BALANCE AND AGREEMENT IN CHILDREN'S SOCIAL PERCEPTION by GLORIA MARGARET GUTMAN B.A., University of British Columbia, 1961 M.A., University of Alberta, 1964 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in the Department of Psychology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September, 1970 In present ing t h i s thes is in p a r t i a l f u l f i l m e n t o f the requirements fo r an advanced degree at the Un ivers i ty of B r i t i s h Columbia, I agree that the L ibrary sha l l make i t f r e e l y a v a i l a b l e for reference and study. I fu r ther agree that permission for extensive copying of th is thes is for scho la r ly purposes may be granted by the Head of my Department or by h is representa t ives . It is understood that copying or p u b l i c a t i o n of th is thes is f o r f i n a n c i a l gain sha l l not be allowed without my wr i t ten permission. The Univers i ty of B r i t i s h Columbia Vancouver 8, Canada ABSTRACT The primary purpose of this study was to investigate the structural bases of pleasantness and consistency ratings and to determine the relation-ship between the two types of judgement in children ranging in age from 5-12 years. A secondary purpose of tie study was to determine whether the results of studies by Atwood (1969) and by Storm and Knox (1969) using a prediction procedure to investigate the developmental course of cognitive balance would generalize to a different dependent measure. Subjects in the study were 80 children, 20 (10 males and 10 females) from each of the following age groups: 5-6, 7-8, 9-10, and 11-12 years. They rated hypothetical social situations both for pleasantness and for con-sistency. The situations were of the P-O-X type, consisting of the subject, another person, and an unspecified, but important "thing." On the assumption that affect influences the social perceptions of younger children more than considerations of consistency i t was predicted that in their ratings of social situations younger children would d i f f e r -entiate l i t t l e between pleasantness and consistency (i.e., situations rated as pleasant would also be rated as consistent). Relative to the youngest children, older children were expected to differentiate more between pleasant-ness and consistency. Thus, i t was predicted that as a function of increas-ing age, correlations between pleasantness and consistency ratings would monotonically decrease across the successive age groups in the study. Further, i t was predicted that children at a l l age levels would attach greater weight to agreement than to balance when making pleasantness ratings and that younger children would also base consistency ratings more on agree-ment than on balance. However, balance was expected to exert greater i n -i i fluence than -agreement on the consistency ratings of older children. This follows from Zajone's (1968) review and i t s extension which suggest that agreement is more important than balance when the dependent measure relates to affect whereas balance exerts greatest influence when the task relates to psychological consistency. The results failed to yield evidence of age differences in differentia-tion between pleasantness and consistency. Correlations between the two types of ratings were high in a l l groups. These were also no age differences in the relative weighting of balance and agreement. Children in a l l groups utilized balance to a slightly greater extent than agreement when pleasantness was the criterion; agreement was used to a slightly greater extent than balance when the children rated for consistency. The effects of balance and agreement were very small, however, in comparison to those of attraction. Children in a l l age groups appeared to base both pleasantness and consistency ratings primarily on attraction (i.e., on the sign of the P/0 bond). A cross-validation study conducted concurrently with the principal study by an independent and , ;naive" IS yielded the same pattern of results. Differences in results obtained with children in the rating situation vs the prediction situation were tentatively attributed to differential task complexity. It was suggested that differences between adults (cf., Zajonc, 1968) and children in the rating situation may be due to differences in in-formation processing a b i l i t i e s and/or to differences in the strength of the balance "schema.1' That i s , the "schema" or implicit code for balance may be more firmly established in adults than in children. This could perhaps account for the fact that although adults u t i l i z e balance to a greater extent than agreement or attraction in the prediction situation and when i i i rating for consistency, strong balance effects among children are obtained only in the easier prediction situation. The balance "schema" in children, in other words, may not be of sufficient strength to withstand the competition of alternative biases such as attraction, agreement, and positivity when the more complex rating task i s used. iv TABLE OF CONTENTS CHAPTER Page I. Introduction . . . . . . . . . . . . . . 1 (a) Basic principles of Balance Theory (b) Empirical studies of Cognitive Balance (i) in adults ( i i ) in children (c) Statement of purpose (d) Hypotheses II. Method 28 (a) Subjects (b) Design (c) Procedure (i) training ( i i ) pleasantness ratings ( i i i ) tests of conservation (iv) consistency ratings III. Assessment of the Procedures 44 (a) Qualitative observations (b) Reliability of the ratings (c) Performance on the interpolated tests of conservation IV. Results and Discussion of Hypotheses 49 V. Ancillary Findings and their Implications 68 VI. Results of the Cross-validation Study 73 VII. Further Discussion 83 VIII. Summary 91 V Page BI BLIOGRAPHY 94 APPENDICES Appendix 1. Pilot studies 98 Appendix 2. A diagrammatic representation of the structures included in level 1 and level 2 of factor 0 . . . . , 104 Appendix 3. Mean pleasantness and consistency ratings for each structure 106 Appendix 4. Multiple regression analysis of the ratings 108 (a) the principal study (b) the cross-validation study Appendix 5. Summary of analyses of variance of pleasantness, consistency, d tension ratings obtained from adult Ss (Gutman, 1969) 116 v i LIST OF TABLES Page Table 1 A comparison of balance, agreement, and attraction (Zajonc, 1968). . 13 Table 2 Reorganization and extension of Zajonc's table comparing balance, agreement, and attraction 15 Table 3 Percent of balanced and transitive responses and explanations in children ranging in age from 5-12 years (Storm and Knox, 1969) . . 21 Table 4 Balance and agreement in prediction of p o l i t i c a l preferences of a liked and a disliked other (Knox and Gutman, 1968) 23 Table 5 Training cards 1-9 35 Table 6 Mean r e l i a b i l i t y coefficients for pleasantness and consistency ratings 46 Table 7 Percent of Ss in each age group showing quantity and volume conservation 47 Table 8 Mean pleasantness-consistency correlations for each age group 50 Table 9 Design of the analyses of variance 53 Table 10 Summary of the analysis of variance of pleasant-XI6 S S ITS t XIl^S « « « e e e a * a o e * * « 4 « « o « o 53 Table 11 Zajonc indices for pleasantness 58 Table 12 Mean pleasantness ratings for structures contain-ing agreement and disagreement 60 Table 13 Summary of analysis of variance of consistency ratings . . . . . 62 Table 14 Zajonc indices for consistency 65 Table 15 Mean pleasantness and consistency ratings relevant to the agreement by structures inter-action 70 Table 16 Mean r e l i a b i l i t y coefficients for pleasantness and consistency (Cross-validation study) . . . . . . 74 v i i Page Table 17 Mean pleasantness-consistency correlations for each age group (Cross-validation study) . . . 75 Table 18 Summary of analysis of variance of pleasant-ness ratings (Cross-validation study) 77 Table 19 Summary of analysis of variance of consistency ratings (Cross-validation study) 80 Table 20 Rank order of mean pleasantness and consistency ratings 85 Table 21 Zajonc indices for Pilot Study II . . . . . . . . 103 Table 22 Mean pleasantness and consistency ratings for each structure . 107 Table 23 Average regression equations for pleasantness and consistency 110 Table 24 Kendall coefficients v>f concordance I l l Table 25 Number of Ss for whom predictors are significant components of pleasantness and/or components of consistency ratings . . . . . . . . 113 Table 26 Average regression equations for pleasantness and consistency (Cross-validation study) . . 114 Table 27a Summary of analysis of variance of pleasantness ratings (Gutman, 1969) . . . 117 Table J7b Summary of analysis of variance of consistency ratings (Gutman, 1969) 118 Table 27c Summary of analysis of variance of tension ratings (Gutman, 1969) . . . 119 v i i i LIST OF FIGURES Page Figure 1 Diagrammatic representation of balanced and imbalanced s;ructures 3 Figure 2 Quadrants uti l i z e d in computing Zajonc indices . . . 10 Figure 3 Diagi ammatic representation of the experimental design . . . . . . . . . . . 30 Figure 4 Rating scale I . . . . . 32 Figure 5 Rating scale II . . . . . . . . . . . . 33 Figure 6 Content of training cards 10-17 36 Figure 7 Rating scale III 37 Figure 8 Rating scale IV 41 ±x ACKNOWLEDGEMENTS I would like to extend sincere thanks to my advisor, Dr. R. Knox, for his generous help and encouragement at a l l stages of the research, I would also like to thank the other members of the doctoral committee Dr. R. Potashin, Dr. E. Signori, Dr. D. Smith, Dr. T. Storm and Dr. F. Valle for their valuable comments and suggestions. Thanks go also to Sister Mary Neva of Holy Trinity School, Mr. C. Siddall of McKechnie School, and Mr. P. Bak of the Vancouver Talmud Torah for providing subjects; to Mr. Dave Malcom for assisting with the s t a t i s t i c a l analysis of the data; and to Miss Carole Hilstrom for typing the manuscript. Special thanks go to Miss Elsie Eccles for conducting the cross-validation study. 1 CHAPTER I INTRODUCTION Considerable attention has been focused, in recent year, on the development and extension of theories of "cognitive consistency." Abelson and Rosenberg (1958), Cartwright and Harary (1956), Festinger (1957), Heider (1946, 1958), McGuire (1960), Newcomb (1953) and Osgood and Tannenbaum (1955) are among those who have proposed theories of this type. Common to a l l i s the postulation of tendencies toward meaningful and harmonious organization of the individual's thought, beliefs, attitudes, and behaviour. In common also is the postulation of tendencies toward reduction of i n -consistencies between elements of the individual's cognitive system. These theories d i f f e r , however, in terminology, in rigor of definition, and i n the type of situation to which they most typically apply. The present study relates most directly to the formulations of Heider (1946, 1958). In Heider's Balance model, analysis is focused on the P-O-X triad, consisting of P (the perceiver), 0 (another person) and X (either a third person or an impersonal entity). Relations between triadic elements are of two types: sentiment relations (like or dislike) and unit relations (associated with or not associated with). . . . separate entities comprise a unit when they are perceived as belonging together. For example, members of a family are seen as a unit; a person and his deeds belong together. (Heider, 1958, p. 176) Units are formed on the basis of perceived similarity, proximity, familiarity, ownership, causality, or kinship. Balance is defined in terms of the number of positive and negative relations in the P-O-X triad. Like (L) and associative relations (U) are classified as positive; dislike (DL) and non-2 associative relations (not-U) are classified as negative. A triadic system is considered to be in a state of balance i f three relations are positive or i f two are negative and one is positive. Structures 1, 2, 7, and 8 in Figure 1 satisfy the cr i t e r i a for balance. Triads consisting of two Figure 1 positive and one negative relation are considered imbalanced. Heider states, though with some equivocation, that imbalance also exists i f a l l three relations are negative. . . . Four [structures] are balanced, containing three positive relations, or one positive and two negative. Four are unbalanced, with three negative or two positive relations. . . . If two negative relations are given, balance can be obtained either when the third relation i s positive or when i t i s negative; although there appears to be a preference for the positive alternative. (Heider, 1958, Pp. 204-206) The fundamental assumptions of Balance Theory are (a) that sentiment and unit relations tend toward balance, and (b) that imbalanced states produce tension and generate forces to restore balance. Heider views the balanced state as "a situation in which the perceived units and experienced sentiments co-exist without stress; there is thus no pressure toward change either in the cognitive organization or in the sentiment" (Heider, 1958, p. 176). Cartwright and Harary (1956) have extended the range of situations to which Balance Theory is applicable by defining balance i n graph theoretical terms. According to their formulation, a system is balanced 3 Balanced structures P *0 P >0 P j O P---*,0 \ / \/ \/ \/ (1) (2) (7) (8) Imbalanced structures p ,0 P---»,(> P — f \ / v v v (3) (4) (5) (6) Figure 1: ijias;rar.: i7atic representation of balanced and imbalanced structures. (Solid lines indicate positive relations; dashed lines indicate negative relations.) 4 i f a l l the semicycles within the system are positive. (A setnicycle is defined as a collection of two or more lines forming a closed path in a graph. For example, the structure Pc'-' "*0 contains three semicycles: PO^ PO; PO,OX~,XP; and PO,OX,XP. The sign of a semicycle i s positive i f the product of the signs of the lines forming the semicycle is positive.) This definition of balance i s applicable to structures containing any f i n i t e number of relations. There is no limitation on the type and number of relations defined on a particular set of elements. In addition, the Cartwright and Harary formulation provides a means of handling non-reciprocal relations between structural elements (e.g., a situation where P likes 0, but 0 does not like P). Heider states that non-reciprocal liking i s imbalanced, but f a i l s to include such situations in his structural definition of balance. The definition of balance in terms of semicycles has the added advantage of permitting degrees of balance to be specified. This is accomplished by computing the proportion of positive semicycles to total number of semicycles in the system. A number of techniques have been devised to test Heider's (1946, 1958) Balance theory, and i t s extensions. One method follows from the assumption that states of imbalance generate tension and are unpleasant (cf., Heider, 1958, p. 207), and requires that subjects rate hypothetical social situations for subjective feelings of "unpleasantness." Jordan (1953), for example, had subjects rate sixty-four triadic structures for unpleasantness. The structures represented a l l possible combinations of L, DL, U, and not-U relations and were of the following form: "I dislike 0; I like X; 0 has Empirical Studies of Cognitive Balance 5 no sort of bond or relationship with X." The subject was instructed to imagine himself in the situation, playing the role of " I " and then to rate the situation for pleasantness or unpleasantness on a ninety-point scale. The results of Jordan's study supported Heider's hypothesis—there was a s t a t i s t i c a l l y significant tendency for balanced structures to be rated more pleasant than imbalanced structures. Contrary to prediction however, subjects tended to distinguish between balanced structures containing positive relationships between P and 0, and those in which the P/0 relation-ship was negative, the latter being rated considerably more unpleasant than the former. In fact, the mean rating for balanced structures with negative P/0 bonds was hardly distinguishable from that given to imbalanced structures. Subsequent studies which have used pleasantness-unpleasantness ratings to test derivatives of the balance hypothesis show a similar discrepancy between prediction and results (e.g., Hershkowitz, 1954; Price, Harburg, and Newcomb, 1966; Rodrigues, 1966; Steiner and Spaulding, 1966). Jordan (1953) attempted to account for this, and other discrepancies between prediction and results, by suggesting that "the original coordinating definition of balance and pleasure is faulty." He recommends instead that balance "be coordinated with the concept of a strong or good 'gestalt.' A strong gestalt i s characterized in practically a l l 'gestalt' theoretical literature as a most proper, inner-necessary state. Balanced situations can therefore be considered to be experienced as more proper than imbalanced situations independent of their degree of experienced pleasantness. Propriety is not synonymous with pleasantness. For many, retribution for sins i s proper, but few i f any consider i t to be pleasant" (p. 282). Following this lead, Knox (1963) had subjects (n=10) rate a number 6 of hypothetical social situations of the P-O-X type for pleasantness and for 'consistency." Consistency was defined as "a state of logical congruity among the persons and thinj» in the [P-O-X] situation. ! i The subject was instructed to rate situations that seemed "to hang together in a log i c a l , sensible, and rational manner" toward the consistent end of the scale. He was instructed to rate situations that seemed '"out of k i l t e r , ' i l l o g i c a l , and irrational,!' toward the inconsistent end of the scale. The results showed a weak, but significant positive relationship between pleasantness ratings and consistency ratings (r = +.20). Knox concludes that pleasantness and consistency are not equivalent terms (p. 81). A subsequent study by Gutman (1969) was designed to replicate Knox's (1963) findings and to determine the relationship or possible equivalence of the concepts of pleasantness, consistency, and tension. The subjects i n this study (n=84) rated the eight structures shoxm in Figure 1 on 9-point scales anchored by the terms unpleasant-pleasant, inconsistent-consistent, and no tension-very strong tension. The results yielded a significant Pearson correlation of +.47 between the pleasantness and consistency ratings.^" The correlations between pleasantness and tension ratings and between consistency and tension ratings were -.81 and. -.42 respectively. These findings indicate that subjects define as unpleasant the same situations that they define as tension-provoking. The results suggest also that ^"Differences in the time interval between pleasantness and consistency ratings may perhaps account for the difference i n magnitude of the correlation between pleasantness and consistency in the Knox (1963) and the Gutman (1969) studies. In Knox's study, pleasantness and consistency ratings were collected in four sessions held on alternate days in an ABBA design. In Gutman's study, pleasantness, consistency, and tension ratings were collected in a single one hour session. 7 subjects differentiate between situations that are unpleasant and tension-provoking and those that are \sychologically inconsistent. Rather than attributing the distinction in the rated pleasantness of balanced structures with positive P/0 bonds and those with negative P/0 bonds to a fault i n the coordinating definition of balance and pleasant-ness, Rodrigues (1965) has suggested that "agreement" may act as an independent source of "cognitive bias," conflicting with tendencies toward balance in the P/0 negative case. As shown in Figure 1, where P likes 0, balance is achieved when P and 0 have the same attitude toward X. Where P dislikes 0, balance results from disagreement regarding X. Thus, in the P/0 positive case preferences for balance and agreement work in the same direction. In the P/0 negative case, agreement and balance work in opposite directions. Jordan's results suggest that subjects' ratings are affected by this conflict of forces. Mean unpleasantness ratings were lowest in balanced structures with agreement and highest (most unpleasant) in balanced structures with disagreement. Imbalanced structures containing agreement were rated somewhat more pleasant than imbalanced structures containing disagreement. Rodrigues (1965) obtained similar results when he had subjects rate triadic structures for tension. His data show an increase in mean tension ratings as one goes from balanced structures with agreement, through imbalanced structures with agreement and imbalanced structures with disagreement, to balanced structures with disagreement. Newcomb (1968), on the other hand, feels that agreement or disagree-ment between P and 0 may have l i t t l e effect on the perceived pleasantness of a situation in which P and 0 dislike one another. He states that the negative P/0 bond "engenders i t s own tension, which is independent of the 8 kind of tension that is intrinsic to the notion of balance, as defined by Heider and by others who have followed him" (p. 33). Consistent with this line of reasoning, Jordan (1966) reports that a considerable portion of the variance in the 1953 study is accounted for by the sign of the P/0 bond. The results of studies by Steiner and Spaulding (1966) and Hershkowitz (1954) also suggest that the sign of the P/0 bond could be an important determinent of pleasantness ratings. The question that remains is which factor contributes most to social perception in the triadic situation: balance, agreement, or attraction (i.e., sign of the P/0 bond)? Zajonc (1968) analyzed data from a number of empirical studies of social perception and reports that approximately half favor agreement over balance as the more c r i t i c a l determinant of responses. None favor attraction. Included in his analysis were several studies involving pleasantness-unpleasantness ratings (Hershkowitz, 1954; Jordan, 1953; Price, Harburg, and Newcomb, 1966; Rodrigues, 1966; Steiner and Spaulding, 1966), a study which required prediction of missing relations (Morrissette, 1958), and two so-called "ease-of-learning" studies (Zajonc and Burnstein, 1965a, 1965b). In the prediction study, subjects were required to role-play a move into an apartment. Sentiments among some of the roommates were given. The subject's task was to predict the remaining sentiments and to rate how much tension he would feel in such a situation. The two ease-of-learning studies u t i l i z e d a paired-associates technique. The dependent measure was the number of errors made in learning the signs of relations in P-O-X and P-O-X-Y structures. (Structures of type P-O-X-Y involve two persons and their attitudes toward two issues.) The method of analysis utilized by Zajonc involved calculation of 9 separate indices for balance, agreement, and attraction. These indices are derived by ordering the eight triads shown in Figure 1 into four quadrants. As shown in Figure 2, the quadrants are derived by considering the sign of the P/0 bond (+ or -) and the presence or absence of agreement between P and 0 with regard to X. Quadrant A contains structures in which there Figure 2 is a positive bond between P and 0 and agreement with regard to X (i.e., structures P *0 and P~——40) ; quadrant B contains structures in * x * x -which there is a negative bond between P and 0, and agreement concerning X (structures P $0 and P >0). Structures P*> >0 and P ~»0 are V v x * V V represented i n quadrant C; structures P- - -» 0 and P — -j 0 are represented X X in quadrant D. Only quadrants A and D contain balanced structures. The effects of balance are estimated by the ratio of average scores (on whatever dependent measure was used in a particular study) for structures in balanced quadrants to the average scores for structures in imbalanced quadrants A + D (i.e., quadrants „ ). The effects of agreement are estimated by the ratio of quadrants containing agreement to those containing disagreement A + B ( c + p ) • The effects of attraction are estimated by the ratio of P/0 A + C positive to P/0 negative quadrants (_,_.). The relative strength of each D T D structural factor is determined by comparing the magnitude of indices computed from these ratios. If the effects of agreement are stronger than 10 Attraction PL+0 PL~0 Agreement Disagreement A B C D Figure 2: Quadrants utilized in computing Zajonc indices. 11 those of balance, the agreement index is greater than that for balance. The situation i s reversed i f the effects of balance are stronger than those of agreement. The index for attraction would be greatest i f this were the most important determinant of the subjects' responses. A concrete example may perhaps c l a r i f y the procedure following i n computing Zajonc indices. Data in the example come from Jordan's (1953) study. Quadrant Structures in Quadrant Mean rating for each structure Mean rating for each quadrant A +++ +— 22.1 33.8 27.9 B -++ 59.1 69.9 64.5 C ++-+-+ 64.5 67.1 65.8 D -+-—+ 64.2 71.5 67.8 Attraction •P PL +0 PL* "o Agreement 27.9 64, .5 Disagreement 65.8 67, .8 \ Index of Agreement: 27.9 + 64.5 65.8 + 67.8 " Index of Attraction: . 27.9 + 65.8 _ 64.5 + 67.8 Index of Balance: 27.9 + 67.8 64.5 + 65.8 " 12 The results of Zajonc's (1968) analysis are shown in Table 1. As he reports, approximately half of the studies in the'table favor agreement over balance as the more c r i t i c a l determinant of social perception. As Knox (1969) has pointed out, however, the bulk of evidence in favor of agreement in the table comes from studies which u t i l i z e pleasantness ratings. Reorganization and extension of Zajonc's table (see Table 2) Table 1 shows that indices of balance are greater than those of agreement and attraction in tasks involving prediction of missing relations, ease-of-learning, consistency ratings, or st a b i l i t y assessment. (Studies of the latter type require the subject to indicate whether he would expect a particular structure to remain the same or to change over time.) Thus, balance seems to be the more important factor when the dependent measure relates to psychological consistency. Agreement would seem to be more Table 2 important when the measure is based on affect. 13 Table 1: A comparison of balance, agreement, and attraction (Zajonc, 1968, p. 348) Study, measure, and condition p and o agree p and ,o disagree + pL o pL o pL o pL o Effect ratio c o C •H 0) J-i 0 a o 8 cd c CO < < « Hershkowitz (1954) Unpleasantness scores on a 100-point rating scale a) Values b) Objects Jordan (1953) Unpleasantness scores on a 90-point rating scale Morrissette (1958) Percent of Ss predicting a positive or negative relation between o and x. Price, Harburg, and Newcomb,(1966) Percent of j>s reporting unpleasant affect Rodrigues (1966) Unpleasantness scores on a 90-point rating scale 20.1 54.2 24.3 54.7 27.9 64.5 66.6 62.9 1.74 1.35 1.46 60.6 62.9 1.56 1.38 1.32 65.8 67.8 1.44 1.41 1.36 73% 37% 6% 41% 27% 63% 1.22 — 2.12 87% 36% 2.61 — 3.04 a) Control (replication of Jordan) 27.5 65.6 63.0 57.0 1 .29 1 .35 1.52 b) Strong relation among peers 22.5 55.0 73.1 64.5 1 .77 1 .25 1 .47 c) Weak relation among peers 34.3 45.0 61.7 58.9 1 .52 1.08 1 .14 d) Strong relation between p and an expert 21.9 58.8 64.9 73.3 1 .71 1 .52 1 .40 e) Weak relation between p and an expert 33.2 50.9 60.1 67.7 1 .52 1 .27 1 .10 Zajonc and Burnstein (1965a) Errors in learning of structures a) Important issue 1.87 2.58 3.07 1.85 1 .10 0 .90 1 .53 b) T r i v i a l issue 3.87 2.47 2.80 3.00 0 .91 0 .82 0 .77 Zajonc and Burnstein (1965b) Errors in learning of structures 2.45 3.35 3.06 1.91 0 .85 0 .95 1 .47 14 Table 1 Continued c u o C •H cu 4J e CJ o cd c I-I CO t-1 J J rH an 4-1 CO < <5 Effect ratio p and o p and o Study, measure, and condition agree disagree + - + -pL o pL o pL o pL o Steiner and Spaulding (1966) Pleasantness ratings from 1 to 18 (sum of four items) a) Wisconsin sample males 31.20 17.68 17.88 22.37 1.21 1.22 1.51 b) Wisconsin sample females 30.47 14.79 19.47 22.47 1.08 1.34 1.54 c) I l l i n o i s sample (8 items) males 55,93 33.75 33.23 43.25 1.17 1.16 1.48 d) I l l i n o i s sample (8 items) females 58.32 34.32 34.00 42.68 1.21 1.20 1.47 15 Table 2: Reorganization and extension of Zajonc's table comparing balance, agreement, and attraction Effect Ratio Study, measure, and condition p and o agree p and o disagree pL o pL o pL o pL o C 4-1 o c •H 0) 4-1 e O O i-H ID 4J rt < < CO I. Pleasantness, tension, or consistency ratings Gutman Q.969) a) Pleasantness scores oh a 9-point scale b) Consistency scores on a 9-point scale c) Tension scores on a 9-point scale Hershkowitz (1954) Unpleasantness scores on a 100-point scale a) values b) objects Jordan (1953) Unpleasantness scores on a 90-point scale Knox (1963) 6.91 3.40 4.17 1.74 6.77 3.55 4.16 4.46 .76 3.98 3.66 6.14 20.1 54.2 66.6 62.9 24.3 54.7 60.6 62.9 27.9 64.5 65.8 67.8 a) Pleasantness scores on a 9-point scale (reciprocated sentiments only) 1) positive unit relations 2) negative unit relations b) Consistency scores on a 9-point scale (reciprocated sentiments only) 1) positive unit relations 2) negative unit relations Price, Harburg, and Newcomb (1966) Percent of S_9 reporting unpleasant affect 8.69 5.97 2.90 4.45 6.45 1.33 5.10 3.94 8.52 6.74 6% 3.60 4.39 41% 4.41 7.35 4.28 6.92 87% 36% 1.74 2.15 1.14 1.20 1.36 1.46 2.07 2.29 1.11 1.74 1.35 1.46 1.56 1.38 1.32 1.44 1.41 1.36 1.49 3.58 1.07 1.15 1.32 1.04 1.03 1.18 1.98 .99 .97 1.58 2.61 — 3.04 16 Table 2 Continued Effect Ratio Study, measure, and condition p and o agree p and o disagree pL +o pL o pL +o pL o c 4J o C •rl QJ 4J a> s CJ u CO a Qi U CO r-l 4-1 r H OC 4-1 CO < -Retest Task 4- 4-Volume F2 II. n=5 Pleasantness Test—^-Conservation Hletest Task 4 - 4 - 4-F2 Volume F l Consistency Test 'Conservation >-Retest Task -j- 4- 4-F2 Quantity F l III. n=5 Consistency Test ^Conservation *-Retest Task 4 - 4 - 4-F l Volume F2 Pleasantness Test ^Conservation—•Retest Task 4 - 4 - 4-F l Quantity F2 IV. n=5 Consistency Test ^-Conservation—>-Retest Task 4 - 4 - 4-F2 Quantity F l Pleasantness Test—^Conservation—>Retest Task 4 - 4 - 4-F2 Volume F l Figure 3: Diagrammatic representation of the experimental design. 31 (mean age 6.0 years), 7-8 (mean age 7.7 years), 9-10 (mean age 10.0 years), and 11-12 (mean age 12.0 years). A l l attended Dr. R. E. McKechnie School, a public school of moderate size, in Vancouver, B. C. Subjects in the 5-6 group were enrolled in kindergarten. Subjects in the 7-8, 9-10, and 11-12 groups were in grades 2, A, and 6 respectively. was blind as to JEj/s specific purposes and hypotheses. Procedure Two sets of instructions were developed—one suitable in language and content for use with younger children (i.e., those in the 5-6 and 7-8 age groups) and another set, identical in purpose, but more suited to the interests and a b i l i t i e s of older children (i.e., those in the 9-10 and 11-12 age groups). Procedure for testing younger children (groups 5-6 and 7-8). (a) Training Session I began with a training period designed to ensure understanding of the rating procedure. Six size-graded blocks were placed in disarray on the table in front of the subject. _E said: "See these blocks—they are a l l different sizes. I'm going to make a stairway out of them. Watch carefully how I do i t . " E then constructed a stairway (starting with the smallest block), making sure that the child watched the procedure. E then mixed the blocks and said—"Now, I want you to make a stairway just like the one I made." If. the child i n i t i a l l y failed the task, E repeated the demonstration. E was prepared to discontinue testing i f the child failed to seriate correctly following the second demonstration. There were, however, no cases of failure following the second demonstration. E was therefore able to; proceed with a l l subjects to rating scale I which 32 consisted of a 26" x 11%" sheet of white shelf paper on which were drawn six points, three inches apart. Intervals delineated by scale points were numbered from 1 through 6. As shown in Figure 4, squares of increasing 4 size paralleled the increase in scale numbers. Js introduced the scale as follows: "I'm going to build another stairway but this time I'm going to put my blocks on this line. See the l i t t l e wee box at this end of the Figure 4: Rating Scale I line (low end of the scale)—well, I'm going to put the l i t t l e s t block right here on top of the l i t t l e s t box. I'm going to put this block, which is a l i t t l e b i t bigger, in the next box i n the line . . .I'm going to put this block, the biggest block, here at this end of the line i n the biggest box. See, we have a stairway again." E then scrambled the blocks and asked the subject to make a stairway on the line . Errors were The rating scales and training procedures employed in this study were specially designed to meet the needs of young, pre-literate children. They were inspired to some extent by the work of Walster, Berscheid, and Barclay (1967) and Elkind (1964). Walster, Berscheid, and Barclay (1967) investigated post-choiee dissonance reduction i n nursery-school children. They had subjects rate toys, before and after choosing one, on a scale which consisted of five squares which increased in size from one square centimeter to five square centimeters. Elkind (1964) used a "stairway" procedure to investigate seriation in children aged 4-6 years. 33 drawn to the subject's attention and he was asked to correct them. Rating scale I was then removed and the child was presented with the following items: a pencil, a small plastic bracelet, a harmonica, a viewmaster, and a scotty dog. jE drew attention to each item and allowed the child to handle i t . Rating scale II was then presented. Rating scale II consisted of nine points, five inches apart. Scale intervals delineated by these points were numbered from 1 through 9. Squares of increasing size paralleled the increase in scale numbers. As shown in Figure 5, rating scale II was anchored at the high end by a large red heart and at the low end by a large blue cross. E said: "See this line, i t ' s just li k e the one we made the Figure 5: Rating Scale II stairway on before, only i t ' s a l i t t l e b i t bigger. What I want you to do this time i s to put the toys on the line so the toys make a stairway. The toy you like the very, very best—the one you would like to play with most of al l — s h o u l d go at this end of the line, in the big box above the red valentine. The red valentine means that you like the toy very, very much. Now, where would you put the toy you lik e the very, very best? . . . very good. Now, see this blue cross at this end of the line? The blue cross means that you don't like the toy. The toy that you really don't like very much and wouldn't want to play with at a l l should go at 34 this end of the line in the l i t t l e box by the blue cross. Now, where would you put the toy you dislike the most? . . . very good. What I want you to do is to start with the toy you lik e the very, very best and put i t at this end of the l i n e — b y the red valentine. Then take the toy you like next best arid put i t here. The last toy, the one you put in the l i t t l e box by the blue cross, should be the one you like the least. O.K.? If you like two toys the same you can put them in the same box. Now let's play the toy game. Show me the toy you l i k e the best—now, where w i l l you put it? etc." When the subject finished, E removed the toys and rating scale and stated: "Before we play any more stairway games I'm going to show you some cards. Each card t e l l s a story, but the story is told in a secret code. Would you like to learn the secret code?" IS then presented training cards 1-9. These cards, described in Table 5, were designed to teach the child Table 5 to associate a red heart with "likes," a blue cross with "dislikes," and a box with "something very, very important." The subject was then told that he would be shown some cards containing stories written in the "secret code." He was instructed to examine each card and to t e l l _E the "story" on i t . Cards 10-17 (series A) were then presented in succession. Any errors or omissions in "t e l l i n g the story" were drawn to the subject's attention as they occurred. The structures represented in series A (and in a l l subsequent series) are shown in Figure 6. Structures 1, 2, 7, and 8 are balanced; structures 3, 4, 5, and 6 are imbalanced; a l l relations between persons are reciprocal. 35 T a b l e 5: T r a i n i n g c a r d s 1-9 Card No. 1 C o n t e n t o f C a r d o 1 V 7 / — - > O A o r X _E's e x p l a n a t i o n A r e d v a l e n t i n e means t h a t somebody l i k e s s o m e t h i n g . A b l u e c r o s s means t h a t some-body d o e s n ' t l i k e s o m e t h i n g . A box means some t h i n g v e r y , v e r y i m p o r t a n t . T h i s c h i l d l i k e s t h e i m p o r t a n t t h i n g . See, t h e r e i s a r e d v a l e n t i n e . T h i s c h i l d d o e s n ' t l i k e t h e v e r y i m p o r t a n t t h i n g . See, t h e r e i s a b l u e c r o s s . T h i s c h i l d l i k e s t h e i m p o r t a n t t h i n g . See, t h e r e i s a r e d v a l e n t i n e . T h i s c h i l d d o e s n ' t l i k e t he i m p o r t a n t t h i n g . See, t h e r e i s a b l u e c r o s s . These two c h i l d r e n l i k e each o t h e r . T h i s c h i l d l i k e s t h i s one and t h i s c h i l d l i k e s t h i s one. See, t h e r e a r e r e d v a l e n t i n e s h e r e and h e r e . These two c h i l d r e n don't l i k e each o t h e r . T h i s a r r o w means t h a t he d o e s n ' t l i k e him because t h e r e i s a b l u e c r o s s h e r e . T h i s arrow means t h a t t h i s c h i l d d o esn't l i k e t h i s one because t h e r e i s a b l u e c r o s s h e r e . 36 Each structure was presented in pi c t o r i a l form on a separate 3" x 5" card. As shown in the example in Figure 6 stick figures were used to indicate persons "P" and "0" and a box was used to symbolize "X," "something very, very important to both children." The affective relations between elements were indicated by red hearts (likes) and blue crosses (dislikes). The direction of relations between elements were indicated by black arrows. Separate sets of cards illustrating the structures were prepared for males and females. Figure 6: Structural content of training cards 10-17 Upon completion of series A, card 18 was presented. On this card, the stick figure in the "P" position was circled. E stated, "Let's pretend that you are one of the boys (girls) in the story. You are this boy (girl) with the c i r c l e around him (her). Now, I want you to look at some stories and whenever you see a boy (girl) with a c i r c l e around him (her) pretend that boy (girl) i s you." Cards 19-26 (series B) were then presented in succession. The structures represented in series B were the same as those 37 in series A but the presentation sequence was varied and figures in the "P" position were circled. The subject's task was to describe the structure on each card, placing himself in the "P" position. (°) Pleasantness ratings: test and retest Pleasantness ratings were made on scale III. Scale III contained fifteen points, five inches apart, anchored at the high end by a smiling face and at the low end by a frowning face (see Figure 7). Squares of increasing size again paralleled the increase in number value of scale intervals. E_ introduced the scale as follows: "Remember when we played O I C l I I—I I • T 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Figure 7: Rating Scale III the stairway game before, I showed you some toys and you told me which one you liked the very, very best and we put i t in the biggest box on the line? Then you showed me the toy you liked next best, and we put i t in the next biggest box. What I want you to do now, is to look at the stories on these cards and put them in the boxes on this line. Put the nicest story, the one that makes you feel very, very happy, in the biggest box at this end of the line by the picture of the happy, smiling face. The one that you like next best, that doesn't make you feel quite as happy, should go in one of these boxes . . . the story that makes you feel very, very unhappy should go at this end of the line by the picture of the sad, 38 unhappy face. O.K.? When you are a l l finished today you'll get your red (or green) ticket for the class party." IS placed cards 27-34 (test series) in a semi-circle in front of the child and ni.id°. "What I want you to do is pretend that this child—the one with the circl e around him (her) is you and this other child i s your partner (Format 1). Do you know what a partner is? Pretend that at the beginning of the year the teacher picked two children—you and this other boy (girl) in the card and said that you two were to be partners for the whole year. You are to help each other with your school work, s i t together, and work on special projects together. Look at a l l the stories and ask youself how happy you would feel i f (pointing to structure 5) you liked your partner and he (she) liked you, and you didn't like the box but he (she) liked i t . " ' Or how happy would you feel (pointing to structure 4) i f you didn't like your partner and he (she) didn't like you and you both didn't like what was in the box. Put the nicest, happiest story at this end of the line by the happy, smiling face, then the next nicest story, and the next, unti l you get down to the most unpleasant, unhappy story." In order to minimize possible experimenter bias effects, went to another part of the room, with her back to the subject while ratings were being made. The subject was instructed to t e l l IS when he had finished his ratings. When the card numbers and scale positions had been recorded, rating scale III was removed. A test of conservation (quantity or volume) was then conducted. The procedure followed in administering these tests was ^The structures are presented in diagrammatic form in the foldout in Appendix 2. The reader may find i t helpful to refer to these diagiams throughout the remainder of the paper. 39 as follows: Conservation of quantity. The subject was shown two glasses (4" in height, 2V1 in diameter) f i l l e d to the same level with colored water. He was asked whether or not the two glasses contained the same amount of water. The water from one glass was then poured into a taller and wider glass (6" in height, 2 3/4" in diameter). The subject was asked whether the two glasses contained more, less, or the same amount of water. He was asked to explain his answer. Conservation of volume. In the conservation of volume task, the subject was shown a glass (6" in height, 2 3/4" in diameter) two-thirds f i l l e d with water. A round piece of Plasticine was placed in the glass. The new water level was marked with a rubber band. The Plasticine was then removed from the water, dried, and rolled into a "sausage." The subject was asked to predict whether the water level would rise more, less, or the same amount i f the "sausage" were placed in the water. He was asked to explain his answer. Rating scale III was then placed once more before the child and the retest series (cards 35-46) was presented. 12 stated: "We are going to play one more line game before you get your red (or green) ticket. Pretend that this is you (E indicated the figure in the "P" position) and this other child i s someone who lives near you that you play with (Format 2). Think how pleasant or happy you would feel i f you liked this child and he (she) liked you and you both didn't like what was in the box (structure 2). Think how happy you would feel i f you didn't like this child and he (she) didn't lik e you and you didn't like what was in the box, but your playmate did (structure 8). How would you feel then? Put the nicest, happiest 40 story here in the biggest box by the happy, smiling face. Put the saddest, most unhappy story in the l i t t l e s t box by the sad, unhappy face. When you are a l l finished you w i l l get your ticket." When the retest series was completed, 12 probed briefly for a classification rule. The child was asked: "How did you decide where to put the cards? Why did you put these stories here (IS pointed to the low end of the scale) and these stories over here (high end of the scale)?" Subjects were then given their tickets, thanked, and dismissed with the plea not to t e l l their classmates about the task "because i t wouldn't be f a i r . " Subjects were given a red ticket at the close of session I and a green ticket at the end of session II. Session II was conducted 6-11 days after session I. If pleasantness ratings were obtained in session I, consistency ratings were obtained i n session II and vice versa. IS introduced session II as follows: "Remember last week, when we played some stairway games and we put toys and stories on a l i n e — w e l l , we are going to play some more stairway games today. We are also going to play some more games with glasses of water today. When we are a l l finished you w i l l get your green ticket and on (day of week) we are going to have the class party. It should be lots of fun . . . Before we start the stairway games though, I want to see i f you remember the secret code. Can you t e l l me what i t says on this card?" Cards 19-26 were then presented in succession. IS corrected any mistakes the child made in "te l l i n g the stories." (c) Consistency ratings: test and retest Consistency ratings were made on scale IV. Like scale III, i t contained fifteen points, five inches apart; squares of increasing size again paralleled the increase in scale numbers. As shown in Figure 8, i t was anchored at the high end by a large r e d » S" a n d a t t h e l o w e n d b y a 41 red "S" with a black cross over i t . Subjects were introduced to the task of rating structures for consistency in the following manner: E stated: i ° i ° i ° i u i ° IIJ I U I U PPIUDIM -4M 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 c Figure 8: Rating Scale IV O "Iwantyou to read the stories on the cards and decide how much sense they make. Stories that seem to make sense should go in the big boxes at the end of the line by the big red "S." Stories that seem to be "mixed-up" or s i l l y , stories that don't make sense should go i n the l i t t l e boxes at the end of the line by the S with the black cross over i t . Now, suppose I told you that when I was out yesterday I saw a l i t t l e tiny cat eat a great big dog. Where would you put a story like that? Right, you would put i t i n one of the wee boxes by the S with the cross over i t , because i t doesn't make s e n s e — l i t t l e wee cats can't eat big dogs! Suppose I told you that 3 plus 3 i s 8. Where would you put a story lik e that? R i g h t — i n a l i t t l e wee box because that doesn't make sense either. 3 plus 3 i s 6. We make 8 by adding 4 plus 4 or 5 plus 3 or 6 plus 2 or 7 plus 1. (Kinder-garten: Suppose I told you there wera 3 blocks here. Does that make sense? No—because there are 6 blocks here—see 1, 2, 3 . . . 6. What I said didn't make sense.) Now, what i f I told you that Peter likes Joe the very best in the whole world—but, I know that Joe is always mean to Peter. Joe hits Peter and pushes him down and kicks him every time they play together. Does i t make sense for Peter to like Joe best in the 42 whole world when Joe is always mean? Where would you put this story? What i f I told you my best friend was always nice to me—where would you put a story lik e that? Well, what I want you to do is to read the stories on these cards and decide whether or not they make sense and then put them on the line . If the story makes lots of sense, where would you put it? If the story seems very s i l l y or mixed up or wrong, where would you put it? . . . The stories in the cards are like the ones you saw before—they are about the two children who are partners in school (Format 1). See in this story (structure 5) you like your partner and he (she) likes you. You don't like what's in the box but he (she) does. Does that make sense? Or what about this one (structure 4) . . . you don't like your partner and he (she) doesn't like you. He (she) doesn't like what's in the box. You don't lik e what's in the box." When the test items had been rated, scale IV was removed and a conservation task was admiristered. A second (retest) series in the alternate format was then presented. When the retest series was completed, 12 probed briefly for a classification rule. The child was asked "How did you decide where to put the cards? . . . Why did you put these stories here (low end of the scale) and these stories over here (high end)?" The subject was then thanked, given his ticket, and dismissed. Procedure for testing older children (groups 9-10 and 11-12) With the older children, the "secret code" guize was dropped and training cards 10-17 were omitted. Instead of beginning session I with practice in building stairways, E began the session as follows: "I'm going to show you some cards today. Each card t e l l s a story—the story is about two children and something that is very, very important to both of them. In some of the stories, the two children l i k e each other, in other stories 43 they dislike each other. Sometimes both children like or both dislike the important thing; in other stories one child likes the important thing but the other child doesn't like i t . I'm showing these cards to children from kindergarten right up to grade 6. Now l i t t l e kids can't read, so I had to make the stories in pictures. Whenever you see a red heart, i t means . . . ." The introduction to session II also differed for older children. E_ stated: "Remember last week when we did some experiments with cards—well, we are going to do some more today. We are also going to do another experi-ment with the glasses of water. But, before we start, I want to make sure that you remember how to read the cards. Can you t e l l me what i t says on these cards? . . . " The remainder of the procedure was identical to that used with younger children except that no mention was made of a class party. 44 CHAPTER III ASSESSMENT OF THE PROCEDURES Qualitative Observations A l l children in the 5-6 and 7-8 groups correctly seriated six blocks, both off and on the "li n e " (i.e., scale I ) . There were few occasions when E needed to correct errors or repeat the i n i t i a l demonstration of "stairway" building. Subjects indicated, verbally and by placement of toys on scale II, that they understood that their task was to arrange things on "lines" in order of increasing magnitude. Both verbal comments during training and scale placement of examples made i t apparent that children at a l l age levels understood that pleasantness ratings were to be based on "how happy I would feel i f I were really i n the situation." There were similar indications that they understood that consistency ratings were to be made on the basis of how much "sense" the situations made. It was E/s impression that motivation was high in a l l age groups, both on the basis of observation of the children's behaviour during the experiment and from conversations with the children and their teachers at the conclusion of the study. An additional indication of high motivation was the willingness of the children to participate in the second session. Reliability of the Ratings Two r e l i a b i l i t y coefficients were computed for each subject: one for pleasantness and one for consistency. Coefficients of each type were computed by correlating the subject's ratings under formats 1 and 2. Z transformed intra-individual coefficients were then combined to yield average r e l i a b i l i t y coefficients for each age group. These are shown in ^ — — — — — — -Fisher transformations were used in a l l cases where correlations were averaged or tested for significance. 45 Table 6. At a l l age levels r e l i a b i l i t y coefficients for pleasantness and Table 6 and consistency are of sufficient magnitude to warrant the conclusion that the processes underlying the ratings were stable across formats. Performance on the Interpolated Conservation Tests Tests of conservation were selected for use as the interpolated activity between ratings made under formats 1 and 2 in the hope that the results of these tests might be of aid in interpreting the ratings and to fa c i l i t a t e comparison of results with those of Atwood (1969). However, as shown in Table 7, although the majority of subjects in the 5-6 group performed i n a manner consistent with Piagetian theory and research' Table 7 (i.e., 85% failed to conserve quantity or volume), the performance of In Piagetian theory, a number of cognitive changes are hypothesized to take place at the end of the pre-operational stage, the most important of which is manifested in the acquisition of the "schema of conservation." When a child i s able to conserve, he realizes that certain properties of an object (e.g., quantity, weight) remain constant in the face of certain transforma-tions (e.g., changes in the object's shape). His thinking is no longer dominated by physical appearance. However, although an understanding of conservation marks the transition from pre-operational to operational functioning, the various types of conservation described by Piaget and his co-workers (Piaget, 1946; Piaget, 1952; Piaget and Inhelder, 1941; Piaget, Inhelder, and Szeminska, 1960) do not develop a l l in a piece. For example, although conservation of quantity i s generally acquired by age 7-8, con-servation of weight does not appear until 9-10, and conservation of volume is seldom apparent before 11-12 years. Atwood (1969) and others have used the differential growth rate of quantity and volume conservation to further subdivide subjects into the Piagetian stages of concrete operational and formal operational functioning. 46 Table 6: Mean r e l i a b i l i t y coeffi- ients (Pearson r) for pleasantness and consistency ratings 5-6 7-8 9-10 11-12 Pleasantness .84 .93 .97 .91 Consistency .87 .92 .89 .89 47 Table 7: % of subjects In each age group showing quantity or volume conservation (principal study n^O in each age group; cross-validation s-tudy n=8) Quantity conservation Volume conservation Age Gp. 5-6 Failure Success Principal Cross-validation Study Study 85.0% 15.0% 75.0% 25.0% Principal Cross-validation Study Study 85.0% 15.0% 87.5% 12.5% 7-8 Failure Success 15.0% 85.0% 25.0% 75.0% 40.0% 60.0% 62.5% 37.5% 9-10 Failure Success 0 100 % 0 100 % 25.0% 75.0% 12.5% 87.5% 11-12 Failure Success 0 100 % 0 100 % 35.0% 65.0% 12.5% 87.5% 48 children in the three older groups was somewhat anomalous. For example, in the principal study the proportion of volume conservers in the 7-8 and 9-10 groups was inordinately high (60% in the 7-8 group and 75% in the 9-10 group) for a cognitive ab i l i t y not supposed to be reliably found before 11-12 years of age (cf., Piaget and Inhelder, 1941). Furthermore, children in the 11-12 group performed no better on the volume conservation task than 9-10 year old children and l i t t l e better than those of 7-8 years. Thus, in view of the atypicallity of the conservation results and the virtually indistinguishable performance of the three older groups, i t did not seem feasible to classify subjects by conservation level nor to make any further attempt to relate these data to the results of the rating task. 49 CHAPTER IV RESULTS AND DISCUSSION OF HYPOTHESES Hypothesis 1 Hypothesis 1 was concerned with the relationship between perceptions of pleasantness and perceptions of consistency in subjects of varying chronological age. Younger children, hypothesized to differentiate l i t t l e between the pleasantness and the consistency of social situations, were expected to yield high correlations between pleasantness and con-sistency ratings. Older children, hypothesized to differentiate somewhat more between pleasantness and consistency in social situations were expected to yield lower correlations between the two types of ratings. Separate correlation coefficients were computed for each subject by correlating his responses to the same situations when rating for pleasant-ness and for consistency. Intra-individual correlation coefficients were then averaged for each age group. These average correlations are shown in Table 8. Inspection of the table indicates a rather high correlation Table 8 between pleasantness and consistency in a l l age groups. A one-way analysis of variance by age groups of the transformed values of the intra-individual correlations indicated no significant between-group effects (F<1). The results, indicating a lack of clear differentiation between pleasantness and consistency at a l l age levels, thus f a i l to confirm the hypothesis that differentiation would increase between the ages of 5-12 years. 8 These high correlations, while necessary to an inference that children 50 Table 8: Mean pleasantness-consistency correlations for each age group (Pearson r) Age Group r p ( , 5-6 .72 7-8 .84 9-40 .79 11-12 .77 51 The absence of evidence of differentiation between pleasantness and consistency, although predicted in the case of younger children, was un-expected in the case of older children. In the training session, these children indicated that they understood the meaning of consistency. In predictions tasks (e.g., Atwood, 1969; Knox and Gutman, 1968; Storm and Knox, 1969) children above the age of 7 show a strong tendency to perform in accordance with consistency principles although younger children appear to respond on a more affective basis. Further, there is evidence that adults distinguish between pleasantness and consistency when rating hypothetical P-O-X situations (e.g., Knox, 1963; Gutman, 1969). Yet, the present results indicate that children between the ages of 7-12 years do not. The question that remains is why? A possible explanation, relating to the overall complexity of the rating task and to the strength of the balance "schema" in comparison to alternative cognitive biases is offered in Chapter VII. Hypotheses 2 and 3 Hypotheses 2 and 3 concerned the relative importance of agreement and balance for pleasantness and consistency ratings. It was predicted that children at a l l age levels would attach greater weight to agreement than to balance when making pleasantness ratings (hypothesis 2) and that younger do not differentiate clearly between the pleasantness and consistency of triadic structures are not alone sufficient for this inference. It is possible, in other words, that the children were differentiating on some basis masked by the correlations. For example, i t i s conceivable that the children could have utilized different portions of the raxing scale when assessing pleasantness as compared to consistency. Consequently, each child's distribution of ratings on these two c r i t e r i a was examined. The distributional characteristics of the pleasantness and consistency ratings were found to be virtually identical for 90% of the children. 52 children would also base consistency ratings more on agreement than on balance (hypothesis 3). Balance was expected to exert greater influence than agreement on consistency ratings of older children (hypothesis 3). The unanticipated high correlation between pleasantness and consistency ratings at a l l age levels precluded joint confirmation of these two hypotheses. Conceivably, however, the data might s t i l l have supported one or the other hypothesis—a possibility that warranted further s t a t i s t i -cal tests. The s t a t i s t i c a l procedures employed for these tests would also provide information concerning the relative contribution of agreement, balance, and other structural determinants (e.g., attraction) to the two sets of ratings. Thus, analyses of variance were performed on the ratings. A between-within design (Winer, 1962, p.320) was used. Assessed within subjects were the agreement factor (A) with two levels, the attraction factor (B) with two levels, the format factor (C) with two levels, and a structure factor (0) with two levels that reflected high or low positivity. The age factor (D) with four levels, was a between-group factor. (The overall design is summarized in Table 9.) Zajonc indices were also computed in Table 9 order to f a c i l i t a t e comparison of results with those obtained in previous studies. Multiple regression equations were fi t t e d to the data of each subject in order to examine the relative effects of agreement, balance, and attraction at the individual level, (a) Results relevant to hypothesis 2 The results of the analysis of variance of pleasantness ratings are 53 Table 9: Design of the analyses of variance Designation Factor Label Levels Structure No. A Agreement 1) Agreement 1, 2, 3, 4 2) Disagreement 5, 6, 7, 8 B Attraction 1) Positive 1, 2, 5, 6 2) Negative 3, 4, 7, 8 C Formats 1) Partners a l l structures 2) Neighbor-playmates II tt D Age 1) 5-6 years a l l structures 2) 7-8 years ti II 3) 9-10 years n ii 4) 11-12 years II ii 0 Structures 1) high positivity 1, 3, 5, 8 2) low positivity 2, 4, 6, 7 54 summarized i n Tab le 10. As shown i n t h i s t a b l e , the main -e f fec t due to agreement was s i g n i f i c a n t (F=7.42, d f - 1 , 7 6 ; P<.01) . The presence of a s i g n i f i c a n t i n t e r a c t i o n between agreement and a t t r a c t i o n (F=33.46; d f = l , 7 6 ; 9 P<.01) i n d i c a t e s that the e f f e c t s o f ba lance were s i g n i f i c a n t a l s o . Table 10 Examinat ion o f the percent of t o t a l v a r i a b i l i t y ( i . e . , percent o f t o t a l sum o f squares) accounted f o r by these two v a r i a b l e s i n d i c a t e s , however, t ha t ba lance e f f e c t s were s l i g h t l y g rea te r than those of agreement ( the reve rse of t ha t p r e d i c t e d i n hypo thes is 2 ) . The percent of v a r i a b i l i t y accounted f o r by these v a r i a b l e s was, n e v e r t h e l e s s , extremely s m a l l . (Balance accounted f o r on ly 1.36% of the t o t a l v a r i a b i l i t y ; the amount a t t r i b u t a b l e to agreement was .29%.) Sub jec ts i n a l l age groups appear to have based t h e i r r a t i n g s p r i m a r i l y on a t t r a c t i o n . The F r a t i o f o r a t t r a c t i o n was 410.09 ( d f = l , 7 6 ; P<.01) . Th i s f a c t o r accounted f o r 45.51% of the t o t a l v a r i a b i l i t y . * 0 The r e l a t i v e magnitude of Zajonc i n d i c e s f o r b a l a n c e , agreement, and a t t r a c t i o n were c o n s i s t e n t w i t h the f i n d i n g s of the a n a l y s i s of v a r i a n c e . * Ba lance i n v o l v e s the i n t e r a c t i o n of the l i k i n g r e l a t i o n between P and 0 and the agreement between them concern ing X . Balanced s t a t e s are those i n which P and 0 l i k e one another and agree , or a l t e r n a t i v e l y , where they d i s l i k e one another and d i s a g r e e . Imbalanced s t a t e s combine l i k i n g w i th d isagreement , o r d i s l i k i n g w i t h agreement. ^ I n s p e c t i o n of Tab le 10 i n d i c a t e s that e f f e c t s r e l a t i n g to f a c t o r 0 a l s o account f o r a g r e a t e r p r o p o r t i o n of the v a r i a b i l i t y than agreement o r b a l a n c e . (Resu l t s and d i s c u s s i o n of f a c t o r 0 are presented In Chapter V toge ther w i t h e f f e c t s r e l a t i n g to format d i f f e r e n c e s . ) **Mean p leasan tness and cons i s tency r a t i n g s used i n the computat ion o f Zajonc i n d i c e s a re presented i n Appendix 3 . 55 Table 10: Summary of analysis of variance of pleasantness ratings % of varia-b i l i t y due to significant main effects Source of and inter-Variation df MS F actions Between Subjects 79 D (Ag e gps) 3 128.69 3.28* 1.27 Subj . in gps 76 39.19 Within Subjects 1200 A (Agreement) 1 88.73 7.42** .29 DA 3 40.15 3.36* .40 A x Subj. in gps 76 11.95 B (Attraction) . 1 13867.00 410.09** 45.51 DB 3 21.09 .62 B x Subj. in gps 76 33.81 C (Format) 1 3.94 1.02 DC 3 7.36 1.91 C x Subj. in gps 76 3.84 0 (Structure) 1 1123.10 156.23** 3.69 DO 3 14.52 2.02 0 x Subj. in gps 76 7.19 AB (Balance) 1 415.19 33.46** 1.36 DAB 3 9.04 .73 AB x Subj. in gps 76 12.41 AC 1 21.27 3.89* .07 DAC 3 18.78 3.44* .18 AC x Subj. in gps 76 5.46 AO 1 1590.90 206.60** 5.22 DAO 3 20.15 2.62 AO x Subj. in gps 76 7.70 BC 1 17.81 2.85 DBC 3 18.00 2.89* .18 BC x Subj. in gps 76 6.24 BO 1 .49 .08 DBO 3 2.96 .49 BO x Subj. in gps 76 6.09 CO 1 7.97 1.66 DCO 3 2.67 .56 CO x Subj. in gps 76 4.80 56 Source of Variation df MS ABC 1 31.56 DABC 3 3.06 ABC x Subj. in gps 76 4.36 ABO 1 1.19 DABC 3 16.50 ABO x Subj. in 8PS 76 5.26 Residual 240 4.68 Total 1279 7.24** .70 .23 3.14* % of varia-b i l i t y due to significant main effects and inter-actions .11 .16 *p<.05 **p<.01 57 As shown in Table 11, indices for balance were of slightly greater magnitude than those for agreement, in a l l groups. Indices for agreement and balance were small in comparison to those for attraction. The index Table 11 value of largest magnitude at a l l age levels was that for attraction. The same pattern of results emerged when pleasantness ratings were subjected to multiple regression analysis (see Appendix 4). Average beta coefficients for balance were greater than those for agreement in a l l groups, and average beta coefficients for attraction were greater than those for balance. A l l three methods of analysis indicate, in other words, that pleasantness ratings were based primarily on the P/0 relationship. The prediction that agreement would exert stronger influence on pleasantness ratings than balance was based on theoretical considerations outlined in the Introduction. As previously mentioned, however, strong attraction effects in pleasantness ratings were obtained by Ohashi (1964) in children, and by Knox (1963) and Gutman (1969) in adults. The present results are thus not without precedent. The results of the present study (and those of Ohashi, 1964) differ, however, from those obtained by Knox (1963) and by Gutman (1969) in the absolute magnitude of attraction and agreement effects. In Gutman's study, for example, attraction accounted 12 for 33.28% of the vari a b i l i t y ; agreement accounted for 18.41%. In the Summaries of the analyses of variance performed on Gutman's (1969) data are presented in Appendix 5. 58 Table 11: Zajonc indices for pleasantness Effect ratio p and o p and o agree disagree + - + -Age Group pL o pL o pL o pL o a 4-1 o C •H CD 4J CU e O o CD CO c OJ U ca U 4-1 r-4 00 4-> td < . This factor accounted for 31.13% of the*variability. Agreement accounted for 3.07% of 13 the v a r i a b i l i t y . The proportion due to balance was 1.24%. The same pattern of results with respect to the effects of balance, agreement, and attraction obtained when the data were analyzed according to Zajonc's (1968) method. As shown in Table 14, index values for agree-ment are slightly greater than those for balance in a l l but the 7-8 group Table 14 but the most noticeable differences are between balance and agreement on the one hand and attraction on the other. The pattern was repeated when consistency ratings were subjected to multiple regression analysis. As shown in Appendix 4, average beta coefficients for agreement were larger than those for balance in a l l but the 7-8 group. Average beta coefficients for attraction exceed those for agreement and balance in a l l groups. A l l three methods of analysis thus indicate: (a) that agreement exerted slightly greater effect on the ratings than balance but that attraction was the primary component of consistency ratings; and (b) that there was no significant interaction between age and balance, age and agreement, or age and attraction. The prediction that balance and agreement would interact with age, 13 Factor 0 effects were also significant (F=66.19; df=l,76; P<.01). The proportion of variability accounted for by factor 0 was 2.68%. An additional 3*23% of the va r i a b i l i t y is attributable to the interaction between factor 0 and agreement. The contribution of factor 0 to the ratings was thus greater than that of balance. 65 Table 14: Zajonc Indices for consistency Effect Ratio p and o p and o agree disagree § XJ co c + _ + _ CD VJ CO Age group pL o pL o pL o pL o & JJ "CO 0 \/ P v ^ - -»0 *o P < - - -*0 V V (5) (8) C (6) (7) D (a) 01 02 01 02 Al 10.95 6.84 Al 10.78 7.41 A2 8.19 8.55 A2 7.30 7.46 X Pleasantness ratings X Consistency ratings (b) (c) 71 in younger children. However, since older children are capable of u t i l i z i n g more complex strategies, one would expect older children to take greater cognizance of the position in which positive relations occur in the triads. The ratings of older children, in other words, should be based more on the arrangement of positive and negative relations in different triads than on the total number of positive relations contained. The significant interaction of factor 0 with age in the consistency ratings (F=5.48; df=3,76; P<.01) is consistent with this expectation. The interaction with age was not significant for the pleasantness ratings (F=2.02; df=3,76; P=.12). The effect of attraction—the predominant influence on pleasantness and consistency ratings at a l l age levels i s , of course, a positive biasing effect as well, but one specific to the P/0 relationship, which tends to override the preference for positive bonds elsewhere in the triad. 2. Effects involving factor C (Format effects) Main effects due to format differences were not significant in either the analysis of pleasantness ratings (F=1.02; df=l,76; P=.32) or the analysis of consistency ratings (F<1). Only one interaction involving factor C was significant in the case of consistency—that between age, attraction, and formats. In the case of pleasantness, formats were sig-nificant as a factor in interaction with agreement; age and agreement; age and attraction; and agreement and attraction. In general, these interactions suggest that triads involving neighbor-playmates have a more pleasant connotation for younger children than triads involving the more task-oriented relationship of partners. Differences i n the means involved in these interactions are small, however. Effects involving factor C 72 account for less than 1% of the va r i a b i l i t y in the data. 3. Main effects of age Significant main effects of age were obtained both for pleasantness (F=3.28; df=3,76; P<.05) and consistency (F=6.82; df=3,76; P<,01). Means of 8.73, 8.17, 9.49 and 8.13 were obtained for each successive age group when pleasantness ratings were collapsed across structures. Corresponding average consistency ratings for the four age groups were 7.32, 7.78, 9.40 and 8.44. This would seem to indicate slightly different use of the rating scale by subjects of different chronological age.*~* Summary An unanticipated finding in the results was the difference associated with factor 0. In the case of both pleasantness and consistency ratings subjects assigned higher ratings to structures in level 1 (i.e., structures 1, 3, 5, and 8) than to those in level 2 (structures 2, 4, 6, and 7). This effect was tentatively attributed to positivity. That i s , i t was attributed to a preference for structures containing the greatest number of positive relations. There were no significant main effects for formats in either the pleasantness ratings or the consistency ratings. The format factor, although involved in several higher order interactions, apparently contributed l i t t l e to the ratings. Pair-wise comparison of age groups according to the Neuman-Keuls procedure (Winer, 1962, pp; 80-85) indicated no significant between-group differences either in the pleasantness ratings or in the consistency ratings. 73 CHAPTER VI RESULTS OF THE CROSS-VALIDATION STUDY Reliability Average r e l i a b i l i t y coefficients for pleasantness and consistency are shown in Table 16. As i n the p r i n c i p a l study, r e l i a b i l i t y i s lower i n the 5-6 group than in the older groups. This is particularly noticeable Table 16 in the case of consistency ratings where the average value for the 5-6 group is .59 while those for the other age groups range from .75 to .96. Mann-Whitney U tests indicated, however, that none of the differences between age groups were significant. Comparison across studies (see Tables 6 and 16) indicates that r e l i a b i l i t y was generally lower in the cross-validation study but the only significant difference between samples was i n the 5-6 group, and then only in the case of consistency ratings (U=120; z=2.02; P<.04, two-tailed). Hypothesis 1 Hypothesis 1 predicted an increase in differentiation between pleasantness and consistency with increasing age. The results of the principal study failed to support the hypothesis: r p r was high in a l l groups. The results of the cross-validation study also f a i l to support hypothesis 1. As shown in Table 17, instead of decreasing with increasing Table 17 Table 16: Mean r e l i a b i l i t y c o e f f i c i e n t s (Pearson r f o r Pleasantness and Consistency 5-6 7-8 9-10 11-12 easentness .76 .82 .89 .87 Consistency .59 .96 .75 .90 75 Table 17: Mean pleasantness-consistency correlations for each age group (Pearson r) Age Gp 5-6 7-8 9-10 11-12 PC .46 .72 .78 .81 76 age, average PC correlations increase with increasing age. Hypotheses 2 and 3 Hypotheses 2 and 3 were concerned with the relative effects of agreement and balance on pleasantness and consistency ratings. It was predicted that subjects in a l l age groups x, 72-87. 96 Osgood, C. E., and Tannenbaum, P. H. The principle of congruity in the prediction of attitude change. Psychol. Rev., 1955, _62, 42-55. Piaget, J. The child's conception of physical causality. London: Kegan Paul, 1930. Piaget, J. Les notions de mouvement et de vitesse che2 1'enfant. Paris: Presses Universitaires, 1946. Piaget, J. Psychology of intelligence. New York: Harcourt Brace, 1950. Piaget, J. The child's conception of number. New York: Humanities, 1952. Piaget, J., and Inhelder, B. Le developpement des quantites chez 1'enfant. Neuchatel: Delachaux et Niestle, 1941. Piaget, J., Inhelder, B., and Szeminska, A. The child's conception of geometry. New York: Basic Books, 1960. Price, K., Harburg, E., and Newcomb, T. Psychological balance in situations of negative !nterpersonal attitudes. J. pers. soc. Psychol., 1966, J3, 265-270. Rodrigues, A. On the differential effects of some parameters of balance. J. Psychol., 1965, 61, 241-250. Rodrigues, A. Tlie psycho-logic of interpersonal relations. Unpublished doctoral dissertation. Univ. of California, Los Angeles, 1966. Siegel, S. Non-parametric s t a t i s t i c s . New York: McGraw-Hill, 1956. Singer, J. E. Motivation for consistency. In S. Feldman (Ed.), Cog i t i v e consistency: motivational antecedents and behavioral consequences. New York: Academic Press, 1966. Storm, T., and Knox, R. E. Cross-cultural study of the development of social perception. Unpublished, 1967. Storm, T.s and Knox, R. E. A cross-cultural study of balance and 97 transitivity. Unpublished, 1969. Steiner, I. D., and Spaulding, J. Preference for balanced situations. Report No. 1 Dept. of Psychol. Univ. of I l l i n o i s , 1966. Walster, E., Berscheid, E., and Barclay, A. M. A determinant of preference among modes of dissonance reduction. J. pers. soc. Psychol., 1967, _7_, 211-216. Winer, B. J. Sta t i s t i c a l principles in experimental design. New York: McGraw-Hill, 1962. Zajonc, R. B. Cognitive theories of social behaviour. In G. Lindzey and B. Aronson (Eds.), Handbook of social psychology (Rev. ed.), Cambridge: Addison-Wesley, 1968. Zajonc, R., and Burnstein, E. The learning of balanced and unbalanced social structures. J. Pers., 1965, 33, 153-163 (a). Zajonc, R., and Burnstein, E. Structural balance, reciprocity, and positivity as sources of cognitive bias. J. Pers., 1965, 33, 570-583 (b). APPENDIX 1 PILOT STUDIES 99 Two pilot studies were conducted as fea s i b i l i t y tests for the main study and provided vehicles for developing a suitable procedure for young children. The f i r s t study required pleasantness ratings from subjects aged 4-6 years (n=ll). Subjects were children of E/s friends and relatives. The children were tested in their homes in two sessions. In the f i r s t session, they were shown how to use the rating scales and then rated two sets of P-O-X structures for pleasantness* there were 8 structures in each set, representing the 8 triads shown in Figure 6. Each structure was presented in pi c t o r i a l form on a separate 3" x 5" card. Stick figures were used to indicate the two persons P and 0, in each situation. A box was used to symbolize X, "something very, very important to both children." Affective relations between elements were indicated by red hearts (like) and blue crosses (dislike). The direction of relations between elements was indicated by black arrows. The subject was instructed to assume the role of P in each structure in the f i r s t set. He was asked to rate structures involving two other children in the second set. Ratings were made on a 15-Rpoint scale, 6% feet in length, anchored at the high end by a smiling face and at the low end by a frowning face. One week later, JE returned to the child's home and asked him to rec a l l the contents of the cards and to describe the rating procedure. Pilot Study II was conducted at the Vancouver Talmud Torah, a Hebrew Day School. In this, study, children in grades 2, 4, and 6 (n=16 in each grade) rated the two sets of P-0TX structures used i n Pilot Study I. They rated the structures f i r s t for pleasantness and then consistency. Con-sistency ratings were made on a 15-polnt scale identical to that used 100 for pleasantness ratings in Pilot Study I except that this scale was anchored at the high end by a large red S indicating that situations placed at this end made "lots of sense." At the low end of the scale there was a large red S covered by a black cross indicating that situations made "very l i t t l e sense." Testing was conducted in one session of approximately 30 minutes duration, prefaced by a training period. The subject made four ratings during the session, two for pleasantness (test and retest) and two for consistency. A five minute rest was given between the pleasantness and consistency rating tasks. In the second pilot study, an attempt was made to establish unit relations between the two persons in each P-O-X situation. This was accomplished by presenting the various structures in the following contexts: Format 1: The stories that I'm going to show you are about two school children. At the beginning of the year, the teacher told these two children that they were to be partners. They were to help each other with their school work, go to swimming together, work together on special projects, etc. You are one of these children—the one with the black c i r c l e around him (her). Format 2: This time I want you to pretend that you and the other child are in Israel. You are staying at a Kibbutz for the summer. You and the other child are the only Canadians on the Kibbutz. You and the other child sleep in the same room. You are the child with the black c i r c l e around him (her); the other child is your roottcjate. Format 1 was used to present structures in the test series. Format 2 was used to present the retest series. The subject was instructed to view himself as a participant in a l l situations. The results of the recall session in Pilot Study I indicated that 101 subjects understood the basic elements of the situations and the procedure involved in making ratings. Observation of the subjects' behaviour in both sessions, however, suggested certain changes in the training procedure and in the wording of instructions. For example, at one point in the training procedure the child was required to seriate 6-size graded blocks on a 6-point scale, V% feet in length. A small square was drawn below scale position 1 and a large one was drawn below scale position 6. These squares proved confusing to the children. They placed the smallest block on the small square and the largest block on the large square but were not sure how to proceed thereafter. A scale containing 6 squares of increasing size, placed above the scale numbers, was substituted. It proved more successful. Squares of increasing size were also added to the pleasantness and consistency rating scales in the principal^study, in order to f a c i l i t a t e scoring. In both pil o t studies subjects sometimes placed structure cards half-way between two scale points. IS had to question the subject in order to determine which scale point was intended. Further procedural changes derive specifically from Pilot Study II. Verbal comments by subjects in Pilot Study II indicated, for example, that retest ratings may have been influenced to some extent by memory of ratings made in the test series. Some form of activity should have been inter-polated between the test and retest series. Observation-of subjects' behaviour during Pilot Study II indicated that motivation tended to wane toward the end of the session. This was partially alleviated by shorten-ing the training procedure and by changing the wording of instructions. Discussion with subjects at the conclusion of the study suggested, however, that a further reduction in the total duration of the training and testing 102 session was necessary. It also became apparent during the course of Pilot Study II that the order in which formats were presented should have been varied, as should the order i n which subjects rated for pleasantness and consistency. Ratings obtained in Pilot Study I were analyzed according to Zajonc's (1968) method. Index values for attraction, agreement, and balance were 1.53, .99, and 1.11 respectively, indicating that attraction was the more important determinant of subjects' pleasantness ratings. The same result was obtained in Pilot Study II.. As shown in Table 21, Zajonc indices for Table 21 attraction exceed those for agreement and balance in a l l age groups, when computed from pleasantness ratings. In the case of consistency ratings, the index value for attraction was larger than that for agreement or balance in the 2nd grade group; values for agreement were of greatest magnitude among 4th and 6th grade children. 103 Table 1: Zajonc indices for Pilot Study II Effect Ratio a) Pleasantness Grade ratings p and o agree PL 0 PL 0 p and o disagree PL 0 PL~0 Agreement Attraction Balance 2 12.08 4.34 11.75 3.54 1,0,7 3.03 .97 4 12.88 4.79 10.82 5.43 1.09 2.32 1.17 6 13.36 6.01 11.60 5,91 1.11 2.09 1.09 b) Consistency Grade ratings 2 12.88 5.48 9,22 4.25 1.36 2.27 1.17 4 12.66 7.77 6.86 7.46 1,43 1.28 1.38 6 12.66 9.11 3.54 8.02 1.31 1.24 1.17 Grade 2 (n=16), Age range 91-107 months; X age=95 months (7.9 years) Grade 4 (n=16), Age range 114-122 months; X age=118 months (9.8 years) Grade 6 (n=16), 4ge range 136-146 months; X age=i42 months (11.8 years) APPENDIX 2 DIAGRAMMATIC REPRESENTATION OF THE STRUCTURES INCLUDED IN LEVEL 1 AND LEVEL 2 OF FACTOR 0 Level 1 of Factor O Level 2 of Factor O (8) (7) NOTE: Structures 1,2,8, and 7 are balanced; the remainder are imbalanced. APPENDIX 3 MEAN PLEASANTNESS AND CONSISTENCY RATINGS FOR EACH STRUCTURE (BY AGE) 107 ••'fable 22: Mean pleasantness and consistency ratings for each structure. Structure No. Age Gp. Pleasantness Consistency 1 5-6 14.63 14.20 7-8 15.00 14.45 3-10 14.83 14.93 11-12 14.78 14.98 2 5-6 1C.85 8.00 7-8 9.33 9.15 9-10 12.08 12.03 11-12 10.53 11.23 3 5-6 6.75 7.05 7-8 6.50 6.60 9-10 8.08 8.10 11-12 6.98 5.90 4 5-6 3.00 2.38 7-8 1.53 3.08 9-10 4.25 7.20 11-12 3.20 6.25 5 5-6 11.50 9.10 7-8 10.65 9.60 9-10 11.83 9.88 11-12 9.48 9.28 6 5-6 11.25 9.18 7-8 12.15 9.50 9-10 12.15 11.03 11-12 9.73 8.95 7 5-6 6.15 4.38 7-8 5.00 5.28 9-10 6.85 5.93 11-12 5.10 5.43 8 5-6 5.68 4.25 7-8 5.23 4.63 9-10 5.88 6.15 11-12 5.30 5.53 The values in each c e l l of the above table represent an average of ratings made under formats 1 and 2. A high value indicates that the structure was perceived to be pleasant (or consistent); low values indicate that on the average, Ss perceived the structure to be un-pleasant (or inconsistent). APPENDIX 4 MULTIPLE REGRESSION ANALYSIS OF THE RATINGS 109 (a) The principal study In order to rule out the possibility that the findings relative to hypotheses 2 and 3 were an artifact of the method of data analysis, pleasantness and consistency ratings were also subjected to multiple regression analysis. In this analysis, multiple regression equations in standard score form were computed for each subject. Predictors were balance, agreement, and attraction. The subject's own ratings served as c r i t e r i a . Beta coefficients were then averaged for each age group. These average coefficients are shown in Table 23. Inspection of this table indicates that average beta weights for attraction exceed those for balance Table 23 and agreement in a l l groups, both for pleasantness and for consistency. Average beta weights for balance are greater than those for agreement in the pleasantness equations but in the consistency equations in a l l groups but 7-8, average beta weights for agreement exceed those for balance. Coefficients of concordance were calculated i n order to estimate the amount of agreement among subjects in each age group in their weighting of the three predictors. The procedure, described by Siegel ( 1 9 5 6 ) , involves ranking the beta coefficients according to magnitude. As shown in Table 24, W values are significant (P<.01) at a l l age levels for both pleasantness and consistency. Significant values of W may be interpreted as meaning Table 24 that subjects within a particular age group tend to use the three 110 Table 23: Average regression equations for pleasantness and consistency. Pleasantness Consistancy Age Agreement Attraction Balance Agreement Attraction Balance 5-6 .0051 .6457 .1434 .1048 .5413 .0977 7-8 -.0163 .7116 .1010 .1196 .6297 .1257 9-10 .0122 .8167 .1661 .1829 .6543 .1333 11-12 .1665 .65S5 .1784 .2414 .5953 .1955 Table 24: Kendall Coefficients of Concordance Age Group 5-6 7-8 9-10 11-12 Pleasantness .593 474.00** .693 554.17** .773 618.00** .438 350.00** Consistency W S_ .552 451.50** .483 386.00** .610 488.00** .333 266.00** *p<.05 **p<.01 112 predictors in similar fashion. The frequency of significance of each predictor i s summarized in Table 25. Attraction was a significant component of pleasantness and consistency ratings for the majority of subjects in a l l groups. From 70-100% of subjects in each group assigned positive weighting to attraction. Table 25 Positive weighting means that i f P and 0 like each other, there i s a tendency of varying intensity, but independent of other predictors, to rate the situation highly pleasant (or highly consistent). Negative weighting implies the reverse meaning. A smaller proportion of subjects (5-40%) in each age group also assigned significant weighting to balance and/or agreement. There is some indication that use of agreement and balance as a basis for rating increases with increasing age. Overall, the results of the regression analyses lead to the same conclusions with regard to hypotheses 2 and 3 as the analysis of variance and the Zajonc indices. That i s , the results f a i l to support either hypothesis. (b) The cross-validation study Average regression equations computed from ratings obtained in the cross-validation study are presented in Table 26. The relative weighting Table 26 of predictors in these equations is the same as in the principal study 113 Table 2?8 No. of J3s for whom predictors are significant components of pleasantness and/or consistency ratings. (n=20 in each age group.) Pleasantness Consistency Balance Agreement Attraction Balance Agreement Attraction Age Gp + - + — + — + — + — + — 5-6 5 0 0 0 14 0 2 0 2 0 14 0 7-8 4 0 0 1 20 0 2 0 2 0 17 0 9-10 6 0 2 1 20 0 2 0 6 0 18 0 11-12 3 0 5 0 17 0 5 0 8 0 16 0 114 Cross Validation Study Table Average regression equations for pleasantness and consistency. Pleasantness Consistency Age Agreement Attraction Balance Agreement Attraction Balance 5-6 .0913 .4394 .0604 .1446 .4085 .0897 7-8 .2160 .5447 .1593 .1872 .4064 .0618 9-10 .1741 .7640 .1132 .2088 .5741 .1165 11-12 .3146 .5412 .2194 .3273 .5855 .1421 115 in the case of consistency. In the regression equations for pleasantness, the relative weighting of balance and agreement is the reverse of that in the principal study (i.e., beta weights for agreement exceed those for balance). As mentioned in Chapter VI however, a subject by subject comparison indicated that beta weights for agreement were of greater magnitude than those for balance in only 50% of the subjects. Thus, the results of the cross-validation study also f a i l to support hypotheses 2 and 3. APPENDIX 5 SUMMARY OF ANALYSIS OF VARIANCE OF PLEASANTNESS, CONSISTENCY, AND TENSION RATINGS OBTAINED WITH ADULT SUBJECTS (GUTMAN, 1 9 6 9 ) 117 Table 27a: Summary of analysis of variance of pleasantness ratings. % of varia-b i l i t y due to significant main effects Source of and inter Variation df MS F actions A (Agreement) 1 1629.80 1317.19** 18.41 B (Attraction) 1 2946.50 2381.39** 33.28 AB (Balance) 1 98.58 79.68** 1.11 C (Format) 1 .50 .41 AC 1 .11 .09 EC 1 .36 .29 ABC 1 3.05 2.46 0 (Structures) 1 97.50 78.80** 1.10 AO 1 56.68 45.81** .64 BO 1 13.36 10.80** ABO 1 6.30 5.09* .07 CO 1 1.86 1.50 ACO 1 .76 .62 BCO 1 9.00 7.28** .10 ABCO 1 .43 .35 S 83 6.60 5.33** 6 C AS 83 8.46 6.84** 7.93 BS 83 11.07 8.94** 10.37 ABS S3 2.72 2.20** 2.55 CS 83 1.28 1.04 ACS 83 1.32 1.07 BCS 83 1.31 1.08 ABCS 83 1.52 1.23 OS 83 1.59 1.29 AOS 83 4.92 3.98** 4.61 BCS 83 1.53 1.24 ABOS 83 1.40 1.13 COS 83 .92 .75 ACOS 83 .97 .79 BCOS 83 1.19 .97 Error 83 1.24 Total 1343 *p<.05 **p<.01 118 Table 27b; Summary of analysis of variance of consistency ratings. Source of Variation A (Agreement) B (Attraction) AB (Balance) C (Format) AC BC ABC 0 (Structures) AO BO ABO CO ACO BCO ABCO S AS BS ABS CS ACS BCS ABCS OS AOS BOS ABOS COS ACOS BCOS Error df 83 83 83 83 83 83 83 83 83 83 83 83 83 83 83 83 MS F 240.89 91.03** 718.97 271.68** 1041.30 393.47** .81 .31 .63 .24 .17 .06 1.94 .73 32.50 12.28** 28.88 10.91** 57.92 21.89** 56.27 21.26** .09 .03 11.63 4.39* 2.77 1.05 4.41 1.67 10.09 3.81** 7.09 2.68** 8.82 3.33** 11.61 4.39** 2.76 1.04 1.77 .67 1.64 .62 1.92 .73 2.29 .87 3.23 1.22 3.32 1.25 2.56 .97 2.37 .90 2.83 1.07 1.90 .72 2.65 % of varia-b i l i t y due to significant main effects and inter-actions 3.11 9.28 13.44 .42 .37 .75 .73 .15 10.81 7.59 9.45 12.43 Total 1343 *p<.05 **p<.01 119 Table 27c: Summary of -.ruilysi3 of variance of tension ratings. > % of varia-b i l i t y due to significant main effects Source of and inter Variation df MS F actions A 1 2160.40 1638.56** 23.67 B 1 2731.40 2071,65** 29.93 AB 1 46.50 35.27** .51 C 1 .19 .14 AC 1 2.50 1.90 BC 1 .36 .27 ABC 1 3.86 2.93 0 1 35.36 26.32** .39 AO 1 11.44 8.68** .13 BO 1 15.43 11.70** .17 ABO 1 4.53 3.43 CO 1 .96 .73 ACO 1 .00 .00 BCO 1 .03 .02 ABCO 1 7.44 5.64* .08 S 83 9.93 7.53** 9.03 AS 83 8.97 6.80** 8.15 BS 83 8.62 . 6.54** 7.83 ABS 83 2.18 1.65* 1.98 CS 83 1.83 1.39* ACS 83 1.12 .90 BCS 83 1.96 1.49* 1.78 ABCS 83 .83 .63 OS 83 2.30 1.74** 2.09 AOS 83 3.83 2.91** 3.48 BOS 83 1.36 1.03 ABOS 83 1.74 1.32 COS 83 1.41 1.07 ACOS 83 1.22 .93 BCOS 83 .81 .61 Error 83 1.32 Total 1343