors than did subjects who were i n the low-valid a c q u i s i t i o n group, but t h i s d i f f e r e n c e i s not s i g n i f i c a n t , F_ (1,48) = .15b'. ; p = .392 . The mean number of errors i n A c q u i s i t i o n Groups 1 and 2 are 2.819 and 2.667 r e s p e c t i v e l y . S i m i l a r l y , subjects who were i n the high-^valid r e v e r s a l group made more errors than did those who were i n the low-valid r e v e r s a l group, but again, t h i s d i f f e r e n c e i s not s i g n i f i c a n t , _F (1,48) = 1.228,"; p = .272.;. The mean number of errors i n Reversal Groups 1 and 2 are 2.840 and 2.646 re s p e c t i v e l y . The f i r s t f i n d i n g was i n the d i r e c t i o n of Hypothesis 2(a) but the second f i n d i n g was not i n the d i r e c t i o n of Hypothesis 2(b). The diffe r e n c e s , however, are very s l i g h t . The a c q u i s i t i o n x rever s a l group i n t e r a c t i o n reaches an acceptable l e v e l of s i g n i f i c a n c e , _F (1,48) = 4.915' ; p = .029 . Figure 4 indicates that when they adjusted t h e i r t r a i t concept while reading low-valid behavioral information, subjects i n the low-valid a c q u i s i t i o n group made fewer errors than did subjects who were i n the h i g h - v a l i d a c q u i s i t i o n group. Conversely, when they adjusted t h e i r t r a i t concept while reading h i g h - v a l i d behavioral information, subjects i n the low-valid a c q u i s i t i o n group made more errors than did subjects i n the high-v a l i d a c q u i s i t i o n . The fewest errors were made when subjects had the same type of information i n both the t r a i t concept formation 109 Table 24 Summary of the Analysis of Variance on Errors Within the F i r s t T r i a l i n Behavior Set 3 Source 3 D.F. S.S. M.S. To t a l 287 798.986 Between-Subj ects 95 231. 648 A 1 1. 680 1. 680 0. 758 0.392 B 1 2. 722 2. 722 1. 228 0.272 C 3 7. 819 2. 606 1. 176 0.328 D/C 8 19. 833 2. 479 1. 119 0.367 AB 1 10. 888 10. 888 4. 915 0.029 AC 3 6. 486 2. 162 0. 976 0.413 BC 3 4. 722 1. 574 0. 710 0.553 AD/C 8 10. 833 1. 354 0. 611 0.765 BD/C 8 27. 222 3. 402 1. 536 0.169 ABC 3 6. 555 2. 185 0. 986 0.408 ABD/C 8 26. 555 3. 319 1. 498 0.182 SS/ABD/C 48 106. 333 2. 215 0. 959 0.554 Within-Subj ects 192 567. 338 G 2 82. 965 41. 482 17. 965 0.001 G X SS 190 484. 373 GA 2 4. 215 2. 107 0. 912 0.407 GB 2 21. 048 10. 524- 4. 557 0.012 GC 6 7. 618 1. 269 0. 549 0.770 GD/C 16 37. 000 2. 312 1. 001 0.462 H/C 8 19. 160 2. 390 1. 034 >.050 LSR 8 17. 84 2. 230 0. 965 >.050 GAB 2 3. 548 1. 774 0. 768 0.470 GAC 6 6. 534 1. 089 0. 471 0.829 GBC 6 35. 090 5. 848 2. 532 0.025 GAD/C 16 61. 500 3. 843 1. 664 0.066 AH/C 8 27. 833 3. 479 1. 506 >.050 LSR 8 33. 666 4. 208 1. 820 >.050 GBD/C 16 56. 444 3. 527 1. 527 0.105 BH/C 8 20. 722 2. 590 1. 080 >.050 LSR 8 35. 722 4. 465 1. 934 >.050 GABD/C 16 22. 611 1. 413 0. 612 0.867 ABH/C 8 9. 388 1. 173 0. 508 >.050 LSR 8 13. 222 1. 652 0. 715 >.050 G X SS/ABD/C 96 221. 666 2. 309 A = a c q u i s i t i o n group, B = r e v e r s a l group. C = p r o f i l e , D = character, SS = subjects, G = condition, H = dimension, and LSR = l a t i n square r e s i d u a l . 110 HVR LVR HVA LVA Figure 4. Mean number of errors i n the f i r s t r e v e r s a l t r i a l of Behavior Set 3 as a function of the a c q u i s i t i o n groups (HVA = h i g h - v a l i d a c q u i s i t i o n group; LVA = low-valid a c q u i s i t i o n groups) and r e v e r s a l groups (HVR = high-v a l i d r e v e r s a l group; LVR = low-valid r e v e r s a l group). I l l part and the t r a i t concept adjustment part of the experiment. The small d i f f e r e n c e s among p r o f i l e means are not s i g n i f i c a n t , F_ (3,48) = 1,176; p = .328. The mean number of errors i n P r o f i l e s 1 through 4 are 2.972, 2.528, 2.667, and 2,800 re s p e c t i v e l y . F i n a l l y , the main e f f e c t due to the character v a r i a b l e i s not s i g n i f i c a n t , _F (8,48) = 1.119; p = .367. The mean number of errors f o r each character are presented i n Table 26. The differences among means are very small.-The main e f f e c t due to the conditions v a r i a b l e i s highly s i g n i f i c a n t , F (2,96) = 17.965; p = .0001. The mean number of errors i n Conditions 1 to 3 are 3.468, 2.188 and 2.573 re s p e c t i v e l y . Two a p r i o r i orthogonal contrasts were made and yielded the following r e s u l t s . S i g n i f i c a n t l y more errors were made i n Condition 1 (the reversed condition) than were made i n Conditions 2 (the unchanged condition) and 3 (the new condition)together, F (1,96) = 33.93; p = ,00Qi. This f i n d i n g supports Hypothesis 3(a). In support of Hypothesis 3(b), more errors were made i n Condition 3 than were made i n Condition 2, but t h i s d i f f e r e n c e i s not s i g n i f i c a n t , F (1,96) = 1.07; p >.05. The r e v e r s a l group x condition i n t e r a c t i o n i s s i g n i f i c a n t , F_ (2,96) = 4.557; p = .012. In Figure 5 are comparisons among condition means for the hi g h - v a l i d and the low-valid r e v e r s a l groups. A l l possible pairwise comparisons among condition means for each r e v e r s a l group were made using the Newman Keuls t e s t (with the l e v e l of s i g n i f i c a n c e set at .05). Two fi n d i n g s emerged from t h i s a n a l y s i s . Table 25 Mean Errors to C r i t e r i o n i n the F i r s t T r i a l f o r Characters i n Behavior Set 3 P r o f i l e Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 Character 2 Character 3 3.042 2,875 3.000 Character 4 Character 5 Character 6 2.583 2.167 2.833 Character 7 Character 8 Character 9 2,750 2.500 2.750 Character 10 Character 11 Character 12 2.292 2.792 3.333 a The number of observations f o r each mean i s 8. 113 4 pi o 1 2 3 CONDITION Figure 5. Mean number of errors i n Behavior Set 3 as a function of the conditions and the r e v e r s a l groups (HVR = high-v a l i d r e v e r s a l group; LVR = low-valid r e v e r s a l group). Condition 1 i s the reversed dimension; Condition 2 i s the unchanged dimension, and Condition 3 i s the new dimension. 114 F i r s t , i n the hi g h - v a l i d r e v e r s a l group, s i g n i f i c a n t l y more errors were made i n Condition 1 than were made i n Conditions 2 and 3 which do not d i f f e r from each other. Secondly, i n the low-valid reversal group, s i g n i f i c a n t l y fewer errors were made i n Condition 2 than were made i n Conditions 1 and 3 which do not d i f f e r from each other. These findings suggest that the rever s a l group x condition i n t e r a c t i o n i s due mainly to 'differences i n the r e l a t i v e l e v e l of d i f f i c u l t y of Condition 3 across the rever s a l groups. By inspection of Figure 5, we see that the r e l a t i o n s h i p of Conditions 2 and 3 d i f f e r across the reversal groups. In the low-valid r e v e r s a l group, subjects made more errors i n Condition 3 than they made i n Condition 2. But i n the hi g h - v a l i d r e v e r s a l group, the r e l a t i o n s h i p i s reversed. This suggests that the new dimension was easier to assimilate when subjects read h i g h - v a l i d behavioral information than i t was when they read low-valid behavioral information. F i n a l l y , one t r i p l e i n t e r a c t i o n , the re v e r s a l group x p r o f i l e x condition i n t e r a c t i o n , i s s i g n i f i c a n t , _F (6,96) = 2.532; p = .025-. In Figure 6, the condition means for each p r o f i l e are shown separately for the two re v e r s a l groups. A Newman Keuls test (with the l e v e l of s i g n i f i c a n c e set at .05) was used to make a l l possible pairwise comparisons among condition means for each p r o f i l e i n the hig h - v a l i d and the low-valid r e v e r s a l groups separately. The following findings emerged. In P r o f i l e s 2, 3, and 4 of the h i g h - v a l i d r e v e r s a l group, s i g n i f i c a n t l y more errors were made i n Condition 1 than were made i n Conditions 2 and 3 which do not 115 ©=Profile 1 • = P r o f i l e 2 • = P r o f i l e 3 T = P r o f i l e 4 1 2 3 1 2 3 CONDITION CONDITION Figure 6. Mean number of errors i n Behavior Set 3 as a function of the conditions, p r o f i l e groups and re v e r s a l groups (HVR = hi g h - v a l i d r e v e r s a l group; LVR = low-valid r e v e r s a l group). Condition 1 i s the reversed dimension; Condition 2 i s the unchanged dimension, and Condition 3 i s the new dimension. 116 d i f f e r s i g n i f i c a n t l y from each other.. The greatest difference i s i n P r o f i l e 2. The differences among the condition means i n P r o f i l e 1 are not s i g n i f i c a n t . In the low-valid r e v e r s a l group, the r e l a t i o n s h i p of the con-d i t i o n means d i f f e r s for each p r o f i l e . In P r o f i l e 1, s i g n i f i c a n t l y more errors were made i n Condition 1 than were made i n Condition 3 which i n turn had s i g n i f i c a n t l y more errors made i n i t than had Condition 2. In P r o f i l e 2, s i g n i f i c a n t l y more errors were made i n Condition 3 than were made i n Condition 1 which i n turn had s i g n i f i c a n t l y more errors i n i t than had Condition 2. In P r o f i l e 3, s i g n i f i c a n t l y more errors were made i n Condition 1 than were made i n Condition 2, but not i n Condition 3. Condition 2 and 3 do not d i f f e r s i g n i f i c a n t l y from each other. F i n a l l y , i n P r o f i l e 4, s i g n i f i c a n t l y more errors were made i n Condition 1 than were made i n both Conditions 2 and 3 which do not d i f f e r s i g n i f i c a n t l y from each other. It i s i n t e r e s t i n g to see how the p r o f i l e groups measure up to the t h i r d hypothesis which yielded the p r e d i c t i o n that: :(a) the the reversed dimension was more d i f f i c u l t to learn than was ei t h e r the unchanged .or the new dimension; and (b) the unchanged dimension was easier to learn than was the'new dimension. In the hi g h - v a l i d r e v e r s a l group, part (a) received s i g n i f i c a n t support'in .three of four p r o f i l e s . Although they were not s i g n i f i c a n t , the r e s u l t s i n three of four p r o f i l e s i n t h i s r e v e r s a l group are i n the d i r e c t i o n of part (b). . In every p r o f i l e 117 the control dimension was the easiest to learn. In the low-valid r e v e r s a l group, the r e s u l t s are very complex. In P r o f i l e l , p a r t (aT but hot p a r t (b-)-' received supp'ort'. -In P r o f i l e 2, only part (b) of Hypothesis 3 was supported. Note that the r e s u l t s contradict part (a) i n that the new dimension (the control) was s i g n i f i c a n t l y harder to learn than was the reversed dimension. In t h i s p a r t i c u l a r p r o f i l e , the co n t r o l dimension was C-U. I t seems that information about t h i s dimension i s the most d i f f i c u l t to integrate with the unchanged and the reversed dimension no matter what they were. In P r o f i l e 3, Hypothesis 3(a) i s p a r t i a l l y supported i n that the reversed condition was s i g n i f i c a n t l y harder to learn than was the unchanged dimension, but i t was not s i g n i f i c a n t l y harder to learn than was the co n t r o l dimension. In addition Hypothesis 3(b) received support i n the t h i r d p r o f i l e . F i n a l l y , i n P r o f i l e 4, both parts of Hypothesis 3 were supported. In every p r o f i l e i n t h i s r e v e r s a l group, the unchanged dimension was the easiest to learn. It i s important to remember that these r e s u l t s indicated the i n i t i a l e f f e c t of the changes i n Behavior Set 3 from Behavior Set 2. It i s i n t e r e s t i n g to see i f the r e s u l t s are present at the end of f i v e t r i a l s . The analysis of errors within fixed t r i a l s . When the number of errors over a f i x e d number of t r i a l s was used as the dependent measure, somewhat d i f f e r e n t e f f e c t s were found. 118 Table 26 presents a summary of the analysis of variance on errors within f i x e d t r i a l s . Subjects who were i n the h i g h - v a l i d a c q u i s i t i o n group made more errors than did subjects i n the low-valid reversal group, but t h i s difference i s not r e l i a b l e , _F (1,48) = 1,112; p = 297. In order, the a c q u i s i t i o n group means are 6.139 and 5.625. S i m i l a r l y , subjects i n the h i g h - v a l i d r e v e r s a l group made more errors than did subjects i n the low-valid r e v e r s a l group but t h i s d i f f e r e n c e does not reach an acceptable l e v e l of s i g n i f i c a n c e , _F (1,48) = 1.236;' p = .271. The d i r e c t i o n of t h i s r e s u l t i s opposite to that predicted by Hypothesis 2(b). In order, the rever s a l group means are 6.153 and 5.611. The a c q u i s i t i o n x rever s a l group i n t e r a c t i o n i s not s i g n i f i c a n t , j? (1,48) = 2.282;' 'p = 1.33. • The number of errors made by subjects d i f f e r s s i g n i f i c a n t l y across the four p r o f i l e s , F (3,48) = 2.852; p = .046.;% The mean number of errors i n P r o f i l e s 1 through 4 are 6.903, 4.889, 5.833 and 5.903 res p e c t i v e l y . The Newman Keuls test was used to make a l l possible pairwise comparisons among p r o f i l e means. No s i g n i f i -cant differences were found. The o v e r a l l s i g n i f i c a n c e of the p r o f i l e e f f e c t i s probably due to the d i f f e r e n t number of errors i n P r o f i l e s 1 and 2 with P r o f i l e 1 having approximately one-third as many errors i n i t than P r o f i l e 2 has. The main e f f e c t due to character(D/C) does not reach art acceptable l e v e l of s i g n i f i c a n c e , F (8,48) = 1.536; ' p = .169/ Table 27 shows the mean number of errors f o r each character. The most 119 Table 26 Summary of the Analysis of Variance on Errors Within Fixed T r i a l s (5) i n Behavior Set 3 Source 3 D.F. S.S. M.S. Dtal 287 4871. 984 Between-Subjects 95 1999. 981 A 1 19. 013 19. 013 1. 112 0. 297 B 1 21. 125 21. 125 1. 236 0. 271 C 3 146. 236 48. 745 . 2. 852 0. 046 D/C 8 210. 083 26. 260 1. 536 0. 169 AB 1 39. 013 39. 013 2. 282 0. 133 AC 3 11. 625 3. 875 0. 226 0. 876 BC 3 60. 569 20. 189 1. 181 0. 326 AD/C 8 125. 527 15. 691 0. 918 0. 510 BD/C 8 302. 972 37. 871 2. 216 0. 042 ABC 3 39. 402 13. 134 0. 768 0. 520 ABD/C 8 204. 083 25. 510 1. 492 0. 184 SS/ABD/C 48 820. 333 17. 090 Within-Subj ects 192 2872. 003 G 2 119. 506 59. 753 3. 984 0. 021 G X SS 190 2752. 497 GA 2 17. 881 8. 941 0. 596 0. 558 GB 2 54. 395 27. 197 1. 813 0. 166 GC 6 43. 576 7. 262 0. 484 0. 819 GD/C 16 254. 333 15. 895 1. 060 0. 403 H/C 8 118. 750 14. 840 0. 990 >. 050 LSR 8 135. 583 16. 940 1. 130 >. 050 GAB 2 36. 798 18. 399 1. 226 0. 297 GAC 6 79. 145 13. 191 0. 879 0. 514 GBC 6 76. 743 12. 790 0. 852 0. 533 GAD/C 16 281. 055 17. 566 1. 171 0. 304 AH/C 8 140. 861 17. 600 1. 174 >. 050 LSR 8 140. 194 17. 520 1. 169 050 GBD/C 16 215. 444 13. 465 0. 897 0. 573 BH/C 8 128. 305 16. 038 1. 069 >. 050 LSR 8 87. 139 10. 890 0. 726 >. 050 GABD/C 16 187. 833 11. 739 0. 782 0. 701 ABH/C 8 61. 417 7. 677 0. 512 >. 050 LSR 8 126. 416 15. 800 1. 054 050 G X SS/ABD/C 96 1439. 667 14. 996 A = a c q u i s i t i o n group, B = r e v e r s a l group, C = p r o f i l e , D = character, SS = subjects, G = condition, H = dimension, and LSR = l a t i n square r e s i d u a l . 120 Table 27 Mean Errors Within a Fixed Number of T r i a l s . f o r the Characters i n Behavior Set 3 a P r o f i l e Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 7,542 Character 2 5.625 Character 3 7.542 Character 4 4.875 Character 5 3.583 Character 6 6.208 Character 7 6.125 Character 8 5.167 Character 9 6.208 Character 10 4.750 Character 11 6.167 Character 12 6.792 a The number of observations for each mean i s 8. 121 errors were made i n Characters 1 and 3 and the fewest were made i n Character 5. The main e f f e c t due to the conditions v a r i a b l e i s s i g n i f i c a n t , F (2,96) =3.984; p = .021. The mean number of errors i n Conditions 1 to 3 are 6.792, 5.385 and 5.469. Two a p r i o r i orthogonal contrasts were made and yielded the following r e s u l t s . S i g n i f i c a n t l y more errors were made i n Condition 1 than were made i n Conditions 2 and 3 together, F (1,96) = 7.954; p< .01. This r e s u l t supports Hypothesis 3(a). S l i g h t l y more errors were made i n Condition 3 than were made i n Condition 2, but t h i s difference i s not s i g n i f i c a n t , _F (1,96) = .0001..' One i n t e r a c t i o n , the re v e r s a l group x character "(B D/G) i n t e r a c t i o n i s s i g n i f i c a n t , _F (8,48) = 2.21; p = .042. In Figure 7 are comparisons of the rever s a l group means f o r each character. Because the character v a r i a b l e was nested within the p r o f i l e v a r i a b l e , there are four parts to Figure 7. The Newman Keuls test was used to test the comparisons of the high- v a l i d and low-valid r e v e r s a l groups f o r each character. Four s i g n i f i c a n t differences emerged. In Character 2, subjects who were i n the hi g h - v a l i d r e v e r s a l group made s i g n i f i c a n t l y more errors than did subjects i n the low-valid r e v e r s a l group. A s i m i l a r f i n d i n g was found i n Character 11. In Characters 4 and 10, the fi n d i n g was reversed. Subjects who were i n the low-valid r e v e r s a l group made s i g n i f i c a n t l y more errors than did those who were i n the hi g h - v a l i d r e v e r s a l group. It appears that the e f f e c t of the behavioral 122 V = CH 1 T = CH 4 • = CH 7 O = CH 10 A = CH 2 A = CH 5 • = CH 8 • = CH 11 O = CH 3 • = CH 6 • = CH 9 < = CH 12 P r o f i l e 1 CO Pi o Pi Pi W P r o f i l e 2 8 CO Pi o Pi 3 5 w £ 4 P r o f i l e 3 HVR LVR HVR LVR Figure 7. Mean number of errors within f i x e d t r i a l s i n Behavior Set 3 as a function of the character (CH) and re v e r s a l groups (HVR = h i g h - v a l i d r e v e r s a l group; LVR = low-valid r e v e r s a l group) Note that the characters are graphed i n four groups. 123 information subjects read to adjust t h e i r t r a i t concept can be ei t h e r f a c i l i t i v e or i n h i b i t i v e depending upon the s p e c i f i c character involved. It i s important to note that there are some differences between the analysis of the fi x e d t r i a l s data and the analysis of the f i r s t t r i a l data. F i r s t , two in t e r a c t i o n s which were s i g n i f i c a n t i n the f i r s t t r i a l analysis are not s i g n i f i c a n t i n the fi x e d t r i a l s a n a l y s is. These i n t e r a c t i o n s are the a c q u i s i t i o n x r e v e r s a l group and the rever s a l group x condition i n t e r a c t i o n s . Secondly, two new findings emerged i n the analysis of the fi x e d t r i a l s data. F i r s t , the successful adjustment of t r a i t concepts was influenced by the combination of t r a i t dimensions which formed the content of t r a i t concept adjustment. The second f i n d i n g was that the learning d i f f i c u l t y of the characters d i f f e r e d as a function of the type of information subj ects were reading during the reve r s a l part of the experiment. These differences suggest that the i n i t i a l e f f e c t of the revers a l s h i f t i s somewhat d i f f e r e n t from the e f f e c t a f t e r f i v e t r i a l s . Perhaps, as subjects progress toward t r a i t concept adjust-ment, d i f f e r e n t processes become important. Behavior Set 4: Design 3 When they had reached the c r i t e r i o n on Behavior Set 3, subjects were given the opportunity to test t h e i r adjusted t r a i t concept on a new set of behaviors which were c a l i b r a t e d and assembled i n l i k e manner to Behavior Set 3. The chief purpose of t h i s fourth set of 124 behaviors was to determine the e f f i c i e n c y with which subjects could apply t h e i r adjusted t r a i t concept to a new set of behaviors. A second purpose was to discover any tran s f e r traces from the i n i t i a l and/or the adjusted t r a i t concept. For t h i s a n a l y s i s , two dependent measures were used: (a) t r i a l s to c r i t e r i o n ; and (b) errors to c r i t e r i o n . F i r s t , the t r i a l s to c r i t e r i o n data are described. To analyze the number of t r i a l s to c r i t e r i o n , a 2 ( a c q u i s i t i o n groups) x 2 (reversal groups) x 4 ( p r o f i l e s ) x 3 (characters) between-subjects analysis of variance was performed. This design i s the same design that was described i n Table 20. Once again, notice that the fourth f a c t o r i s nested within the t h i r d f a c t o r . No differences among the between-subject groups were detected. It appears that once they reached the c r i t e r i o n on Behavior Set 3, subjects i n a l l conditions could e a s i l y apply t h e i r adjusted t r a i t concept to a new set of behaviors. In Table 28 i s a summary of the analysis of variance which was performed on the number of t r i a l s to c r i t e r i o n on the fourth set of behaviors. The small difference between the a c q u i s i t i o n groups i s not s i g n i f i c a n t , _F (1,48) = .160; p = .691. The mean t r i a l s to c r i t e r i o n for the a c q u i s i t i o n groups are 1.958 and 1.833 r e s p e c t i v e l y . S i m i l a r l y , the small difference between the r e v e r s a l groups i s not s i g n i f i c a n t , F (1,48) = .875 5 p = .357-' The mean t r i a l s to c r i t e r i o n f o r the r e v e r s a l groups are 1.750 and 2.042 re s p e c t i v e l y . The p r o f i l e e f f e c t i s not s i g n i f i c a n t , F_ (3,48) = .767; 125 Table 28 Summary of the Analysis of Variance on T r i a l s to C r i t e r i o n i n Behavior Set 4 a Source D.F. S. S. M.S. F P Tot a l 95 212. 958 A 1 0. 375 0. 375 0. 160 0. 691 B 1 2. 041 2. 041 0. 875 0. 357 C 3 5. 375 1. 791 0. ,767 0. 520 D/C 8 13. 083 1. 635 0. ,700 0. 690 AB 1 6. 041 0. 041 0. ,017 0. 863 AC 3 4. 375 1. 458 0. .625 0. 606 BC 3 3. 208 1. 069 0. ,458 0. 7,16 AD/C 8 26. 750 3. 343 1. .433 0. 206 BD/C 8 18. 250 2. 281 6. .977 0. 465 ABC 3 15. 875 5. 291 2. .267 0. 091 ABD/C 8 11. 583 1. 447 0, .620 0. 757 SS/ABD/C 48 112. 000 2. 333 a A = a c q u i s i t i o n group, B = r e v e r s a l group, C = p r o f i l e , D = character and SS = subjects. 126 p = .520. The mean number of t r i a l s to c r i t e r i o n f o r P r o f i l e s 1 through 4 are 1.625, 1.750, 2.250 and 1.958 r e s p e c t i v e l y . S i m i l a r l y , the main e f f e c t due to character QD/C )is not s i g n i f i c a n t , _F ( 8 , 4 8 ) = .700;; p = .600. Table 29 presents the mean t r i a l s to c r i t e r i o n for each of the characters. F i n a l l y , there are no s i g n i f i c a n t i n t e r a c t i o n s . Behavior Set 4: Design 4 For the analysis of the number of errors to c r i t e r i o n , a 2 ( a c q u i s i t i o n groups) x 2 (reversal groups) x 4 ( p r o f i l e s ) x 3 (characters) x 3 (conditions) mixed analysis of variance was performed. This design was described i n Table 23. The va r i a b l e s which were nested and l a t i n i z e d are the same as they were for the analysis of Behavior Set 3. Once again, no s i g n i f i c a n t differences among the between-subject groups were detected. However, the within-subject part of the analysis yielded some i n t e r e s t i n g r e s u l t s . In Table 30 i s a summary of t h i s mixed analysis of variance. The r e s u l t s from the between-subject part of the analysis were almost i d e n t i c a l to those found when using the number of t r i a l s to c r i t e r i o n was used as the dependent measure. Therefore, they w i l l not be repeated here. The r e s u l t s from the within-subject part of the analysis are new. A s i g n i f i c a n t main e f f e c t due to the conditions v a r i a b l e i s e v i -dent, F (2,96) = 3.680; p = .028. The mean number of errors i n Conditions 1 to 3 are 1.396, 1.396, and 1.010 re s p e c t i v e l y . A l l Table 29 Mean T r i a l s to C r i t e r i o n f o r the Characters i n Behavior Set 4 a P r o f i l e s Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 1.875 Character 2 1.375 Character 3 . 1.625 Character 4 2.000 Character 5 1.625 Character 6 1.625 Character 7 2.000 Character 8 1.625 Character 9 3.125 Character 10 1.625 Character 11 2.250 Character 12 2.000 3 The number of observations f o r each mean i s 8. 128 Table 30 Summary of the Analysis of Variance on Errors to C r i t e r i o n i n Behavior Set 4 Source 3 D.F. S.S. M.S. o t a l 287 708. 413 Between-Subj ects 95 391. 075 A 1 1. 253 1. 253 0. 324 0. 578 B 1 1. 531 1. 531 0. 396 0. 539 C 3 7. 982 2. 660 0. 688 0. 566 D/C 8 48. 472 6. 059 1. 567 0. 159 AB 1 0. 281 0. 281 0. 072 0. 778 AC 3 7. 204 2. 401 0. 621 0. 608 BC 3 3. 816 1. 272 0. 329 0. 806 AD/C 8 46. 416 5. 802 1. 501 0. 181 BD/C 8 40. 861 5. 107 1. 321 0. 255 ABC 3 23. 732 7. 910 2. 047 0. 118 ABD/C 8 24. 027 3. 003 0. 777 0. 625 SS/ABD/C 48 185. 500 3. 864 Within-Subj ects 192 317. 338 G 2 9. 506 4. 753 3. 680 0. 028 G X SS 190 307. 832 GA 2 9. 881 4. 941 3. 825 0. 024 GB 2 5. 395 2. 697 2. 088 0. 127 GC 6 10. 548 1. 758 1. 361 0. 237 GD/C 16 47. 0271' 2. 939 2. 275 0. 007 H/C 8 14. 138 1. 767 1. 370 >. 050 LSR 8 32. 889 4. 111 3. 180 <, 050 GAB 2 15. 270 7. 635 5. 911 0. 004 GAC 6 2. 618 6. 436 0. 337 0. 915 GBC 6 10. 381 1. 730 1. 339 0. 246 GAD/C 16 26. 750 1. 671 1. 294 0. 216 AH/C 8 15. 416 1. 927 1. 490 >. 050 LSR 8 11. 334 1. 416 1. 098 >. 050 GBD/C 16 34. 138 2. 133 1. 651 0. 069 BH/C 8 15. 694 1. 961 1. 520 >. 050 LSR 8 18. 444 2. 305 1. 787 >. 050 GABD/C 16 16. 972 1. 060 0. 821 0. 659 ABH/C 8 5. 194 0. 649 0. 503 050 LSR 8 11. 778 1. 470 1. 141 050 G X SS/ABD/C 96 124. 000 1. 291 A = a c q u i s i t i o n group, B = r e v e r s a l group, C = p r o f i l e , D = character, SS = subjects, G = condition, H = dimension, and LSR = l a t i n square r e s i d u a l . 129 possible pairwise comparisons among the condition means were made with • the Newman Keuls t e s t . Subjects made s i g n i f i c a n t l y fewer errors i n condi-t i o n 3 than they made i n Conditions 1,and 2 x^hich did not d i f f e r r e l i a b l y from each other. I t appears that when the c r i t e r i o n was reached on Behavior Set 3, subjects have learned the control condition better than they had learned the other two conditions. There are three s i g n i f i c a n t i n t e r a c t i o n s . F i r s t the a c q u i s i t i o n group x condition i n t e r a c t i o n i s s i g n i f i c a n t , _F (2,96) = 3.680; p = .028. Figure 8 shows t h i s i n t e r a c t i o n . As i s indicated i n that f i g u r e , the r e l a t i o n s h i p of the three conditions to one,another i s very d i f f e r e n t i n the high- and low-valid a c q u i s i t i o n groups. The Newman Keuls test was used to make a l l possible pairwise com-parisons among condition means f o r each a c q u i s i t i o n group. The following differences were found. In the h i g h - v a l i d a c q u i s i t i o n group, subjects made s i g n i f i c a n t l y more errors i n Condition 2,-.than they made i n Conditions 1 and 3 which do not d i f f e r s i g n i f i c a n t l y from each other. In the low-valid a c q u i s i t i o n group, subjects made s i g n i f i c a n t l y more errors i n Condition 1 than they made i n Conditions 2 and 3 which do not d i f f e r s i g n i f i c a n t l y from each other. The r e l a t i o n s h i p i n the low-valid a c q u i s i t i o n group follows Hypothesis 3(a), and 3(b). But the r e l a t i o n s h i p i n the high- v a l i d a c q u i s i t i o n group does not support any part of Hypothesis 3. However, the differences indicated, i n Hypothesis 3(a and b) were not expected to appear when the adjusted t r a i t concept was applied to a new set of behaviors. 130 • 2 1 2 3 CONDITION Figure 8. Mean number of errors i n Behavior Set 4 as a function of the condition and a c q u i s i t i o n groups (HVA = h i g h - v a l i d a c q u i s i t i o n ; LVA = low-valid a c q u i s i t i o n ) . Condition 1 i s the reversed dimension; Condition 2 i s the unchanged dimension, and Condition 3 i s the new dimension. 131 The second s i g n i f i c a n t i n t e r a c t i o n i s the a c q u i s i t i o n group x rev e r s a l group x condition i n t e r a c t i o n , J? (2,96) = 5.911.. p = .004. Figure 9 shows the rever s a l group x condition i n t e r a c t i o n f o r each a c q u i s i t i o n group. By inspection, one can see that i n the high-v a l i d r e v e r s a l group, the r e l a t i o n s h i p among the conditions i s nearly i d e n t i c a l across both a c q u i s i t i o n groups. The Newman Keuls test was used to test a l l possible pairwise comparisons among condition means f o r the hi g h - v a l i d r e v e r s a l group i n both the high-and low-valid a c q u i s i t i o n groups. In both a c q u i s i t i o n groups, s i g n i f i c a n t l y fewer errors were made i n Condition 3 than were made i n Conditions 1 and 2 which do not d i f f e r s i g n i f i c a n t l y from each other. This r e s u l t supports Hypothesis 3(a) but not 3(b). In the low-valid r e v e r s a l group, the re l a t i o n s h i p s among the condition variables changes from one a c q u i s i t i o n group to the other. In the h i g h - v a l i d a c q u i s i t i o n group, subjects made s i g n i f i c a n t l y fewer errors i n Condition 1 than they made i n Conditions 2 and 3. Secondly, s i g n i f i c a n t l y more errors were made i n Condition 2 than were made i n Condition 3. In the low-valid a c q u i s i t i o n group, s i g n i f i c a n t l y more errors were made i n Condition 1 than were made i n Conditions 2 and 3 which did not d i f f e r s i g n i f i c a n t l y from each other. The biggest d i f f e r e n c e between the hi g h - v a l i d and the low-valid a c q u i s i t i o n groups occurs with Conditions 1 and 2. Subjects who formed t h e i r i n i t i a l t r a i t concept by reading h i g h - v a l i d behavioral information had more d i f f i c u l t y learning the reversed condition i f they 132 HVR HVA LVR LVA 1 2 3 CONDITION 1 2 3 CONDITION Mean number of errors i n Behavior Set 4 as a function of the conditions, a c q u i s i t i o n groups (HVA = h i g h - v a l i d a c q u i s i t i o n group; LVA = low-valid a c q u i s i t i o n groups) and reve r s a l groups (LVR= hi g h - v a l i d r e v e r s a l groups; LVR = low-valid r e v e r s a l groups). The conditions are the same as i n Figure 8. 133 t r i e d to adjust t h e i r t r a i t concept by reading h i g h - v a l i d behavioral information than i f they t r i e d to adjust t h e i r t r a i t concept by reading low-valid behavioral information. For subjects i n the low-valid a c q u i s i t i o n group, the r e l a t i o n s h i p was the opposite. In contrast, subjects who formed t h e i r i n i t i a l t r a i t concept by reading h i g h - v a l i d behavioral information had more d i f f i c u l t y learning the unchanged condition i f they adjusted t h e i r t r a i t concept by reading low-valid behavioral information than i f they adjusted t h e i r t r a i t concept by reading h i g h - v a l i d behavioral informa-t i o n . For subjects i n the low-valid a c q u i s i t i o n group, the r e l a t i o n s h i p of the learning d i f f i c u l t y on the unchanged condition was the opposite. F i n a l l y , i n both a c q u i s i t i o n groups, the control condition was easier to learn i f subjects read h i g h - v a l i d behavioral informa-t i o n . The t h i r d s i g n i f i c a n t i n t e r a c t i o n i s the character x condition (D G/C) i n t e r a c t i o n , F (6,96) = 2.275;, p = .007. Figure 10 presents t h i s complicated i n t e r a c t i o n . Notice the comparatively high number of errors i n Conditions 2 and 3 f o r Character 9 and the comparatively low number of errors i n Condition 2 for Character 5. Secondly, note that the r e l a t i o n s h i p among the three conditions d i f f e r s considerably across the 12 characters. These differences are very complex and not much i s gained by describing the r e s u l t s f u rther. It i s reasonable to conclude that which dimension i s i n which contextual arrangement and i n which condition makes a considerable difference i n the a p p l i c a t i o n of the adjusted t r a i t concept to a 134 CONDITION Figure 10. Mean number of errors i n Behavior Set 4 as a function of the condition and character (CH) groups. Note.that the characters are separated into four groups. The conditions are the same as i n Figure 8. 135 a new set of behavioral information. One aspect which i s not described i n Figure 10 i s the dimension (H/C) v a r i a b l e which was purposely confounded within the character x condition (DG/C) i n t e r a c t i o n i n the form of an incomplete l a t i n square. To sort out the e f f e c t s of the dimension (H/C) v a r i a b l e , the character x condition (DG/C) i n t e r a c t i o n must be broken down into two components: (a) the dimension v a r i a b l e (H/C); and (b) the l a t i n square r e s i d u a l (LSR). This breakdown i s shown i n Table 30. Table 30 shows that the dimension (H/C) e f f e c t i s not s i g n i f i c a n t , F (8,96) = 1.370; p >.05, but that the l a t i n square r e s i d u a l (LSR) i s s i g n i f i c a n t , F (8,96) = 3.180; p <.05. The l a t t e r f i n d i n g indicates that the assumption of no r e l a t i o n s h i p between the l a t i n i z e d v a r i a b l e and either component of the i n t e r a c t i o n has been v i o l a t e d i n some way. In order to specify the s i g n i f i c a n t l a t i n square r e s i d u a l further, the character x condition (DG/C) i n t e r a c t i o n was separated into i t s component parts for each p r o f i l e (D). In Table 31 are the component parts for each p r o f i l e . As Table 31 shows, when the component parts are separated according to the appropriate p r o f i l e group, the l a t i n square r e s i d u a l was not found to be s i g n i f i c a n t i n any group. I t appears that when the component parts are broken down i n t h i s manner, the s i g n i f i c a n c e test i s not powerful enough to indicate a s i g n i f i c a n t l a t i n square r e s i d u a l i n any p r o f i l e group. I t should be noted that the analysis of the data from Behavior Set 4 i s not c r u c i a l to any of the hypotheses i n t h i s i n v e s t i g a t i o n . 136 Table 31 Summary of the Analysis of the Latin Square Residual f or Each p r o f i l e r-Source D.F. S.S. M.S. F DG/C, 4 2.416 H/C1 2 1.083 0.541 0.268 >.05 LSR 2 1.333 0.666 0.333 >.05 SS/DG/C1 24 48.333 2.013 DG/C2 4 15.584 H/C2 2 8.583 4.292 1.872 >.05 LSR 2 7.000 3.500 1.527 >.05 SS/DG/C2 24 55.000 2.291 DG/C3 4 16.140 H/C3 2 0.777 0.388 0.145 >.05 LSR 2 15.363 7.681 2.873 >.05 SS/DG/C3 24 64.165 2.673 DG/C. 4 4 17.333 H/ C4 2 3.694 1.847 0.527 >.05 LSR " 2 13.639 6.846 1.956 >.05 SS/DG/C. 4 24 84.083 3.503 C - p r o f i l e , D = character, G = condition, SS = subjects and LSR = latLn square r e s i d u a l 137 However, i t i s of i n t e r e s t to discover that the a p p l i c a t i o n of the adjusted t r a i t concept to a new set of behaviors/, i s either f a c i l i t a t e d or i n h i b i t e d depending upon which a c q u i s i t i o n group, r e v e r s a l group, and/ or condition subjects were placed.in. Up to t h i s point, the r e s u l t s from the analyses of the learning data i n both the a c q u i s i t i o n and the r e v e r s a l parts of the experiment have been presented. We now turn to the analysis of the a t t r i b u t i o n data. A t t r i b u t i o n Data Aft e r they had reached c r i t e r i o n on the' fourth behavior set, subjects were asked to give t h e i r o v e r a l l impression of both f i c t i t i o u s people by responding to a seven-place scale. As described i n the chapter on method, t h i s scale consisted of a seven-place separation between the upper and the lower pole of each of the four dimensions. There were three purposes for t h i s procedure. F i r s t , i t was designed to determine the i n t e r n a l v a l i d i t y of the study - that i s , to see i f subjects c l a s s i f i e d the f i c t i t i o u s people i n a manner appropriate to the preset a t t r i b u t e s of these f i c t i t i o u s people. The second purpose was to assess the e f f e c t s of the v a l i d i t y of behavioral information on post-experimental a t t r i b u t i o n . In other words, would subjects c l a s s i f y the f i c t i t i o u s people more accurately on the four dimensions i f the behavioral information they read was high - v a l i d as opposed to low-valid? T h i r d l y , based on the amount of information subjects had i n each condition, some s p e c i f i c p r edictions 138 about the impressions of the f i c t i t i o u s people were made. These predictions are tested here. Design 5. A 2 ( a c q u i s i t i o n groups) x 2 (reversal groups) x 4 ( p r o f i l e s ) x 3 (characters) x 2 (poles) x 4 (conditions) mixed analysis' of variance was performed on the scale score data. Table 32 presents t h i s design. Notice that the fourth between-subject f a c t o r (characters) i s nested within the t h i r d between-subject f a c t o r ( p r o f i l e s ) . In addition, notice that the dimension v a r i a b l e i s purposefully confounded within the character x condition i n t e r a c t i o n i n the form of an incomplete l a t i n square. The l a t i n square i s incomplete because the characters and the dimensions are nested within the p r o f i l e s . However, across the e n t i r e set of characters, the l a t i n square becomes complete such that each dimension appears i n each condition an equal number of times. In Table 33. i s a summary of the analysis of variance performed on the a t t r i b u t i o n data. Subjects who were i n the h i g h - v a l i d a c q u i s i t i o n group scored s l i g h t l y but not s i g n i f i c a n t l y higher on the scales than did subjects i n the low-valid a c q u i s i t i o n group, J£ (1,48) = 1.364' ; p = .247. . The mean scale scores f o r A c q u i s i t i o n Groups 1 and 2 are 4.307 and 4.211 re s p e c t i v e l y . S i m i l a r l y , the differences i n scale scores by subjects i n Reversal Groups 1 and 2 do not reach an acceptable l e v e l of s i g n i f i c a n c e , _F (1,48) = • 439\_; p = .517' . The mean scale scores f o r Reveral Groups 1 and 2 are 4.286 and 4.232 res p e c t i v e l y . The a c q u i s i t i o n and re v e r s a l groups do not i n t e r a c t s i g n i f i c a n t l y , _F (1,48) = .359 ; p = .558':. 139 Table 32 Analysis of Variance: Design 5 Factors Gl G2 HI H2 H3 H4 HI H2 H3 H4 Dl F 1-2 11 12 13 14 11 12 13 14 C l D2 13 11 13 X4 13 11 12 14 D3 12 13 11 14 12 13 11 14 D4 11 12 14 13 11 12 14 13 C2 D5 14 11 12 13 14 11 12 13 D6 12 14 11 13 12 14 11 13 D7 11 13 14 12 11 13 14 12 C3 D8 14 11 13 11 14 11 13 12 D9 13 14 11 12 13 14 11 12 D10 12 13 14 11 12 13 14 11 C4 D l l 14 12 13 11 14 12 13 11 D12 13 14 11 11 13 14 12 11 Dl 11 12 14 14 11 12 13 14 Cl D2 13 11 13 14 13 Il- 12 14 D3 12 13 12 14 12 IS 11 14 D4 11 12 13 13 11 12 14 13 C2 D5 14 11 12 13 14 11 12 13 D6 12 14 11 13 12 14 11 13 D7 11 13 14 12 11 13 14 12 C3 D8 14 11 13 12 14 11 13 12 D9 13 14 11 12 13 14 11 12 D10 12 13 14 11 12 13 14 11 C4 D l l 14 12 13 11 14 12 13 11 D12 13 14 12 11 13 14 12 11 F95-96 The factors are as follows: A i s a c q u i s i t i o n groups; B i s rever s a l groups; C i s p r o f i l e s ; D i s characters; F i s subjects; G i f poles; H i s conditions; and I i s dimensions. The factor arrangement i n A2 i s exactly the same as i t i s i n A l . 140 Table 33 Summary of the Analysis of Variance on the A t t r i b u t i o n Data Source 3 D.F. S.S. M.S. F P Tota l 767 2401.436 Between-Subj ects 95 109. 308 A 1 1. 782 1. 782 1. 364 0. 247 B 1 0. 574 0. 574 0. 439 0. 517 C 3 2. 181 0. 727 0. 556 0. 650 D/C 8 17. 583 2. 197 1. 683 0. 126 AB 1 0. 470 0. 470 0. 359 0. 558 AC 3 2. 191 0. 730 0. 559 0. 648 BC 3 2. 108 0. 702 0. 538 0. 662 AD/C 8 4. 166 0. 520 0. 398 0. 915 BD/C 8 9. 583 1. 197 0. 917 0. 511 ABC 3 1. 608 0. 536 0. 410 0. 749 ABD/C 8 4. 375 0. 546 0. 418 0. 904 SS/ABD/C 48 62. 687 1. 306 Within-Subj ects 672 2292. 128 G 1 356. 157 356. 157 103. 024 0. 001 H 3 8. 795 2. 931 3. 999 0. 009 GH 3 214. 306 71. 435 17. 135 0. 001 G X SS 95 378. 212 GA 1 0. 032 0. 032 0. 009 0. 885 GB 1 0. 470 0. 470 0. 136 0. 713 GC 3 3. 587 1. 195 0. 345 0. 794 GD/C 8 73. 270 9. 158 2. 649 0. 017 GAB 1 1. 418 1. 418 0. 410 0. 532 GAC 3 5. 358 1. 786 0. 516 0. 676 GBC 3 25. 878 8. 626 2. 495 0. 070 GAD/C 8 30. 812 3. 851 1. 114 0. 370 GBD/C 8 32. 354 4. 044 1. 169 0. 336 GABC 3 22. 826 7. 608 2. 201 0. 098 GABD/C 8 16. 270 2. 033 0. 588 0. 783 G X SS/ABD/C 48 165. 937 3. 457 (continued) 141 Table 33 (continued) e D.F. S. S. M. S. F p a x ss 285 225. 824 HA 3 5. 722 1. 907 2. 602 0. 053 HB 3 3. 347 1. 115 1. 522 0. 209 HC 9 18. 522 2. 058 2. 807 0. 004 HD/C 24 36. 791 1. 533 2. 091 0. 004 I/C 12 35. 650 2. 970 4. 052 <. 010 LSR 12 1. 141 0. 095 0. 129 >. 050 HAB 3 2. 389 0. 796 1. 086 0. 357 HAG 9 4. 199 0. 466 0. 636 0. 765 HBC 9 3. 615 0. 401 0. 548 0. 838 HAD/C 24 21. 375 0. 890 1. 214 0. 238 AI/C 12 10. 505 0. 875 1. 194 >. 050 LSR 12 10. 870 0. 905 1. 235 >. 050 HBD/C 24 11. 958 0. 498 0. 679 0. 865 BI/C 12 5. 548 0. 462 0. 630 >. 050 LSR 12 6. 410 0. 534 0. 728 >. 050 HABC 9 3. 511 0. 390 0. 532 0. 850 HABD/C 24 8. 833 0. 368 0. 502 0. 974 ABI/C 12 3. 753 0. 311 0. 424 >. 050 LSR 12 5. 080 0. 423 0. 577 >. 050 H X SS/ABD/C 144 105. 562 0. 733 (continued) 1 4 2 Table 33 (continued) > D . F . S. S. M. S. F p ;H x ss 2 8 5 1 1 0 8 . 8 1 3 G H A 3 6. 0 7 6 2. 0 2 5 0. 4 8 5 0. 6 9 6 G H B 3 3 5 . 6 8 1 1 1 . 8 9 3 2 . 8 5 3 0. 0 3 8 GHG 9 4 9 . 3 4 5 5. 4 8 2 1. 3 1 5 0. 2 3 3 GHD/C 2 4 9 2 . 0 2 0 3. 8 3 4 . 0. 9 1 9 0. 5 7 5 G H I / C 1 2 8 1 . 8 7 5 6. 8 2 2 1. 6 3 6 >. 0 5 0 L S R 1 2 1 0 . 1 4 5 0. 8 4 5 0. 2 0 2 >. 0 5 0 GHAB 3 1. 4 2 0 0. 4 7 3 0. 1 1 3 0. 9 4 7 GHAC 9 1 3 . 7 6 1 1. 5 2 9 0. 3 6 6 0. 9 4 8 GHBC 9 1 8 . 6 1 5 2 . 0 6 8 0. 4 9 6 0. 8 7 5 GHAD/C 2 4 8 9 . 6 4 5 3 . 7 3 5 0. 8 9 6 0. 6 0 7 G H A I / C 1 2 5 5 . 0 0 5 4. 5 4 3 1. 0 9 0 ># 0 5 0 L S R 1 2 3 4 . 6 4 0 2 . 8 8 6 0. 6 9 2 >. 0 5 0 GHBD/C 2 4 1 2 4 . 4 3 7 5 . 1 8 4 1. 2 4 3 0. 2 1 4 G H B I / C 1 2 4 9 . 2 9 6 4. 1 0 8 0. 9 8 5 >. 0 5 0 L S R 1 2 7 5 . 1 4 1 6. 2 6 1 1. 5 0 2 >. 0 5 0 G H A BC 9 2 7 . 6 4 7 3. 0 7 1 0. 7 3 6 0. 6 7 5 GHABD/C 2 4 4 9 . 8 5 4 2 . 0 7 7 6. 4 9 8 0. 9 7 5 G H A B I / C 1 2 3 3 . 9 9 0 2 . 8 3 2 0. 6 7 0 >. 0 5 0 L S R 1 2 1 5 . 8 6 4 1. 3 2 2 0. 3 1 7 >. 0 5 0 G H X S S / A B D / C 1 4 4 6 0 0 . 3 1 2 4. 1 6 8 A = acquisition group, B = reversal group, C = profile, D = character, S S = subjects, G = pole, H = condition, I = dimension and L S R = l a t i n square residual. 143 There are s l i g h t d i f f e r e n c e s i n scale scores by subjects i n the p r o f i l e groups, but these differences are not s i g n i f i c a n t , F_ (3,48) = .556; p = .650. The mean scale scores i n P r o f i l e s 1 through 4 are 4.214, 4.344, 4.214 and 4.266 r e s p e c t i v e l y . S i m i l a r l y , the main e f f e c t due to the character v a r i a b l e (D/C) i s not s i g n i f i c a n t , F_ (8,48) = 1.683; p = .126. In Table 34 are the mean scale scores for the 12 characters. The scale scores on the upper pole are s i g n i f i c a n t l y higher than those on the lower pole, F (1,48) = 103.024; p = .001. The mean scale scores on Poles 1 and 2 are 4.940 and 3;578 r e s p e c t i v e l y . Notice that the average scale score on the lower pole i s only .578 • units from the center of the scale whereas the average "scale score on the upper pole i s 1.940 units from the center. This suggests an o v e r a l l tendency of subjects to rate the upper pole a t t r i b u t e s further toward the extremes than the lower pole a t t r i b u t e s . The main e f f e c t due to the conditions v a r i a b l e i s s i g n i f i c a n t , F (3,144) = 3.999; p = .009. The mean scale scores of Conditions 1 through 4 and 4.089, 4.250, 4.365, and 4.333 re s p e c t i v e l y . The Newman Keuls test was used to make a l l possible pairwise comparisons among the condition means. Only one d i f f e r e n c e i s s i g n i f i c a n t . The^mean %l*aie^%bbTe>'vih"^o1adil?±^n" "lt-('th1e r ever^ed=~d^im%nsabiT)s i s * s i g n i f i c a n t l y lower ^th'aff t>ti&%eah-.§i?a$e %"dore -Irf:Gondit-io^i--2-2(-"t•he, unchanged" dimension) , 3 (the, -replaced dimension), and 4 (the new dimension) which do not d i f f e r s i g n i f i c a n t l y from each other. Not much can be said about t h i s effeet.because the scale scores on both poles were averaged to obtain t h i s r e s u l t . A l l 144 Table 34' Mean Scale Scores for Each Character 3. P r o f i l e Character Mean P r o f i l e 1 P r o f i l e 2 P r o f i l e 3 P r o f i l e 4 Character 1 Character 2 Character 3 4.609 3,969 4.063 Character 4 Character 5 Character 6 4.469 4.281 4.281 Character 7 Character 8 Character 9 4; 141 4.281 4.219 Character 10 Character 11 Character 12 4.266 4.297 4.234 I cl The number of observations f o r each mean i s 64. 145 that can be said i s that the average i n the reversed condition i s lower ( s i g n i f i c a n t l y lower) than i t i s i n the other three conditions. The pole x condition i n t e r a c t i o n i s s i g n i f i c a n t , F_ (3,144) =17.135'; p = .001 . In Figure 11 are comparisons among condition means f o r each pole. The Newman Keuls test was used to make a l l possible pairwise comparisons among condition means f o r each pole. A l l of the means i n both poles d i f f e r e d s i g n i f i c a n t l y from one another. Of i n t e r e s t i s the f i n d i n g that, f o r both poles, the reversed condition has a score which i s very near the midpoint of the scale. This f i n d i n g supports part (a) of Hypothesis 5. S i m i l a r l y , the scale scores i n the remaining conditions are i n the hypothesized d i r e c t i o n . That i s , the most extreme scores occur i n the unchanged condition where a t o t a l of eight behaviors i n the same d i r e c t i o n were . read. The next-most extreme scores occur i n the cont r o l condition about which four behaviors were read i n the re v e r s a l part of the experiment. . F i n a l l y , the le a s t extreme scores are i n the replaced condition where four behaviors were read i n the a c q u i s i t i o n part of the experiment. Some a d d i t i o n a l variables interacted with the condition v a r i a b l e . F i r s t , the a c q u i s i t i o n group x condition i n t e r a c t i o n very nearly reaches an acceptable l e v e l of s i g n i f i c a n c e , F_ (3,144) = 2.602;i; p = .053 . In Figure 12 i s a comparison of the condition means for each a c q u i s i t i o n group. The Newman Keuls t e s t was used to make a l l possible pairwise comparisons among the condition means for each 146 1 2 3 4 CONDITION Figure 11. Mean scale scores as a function of the conditions and the poles. Condition 1 i s the reversed dimension; Condition 2 i s the unchanged dimension, Condition 3 i s the replaced dimension, and Condition 4 i s the new dimension. 147 HVA • - • LVA 1 2 3 4 CONDITION Figure 12. Mean scale score as a function of the condition and a c q u i s i t i o n groups (HVA - h i g h - v a l i d a c q u i s i t i o n group, LVA = low-valid a c q u i s i t i o n group). The conditions are the same as i n Figure 11. 148 a c q u i s i t i o n group. No s i g n i f i c a n t differences were found i n the high- v a l i d a c q u i s i t i o n group. However, i n the low-valid a c q u i s i t i o n group, the scale score i n Condition 1 i s s i g n i f i c a n t l y lower than the scale scores i n Conditions 2, 3, arid 4 which do not d i f f e r s i g n i f i c a n t l y from each other. The p r o f i l e x condition i n t e r a c t i o n i s also s i g n i f i c a n t , ¥_ (9,144) = 2.807;; p = .004 . In Figure 13 are comparisons of the condition means for each p r o f i l e . The Newman Keuls t e s t was used to make a l l possible pairwise comparisons among condition means for each p r o f i l e . Several s i g n i f i c a n t differences were found. Within P r o f i l e 1, the scale scores on Conditions 2 and 3 (which do not d i f f e r s i g n i f i c a n t l y from each other) are s i g n i f i c a n t l y higher than those i n Conditions 1 and 4 (which do not d i f f e r s i g n i f i c a n t l y from each other). Secondly, i n P r o f i l e 2, the scale score i n Condition 4 i s s i g n i f i c a n t l y higher than the scores i n Conditions 1,2, or 3 which do not d i f f e r s i g n i f i c a n t l y from each other. F i n a l l y , the scale scores i n both Conditions 2 and 3 (which do not d i f f e r s i g n i f i c a n t l y from each other) are s i g n i f i c a n t l y higher than the scale scores i n Condition 1. No s i g n i f i c a n t differences are found i n P r o f i l e s 3 and 4. The i n t e r a c t i o n appears to be mainly due to the comparatively low scale score i n Condition 4 i n P r o f i l e 1, to the comparatively high score on Condition 3 i n the same p r o f i l e and to the comparatively high 149 5.0 o CO w o CO 4.5 4.0 • — B Y — • A—A, = P r o f i l e 1 = P r o f i l e 2 = P r o f i l e 3 = P r o f i l e 4 Figure 13. 1 2 3 4 CONDITION Mean Scale Scores as a function of conditions and p r o f i l e s . The conditions are the same as i n Figure 11. 150 scale score i n Condition 4 i n P r o f i l e 2. A t h i r d i n t e r a c t i o n , the character x condition (DH/C) i n t e r a c t i o n i s s i g n i f i c a n t , F (8,48) = 2.649; p = .017. In Figure 14 are comparisons among condition means for the 12 characters which are nested within the appropriate p r o f i l e s . Notice that the r e l a t i o n s h i p of the condition means to one another d i f f e r s greatly from character to character. Because information from both poles was averaged to obtain the data, very l i t t l e can be deduced from the i n t e r a c t i o n of the p r o f i l e s , characters and a c q u i s i t i o n groups with the condition v a r i a b l e . In these analyses, i t i s impossible to compare the d i s t r i b u t i o n of scale scores on both poles across the d i f f e r e n t conditions. One aspect of the character x condition i n t e r a c t i o n , which i s not described i n Figure 15, i s the dimension v a r i a b l e which was purposefully confounded within that i n t e r a c t i o n . To sort out the character x condition (DG/C) i n t e r a c t i o n , i t was broken, down into two components: (a) the dimension (I/C) v a r i a b l e ; and (b) the l a t i n square r e s i d u a l (LSR). This breakdown i s shown i n Table 33. As Table 33 "shows, the dimension (I/C) e f f e c t i s s i g n i f i c a n t , F (12,144) = 4.052; p < .01. This f i n d i n g suggests that, i n at l e a s t one p r o f i l e , the scale scores on some dimensions were s i g n i f i c a n t l y higher than they were on other dimensions. Note that the l a t i n square r e s i d u a l i s not s i g n i f i c a n t , (F12,144) = 0.129; p > .05. 151 v = CH 1 T = CH 4 B = CH 7 O = CH 10 A = CH 2 A = CH 5 • = CH 8 • = CH 11 O = CH 3 • = CH 6 • = CH 9 < = CH 12 1 2 3 4 1 2 3 4 CONDITION CONDITION Figure 14. Mean Scale Score as a function of conditions and character (CH) groups. Note that the characters are graphed i n four groups. The conditions are the same i n Figure 11. 152 In order to specify the s i g n i f i c a n t dimension (l/C) e f f e c t further, the dimension (I/C) v a r i a b l e and the l a t i n square r e s i d u a l were separated according to the appropriate p r o f i l e group. Table 35 presents t h i s separa-t i o n . As Table 35 shows,- the dimension (I/C) v a r i a b l e i s s i g n i f i c a n t only i n the f o u r t h p r o f i l e group, F (3,36) = 8.614; p < .05. In t h i s p r o f i l e , the mean scale scores f o r Dimensions 1 through 4 are 4.333, 4.104, 4.625, and 4.000. The Newman Keuls test was used to make a l l possible pairwise comparisons among the dimension means i n the fourth p r o f i l e . Two differences are s i g n i f i c a n t . The mean scale score on Dimension _3 (conscientiousness-unconscientiousness) was s i g n i f i c a n t l y higher than the mean scale score on Dimension h_ (emotional s t a b i l i t y -emotional i n s t a b i l i t y ) and Dimension 2_ (agreeableness-disagreeableness) which do not d i f f e r s i g n i f i c a n t l y from each other. Note that the new dimension i n t h i s group i s extroversion-introversion. One f i n a l double i n t e r a c t i o n i s s i g n i f i c a n t , _F (8,48) = 2.649; p = .017. This i s the character x pole (DC/C) i n t e r a c t i o n . In Figure 15 are comparisons of the character means f o r each pole. Since the characters are nested within the p r o f i l e s , the Newman Keuls test was used to make a l l possible pairwise comparisons among character means within each p r o f i l e and f o r each pole. No s i g n i f i c a n t d i f -ferences among characters are found i n eit h e r the upper or the lower pole. By inspection of Figure 15, we see that except f o r the lower pole i n Character 1, a l l scores i n the lower pole f a l l between the center of the scale and that lower pole. A l l scores i n the upper pole f a l l at least .4 units above the center point of the 153 Table 35 Summary of the Analysis of the Dimension Variable (I/C) for Each P r o f i l e : A t t r i b u t i o n Data Source D.F. S.S. M.S. F P i / c 1 3 5.765 1.921 2.654 >.05 s s / i / c 1 36 26.062 0.723 i / c 2 3 12.687 4.229 3.956 >.05 s s / i / c 2 36 38.500 1.069 i / c 3 3 5.040 1.713 2.563 >.05 s s / i / c 3 36 24.062 0.668 i / c 4 3 12.157 * 4.052 8.614 <.05 SS/I/C. 4 36 16.937 0.470 I = dimension, C = p r o f i l e , and SS = subjects. 154 1 2 3 4 5 6 7 8 9 1 0 1 1 CHARACTERS Mean scale scores as a function of the poles and character groups. Note that the characters are separated into four groups. 15,o scale i n the d i r e c t i o n of the upper pole of the dimensions. However,the distance from the center of the scale d i f f e r s f or each character. Secondly, Characters 3 and 10 show the biggest separation between the two poles. It i s l i k e l y that the s i g n i f i c a n t i n t e r -a ction i s due to the quantitative differences among scale scores for each pole. F i n a l l y , there i s one s i g n i f i c a n t t r i p l e i n t e r a c t i o n , F_ (3.144) = 2.853 ; p = .038 :. This i s the r e v e r s a l group x pole x condition i n t e r a c t i o n . In Figure 16 are comparisons among the means of the conditions for each pole but f o r the high- and low-v a l i d r e v e r s a l groups separately. Within each reve r s a l group, the Newman Keuls test was used to make a l l possible comparisons among conditions means f o r each pole. Several s i g n i f i c a n t differences were found. F i r s t are the differences i n the h i g h - v a l i d r e v e r s a l group. In the upper pole, the scale scores i n Conditions 2 and 4 (which do not d i f f e r s i g n i f i c a n t l y from each other) are s i g n i f i c a n t l y higher than are the scale scores i n Conditions 1 and 3 which do not d i f f e r s i g n i f i c a n t l y from each other. No s i g n i f i c a n t differences among condition means were found i n the lower pole, but the pattern of the r e s u l t s appears opposite to those found i n the upper pole. By inspection of Figure 16, we see that there i s very l i t t l e d i fference between the scale scores on the two poles i n the 156 • l i t 1 2 3 4 CONDITION Mean scale score as a function of the poles, the conditions and the r e v e r s a l groups (HVR = h i g h - v a l i d r e v e r s a l group; (LVR = low-valid r e v e r s a l groups). Condition 1 i s the reversed dimension'; Condition 2 i s the unchanged dimension, Condition 3 i s the replaced dimension, and Condition 4 i s the new dimension. CONDITION 157/ reversed condition. S i m i l a r l y , there i s very l i t t l e d i f f e r e n c e between the scale scores of the two poles i n the replaced condition. However, i n both the same condition and the new condition, there i s a s u b s t a n t i a l spread between the scale scores on each dimension. The pattern of r e s u l t s i n the low-valid r e v e r s a l group i s somewhat d i f f e r e n t . In the upper pole, the scale scores i n Condition 1 are s i g n i f i c a n t l y lower than the scale scores i n Conditions 2, 3, and 4 (which do not d i f f e r s i g n i f i c a n t l y from each other). In the lower pole, the reverse i s the case. The mean scale score on Condition 1 i s s i g n i f i c a n t l y higher than the mean scale score on each of the remaining conditions which do not d i f f e r s i g n i f i c a n t l y from each other. When the r e v e r s a l groups are compared on each'condition, i t appears that the e f f e c t of the v a l i d i t y of the information upon which t r a i t concept adjustment takes place d i f f e r s q u a n t i t a t i v e l y i n Conditions 1 and 3, but not i n Conditions 2 and 4. When they ,read low-valid behaviors, subjects appear to place more emphasis upon the information received during r e v e r s a l than they do when they read h i g h - v a l i d behaviors. More importantly, the f i f t h hypothesis receives support from t h i s i n t e r a c t i o n as w e l l . The difference; among the re v e r s a l groups i s a quantitative d i f f e r e n c e . The basic r e l a t i o n s h i p among the conditions remains as predicted. In order to tes t Hypothesis 6, I completed two a d d i t i o n a l analyses. These analyses concerned the re v e r s a l condition f o r both 158; the high- and low-valid r e v e r s a l groups. R e c a l l that f o r the analysis of the a t t r i b u t i o n data, the scale scores were t a l l i e d according to the preset pole pos i t i o n s on each condition. For the reversed condition, the data were scored according to the pole p o s i t i o n which was set for the t r a i t concept formation part of the study. Hence, i n the t r a i t concept adjustment part of the study, the t a l l y f o r the reversed condition i s opposite to the pole designations. The f i r s t analysis concerned the upper pole of the reversed condition. The average scale score on the reversed condition of the h i g h - v a l i d r e v e r s a l group was higher but not s i g n i f i c a n t l y higher than the average scale score on the reversed condition of the low-valid r e v e r s a l group, _t (96) = 1.717; p > .05 (two t a i l e d t e s t ) . The mean scale scores on the reversed condition f o r the hig h - v a l i d and the low-valid r e v e r s a l groups are 4.2708 and 3.6458 res p e c t i v e l y . The second analysis concerned the lower pole of the reversed condition. The average scale score on the reversed condition of the h i g h - v a l i d r e v e r s a l group was lower but not s i g n i f i c a n t l y lower than the scale score on the reversed condition of the low-valid r e v e r s a l group, _t (96) = .16; p > .05. The mean scale scores on the reversed condition f o r the h i g h - v a l i d and the low-valid r e v e r s a l groups are 4.083 and 4.145f,; r e s p e c t i v e l y . Hypothesis 6 was not supported i n eit h e r a n a l y s i s . In f a c t , '15$ the d i r e c t i o n of the r e s u l t s of the upper pole analysis was opposite to Hypothesis 6. Perhaps on the upper pole, the high-v a l i d behavioral information which was read during the t r a i t concept adjustment part of the study was ignored, was discounted or was simply not strong enough to overcome the trend of the i n i t i a l t r a i t concept. 160 Chapter 4 Discussion Several findings emerged from the present i n v e s t i g a t i o n . The f i r s t part of t h i s chapter presents a review of the hypotheses and an overview of the findings which r e l a t e to them. Following the overview i s a discussion of the findings i n r e l a t i o n to the pertinent l i t e r a t u r e . The f i r s t hypothesis y i e l d s the p r e d i c t i o n that t r a i t concepts are easier to form when t h e i r basis i s high - v a l i d behavioral information. The second hypothesis has two parts. I t states that: (a) i n i t i a l t r a i t concepts are easier to adjust ( i . e . , are l e s s r e s i s t a n t to change) when they were formed by reading low-valid behavioral information; and (b) t r a i t concept adjustment i s easier when i t occurs while reading h i g h - v a l i d behavioral information. The t h i r d hypothesis has two parts. I t states that (a) when t r a i t concepts are adjusted to accommodate new information, the dimension which was reversed i s harder to le a r n than are both the unchanged and the new (control) dimensions; and (b) the unchanged dimension i s easier to learn than i s the new dimension. The fourth hypothesis y i e l d s the p r e d i c t i o n that subjects 1' t r a i t impressions are i n accordance with the preset personality a t t r i b u t e s of the f i c t i t i o u s people. The f i f t h hypothesis y i e l d s the pr e d i c t i o n that t r a i t a t t r i b u t i o n r e f l e c t s the amount and the consistency of behavioral information which i s suggestive of p a r t i c u l a r 161 t r a i t dimensions. The s i x t h hypothesis y i e l d s the p r e d i c t i o n that t r a i t a t t r i b u t i o n on the reversed component of the adjusted t r a i t concept v a r i e s as a function of the type of behavioral information subjects read i n both the t r a i t concept formation and the t r a i t concept adjustment parts of the i n v e s t i g a t i o n . In general, t r a i t a t t r i b u t i o n r e f l e c t s the more v a l i d behavioral information. The major findings from the present i n v e s t i g a t i o n were as follows. F i r s t , i n opposition to Hypothesis 1, t r a i t concept formation was not found to vary s i g n i f i c a n t l y as a function of the v a l i d i t y of the behavioral information upon which the t r a i t concepts were formed. However, t r a i t concept formation was found to vary s i g n i f i c a n t l y as a function of the character groups which r e f l e c t d i f f e r e n t contextual arrangements of d i f f e r e n t t r a i t a t t r i b u t e s . Secondly, i n support of Hypothesis 2(a), there i s evidence that t r a i t concepts which were formed by reading h i g h - v a l i d behavioral information are more r e s i s t a n t to change than are t r a i t concepts which were formed by reading low-valid behavioral information. T h i r d l y , t r a i t concept adjustment varied as a function of the v a l i d i t y of the behavioral information which required an adjustment to be made on the i n i t i a l t r a i t concept. However, i t did not vary as predicted by Hypothesis 2(b). Most of the findings which re l a t e d to t h i s hypothesis emerged i n s i g n i f i c a n t i n t e r a c t i o n s . Hence, the e f f e c t s of the v a l i d i t y of the behavioral information upon which 162 t r a i t concept adjustment took place are very complex. Fourthly, the reversed component of the adjusted t r a i t concept was s i g n i f i c a n t l y harder to learn than were the unchanged and the new components. This f i n d i n g supports Hypothesis 3(a). However, i n opposition to Hypothesis 3(b), the unchanged component was not s i g n i f i c a n t l y easier- to learn than was the new component. F i f t h l y , i n support of Hypotheses 4 and 5, t r a i t a t t r i b u t i o n r e f l e c t e d the amount and the consistency of the behavioral information that subjects received throughout the en t i r e study. That i s , the extremity of t r a i t a t t r i b u t i o n increased as a function of an increase i n the amount of consistent behavioral information which the subjects read. Secondly, when subjects read an equal amount of behavioral information which suggested a t t r i b u t e s f o r each f i c t i -tious person on both poles of a p a r t i c u l a r dimension, t r a i t a t t r i b u t i o n r e f l e c t e d an averaging of the t r a i t information. F i n a l l y , i n opposition to Hypothesis 6, t r a i t a t t r i b u t i o n on the reversed component of the adjusted t r a i t concept did not vary s i g -n i f i c a n t l y as a function of the v a l i d i t y of the behavioral information upon which t r a i t concept adjustment took place. In many of the analyses, t r a i t concept formation and change varied s i g n i f i c a n t l y as a function of the content of the t r a i t a t t r i -butes and as a function of the arrangement of t r a i t a t t r i b u t e s into characters. It appears that information which suggests c e r t a i n combinations of t r a i t a t t r i b u t e s i s more or l e s s d i f f i c u l t to integrate depending upon which t r a i t dimensions are used and depending 163 upon which pole positions along these dimensions exist i n combina-t i o n with one another. The r e s u l t s from the present i n v e s t i g a t i o n are pertinent to several t h e o r e t i c a l ideas. With regard to the f i r s t hypothesis, i t was thought that h i g h - v a l i d behavioral information would cause cognitive categories to be formed quickly and, consequently, would enable subjects to d i f f e r e n t i a t e the a t t r i b u t e s of the two f i c t i -t i o us people very e a r l y i n the task (Jones & Goethals, 1972). Hence subjects who read h i g h - v a l i d information should r e a d i l y form conjunctive t r a i t concepts. On the other hand, subjects were expected to reserve t r a i t judgments concerning f i c t i t i o u s people who performed low-valid behaviors because t h i s information i s l e s s r e l i a b l e or v a l i d as a suggestor of s p e c i f i c t r a i t dimensions. These reservations should i n h i b i t t r a i t concept formation (Jones & Goethals, 1972). The r e s u l t s from the analysis of the t r a i t concept formation data do not support these contentions. In f a c t , the trend of the r e s u l t s i n d i c a t e s that t r a i t concept formation occurred s l i g h t l y f a s t e r i n the l o w - v a l i d - a c q u i s i t i o n group. It appears that when they read low-valid behavioral information, subjects did not reserve t r a i t a t t r i b u t i o n judgments any more than did other subjects who read h i g h - v a l i d behavioral information. Secondly, when they read h i g h - v a l i d behavioral information, subjects did not more re a d i l y form cognitive categories than d i d other subjects who read low-valid behavioral information. 164 Two explanations f o r the r e s u l t s of the analysis of the t r a i t concept formation data are as follows. F i r s t , the d i s t i n c t i o n between the two v a l i d i t y groups of behavioral information may not be s u f f i c i e n t to i l l u s t r a t e a d i f f e r e n c e i n the i n t e g r a t i o n of d i f f e r e n t i a l l y v a l i d behavioral information ( c f . Anderson, 1974a, 1974b). Note that there i s much variance among the v a l i d i t y indices of the behavioral items which suggested the t r a i t dimensions (see Appendix A). For example, some h i g h - v a l i d items for ES-EI have v a l i d i t y indices s i m i l a r to the low-valid items f o r C-U and A-D. The fact that items with varying v a l i d i t y indices were combined to suggest a conjunctive t r a i t concept may obscure the e f f e c t . A new study which would make f i n e r d i s t i n c t i o n s between the two v a l i d i t y groups of items may i n d i c a t e a clearer e f f e c t of the v a l i d i t y of behavioral information on t r a i t concept formation. Secondly, the t r a i t concept formation task may not have been s u f f i c i e n t l y d i f f i c u l t to d i f f e r e n t i a t e the two v a l i d i t y groups. Because the a t t r i b u t e s of the two f i c t i t i o u s people were described together subjects received information about the a t t r i b u t e s of both f i c t i t i o u s people from reading about the actions of only one person. This experimental procedure may have greatly f a c i l i t a t e d t r a i t concept formation. Perhaps a procedure where the a t t r i b u t e s of the f i c t i t i o u s people are more independent of one another would show stronger e f f e c t s due to the v a l i d i t y of the behavioral information. It should be noted that the small difference between the v a l i d i t y groups appeared i n the d i r e c t i o n opposite to the f i r s t hypothesis. 165 Hence, a more d i f f i c u l t task or f i n e r d i s t i n c t i o n s between the two v a l i d i t y groups of items may make t h i s difference even greater. If t h i s f i n d i n g should occur, i t would require a new t h e o r e t i c a l basis for i t s explanation. One a d d i t i o n a l f i n d i n g from the analysis of the t r a i t concept formation data i s important. T r a i t concept formation varied s i g n i -f i c a n t l y as a function of the d i f f e r e n t character patterns. These character patterns r e f l e c t e d s p e c i f i c content and context factors which defined the type of t r a i t concepts used i n t h i s study The d i f f e r i n g learning rates suggest that some combinations of t r a i t a t t r i b u t e s are more d i f f i c u l t to integrate than are other combina-ti o n s . In order to discover which aspects of the characters contributed most to the d i f f i c u l t y of t r a i t concept formation, I broke the characters down into t h e i r component t r a i t dimensions. By inspection of the appropriate tables, I discovered that, i n the f i r s t behavior set, the contextual arrangement of two t r a i t dimensions seemed to be an important factor i n t r a i t concept formation. These dimensions were A-D and C-U (see Table 5). In p a r t i c u l a r , i t seemed that, i n any combination of t r a i t a t t r i b u t e s , i f the pole p o s i t i o n on each of these t r a i t dimensions was s i m i l a r (e.g., agreeable and conscientious), t r a i t concepts were formed r a p i d l y . In contrast, i f a pole p o s i t i o n on e i t h e r dimension was opposite to the pole p o s i t i o n on the opposite dimension 166' (e.g., disagreeable and conscientious), t r a i t concepts were d i f f i c u l t to form. A s i g n i f i c a n t character x dimension i n t e r a c t i o n demonstrated that the contextual arrangement of A-D and C-U such that the pole p o s i t i o n on A-D was opposite to the pole p o s i t i o n on C-U was most d i f f i c u l t to learn when the t h i r d dimension was ES-EI (e.g., agreeable, unconscientious and e i t h e r emotionally stable or emotionally unstable). In the second behavior set, the contextual arrangement of A-D and ES-EI seemed to a f f e c t the a p p l i c a t i o n of the i n t i t i a l t r a i t concept to a new set of behaviors which were c a l i b r a t e d and arranged i n the same manner.as they were i n the f i r s t behavior set. In p a r t i c u l a r , i f the pole p o s i t i o n on e i t h e r of these dimensions was opposite to the pole p o s i t i o n on the opposite dimension (e.g., emotional s t a b i l i t y and disagreeable), the a p p l i c a t i o n of the i n i t i a l t r a i t concept to a new set of behaviors was adversely affected. That"is, when presented with t h i s contextual arrangement of A-D and ES-EI, subjects found the discrimination of the two f i c t i t i o u s people d i f f i c u l t to master even though the basis f o r the discrimina-t i o n remained the same as i t was i n the f i r s t behavior set. A possible explanation of the observations concerning these t r a i t dimensions comes from the data whichwere-presented by Norman (1963) and which concerned the r e l a t i v e independence of h i s pe r s o n a l i t y f a c t o r s . Table 2 showed that the second and t h i r d f a c t o rs and the second and fourth factors are f a i r l y highly correlated with each other. It appears that s i m i l a r pole p o s i t i o n s on these dimensions are either observed frequently by subjects who rate t h e i r peers or 167 occur more frequently i n other people independently of conceptualiza-tions of these people. In agreement with Norman's (1963) f a c t o r s , the second, t h i r d and fourth t r a i t dimensions i n t h i s study were A-D, C-U, and ES-EI r e s p e c t i v e l y . The learning d i f f i c u l t y of t r a i t concepts which consist of opposite pole posi t i o n s on e i t h e r A-D and C-U or A-D and ES-EI i s probably due to the p o s i t i v e c o r r e l a t i o n s among these dimensions. That i s , from t h e i r experience with combinations of t r a i t a t t r i b u t e s (as indicated by peer r a t i n g s t u d i e s ) , subjects may not expect to encounter i n d i v i d u a l s whose a t t r i b u t e s on these dimensions are opposite i n polar p o s i t i o n . This explanation i s i n agreement with the findings i n the areas of i m p l i c i t personality theory (Kaplan, 1975; Schneider, 1973) and t r a i t i m p l i c a t i o n research (Wiggins, 1973). Although the s p e c i f i c t r a i t terms which were used i n these areas of study are not . i d e n t i c a l to the t r a i t terms used i n the present i n v e s t i g a t i o n , the r e s u l t s from these areas of research show that subjects expect c e r t a i n combinations of t r a i t s to occur i n i n d i v i d u a l s and do not expect to f i n d other combinations of t r a i t a t t r i b u t e s . Perhaps a study which would investigate the i n t u i t i v e consistency and the frequency of occurrence of characters which were used i n t h i s study may illu m i n a t e some factors important to t r a i t concept formation. For the past several years, studies i n the area of impression formation have been concerned with both context and content e f f e c t s on impression formation (Kaplan, 1975). Both content and context 168 e f f e c t s have been found (Birnbaum, 1974; Hamilton & F a l l o t , 1974; Hamilton & Zanna, 1974; Wyer, 1974) but there i s a controversy concerning the best explanation of these findings (Kaplan, 1974). The a l t e r n a t i v e s are the meaning-change hypothesis and the weighted-average hypothesis. According to Kaplan (1974), the major obstacle to the d i f f e r e n t i a t i o n of these explanations i s that the response of subjects to the person and the response to the t r a i t are d i f f i c u l t to separate. That i s , the evaluation of people i s hard to d i f f e r e n t i a t e from the denotative meaning of t r a i t terms. I t should be noted that the terms, context and content have s p e c i f i c meaning when they r e f e r to impression formation research. The term, context, usually r e f e r s to the evaluative context, i . e . , the s o c i a l d e s i r a b i l i t y or likeableness of t r a i t s which appear along with the target t r a i t . The term, content, i s usually used i n r e l a t i o n to information salience. According to Hamilton and F a l l o t (1974), information salience r e f e r s to the i m p l i c a t i o n a l r e l a t i o n s h i p between the content of the stimulus a t t r i b u t e s describing a person and the judgment of the person that i s being made. For example, one can judge whether he l i k e s another person or whether he respects that person. The term i n t e l l i g e n t , may be s a l i e n t f o r the judgment concerning respect but not for the judgment concerning likeableness. Hence, the content of t h i s t r a i t term, i n t e l l i g e n t , i s relevant only to the j udgment concerning respect. In t h i s i n v e s t i g a t i o n , the terms, content and context, have meanings which d i f f e r from t h e i r meanings i n the impression formation 169 l i t e r a t u r e . Here the term, content, r e f e r s to the d i f f e r e n t t r a i t dimensions which categorize the behaviors of a p a r t i c u l a r f i c t i t i o u s person and which represent the components of a conjunctive t r a i t concept. Each t r a i t term r e f e r s to a s p e c i f i c subset of the t o t a l a t t r i b u t e s of the f i c t i t i o u s people. The term, context, ref e r s to the p a r t i c u l a r arrangement of upper and lower pole positions on each of the t r a i t dimensions which described d i f f e r e n t a t t r i b u t e s of the f i c t i t i o u s people. To a large extent, the pole p o s i t i o n s are confounded with s o c i a l d e s i r a b i l i t y ; upper poles are desirable and lower poles are undesirable. Consequently, evaluation plays an important r o l e i n t r a i t concept formation. However, i t does not seem to play the only r o l e . With the present design, s p e c i f i c content- and context-r e l a t e d f a ctors which influence t r a i t concept formation and change cannot be determined. The f i r s t reason i s that the t r a i t a t t r i b u t e combinations of the two f i c t i t i o u s people were learned together and therefore i t i s impossible to determine which contextual arrangement, the one which described the f i r s t f i c t i t i o u s person or the one which described the second f i c t i t i o u s person, contributed to the d i f f i c u l t y i n forming any p a r t i c u l a r t r a i t concept.. Secondly, i n the present i n v e s t i g a t i o n , the pole p o s i t i o n s along the dimensions are confounded with s o c i a l d e s i r a b i l i t y : the upper poles are s o c i a l l y desirable and the lower poles are s o c i a l l y undesirable. This confusion of s o c i a l d e s i r a b i l i t y with t r a i t terms 170 has been troublesome to most researchers i n the impression formation area of research and i t has been extremely d i f f i c u l t to sort out e f f e c t s due to general evaluations of people and those due to the denotative aspects of the t r a i t terms (Kaplan, 1974). Even though, i n the present i n v e s t i g a t i o n , I cannot spe c i f y what about c e r t a i n a t t r i b u t e combinations caused the d i f f i c u l t y i n learning t r a i t concepts, I can say that the e f f e c t s are d i f f e r e n t for d i f f e r e n t a t t r i b u t e combinations. In p a r t i c u l a r , i t appears that i f f o r one f i c t i t i o u s person the a t t r i b u t e , conscientious, e x i s t s i n a contextual r e l a t i o n s h i p with the a t t r i b u t e , disagreeable, and f o r the other f i c t i t i o u s person the a t t r i b u t e , unconscientious, e x i s t s along with the a t t r i b u t e , agreeable, t r a i t concepts are compara-t i v e l y d i f f i c u l t to form. In part, t h i s r e s u l t may be explained by assuming a general evaluation process which d i r e c t s t r a i t concept formation. In p a r t i c u l a r , subjects evaluate the o v e r a l l a t t r i b u t e s of the f i c t i t i o u s people and then r e s i s t the a s s i m i l a t i o n of e v a l u a t i v e l y inconsistent information ( c f . Kanouse & Hanson, 1972; Kaplan, 1974). If d i f f e r e n t t r a i t a t t r i b u t e combinations lead to d i f f e r e n t o v e r a l l evaluations, then subjects i n d i f f e r e n t character conditions would r e s i s t the a s s i m i l a t i o n of e v a l u a t i v e l y inconsistent information to d i f f e r e n t degrees. These d i f f e r e n t products of 'the general evaluation process would lead to d i f f e r e n t learning rates. On the other hand, one can argue that the denotative information which i s obtained by reading d i f f e r e n t categories of behavioral information 171 may be more d i f f i c u l t to integrate or more l o g i c a l l y inconsistent i f c e r t a i n t r a i t a t t r i b u t e combinations are learned. This explana-t i o n follows from i m p l i c i t personality theory (Kaplan, 1975; Schneider, 1973) and t r a i t i m p l i c a t i o n research (Wiggins, 1973). The r e s u l t s from the present i n v e s t i g a t i o n do not favor one explanation over the other one. A d i f f e r e n t study i s required to i d e n t i f y the causes f o r the d i f f e r e n t learning rate of d i f f e r e n t t r a i t concepts. With regard to the f i r s t two behavior sets, one a d d i t i o n a l observation i s noteworthy. In the f i r s t behavior set, the dimension of the t r a i t concept which had a pole p o s i t i o n opposite to the pole p o s i t i o n on the remaining two dimensions was always the most d i f f i c u l t dimension to learn. Because the pole pos i t i o n s along the dimensions are confounded with s o c i a l d e s i r a b i l i t y , t h i s observa-t i o n suggests that information of opposite valence i n terms of s o c i a l d e s i r a b i l i t y i s d i f f i c u l t to integrate. In the impression formation l i t e r a t u r e , t h i s f i n d i n g i s common '. (Birnbaum, 1974; Kanouse & Hanson, 1972; Kaplan, 1975). However, i n the second behavior set, t h i s pattern of learning appears "only i n -vJ, one-half of the character groups. Perhaps, subjects t r y to form a simple good-bad dichotomy but quickly abandon t h i s simple concept i n favor of a more complex one as the task proceeds. The r e s u l t s from the present i n v e s t i g a t i o n t e l l us very l i t t l e about how subjects go about forming concepts of other people. A 172 study of the strategies which these subjects used when forming a t r a i t concept may illuminate some of the important processes. With regard to the second hypothesis, the r a t i o n a l e f o r the f i r s t part was that h i g h - v a l i d behavioral information should r e s u l t i n strong impressions which are protected against contradic-tory information e i t h e r by the cognitive processes of attention decrement or discounting, or by a strong cognitive press toward the a s s i m i l a t i o n of subsequent information into the e x i s t i n g t r a i t concept. The analysis of the t r i a l s to c r i t e r i o n data indicated support f o r t h i s contention. The r e s u l t s from the analysis of the number of" error© within the f i r s t r e v e r s a l t r i a l and within a f i x e d number of t r i a l s were i n the d i r e c t i o n of t h i s hypothesis a but they did not reach an acceptable l e v e l of s i g n i f i c a n c e . These r e s u l t s may r a i s e suspicions about the r e l i a b i l i t y of the fi n d i n g from the analysis of the t r i a l s data. Perhaps the difference between the high- and low-valid ac-q u i s i t i o n groups had not reached i t s peak within f i v e t r i a l s or the re s o l u t i o n of the inconsistencies beyond some point of moderate int e g r a t i o n was more d i f f i c u l t i n the h i g h - v a l i d r e v e r s a l group. I f t h i s i s the case, one needs to analyze more than f i v e t r i a l s to get a c l e a r p i c t u r e of the r e s u l t s . The r a t i o n a l e behind the second part of Hypothesis 2 was as follows. F i r s t , i t i s more d i f f i c u l t to ignore or to discount subsequent behavioral information i f that information i s h i g h - v a l i d . 173 Such information i s more l i k e l y to demand that some adjustment be made to account f o r the i n c o n g r u i t i e s . Secondly, the contrast caused by the hi g h - v a l i d information should be s u f f i c i e n t f o r subjects not to press f o r the a s s i m i l a t i o n of the new behavioral information into the i n i t i a l t r a i t concept. Hence, subjects should be more prepared to adjust t h e i r t r a i t concept to accommodate the new information. The r e s u l t s of the analysis of both the t r i a l s and the errors data do not support t h i s contention. In f a c t , the r e s u l t s (some of which are s i g n i f i c a n t ) show that more errors were made when subjects adjusted t h e i r t r a i t concept by reading h i g h - v a l i d behavioral information that when they adjusted t h e i r t r a i t concept by reading low-valid behavioral information. This f i n d i n g i s i n the d i r e c t i o n opposite to the p r e d i c t i o n of the second part of Hypothesis 2. One explanation f o r t h i s f i n d i n g i s offered. Concerning the l e v e l of incongruity of t r a i t adjectives, Jones and Goethals (1972) proposed a c u r v i l i n e a r hypothesis to suggest a manner i n which incon-s i s t e n t information i s resolved. According to t h i s hypothesis, behavioral information which was only s l i g h t l y d i f f e r e n t from the early information should be assimilated into the i n i t i a l t r a i t concept. With s l i g h t l y more discrepant information, a contrast should occur such that subjects would begin to adjust t h e i r i n i t i a l t r a i t concept to accommodate that incongruent information. However, beyond some point of moderate discrepancy, processes of i n c r e d u l i t y , d e n i a l or discounting would take over because the contrast i s j u s t too great. 174 Perhaps, the contrast caused by the h i g h - v a l i d behavioral information i n the present i n v e s t i g a t i o n was too great and therefore i n i t i a t e d the processes of d e n i a l , i n c r e d u l i t y or discounting which prohibited t r a i t concept adjustment. If the contrast had not been so extreme, t r a i t concept adjustment would have been f a c i l i t a t e d . This explanation i s unsatisfactory because i t does not explain some a d d i t i o n a l findings. In the analysis of the errors within a f i x e d number of t r i a l s , the r e v e r s a l x character group i n t e r a c t i o n showed that for one-quarter of the characters, the h i g h - v a l i d behavioral information f a c i l i t a t e d t r a i t concept adjustment. In two of these characters (4 and 10) s i g n i f i c a n t l y fewer errors were made i n the h i g h - v a l i d r e v e r s a l group than were made i n the low-valid rev e r s a l group. These two r e s u l t s support the second part of Hypothesis 2. In the majority of the characters, however, the f i n d -ings (some of which were s i g n i f i c a n t ) were opposite to the predicted d i r e c t i o n . Consequently, f o r some t r a i t concepts, the pattern of t r a i t a t t r i b u t e s seems to be j u s t as important as or more,important than the type of behavioral information which presented the contra-dic t o r y information. Perhaps, there i s some kind of i n t u i t i v e consistency to, or a high e x p e r i e n t i a l frequency of, c e r t a i n combinations of the t r a i t a t t r i b u t e s which were used i n t h i s study }whereas c e r t a i n other combinations are r a r e l y experienced or do not seem to be l o g i c a l . As was mentioned e a r l i e r , i n the r e l a t e d areas of i m p l i c i t personality theory(Kaplan, 1975; Schneider, 1973) and t r a i t 175 implication research (Wiggins, 1973), c e r t a i n t r a i t s have been found to be strongly implied by c e r t a i n other t r a i t s ^ whereas other t r a i t s appear to be more independent of one another. Perhaps a s i m i l a r r e l a t i o n s h i p e x i s t s among the t r a i t dimensions which were used i n the present i n v e s t i g a t i o n . Further, perhaps the v a l i d i t y of the inconsistent behavioral information influences t r a i t concept adjustment to 1 the extent that the adjusted t r a i t concept i s i n t u i t i v e l y consistent or has a high ex-p e r i e n t i a l frequency. If the adjustment i s to a more i n t u i t i v e l y consistent or to a more frequently experienced combination of t r a i t a t t r i b u t e s , h i g h - v a l i d behavioral information f a c i l i t a t e s the adjustment. On the other hand, i f the required adjustment i s to an i n t u i t i v e l y inconsistent or to a rare combination of a t t r i b u t e s , the h i g h - v a l i d behavioral information i s denied or rendered i n c r e d i b l e . Hence, t r a i t concept adjustment i s prolonged. Future investigations concerning the i n t u i t i v e consistency or the e x p e r i e n t i a l frequency of the various combinations of t r a i t a t t r i b u t e s would appear to be b e n e f i c i a l i n assessing the e f f e c t s of h i g h - v a l i d behavioral information on t r a i t concept adjustment. Such research should be easy to carry out. The influence of the v a l i d i t y of the behavioral information upon t r a i t concept adjustment i s complicated i n the analysis of the errors within the f i r s t r e v e r s a l t r a i l by a s i g n i f i c a n t a c q u i s i t i o n x r e v e r s a l group i n t e r a c t i o n . This i n t e r a c t i o n showed that, a f t e r they formed t h e i r i n i t i a l t r a i t concept by reading low-valid behavioral information, 176 subjects who adjusted t h e i r t r a i t concept by reading low-valid behavioral information made fewer errors than did subjects who adjusted t h e i r t r a i t concept by reading h i g h - v a l i d behavioral information. In contrast, a f t e r they formed t h e i r i n i t i a l t r a i t concept by reading h i g h - v a l i d behavioral information, subjects who adjusted t h e i r t r a i t concept by reading h i g h - v a l i d behavioral information made fewer errors than did subjects who adjusted t h e i r t r a i t concept by reading low-valid behavioral information. This a c q u i s i t i o n x reve r s a l group i n t e r a c t i o n shows that the fewest number of errors i n t r a i t concept adjustment were made when both t r a i t concept formation and t r a i t concept adjustment occurred while subjects read low-valid behavioral information. It was expected that i n i t i a l t r a i t concepts which were formed by reading low-valid behavioral information would not be r e s i s t a n t to change because the informational base i s low-valid (Hypothesis 2a). But i t was not expected that t r a i t concept adjustment would be the f a s t e s t when subjects read new behavioral information of low-validity,because t h i s new information was not expected to present much of a contrast to the i n i t i a l t r a i t concept (Hypothesis 2b). In addition, the i n t e r a c t i o n shows that the greatest number of errors within the f i r s t r e v e r s a l t r i a l occurred under conditions of low-valid a c q u i s i t i o n and h i g h - v a l i d r e v e r s a l . Again t h i s f i n d i n g was unexpected. The i n i t i a l t r a i t concept which was based on low-valid behavioral information was not expected to be very r e s i s t a n t to change (Hypothesis 2a). Moreover, the subsequent 177 behavioral information which was h i g h - v a l i d was expected to contrast with the i n i t i a l t r a i t concept such that subjects would make immediate adjustments to the i n i t i a l t r a i t concept (Hypothesis 2b). These two conditions were expected to enable optimal t r a i t concept adjustment. However, subjects had the most d i f f i c u l t time adjusting t h e i r t r a i t concepts when they were given t h i s combination of v a l i d i t y . F i n a l l y , the remaining v a l i d i t y combinations were not much easier to learn than was the immediately preceeding combination. Of the two which remained, the more d i f f i c u l t adjustment took place when subjects read h i g h - v a l i d behavioral information to form t h e i r i n i t i a l t r a i t concept but read low-valid items to adjust that concept. Given t h i s arrangement of v a l i d i t y , I expected t r a i t concept adjustment to be very d i f f i c u l t . In t h i s more d i f f i c u l t combination, the i n i t i a l t r a i t concept which was based on h i g h - v a l i d behavioral information was expected to be very r e s i s t a n t to change (Hypothesis 2a). Because the subse-quent behavioral information was low-valid, subjects were expected to ignore, discount or assimilate that information (Hypothesis 2b). Hence, t r a i t concept adjustment was expected to be very d i f f i c u l t . The remaining condition of adjustment took place when subjects read h i g h - v a l i d behavioral information during both the t r a i t concept formation and the t r a i t concept adjustment parts of the study. Given t h i s combination of v a l i d i t y , I expected t r a i t concept adjustment to be r e l a t i v e l y easy. Even though the i n i t i a l t r a i t 178 concept was expected to be quite r e s i s t a n t to change (Hypothesis 2a), the subsequent information (which was high-valid) was expected to contrast with that i n i t i a l information such that t r a i t concept adjustment would be i n i t i a t e d immediately (Hypothesis 2b). These findings suggest a very complex i n t e r p l a y of behavioral information of d i f f e r e n t v a l i d i t i e s . I t seems that subsequent behavioral information influences t r a i t concept adjustment as a function of the type of contrast i t makes with the anchor which was established during t r a i t concept formation. If subsequent behavioral information greatly contrasts with the anchor (as i t does when i n i t i a l t r a i t concepts are formed by reading low-valid behavioral information but are adjusted by reading h i g h - v a l i d behavioral information), the subsequent information i s denied or discounted. Consequently, t r a i t concept adjustment becomes very d i f f i c u l t . Secondly, i f subsequent behavioral information moderately contrasts the anchor (as i t does (a) when t r a i t concept are formed and are adjusted by reading low-valid behavioral information; and (b) when t r a i t concepts are formed and adjusted by reading h i g h - v a l i d behavioral information), the contrast i s s u f f i c i e n t to i n i t i a t e t r a i t concept adjustment immediately. F i n a l l y , i f the subsequent behavioral information only s l i g h t l y contrasts the i n i t i a l t r a i t concept (as i t ' does when t r a i t concepts are formed by reading h i g h - v a l i d behavioral information, but are 179 adjusted by reading low-valid behavioral information), processes of ignoring, discounting or a s s i m i l a t i n g take place. Consequently, t r a i t concept adjustment i s postponed u n t i l more evidence i s brought f o r t h . These explanations which are based on the t h e o r e t i c a l statements of Jones and Goethals (1972) are very t e n t a t i v e and require a d d i t i o n a l research for t h e i r confirmation. I t i s important to note that a s i g n i f i c a n t a c q u i s i t i o n x re v e r s a l group i n t e r a c t i o n was not found i n the analysis of the errors within a fixed number of t r i a l s . I t appears that, as subjects progress i n t h e i r adjustment of t r a i t concepts, the influence of d i f f e r e n t i a l l y v a l i d behavioral information fades out. The f i r s t part of Hypothesis 3 yielded the p r e d i c t i o n that when t r a i t concepts are adjusted to accommodate new and contradictory behavioral information, the component of the adjusted t r a i t concept which was reversed i s harder to adjust than i s either the unchanged or the new component. The r a t i o n a l e for t h i s p r e d i c t i o n i s that negative transfer from the i n i t i a l t r a i t concept should be the greatest on the reversed component. The r e s u l t s from the analysis of the errors within the f i r s t r e v e r s a l t r a i l and within a fi x e d number of t r i a l s confirmed t h i s p r e d i c t i o n . Subjects made more errors on the reversed component of the t r a i t concepts than they made on eit h e r the unchanged or the new component. 180 Two findings from the analysis of the errors within the f i r s t r e v e r s a l t r i a l q u a l i f y t h i s r e s u l t somewhat. F i r s t , the rev e r s a l x condition group i n t e r a c t i o n indicated that there was one exception to the f i r s t part of Hypothesis 3. In the low-valid r e v e r s a l group, the reversed component of the adjusted t r a i t concept was harder but not s i g n i f i c a n t l y harder to learn than was the new component. Secondly, the reve r s a l x p r o f i l e x condition group i n t e r a c t i o n showed that, i n the low-valid r e v e r s a l group, the pattern of learning on the three components of the t r a i t concepts d i f f e r e d as a function of the p r o f i l e groups. In some p r o f i l e groups, the reversed components was s i g n i f i c a n t l y easier to learn than was the new component. This f i n d i n g contradicts the f i r s t part of Hypothesis 3. Perhaps the contrast which was presented by low-valid behavioral information was too weak to exert a consistent influence upon the d i f f e r e n t components of the adjusted t r a i t concepts and therefore the content (as represented by the d i f f e r e n t p r o f i l e groups) became a more important determinant of t r a i t concept adjustment. Research concerning the e f f e c t of the content of t r a i t concepts on the adjustment of those t r a i t concepts may ill u m i n a t e t h i s f i n d i n g . The second part of Hypothesis 3 yielded the p r e d i c t i o n that the new component of the adjusted t r a i t concept would be s i g n i f i c a n t l y harder to learn than the unchanged component. The r a t i o n a l e behind t h i s p r e d i c t i o n was that there should be p o s i t i v e t r a n s f e r from the t r a i t concept forma-t i o n part of the study which would make the unchanged dimension easy to learn. Because i t i s new, learning on the 181 new component should not be i n h i b i t e d or enhanced by any type of learning on the i n i t i a l t r a i t concept. The r e s u l t s from the analysis of the errors within the f i r s t r e v e r s a l t r i a l and within a f i x e d number of t r i a l s supported t h i s hypothesis. That i s , o v e r a l l , subjects made more errors (but not s i g n i f i c a n t l y more errors) on the new component than they made on the unchanged component. It appears that i f any negative transfer was displaced on to the unchanged component, i t was very quickly d i s p e l l e d and subjects responded according to t h e i r previous learning on t h i s component. Two findings from the analysis of the f i r s t r e v e r s a l t r i a l indicated that the r e l a t i o n s h i p between the learning on the unchanged component and learning on the new component changed r a d i c a l l y depending upon the v a l i d i t y of the behavioral information arid upon the content of the t r a i t concepts. With some combinations of the v a l i d i t y of behavioral information which was read to adjust t r a i t concepts and of the content of the t r a i t concept, the r e s u l t s appeared i n the hypothesized d i r e c t i o n . In d i f f e r e n t combinations, the differences (some of which were s i g n i f i c a n t ) were opposite to the p r e d i c t i o n . Because appropriate controls were not used i n the present i n v e s t i g a t i o n , one cannot specify which aspects of the task f a c i l i t a t e d or i n h i b i t e d the learning of the d i f f e r e n t components to the adjusted t r a i t concept. The pattern of transfer depends upon the p a r t i c u l a r combination of personality a t t r i b u t e s which are r e f l e c t e d i n 182 the combinations of pole positions along the t r a i t dimensions. A new between-subject design may contribute knowledge i n t h i s regard. Af t e r they reached the c r i t e r i o n on the t h i r d behavior set, subjects were given the fourth behavior set which was c a l i b r a t e d and assembled i n a manner i d e n t i c a l to the t h i r d set. The purpose of t h i s procedure was to determine whether subjects i n d i f f e r e n t experimental conditions could e a s i l y apply t h e i r adjusted t r a i t concept to a new set of behaviors and to ensure that subjects read an equal amount of behavioral information during both the t r a i t concept formation and the adjustment parts of the experiment. The r e s u l t s from the analysis of the t r i a l s data did not uncover any s i g n i f i c a n t differences among the several groups of subjects who applied t h e i r adjusted t r a i t concept, to a new set of behaviors. However the analysis of the errors to c r i t e r i o n i n the fourth behavior set showed that subjects had a more d i f f i c u l t time applying the reversed and the unchanged components of the adjusted t r a i t concept than they had applying the new component. Further, the pattern of learning d i f f i c u l t y on the three components d i f f e r e d as a function of the v a l i d i t y combinations across t r a i t concept formation and adjustment, and as a function of the character groups. These differences are very complex. Because the appropriate controls were not used i n t h i s design, i t i s not possible to specify the p a r t i c u l a r sources of i n t e r -ference which was i n i t i a t e d by learning i n the e a r l i e r parts of the experiment. Therefore not much i s gained by t r y i n g to sort 183 out which t r a i t a t t r i b u t e patterns and v a l i d i t y combinations contributed to the learning d i f f i c u l t y of which component. The complex e f f e c t does seem worth pursuing i n future research projects which would deal more d i r e c t l y with t r a i t a t t r i b u t e combinations. The fourth hypothesis yielded the p r e d i c t i o n that subjects' impressions of the f i c t i t i o u s people would be i n accordance with the preset personality a t t r i b u t e s of the f i c t i t i o u s people. This p r e d i c t i o n was based on the assumption that subjects who read behavioral i n f o r -mation which was generated by an independent group of subjects on the basis of c e r t a i n t r a i t dimensions should be able to recognize those t r a i t dimensions. The r e s u l t s from the analysis of the a t t r i b u t i o n data showed that, o v e r a l l , subjects did respond to the scale dimensions appropriate to the d i r e c t i o n which was suggested by the behaviors they read. This f i n d i n g suggests a high degree of i n t e r n a l v a l i d i t y to the experiment. I t i s noteworthy that subjects tended to rate t h e i r impressions of the f i c t i t i o u s people more toward the upper pole of the dimensions than they did toward the lower pole. This tendency suggests some reluctance on the part of subjects to rate people extremely on the s o c i a l l y undesirable end of the scale. The average of the scale scores on the two poles was s i g n i f i c a n t l y lower i n the reversed component than i t was i n the remaining com-ponents of the t r a i t concept. I t i s d i f f i c u l t to specify the s i g n i -ficance of t h i s f i n d i n g because the d i s t r i b u t i o n of the data upon which the average on the reversed component was obtained d i f f e r s 184 from the d i s t r i b u t i o n upon which the average on the remaining com-ponents was obtained. In p a r t i c u l a r , on the unchanged, the replaced, and the new components, the average was obtained by summing extreme scores on each of the two poles of that p a r t i c u l a r dimension, whereas most of the scores which were averaged to obtain the average on the reversed component were near the midpoint of the scale. Perhaps subjects are l e s s lenient i n t r a i t a t t r i b u t i o n when they have been given contradictory information. That i s , they tend to rate an inconsistent person as being le s s s o c i a l l y desirable than they do a consistent person who possesses s o c i a l l y un-desirable t r a i t s . Once again, the evidence for t h i s tendency i s very weak, but i t does suggest a possible area of inquiry. With.regard to Hypothesis 5, the supporting r a t i o n a l e was based on the notion of how the amount and the consistency of behavioral information subjects read about each component of the t r a i t concept would influence t r a i t a t t r i b u t i o n . In general, i t was expected that the more consistent information subjects read, the stronger would be t h e i r impressions along the dimensions which were suggested by those behaviors. Secondly, i t was expected that i f they were given an equal amount of evidence which suggested opposite poles of a p a r t i c u l a r t r a i t dimension, the subjects would rate t h e i r impressions near the midpoint between the poles of that dimension. The r e s u l t s from the analysis of the a t t r i b u t i o n data supported a l l of the predictions i n Hypothesis 5. The pole x condition 185 i n t e r a c t i o n indicated that the impressions on the reversed component were near the midpoint between the two poles of the appropriate dimensions, that the impressions on the unchanged component were extreme i n the appropriate d i r e c t i o n , and that the impressions on the replaced and the new components were moderately extreme i n the appropriate d i r e c t i o n . The impressions on the new component were s l i g h t l y (but not s i g n i f i c a n t l y ) more extreme than they were on the replaced dimension. These r e s u l t s suggest that laymen a t t r i b u t e t r a i t concepts i n proportion to the amount of consistent information they have about behavioral performances which can be s i m i l a r l y classed. Secondly, the r e s u l t s suggest that, i f they have an equal amount of information which suggests each pole of the t r a i t dimensions, subjects' im-pressions r e f l e c t an average of that behavioral information (Anderson, 1974a, 1974b; Kaplan, 1975). These findings support the summary view of t r a i t s which holds that t r a i t concepts are c a t e g o r i c a l summaries of past conduct to date. Let i t be clear, however, that these r e s u l t s do not necessa r i l y bring negative evidence to bear on the psychodynamic model of t r a i t concept usage. I t was pointed out e a r l i e r that i t i s very d i f f i c u l t to d i f f e r e n t i a t e the two models of t r a i t concept usage i n terms of t h e i r predictions concerning t r a i t a t t r i b u t i o n . The condition v a r i a b l e interacted with each of the following variables;, a c q u i s i t i o n groups, p r o f i l e groups, and character groups. Very l i t t l e can be deduced from these i n t e r a c t i o n s because information 186 on the two poles was averaged to obtain these data. The most that can be said i s that the average of scores from the two poles of the component dimensions varied as a function of a c q u i s i t i o n , p r o f i l e , and character groups. A few speculative comments are offered. F i r s t , the p r o f i l e x condition i n t e r a c t i o n may suggest that with some combinations of t r a i t concepts, subjects are more negative i n t h e i r impressions. Secondly, the character x condition (DH/C) i n t e r a c t i o n shows that the quantitative differences between scale scores on the two poles of the t r a i t dimensions varied as a function of the character groups, but these differences did not deviate from the basic r e l a t i o n s h i p of high pole scores to low pole scores. It appears that f o r some characters, subjects were more extreme i n t h e i r ratings than they were f o r other characters. Future research may uncover important content and context-related factors which contribute to the differences i n the extremity of t r a i t a t t r i b u t i o n . T r a i t a t t r i b u t i o n on the four components for each pole d i f f e r e d as a function of the r e v e r s a l groups. The r e v e r s a l x pole x condition group i n t e r a c t i o n showed that the e f f e c t of the v a l i d i t y of the informaton upon which t r a i t concept adjustment took place d i f f e r e d q u a n t i t a t i v e l y i n components 1 (the reversed component) and 3, (the replace component) but not i n components 2 (the unchanged component) and 4, (the new component)- The scale scores on components 1 and 3 were lower i n the low-valid r e v e r s a l group. It appears that when they read low-valid behavioral information, subjects placed more emphasis upon the informa-t i o n received during the reversal part of the experiment than they did 187 when they read h i g h - v a l i d behavioral information. This d i f f e r e n c e , which i s s l i g h t , i s opposite to the inference which was made i n Hypothesis 6. With regard to Hypothesis 6, the supporting r a t i o n a l e was as follows. I f subjects formed t h e i r i n i t i a l t r a i t concept by reading ei t h e r high- or low-valid behavioral information, but adjusted that concept on the basis of h i g h - v a l i d behavioral information, t r a i t a t t r i b u t i o n should r e f l e c t a recency e f f e c t because the new and contradictory information i s too v a l i d to be ignored, discounted, or assimilated. In contrast, i f they formed t h e i r i n i t i a l t r a i t impression on the basis of ei t h e r high- or low-valid behavioral information, but adjusted that t r a i t concept by reading low-valid behavioral information, t r a i t a t t r i b u t i o n should r e f l e c t a primacy e f f e c t because the new behavioral information should tend to be ignored, discounted, or assimilated into that i n i t i a l t r a i t concept. Comparisons of the scale scores f o r each reve r s a l group were made separately f o r subjects who read behavioral information which suggested the appropriate pole of the reversed component. The r e s u l t s from these analyses do not support Hypothesis 6. There were no s i g n i f i c a n t differences among scale scores of subjects.in the re v e r s a l groups f or ei t h e r the upper or the lower pole of the d i -mension which happened to be i n the reversed component. Perhaps . the amount of information subjects read with regard to the polar pos i t i o n s on the t r a i t dimensions was more important than the type of information. 188 One f i n a l t r a i t a t t r i b u t i o n f i n d i n g i s noteworthy. The character x condition (DH/C) i n t e r a c t i o n was s i g n i f i c a n t . L a t i n i z e d within t h i s i n t e r a c t i o n was the t r a i t dimensions which formed the basis f o r the d i f f e r e n t components. When i t was broken down into the e f f e c t s due to the dimension v a r i a b l e the character x condition (DH/C) i n t e r a c t i o n showed that the average scale score on the C-U dimension, i n P r o f i l e 4, was s i g n i f i c a n t l y higher than the scale score on the ES-EI and the A-D dimension. The d i f f e r e n t scale scores on these dimensions may r e f l e c t the d i f f e r e n t mean v a l i d i t y indices of the two groups of items which suggest the respective dimensions. Table 4 indicated that, i n the hi g h - v a l i d group, the average v a l i d i t y index for the C-U items was much higher than the average v a l i d i t y index f o r the ES-EI items. I f subjects a t t r i b u t e t r a i t concepts i n proportion to the v a l i d i t y of the information they are given , then one would expect more extreme ratings on the C-U dimension. A second explanation,which i s sheer speculation, i s that people are more lenient with t r a i t a t t r i b u t i o n s on the C-U dimension than they are on the ES-EI dimension. Future research may suggest aspects about the dimensions which would account for the differences of t r a i t a t t r i b u t i o n . Many of the factors which influenced t r a i t concept formation and change cannot be sorted out by analyzing the data from the present experiment. However, the findings from t h i s i n v e s t i g a t i o n suggest some promising research leads. I suggest that several smaller experiments which would study d i f f e r e n t aspects of the content and 189 and context-related factors which influence t r a i t concept formation and change be carried out. They may uncover important processes which underly these human a c t i v i t i e s . In conclusion, some comments concerning the Fishbein and Ajzen :(1974) approach to v a l i d i t y are offered. While the v a l i d i t y index may be a good psychometric approach to the t r a i t behavior r e l a t i o n s h i p , i t may be an unpsychological approach. That i s , when they apply t r a i t concepts to the behavior of other people, laymen may not focus upon the extent to which a behavior suggests the t r a i t i n question. Rather, they may be more concerned with the amount of si m i l a r behaviors that have been performed and with the d i v e r s i t y of si t u a t i o n s i n which the performances are observed. A study which used these more psychological d e f i n i t i o n s of v a l i d i t y may show a c l e a r e r r e l a t i o n s h i p between behavioral v a l i d i t y and t r a i t concept formation and change. Prospect A few improvements which would enable a better test of the e f f e c t of the v a l i d i t y of behavioral information on t r a i t concept formation are suggested. These improvements concern:.; (a) the item pool; (b) the t r a i t concept formation task; and (c) ad d i t i o n a l aspects of the t r a i t concepts which subjects were expected to form. The item pool may be improved i n two ways. F i r s t , f i n e r d i s t i n c t i o n s can be made among v a l i d i t y indices both within and across t r a i t dimension item pools. Secondly, psychometric properties i n addit i o n to the v a l i d i t y index may be taken into consideration when behavioral items are chosen for a concept formation task. For example, behavioral items may be equated i n terms of the extent to 190 which t h e i r performance suggests the performance of other behaviors. Secondly, the items may be equated i n terms of the extent to which they suggest (both q u a n t i t a t i v e l y and q u a l i t a t i v e l y ) t r a i t dimensions i n a d d i t i o n to the t r a i t dimension from which they were generated. The t r a i t concept formation task may be improved i n two ways. F i r s t , i t can be made more d i f f i c u l t by adding a t h i r d f i c t i t i o u s person who would have to be d i f f e r e n t i a t e d from the f i r s t two f i c t i t i o u s people who i n turn have to be d i f f e r e n t i a t e d from each other. Secondly, one could arrange the characters of the f i c t i t i o u s people such that the people do not ne c e s s a r i l y possess a t t r i b u t e s on opposite poles of the same t r a i t dimension. This procedure would reduce the amount of information about each f i c t i t i o u s person which i s implied by the a t t r i b u t e s of other f i c t i t i o u s people. Some a d d i t i o n a l information about the characters which were used i n the present i n v e s t i g a t i o n may enable an experimenter to co n t r o l for sources of variance due to differences among the t r a i t concepts i n terms of t h e i r i n t u i t i v e consistency, s o c i a l d e s i r a b i l i t y , and e x p e r i e n t i a l frequency i n everyday l i f e . This information may be obtained by giving subjects the behavioral items which correspond to the appropriate characters and asking them to describe (either by a free d e s c r i p t i o n or by responding to a c h e c k l i s t ) a person who performed a p a r t i c u l a r set of behaviors. In t h i s manner, each character could be rated along several dimensions such as per s o n a l i t y type, s o c i a l d e s i r a b i l i t y , a f f i l i a t i o n p u l l , a f f i n i t y , l o g i c a l coherence, incidence i n the general population, etc. This information 191 would a f f o r d many types of con t r o l as well as suggest new parameters of t r a i t concept formation and change. Some v a r i a t i o n s on the t r a i t a t t r i b u t i o n part of the study may suggest important processes of t r a i t a t t r i b u t i o n . F i r s t , one could systematically increase the amount of information subjects read about the d i f f e r e n t t r a i t a t t r i b u t e s and see i f t r a i t a t t r i b u t i o n v aries as a l i n e a r function of the . amount of information which i s obtained. Secondly, one could study the influence of inconsistent information on t r a i t a t t r i b u t i o n by f i r s t having subjects form an i n i t i a l t r a i t concept by reading the same amount of behavioral information and then giving d i f f e r e n t groups of subjects d i f f e r i n g amounts of inconsistent information. The study of t r a i t concept formation and change i s i n i t s very early stages. The focus of the present i n v e s t i g a t i o n was l i m i t e d to one point of view and to only a few parameters. Much can be gained by studying d i f f e r e n t parameters from new perspectives. 192 References Anderson, N. H. Primacy e f f e c t s i n p e r s o n a l i t y impression formation using a generalized order e f f e c t paradigm. Journal of  Per s o n a l i t y and S o c i a l Psychology, 1965, 2^ , 1-9. Anderson, N. H. Cognitive algebra: Integration theory applied to s o c i a l a t t r i b u t i o n . In L. Berkowitz (Ed.), Advances i n  experimental s o c i a l psychology. Vol. 7. New York: Academic Press, 1974. (a) Anderson, N. H. Information i n t e g r a t i o n theory: A b r i e f survey. In D. H. Krantz, R. C. Atkinson, R. D. Luce, & P. Suppes (Eds.), Contempory developments i n mathematical psychology. Vol. 2. New York: Academic Press, 1974. (b) Anderson, N. H., & Jacobson, A. E f f e c t of stimulus inconsistency and discounting i n s t r u c t i o n s i n personality impression formation. Journal of Personality and S o c i a l Psychology, 1965, _2, 531-539. Anderson, N. H., & Norman, A. Order e f f e c t s i n impression formation i n four classes of s t i m u l i . Journal of Abnormal and S o c i a l Psychology, 1964, 60, 467-471. Argyle, M., & L i t t l e , B. R. Do personality t r a i t s apply to s o c i a l behavior? Journal f o r the Theory of S o c i a l Behavior, 1972, 2, 1-35. Asch, S. E. Forming impressions of personality. Journal of Abnormal • and S o c i a l Psychology, 1946, 41, 258-290. A v e r i l l , J . R. The d i s - p o s i t i o n of psychological d i s p o s i t i o n s . Journal of Experimental Research i n Personality, 1973, 6^  275-282. 193 Birnbaum, M. H. The nonadditivity of personality impressions. Journal of Experimental Psychology, 197'4, 192, 543-561. Briscoe, M. E., Woodyard, H. D., & Shaw, M. E. Personality impression as a function of the favorableness of f i r s t impressions. Journal of Personality, 1967, 35, 343-357. Bourne, L. E. J r . Human conceptual behavior. Boston: A l l y n and Bacon, 1966. Bourne, L. E. J r . , Ekstrand, B. R. , & Dominowski, R. L. The psychology of thinking. Englewood C l i f f s : P r e n t i c e - H a l l , 1971. Bourne, L. E. J r . , & M i l l e r , S. Some stimulus v a r i a b l e s a f f e c t i n g s o l u t i o n s h i f t performance. Journal of Experimental Psychology, 1973, 98, 291-296. Bowers, K. B. Situationism i n psychology: An analysis and c r i t i q u e . Psychological Review, 1973, 80, 307-336. Cohen, A. R. Cognitive tuning as a fac t o r a f f e c t i n g impression formation. Journal of Pers o n a l i t y , 1961, 2J), 235-245. Cook, M. Interpersonal perception. Hardmondsworth,.England: Penguin, 1971. Fishbein, M. , & Ajzen, I. Attitudes and opinions. Annual Review of Psychology, 1972,.23, 487-544. Fishbein, M., & Ajzen, I. Attitudes toward objects as predictors of s i n g l e and multiple behavioral c r i t e r i a . Psychological Review, 1974, 81, 59-74. Gormly, J . , & Edelberg, W. V a l i d i t y i n personality t r a i t a t t r i b u t i o n . American Psychologist, 1974, 29, 189-193. 194 Greig, M. UBC Anova. U n i v e r s i t y of B r i t i s h Columbia Documentation: Subject code, 17.3, 1974. Hamilton, D. L., & F a l l o t , R. D. Information salience as a weighting f a c t o r i n impression formation. Journal of Personality and S o c i a l Psychology, 1974, 30, 444-448. Hamilton, D. L., & Zanna, M. P. Context e f f e c t s i n impression formation: Changes i n connotative meaning.. Journal of Personality and S o c i a l Psychology, 1974, 29, 649-654. Hampshire, S. D i s p o s i t i o n s . Analysis, 1953, 14_, 5-11. Hastorf, A. H., Schneider, D. S., & Polefka, J . Person Perception. Reading, Mass.: Addison-Wesley, 1970. Hayden, T., & Mischel, W. Maintaining t r a i t consistency the r e s o l u t i o n of behavioral inconsistency: The wolf i n sheep's cl o t h i n g . Unpublished manuscript, Stanford U n i v e r s i t y , 1973. Hendrick, C. E f f e c t s of stimulus consistency-inconsistency on impressions of personality. Paper presented as a symposium, "Impression formation: Empirical and t h e o r e t i c a l issues and i n t e g r a t i o n s . " Meeting of the Midwestern Psychological Association, Cleveland, 1972. Isaacs, J . D., & Duncan, C. P. Reversal and non-reversal s h i f t s within and between dimensions i n concept formation. Journal of Experimental Psychology, 1962, 64, 580-585. Jaccard, J . J . P r e d i c t i n g s o c i a l behavior from personality t r a i t s . Journal of Research i n Personality, 1974, 7, 358-367. 195 Johnson, P. J . Factors a f f e c t i n g t r a n s f e r i n concept i d e n t i f i c a t i o n problems. Journal of Experimental Psychology, 1966, 72, 655-660. Jones, E. E., & Goethals, G. R. Order e f f e c t s i n impression formation: A t t r i b u t i o n context and the nature of the e n t i t y . In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S, V a l i n s , & . B. Weiner (Eds.), A t t r i b u t i o n : Perceiving the causes of  behavior. •Morrdstowh:,;N..J...:.. General Learning. Press, 1972. Jones, E..E., Kanouse, D. E., Kelley, H. H., Nisbett, R.E., Va l i n s , S.. & Weiner, B. (Eds.), A t t r i b u t i o n : Perceiving  the causes of behavior. Morristown, N.J.: General Learning Press, 1972. Jones, E. E., & Nisbett, R. E. The actor and the observer: Divergent perceptions of the causes of behavior. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. V a l i n s , & B. Weiner (Eds.), A t t r i b u t i o n : Perceiving the causes of behavior. 'Morristbwh,N!.!J.:TjSenera 1972 Kanouse, D. E., & Hanson, L. R. J r . Negativity i n evaluations. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Val i n s & B. Weiner (Eds.), A t t r i b u t i o n : Perceiving the  causes of behavior. Morristown, N.J.: General Learning Press, 1972. Kaplan, M. F. D i s p o s i t i o n a l e f f e c t s and weight of information i n impression formation. Journal of Personality and S o c i a l  Psychology, 1971, 18, 279-284. 196 Kaplan, M. F. The modifying e f f e c t of stimulus information on the consistency of i n d i v i d u a l differences i n impression formation. Journal of Experimental Research i n Personality, 1972, J5, 213-219. Kaplan, M. F. Stimulus inconsistency and response d i s p o s i t i o n s i n forming judgments of other persons. Journal of Personality  and S o c i a l Psychology, 1973, 25_, 58-64. Kaplan, M. F. Context-induced s h i f t s i n personality t r a i t evaluation: A comment on the evaluative halo e f f e c t and the meaning change i n t e r p r e t a t i o n s . Psychological B u l l e t i n , 1974, 81, 891-895. Kaplan, M. F. Information i n t e g r a t i o n i n s o c i a l judgment: Inter-action of judge and informational components. In M. Kaplan, & S. Schwartz (Eds.), Human judgment and decision processes. New York: Academic Press, 1975. Kirk, R. E. Experimental Design: Procedures f o r the behavioral  sciences. Belmont, C a l i f o r n i a : Brooks/Cole Publishing Company, 1968. Mischel, W. Personality and assessment. New York: Wiley, 1968. Mischel, W. Continuity and change i n per s o n a l i t y . American Psychologist, 1969, 24, 1012-1018. Mischel, W. Introduction to personality. New York: Holt, Rinehart & Winston, 1971 197 Mischel, W. Toward a cognitive s o c i a l learning reconceptualization of personality. Psychological Review, 1973, 80, 252-283. Norman, W. T. Toward an adequate taxonomy of personality a t t r i b u t e s : Replicated f a c t o r structure i n peer nomination pe r s o n a l i t y r a t i n g s . Journal of Abnormal and S o c i a l Psychology, 1963, 66, 574-583. Norman, W. T. "To see oursels a s i t h e r s see us!": Relations among self-perceptions, peer-perceptions, and expected peer-perceptions of personality a t t r i b u t e s . M u l t i v a r i a t e Behavioral Research, 1969, h, 417-442. Norman, W. T., & Goldberg, L. R. Raters,ratees and randomness i n personality structure. Journal of Personality.and S o c i a l Psychology, 1966, k_, 681-691. P a s s i n i , F. T., & Norman, W. T. A un i v e r s a l conception of per s o n a l i t y structure? Journal of Personality and S o c i a l Psychology, 1966, 4, 44-49. Richey, M. H., McClelland, L., & Shimkunas, W. M. Relative influence of p o s i t i v e and negative information i n impression formation and persistence. Journal of Personality and S o c i a l Psychology, 1967, J3, 322-327. Rosenbaum, M. E., & Levin, I. P. Impression formation as a function of source c r e d i b i l i t y and the p o l a r i t y of information. Journal  of Personality and S o c i a l Psychology, 1969, Yl_, 34-37. Schmidt, C. F. Personality impression formation as a function of relatedness of information and length of set. Journal of 198 Personality and S o c i a l Psychology, 1969, 12, 6-11. Schneider, D. S. I m p l i c i t personality theory: A review. Psychological B u l l e t i n , 1973, 79_, 294-309. Slamecka, N . J . A methodological analysis of s h i f t paradigms i n human di s c r i m i n a t i o n l e a r n i n g . Psychological B u l l e t i n , 1968, 423-438. Tupes, E. C., & C h r i s t a l , R. E. Recurrent personality f a c t o r s based  on t r a i t ratings. USAF ASD Technical Report, No. 61-97, 1961. Wachtel, P. L. Psychodynamics, behavior therapy, and the implacable experimenter: An inquiry into the consistency of p e r s o n a l i t y . Journal of Abnormal Psychology, 1973, 82, 324-334. Wiggins, J . S. Personality and p r e d i c t i o n : P r i n c i p l e s of p e r s o n a l i t y assessment. Reading, Mass.: Addison-Wesley, 1973. Wiggins, J . S. In defense of t r a i t s . Unpublished manuscript, Uni v e r s i t y of B r i t i s h Columbia, 1974. Wyer, R. S. J r . Information redundancy, inconsistency and novelty and t h e i r r o l e i n impression formation. Journal of Experimental S o c i a l Psychology, 1970, 6^, 111-127. Wyer, R. S. J r . Changes i n meaning and halo e f f e c t s i n personality impression formation. Journal of Personality and S o c i a l Psychology, 1974, 28, 829-835. 1 9 9 Appendix A The behavioral statments which appear on the following pages were chosen for the f i n a l item pool. The items are l i s t e d according to the t r a i t term with which they are associated and according to the v a l i d i t y group within which they belong. A f t e r each statement, the item properties are l i s t e d as follows: (a) i s the code f o r l o c a t i n g the observations upon which the item properties were obtained; (b) i s the v a l i d i t y index; (c) i s p(T-|.B); (d) i s the p(T|B); and (e) i s the p(B). The procedures f o r obtaining the appropriate p r o b a b i l i t i e s and the v a l i d i t y index were described i n the section which was c a l l e d , "Method of obtaining-,/the v a l i d i t y of the behavioral information" i n the f i r s t chapter. 200 A. Extroversion: High-valid 1. When she saw that a f r i e n d was uneasy i n the company of a new boyfriend, she introduced h e r s e l f and i n i t i a t e d a conversation, (a) Code: 1-2; (b) V a l i d i t y Index: .5387; (c) p(T|B): .7779; (d) p(T|B): .2392; (e) p .(B): .38. 2. When one person i n the group was put down, she stuck up f o r that person and made everyone sociable to her. (a) 1-4 (b) .5025 (c) .7382 (d) .2357 (e) .28. 3. During a long t r a i n t r i p , she made friends with a family and showed the ch i l d r e n around the t r a i n . (a) 1-11 (b) .5252 (c) .7323 (d) .2071 (e) .37. 4. When she and her fri e n d s made a big mistake at the a i r p o r t , she cracked a joke and eased the tension. (a) 1-12 (b) .6050 (c) .8264 (d) .2214 (e) .61. 5. One evening when she was v i s i t e d by fatigued f r i e n d s , she made some funny remarks and brightened up the group. (a) 2-12 (b) .5786 (c) .8250 (d) .2464 (e) .52. 6. During the f i r s t day at a new job, she i n i t i a t e d personal conversation with the boss and co-workers a l i k e . (a) 3-3 (b) .4805 (c) .8180 (d) .3375 (e) .36. 7. When she saw someone alone at a party, she went over to chat with that person. (a) 3-4 (b) .6372 (c) .8347 (d) .1975 (e) .51. 201 8. When her escort met an old f r i e n d and h i s escort while waiting for a dining table at a restaurant, she suggested that they f i n d a table f o r four and became a hostess to her new f r i e n d s , (a) 3-11 (b) .5130 (c) .8430 '(d) .33 (e) .36. 202 B. Extroversion: Low-valid 1. During a c r u i s e , she made a point to spend some time i n conversa-t i o n with each person on that ship. (a) 1-8 (b) .4097 (c) .6623 (d) .3312 (e) .38. 2. Several times during a class period, she argued with and extended the points made by the i n s t r u c t o r . (a) 1-9 (b) .3696 (c) .7338 (d) .3642 (e) .31. 3. When she and several friends went out to dinner, she kept everyone singing and doing things as a group. (a) 2-6 (b) .3410 (c) .6972 (d) .3562 (e) .30. 4. While waiting f o r f r i e n d s to j o i n her f o r lunch at the c a f e t e r i a she conversed with many people who walked by. (a) 2-7 (b) .3393 (c) .7361 (d) .3968 (e) .50. 5. During a discussion, she came out with a remark that most people would have been embarrassed to say. (a) 2-11 (b) .4135 (c) .7791 (d) .3656 (e) .22. 6. When she found out that someone i n the group was having a birthday, she began singing Happy Birthday even though she d i d not know that person. (a) 2-10 (b) .4157 (c) .6750 (d) .2593 (e) .31. 7. When informed of an upcoming s t a f f party, she offered to bake something, to help arrange the house and to clean up afterward, (a) 3-9 (b) .2850 (c) .6375 (d) .3525 (e) .36. 203 8. When she found out that her f r i e n d could not go to Jasper with her to f i n d work, she went by h e r s e l f . (a) 3-12 (b) .3061 (c) .7111 (d) .4050 (e) .32. 204 C. Introversion: High-valid 1. During a b u l l , session with f r i e n d s , she refused to di s c l o s e even minor secrets. (a) 1-1 (b) .5656 (c) .6906 (d) .1250 (e) .22. 2. She came to class alone, sat by h e r s e l f and remained speechless, (a) 1-3 (b) .6691 (c) .7691 (d) .1000 (e) .45. 3. When a l l her friends had appointments during lunch hour, she ate lunch by he r s e l f even though there were several people she knew close by.(a) 1-8 (b) .4661 (c) .7161 (d) .2500 (e) .29. 4. When asked by her roommate to do a favor, she complied even though she did not want to do i t . (a) 1-10 (b) .4387 (c) .6279 (d) .1892 (e) .49. 5. While at a party with friends she had not seen f o r a couple of years, she ju s t sat i n the corner and contributed l i t t l e to the party. (a) 2-2 (b) .4111 (c) .7611 (d) .3500 (e) .19. 6. She was a f r a i d to ask for a d i f f e r e n t vegetable than was served with the c a f e t e r i a meal. (a) 2-7 (b) .4111 (c) .7166 (d) .305 (e) .35. 7. In the l i b r a r y , she chose the chair which was the farthest from everyone. (a) 2-11 (b) .4403 (c) .6625 (d) .2222 (e) .27. 8. When her i n s t r u c t o r made a mistake marking her paper, she l e f t i t and accepted the loss of four marks. (c) 3-4 (b) .4305 (c) .5930 (d) .1625 (e) .12. 205 D. Introversion: Low-valid 1. When her f r i e n d discussed something with her, she j u s t nodded and l i s t e n e d - never o f f e r i n g a comment of her own. (a) 1-12 (b). .3378 (c) .6735 (d) .3357 (e) .38. 2. When a stranger was b i l l e t e d i n her home, she asked few questions and responded to those of her guest with as few words as possible. (a) 2-1 (b) .2820 (c) .7125 (d) .4305 (e) .24. 3. She remained s i l e n t about her engagement u n t i l someone noticed a r i n g on her fing e r . (a) 2-4 (b) .3510 (c) .6791 (d) .3281 (e) .27. 4. When there were cooperative a c t i v i t i e s on her f l o o r , she remained i n her room. (a) 2-5 (b) .3541 (c) .6541 (d) .30 (e) .28. 5. During a close basketball game where everyone was cheering and jumping, she sat s t i l l at the end of the row. (a) 2-8 (b) .2444 (c) ..5888 (d) .3444 (e) : .12. 6. When introduced to new people at a party, she blushed and moved on. (a) 3-3 (b) .3091 (c) .6791 (d) .3700 (e) .08. 7. She kept the true condition of her father's health to h e r s e l f , (a) 3-5 (b) .3402 (c) .5902 (d) .2500 (e) .09. 8. She remained s i l e n t about her exam marks even a f t e r her questioner had disclosed her own marks. (a) 3-7 (b) .3316. (c) .6291 (d) .2975 (e) .28. 206 E. Agreeableness: High-valid 1. When asked which places i n Vancouver she would l i k e to v i s i t , she r e p l i e d , "Whichever i s most convenient f o r you." (a) 1-1 (b) .6485 (c) .7735 (d) .2343 (e) .54. 2. Even though i t was out of her way to pick everyone up, she gave several people a r i d e to a community function. (a) 1-2 (b) .7231 (c) .8588 (d) .2437 (e) .40. 3. When her . f r i e n d was i n a hurry to get to a basketball game, she prepared a quick meal so her f r i e n d did not have to go further from the gym to'obtain one at her own home. (a) 1-7 (b) .6023 (c) .7308 (d) .1285 (e) .56. 4. On the spur of the moment, she went for a bike r i d e upon a friend's request. (a) 1-9 (b) .7719 (c) .8647 (d) .1875 (e) .56. 5. When her f r i e n d objected to bringing r e l i g i o n i n t o a l e c t u r e , she explained the o r i e n t a t i o n of the school, helped her f r i e n d work through her reaction and accepted her as she was. (a) 1-10 (b) .6608 (c) .75 (d) .0892 (e) .45. 6. When i t was ra i n i n g , she saw a classmate running to the gym, stopped and gave her a r i d e . (a) 1-11 (b) .7598 (c) .8441 (d) .0843 (e) .72. 7. Even though she was t i r e d , she entertained guests u n t i l 3 a.m. (a) 1-12 (b) .5970 (c) .8720 (d) .2750 (e) .58. 8. A f t e r having persuaded a f r i e n d to park i n a r e s t r i c t e d zone, she i n s i s t e d on paying f o r the t i c k e t which was received. (a) 3-12 (b) .5697 ' (c) .8222 (d) .2525 (e) .55. 207 F. Agreeableness: Low-valid 1. When she obtained her father's car f o r a week, she telephoned her friends and offered them a r i d e to the unversity everyday that week. (a) 1-4 (b) .4227 (c) .6691 (d) .2464 (e) .42. 2. The day a f t e r a major operation, she smiled and laughed even though i t hurt to do so. (a) 1-8 (b) .3271 (c) .6735 (d) .3464 (e) .35. 3. When asked to purchase some medication f o r a fr i e n d , she went immediately. (a) 2-2 (b) .4272 (c) .6972 (d) .2700 (e) .66 4. When her friends said they played a nasty t r i c k on her, she jus t laughed. (a) 2-6 (b) .3577 (c) .6152 (d) .2575 (e) .36. 5. Even though she was bogged down with her own work, she helped a f r i e n d with some assignments. (a) 2-8 (b) .3891 (c) .7916 (d) .4025 (e) .42. 6. She did as much as she could to help a very dependent g i r l with whom many people l o s t t h e i r patience. (a) 3-7 (b) .3273 (c) .7250 (d) .3975 (e) .36. 7. Even though her work was f i n i s h e d , she went with a f r i e n d to the l i b r a r y j u s t so her f r i e n d would not have to walk home alone, (a) 3-9 (b) .3277 (c) .7027 (d) .3750 (e) .42. 8. When her f r i e n d was asked out by the guy she wanted to go out with, she wished her good luck. (a) 3-11 (b) .3875 (c) .7500 (d) .3625 (e) .35. 208 G. Disagreeable: High-valid 1. When a f r i e n d t r i e d to describe a new opera, she drowned her out and claimed that i t was" lousy. (a) 1-1 (b) .5191 (c) .6691 (d) .1500 (e) .24. 2. At a play -rehearsal, i f an idea did not come from her, she groaned and r i d i c u l e d i t . (a) 1-8 (b) .4947 (c) .7161 (d) .2214 (e) .23. 3. Whenever her s i s t e r brought friends home, she found f a u l t with each of them. (a) 1-10 (b) .5388 (c) .6852 (d) .1464 (e) .32. 4. Even though she has as much as anyone could want, she expressed her wish to have what another person had. (a) 1-11 (b) .5813 (c) .6705 (d) .0892 (e) .51. 5. When a f r i e n d went to a movie without her, she rebuffed her. (a) 1-12 (b) .5034 (c) .6926 (d) .1892 (e) .26. 6. She immediately downgraded a g i r l who won the award that she expected to.win instead. (a) 2-1 (b) .5686 (c) .8102 .(d) .2416 (e) .25. 7. When she had a disagreement with her mother, she held a stubborn p o s i t i o n and refused to admit that she was wrong, (a) 2-3 (b) .5859 (c) .7720 (d) .1861 (e) .50. 8. During the f i r s t meeting of a c l a s s , she remarked that she was not going to l i k e being i n that c l a s s . (a) 3-3 (b) .5413 (c) .7513 (d) .2100 (e) .46. 209 H. Disagreeable: Low-valid I. When she had not been i n v i t e d d i r e c t l y , she picked a f i g h t and refused to go f o r dinner. (a) 1-9 (b) .4231 (c) .6838 (d) .2607 (e) .28. 2. When her f r i e n d bought some new clothes, she ignored her. (a) 2-2 (b) .3335 (c) .7029 (d) .3694 (e) .18. 3. While touring with a group i n Europe, she i n s i s t e d that the group go to a c e r t a i n place against t h e i r wishes but l a t e r refused to go where they wanted to go. (a) 2-4 (b) .4000 (c) .7250 (d) .3250 (e) .16. 4. When t o l d about a humorous event, she commented about the i r r e s p o n s i b i l i t y of the action. (a) 2-9 (b) .2522 (c) .6397 (d) .3875 (e) .13. 5. When her f r i e n d t o l d her something e x c i t i n g , she commented, "Is that a l l ? " and changed the t o p i c . (a) 2-12 (b) .4384 (c) .7852 (d) .3468 (e) .15. 6. She got very bossy at work even though she did not hold a po s i t i o n of authority. (a) 3-2 (b) .4080 (c) .7308 (d) .3225 (e) .20. 7. She refused to do a job which was not l i s t e d i n her job-des c r i p t i o n . (a) 3-5 (b) .4263 (c) .6763 (d) .2500 (e) .30. (8) When she entered a room which contained people she d i s l i k e d , she downgraded them and got nasty when they did not leave. (a) 3-8 (b) .4211 (c) .7361 (d) .3510 (e) .18. 210 I. Conscientious: High-valid 1. When decorating a . h a l l f o r a dance, she organized everything that had to be done and saw to i t that everything was done properly, (a) 1-5 (b) .6943 (c) .8514 (d) .1571 (e) .28. 2. When her locker was l e f t messy by her f r i e n d s , she took a few moments and methodically straightened i t out. (a) 1-6 (b) .7143 (c) .7750 (d) .0607 (e) .33. 3. She planned out her day and studied throughout regular time s l o t s . (a) 1-7 (b) .6402 (c) .7544 (d) .1142 (e) .28. 4. When her parents went on a vacation f o r a month, she managed the household which included carrying out business a f f a i r s as well as chi l d c a r e (a) 1-8 (b) .6517 (c) .7588 (d) .1071 (e) .48 5. When she received a l e t t e r that she would need at a l a t e r date, she placed i t i n a s p e c i f i c place f o r things to remember. (a) 1-12 (b) .7700 (c) .8485 (d) .0785 (e) .63. 6. When her s i s t e r threw clothes a l l over the room, she requested that they be hung up. (a) 3-3 (b) .6322 (c) .7472 (d) .1150 (e) .65. 7. When she was making a d i f f i c u l t item f o r someone, she made sure everything was j u s t r i g h t before giving i t to that person. (a) 3-8 (b) .6500 (c) .8500 (d) .2000 (e) .68. 8. When a f r i e n d s p i l l e d strawberry shortcake on the seat of her father's car, she began clean-up operations immediately. (a) 3-10 (b) .6000 (c) .8125 (d) .2125 (e) .52. 211 J. Conscientious: Low-valid 1. The night before an exam, she studied u n t i l 3 a.m. (a) 1-4 (b) .2880 (c) .6558. (d) .3678 (e) .50. 2. She did three times more work i n her assignment than was required (a) 1-9 (b) /"3185 (c) .6970 (d) .3785 (e) .25. 3. The night before the j o i n t term paper was due, she rewrote every page of rough notes to make sure i t was cl e a r enough f o r her partner to type quickly. (a) 2-4 (b) .3230 (c) .6323 (d) .3093 (e) .48. 4. When someone borrowed a cooking u t e n s i l , she s p e c i f i e d how to use i t and how to clean i t . (a) 2-6 (b) .2746 (c) .6058 (d) .3312 (e) .43. 5. When her f r i e n d was on holidays, she checked that friend ' s apartment r e g u l a r l y to remove mail and other signs of her friend's absence. (a) 2-9 (b) .3761 (c) .6823 (d) .3062 (e) .60 6. On a camping t r i p , she .cleaned the tent every day, made sure >< that everyone was comfortable and made a point to know where everyone was at a l l times. (a) 2-12 (b) .3379 (c) .6441 (d) .3062 (e) .2908. 7. Even though her art work looked extremely good, she co n t i n u a l l y t r i e d to patch i t up. (a) 3-1 (b) .425 (c) .7875 (d) .3750 (e) .32. 8. On a Friday evening before going to a party, she studied f o r two hours and then had a good time. (a) 3-11 (b) .2872 (c) .6597 (d) .3725 (e) .18. 212 K. Unconscientious: High-valid 1. Even though she promised to relay ah' important message, she forgot- arid -afriend missed out on an important function. (a) 1-2 (b) .4802 (c) .7087 (d) .2285 (e) .21. 2. Immediately a f t e r a f r i e n d had confided i n her, she t o l d someone else about i t . (a) 1-5 (b) .6193 (c) .8014 (d) .1821 (e) .41. 3. She began washing windows, l e f t most of them streaky and quit before she had f i n i s h e d . (a) 1-6 (b) .4822 (c) .8250 (d) .3428 (e) .11. 4. She terminated employment without s u f f i c i e n t notice, (a) 1-8 (b) .4946 (c) .8088 (d) .3142 (e) .21. 5. She l e f t everything disorganized, promised to clean i t up shortly, but vanished. (a) 1-12 (b) .5031 (c) .8352 (d) .3321 (e) .39. 6. Aft e r having obtained a job that she r e a l l y needed, she was l a t e one morning and c a l l e d i n s i c k the next morning when she did not want to work. (a) 2-1 (b) .5282 (c) .7750 (d) .2468 (e) .25. 7. She missed a prearranged meeting and when questioned, denied i t s importance. (a) 2-2 (b) .5458 (c) .7708 (d) .2250 (e) .20 8. She promised to bring some equipment f o r a friend' s r o l e i n a play, but c o n t i n u a l l y put i t o f f u n t i l there wasn't time to make use of i t . (a) 3-2 (b) .4766 (c) .7916 (d) .3150 (e) .24. 213 L. Unconscientious: Low-valid 1. Even though she had indicated that she would go out, she c a l l e d up that Saturday afternoon and said she no longer f e l t l i k e going out. (a) 2-7 (b) .3299 (c) .7486 (d) .4187 (e) .28. 2. She forgot to get some cigaret t e s f o r a f r i e n d and f a i l e d to return that friend's money (a) 3-1 (b) .3616 (c) .6291 (d) .2675 (e) .14. 3. She borrowed a ten-speed bike from a f r i e n d , forgot to lock i t and had i t stolen. (a) 3-3 (b) .3575 (c) .7125 (d) .3550 (e) .22. 4. When she had promised to give friends a r i d e to the u n i v e r s i t y and discovered that t h e i r schedule was d i f f e r e n t from hers, she t o l d them not to count on her f o r a r i d e . (a) 3-4 (b) .2927 (c) .7027 (d) ..4100(e) .46, 5. She went with a f r i e n d to buy a birthday g i f t f o r another f r i e n d , signed the card but forgot to pay her share. (a) 3-7 (b) .2675 .(c) .6625 (d) .3950 (e) .17. 6. She agreed to meet her friends at The Bayistore but showed up two hours l a t e r and excused h e r s e l f f o r browsing too long. (a) 3-8 (b) .3841 (c) .7291 (d) .3450 (e) .20. 7. When she had a s l i g h t stomach ache, she stayed home from classes, (a) 3-9 (b) .2972 (c) .6847 (d) .3875 (e) .12. 8. Even though she was asked not to t a l k about a friend's assign-ment i n pu b l i c , she continued to r i d i c u l e i t . (a) 3-11 (b) .3708 (c) .6458 (d) .2750 (e) .16. 214 M. Emotional S t a b i l i t y : High-valid 1. At work, she concentrated on her goal without l e t t i n g other things d i s t r a c t her. (a) 1-2 (b) .5477 (c) .7191 (d) .1714 (e) .32. 2. When a f r i e n d broke an arm l a t e one night, she stayed with her u n t i l i t was reset. (a) 1-5 (b) .6454 (c) .8132 (d) .1678 (e) .61. 3. When a customer was holding up a l i n e , she waited p a t i e n t l y and smiled. (a) 1-10 (b) .5313 (c) .7205 (d) .1892 (e) .33. 4. When they were being followed one night, she kept d r i v i n g around l i g h t e d areas u n t i l the other car gave up. (a) 2-5 (b) .5952 (c) .8902 (d) .2950 (e) .45. 5. When she was i n c o r r e c t l y accused of s t e a l i n g , she went promptly to the correct authority, explained her p o s i t i o n and cleared her name. (a) 2-6 (b) .5050 (c) .8000 (d) .2950 (e) .63. 6. When she did not know where to get o f f the subway i n P a r i s , she asked around u n t i l she found someone who knew. (a) 2-8 (b) .5905 (c) .8180 (d) .2275 (e) .54. 7. When an innocent party at her house turned into a w i l d a f f a i r , she t r i e d to calm things down without offending anyone, (a) 2-9 (b) .5550 (c) .7000 (d) .1450 (e) .47. 8. When her landlord y e l l e d at her and made impossible demands, she waited out the harangue and proposed a s o l u t i o n . (a) 3-9 (b) .5086 (c) .7361 (d) .2275 (e) .39. 215 N. Emotional S t a b i l i t y : Low-valid 1. When a f r i e n d voiced d i s s a t i s f a c t i o n with a course, she explained that most people are d i s s a t i s f i e d with some courses they must take but one must carry on. (a) 1-11 (b) .3852 (c) .6352 (d) .2500 (e) .47. 2. When robbed i n Europe, she knew exactly what to do and did i t . (a) 1-12 (b) .4128 (c) .7235 (d) .3107 (e) .35. 3. When the members of her guitar group could not begin together, she pleasantly suggested that they t r y d i f f e r e n t things. (a) 2-3 (b) .3239 (c) .7194 (d) .3950 (e) .51. 4. When she was entertaining one boyfriend and another came to the door, she explained her s i t u a t i o n and excused h e r s e l f . (a) 2-10 (b) .2458 (c) .6583 (d) .4125 (e) .27. 5. When a tragedy happened i n her family, she made he r s e l f continue with her work. (a) 2-11 (b) .3230 (c) .6680 (d) .3450 (e) .53. 6. When confronted about her possible use of marijuana, she kept quiet about her practices and commented that the questioner's opinion of her was unimportant. (a) 3-6 (b) . .3080 (c) .6180 (d) .3100 (e) .31. 7. She stepped into a f i g h t , calmed both p a r t i e s and helped them resolve t h e i r dispute. (a) 3-11 (b) .2913 (c) .5888 (d) .2975 (e) .52. 8. When she. discovered that her boyfriend was dating another g i r l , she i n i t i a t e d conversation with them and worked i t out. (a) 3-3 (b) .4350 (c) .7625 (d) .3275 (e) .22. 216 0. Emotional I n s t a b i l i t y : High-valid 1. When she f e l t i n danger while i n an automobile, she gasped and braced h e r s e l f . (a) 1-1 (b) .4005 (c) .6897 (d) .2892 (e) .41. 2. Just before an exam, she became concerned about her performance and developed a headache. (a) 1-3 (b) .6906 (c) .8441 (d) .1535 (c) .28. 3. On her way to a s o c i a l function, she wondered i f she wore the rig h t clothes, i f she would make a f o o l of h e r s e l f and i f she would know anyone there. (a) 2-2 (b) .4075 (c) .7985 (d) .3610 (e) .53. 4. She depreciated h e r s e l f when the s i t u a t i o n took a turn f o r the worse. (a) 2-10 (b) .3930 (c) .7652 (d) .3733 (e) .25. 5. When she was asked out by a guy, she got hyper and blew i t by not being h e r s e l f . (a) 3-5 (b) .3177 (c) .6277 (d) .3100 (e) .51. 6. She got anxious and dis t r e s s e d while t a l k i n g one moment and became elated the next but was s t i l l t a l k i n g about the same thing, (a) 3-9 (b) .3150 (c) .6625 (d) .3475 (e) .06. 7. When someone approached her from behind and said , "Hello."', she jumped and quickly turned around. (a) 3-10 (b) .3661 (c) .6111 (d) .2450 (e) .41. 8. When she found out that a f r i e n d whom she had not seen f o r a couple of months was coming to v i s i t , she talked r a p i d l y i n a high-pitched voice and was i n constant motion. (a) 1-5 (b) .4138 (c) .6102 (d) .1964 (d) .27. 217 P. Emotional I n s t a b i l i t y : Low-valid 1. She got so excited about the upcoming weekend that she could not eat. (a) 1-6 (b) .2892 (c) .6308 (d) .3416 (e) ,26. 2. When her actions were questioned, she immediately burst into tears. (a) 1-7 (b) .2616 (c) .6044 (d) .3428 (e) .16. 3. When d r i v i n g behind a slow car, she y e l l e d at the slow d r i v e r to speed up and then changed lanes. (a) 1-8 (b) .2061 (c) .6132 (d) .4071 (e) .32. 4. When there was a l i t t l e noise i n the l i b r a r y , she f a i l e d to concentrate on her work. (a) 1-12 (b) .2437 (c) .5794 (d) .3357 (e) .33. 5. When her parents had a f i g h t , she screamed and y e l l e d at them, (a) 2-3 (b) .2486 (c) .6319 (d) .38 (e) .20. 6. Af t e r being confronted by her father, she got upset and drank a b o t t l e of medicine to calm h e r s e l f down. (a) 2-7 (b) .2180 (c) .5930 (d) .3750 (e) .04. 7. When two friends wanted to go to d i f f e r e n t places, she began to cry when she could not decide which place to go. (a) 3-2 (b) .1736 (c) .6013 (d) .4277 (e) .08. 8. When her boyfriend did not show up at the designated time, she became f r a n t i c , speculated about an accident and wanted to c a l l the h o s p i t a l (a) 3-12 (b) .2469 (c) .6194 (d) .3725 (e) .26. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0100054/manifest

Comment

Related Items