o‘l’u‘. —— , . . Q ! ’ ’. OIO-OVUOOO -.o "..""<‘I> V - -n.I. 0' - n‘. -.-.~- - . --- ‘- ~ ,- O'--. - Q . .-‘ o.‘». -0..-q.0v~ -- 0-4-0-‘. q"~[“-."..'..3...' 0V...OQ—‘~.-u..~..."f.'O'-'ONOO-.O.Qgh......"”-‘ 9 THE UMITS 0F CONFGRMHY . Thesis for the Degree of «M..A.. ,MECH‘IGAN STATE UNIVERSlTY HERBERT! L. TYSON, JR. ' 1975. ........ ...... '''''' a ooooo .................... I A“ 1 19 A 154.1 ""1 m ABSTRACT THE LIMITS OF CONFORMITY By Herbert L. Tyson, Jr. A great deal of research on conformity has indicated that many individuals are more than willing to adopt the judgments or opinions of a majority group, even when.the majority is wrong. All of the previous research, however, has failed to demonstrate whether the source of the con- formity is external group pressure or some internalized value for conformity. This thesis sought to discover the source of conformity by eliminating the possibility of external pressure of any kind. Using sixty Michigan State University undergraduates as subjects, the researcher posed as a pollster asking for responses to six statements on public issues. The falsi- fied results of an earlier "national college sample" were ‘used as the majority consensus stimulus. A written ques- tionnaire was administered in secret, and then placed in a locked ballot box. This study found that subjects do not tend to conform in any significant way when responding anonymously. This finding supports the idea that conformity is externally en- forced, rather than internalized. Introducing controls for sex, religion, and identification with the false sample frame failed to yield any significant conformity, further supporting the external pressure hypothesis. THE LIMITS OF CONFORMITY By Herbert L. Tyson, Jr. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Sociology 1975 TABLE OF List of Tables . . . . . . . List of Figures . . . . . . Introduction . . . . . . . . The Survey . . . . . . . . . Design . . . . . . . . The Subjects . . . . . Results . . . . . . . . . . manipulations . . . . . The Hypothesis . . . . Testing the Hypothesis Discussion . . . . . . . . . Main Hypothesis . . . . Other Factors . . . . . Summary and Conclusion . . . Recommendations . . . . . . Footnotes . . . . . . . . . Appendices Appendix 1. Additional Appendix 2. Combined T Appendix 3. F Tests of CONTENTS T Tests Tests . variance Appendix 4. The Questionnaire Blbliogmphy............. ii page page page page page page page page page page page page page page page page page page page page page iii iv 14 14 1a 21 21 21 23 34 34 35 38 4o 42 43 44 45 46 48 Table Table Table Table Table Table 1. 2. 3. 4. 5. LIST OF TABLES Analysis of variance . Means and variances . Individual T Tests . . Combined T Test . . . T Tests for Controls . Distribution of Subjects 111 P389 page page page page page 24 27 28 28 31 33 LIST OF FIGURES Figure 1. Latin Squares Design . . . . . .page 23 iv INTRODUCTION Considerable research has been done on the issue of conformity, much of which would indicate that many of us are more than willing to sacrifice our own judgments or opinions to a majority consensus, even when under considerable pres- sure to do otherwise. One of the pioneers of such study was Asch. In his experiments, Asch exposed individuals to lines of varying lengths, asking the subjects to match a test line with a line of equal length (1951). He found that when confederates give incorrect judgments about the length of lines, subjects respond by giving incorrect responses themselves. Depending on the amount of unanimity and the degree of support on the subject's side, a remarkable num- ber of individuals gave in to the group on a large number of judgments. Specifically, this demonstrated that some individuals who are exposed to the knowledge that others overwhelmingly disagree with him/her, abandon their own conceptions about a particular phenomenon, and agree with the group. The ex- planation which was posited for this behavior was that individuals have such a fear of appearing wrong or foolish, that they find it safer to be wrong when a group is also 1 wrong. It also was taken to indicate that, for many indivi- duals, agreement with a group is more important than being singularly right, especially in terms of social risk. There- fore, individuals succumb to social pressure. Asch's idea was.adapted by Crutchfield in a manner which did not require the use of confederates (1954). With each subject seated in a separate booth, the experimenters controlled a panel of lights in each booth visible to each subject. Each subject was told that his panel of lights would indicate the responses of the other subjects to the question being asked, and that his own responses were also visible to the others. The researchers then controlled the lights in such a way that each individual was made to believe that the others were in solid agreement, but had reached a different decision than the subject. under the illusion that the lights indicated the responses of the other four subjects, each of the subjects was asked to respond. Confirming the results of the Asch study, a number of subjects agreed with the false consensus. Moreover, by using an "artificial" group agreement, Crutchfield was able to greatly multiply the number of usable observations, since, unlike Asch's study, all people present were genuine subjects. Even with this change, Crutchfield's study resulted in the same conclusions as Asch's eXperiments, with respect to group pressure and conformity. Following Crutchfield, Tuddenham (1961) attempted to show the effects of the presence of a known group norm with less bizarre distortions in judgment than those tested by Crutchfield and Asch. Subjects were tested to find their "acceptable range of variation" on particular visual, gene- ral information, and opinion stimuli, by observing a number of responses given by subjects with no peer pressure present. That is, subjects were initially tested to see how far in either direction they would vary their judgments, when al- lowed to respond under no social pressure. A distorted norm, i.e. the falsified response of others on the same issue, was then introduced. Where the distortions were within the ”acceptable range," conformity was induced. When.distortions were of greater magnitude, the subjects conformed as expected, but usually to a lesser degree and still within their own range potenp tials. A significant portion, however, did shift beyond their range potentials, and sometimes to "grotesque limits." Luchins and Luchins were interested in the possible implications of Asch's and others' findings on test results (1968). They attempted to find the circumstances or ex- perimental conditions under which subjects would accurately report perceptions. They assumed that the presence of peer pressure leads to distortions and misreporting on the part of subjects. In one experiment they found that a motivation to tell the truth (three lectures stressing the importance and neces- sity of accurate reporting of perceptions, plus a grade increment reward for reporting accurately) was ignored in a significant number of situations, when subjects were confronted with the disagreement of both peers and the experimenters. They next allowed some peers to agree with the subjects, while keeping the experimenter in disagreement, and found that the experimental motivation did appear to have a sig- nificant effect, when tested against an unmotivated control group. Conformity was still present in this set of line experiments, even though the addition of some peer agree- ment did reduce the subject's conformity with the group. In another conformity experiment (1954) Goldbeng tested the effects of three situational variables-group size, frequency of exposure to the stimulus, and the extent to which the individual sees himself/herself as different from a specified normp-on.the conformity behavior of voluns teer subjects. Goldberg asked subjects to repeatedly judge the intelligence of nine individuals from their photographs, each time giving them.a false report about the group's judgment. He found that the greater the distance from the norm (group mean guess minus the subject's guess), the more the individual conformed. Group size and frequency of exposure had no noticeable effect. Additionally, by some reasoning which was not clearly explained, Goldberg concluded that a tendency for individuals to conform may not be generalized. That is, conformity was seen as a function of specific situational variables, not as a function of the individual. He asserted this result as evidence against any general personality characteristics of suggestibility or conformity. Walker and Heyns (1967) have done some more recent studies which lead them to an entirely different conclusion. In one set of experiments, two groups of subjects were placed in situations where conformity and nonpconformity, respec- tively, were rewarded. One group was consistently rewarded for conformity while the other was rewarded for nonpconformity. Each was then tested to determine whether or not they cone tinued to confomm or nonpconform, accordingly, in a nonpreward situation. Their manipulations were successful in producing the appropriate conformity and nonpconformity. Those who were rewarded for conformity, as well as those who were rewarded for nonpconformity, continued to exhibit the same behavior, respectively, even when the rewards were absent. This finding was interpreted as meaning that conformity‘ig generalizable. Specifically, Walker and Heyns are saying that whether or not an individual conforms will be determined by the pattern of rewards he/she has received in the past. If the individual has been rewarded most frequently for conformity, then that person will tend to generalize conformity to non- reward situations. Conversely, if a person has been re- warded most frequently for nonaconformity, then he/she will tend to non-conform in non—reward situations. In contrast, Goldberg claimed that such individual differences are not a factor. His results reportedly show that situational differences will account for conformity and that individual differences are not significant deter- minants. Actually, neither Goldberg nor Walker and Heyns make a very convincing case for their arguments. Goldberg, by using a conclusion, the foundations of which are not ade- quately spelled out, does not appear to test the question of conformity in any general sense. Rather, the influence of a few situational variables is apparently taken to be evidence that conformity is not a function of individual characteristics. In this case, contrary to Goldberg's con- clusion, the presence of one type of influencing factor certainly does not rule out the other. Walker and Heyns claim, on the other hand, that they found that individuals gg.generalize not only conformity, but nonpconformity as well. While their results, i.e. that subjects continued to conform and nonpconform.in a nonpreward situation, may not be disputed, whether or not the situation is sufficiently "general" to warrant their conclusion is highly questionable. Certainly, the previous behavior was extended to a nonpreward situation. However, there was a sufficient number of similarities between.the situations to question the use of the term "generalization." Both were conducted in.a laboratory and in an experimental setting with observation by the same experimenter. Both situations gave the subjects similar stimuli and information. The only difference was the reward. 7 In fact, Walker and Heyns have shown only that the be- havior which has been rewarded in.a specific situation.will be extended to other similar situations, not that it will be extended (or generalized) to non-similar or general situa- tions. Thus, the question remains unanswered. By combining both arguments, i.e. the situational and the individual (generalized), a reasonable synthesis can be presented. That is, Walker and Heyns have demonstrated that behavior will be extended, but not necessarily generalized. In other words, for situations in which a particular distri- bution of rewards has already been established, one would expect that the individuals will exhibit the behavior appro- priate in those specific situations. This is somewhat compatible with Goldberg's findings. Thus, in experiments of this nature, the individuals will continue to behave as their rewards dictate. In different situations, however, individuals may revert to whatever behavior has been pre- viously rewarded in such contacts. The generalization question is most clearly an issue when the person has neither been rewarded nor told to expect a reward. Otherwise, the individual is conforming (or none conforming) with respect to a specific set of expectations. It is in the realm of new, unfamiliar, or unrewarded situa- tions that the individual might generalize past conformity or nonpconformity. Having no tailored set of expectations as to which behavior will bring a reward, an individual in a new situation may tend to conform (or nonaconform) simply because that behavior has yielded consistently better results in the past. Now, one might easily infer from a casual examination of the society which surrounds us, that conformity is more frequently rewarded than nonpconformity. If this is true, then.it is not entirely unreasonable to propose that in new or unrewarded situations, in which the individual is not aware of the basis upon which rewards or penalties, if any will be assessed, individuals will generally tend to conform to the behavior of others, provided that the behavior of others is known.to the individual. using generalization as a basis, the presence of rewards and penalties in such situations need not be a factor in the conformity. Here, it is the generalization which produces the conformity, since the individual does not specifically know what kind of be- havior will be rewarded. It should be emphasized that the foregoing argument would additionally apply in the same manner to a person who has been consistently rewarded for nonpconformity in most situations. However, such cases would be quite unusual, and consequently, we expect that conformity will be the norm. This expectation of conformity as a norm depends upon several factors. The first is that individuals are usually rewarded for conformity rather than nonpconformity. As suggested above, this does not seem to be an unreasonable assumption. A second assumption is that individuals inter- nalize the desire for conformity in such a way that, even when.external rewards and constraints are not present, they desire to conform.with "most other people.” Through such a process, individuals associate conformity with rewards, creating internal rewards which effect conformity. Walker and Heyns, it seems, were trying to demonstrate this phenomenon. That is, to say that conformity (or none conformity) will be generalized to a situation wherein re- wards are not present, is to say that a value for such cone formity has been internalized. However, walker and Heyns erred in one specific and crucial respect which is cause for considerable misgivings about their experiment. They failed to completely isolate the subjects from observation. Such observation, I believe, constitutes a situation in which the subject may still be behaving so as to please the experimenter. Furthermore, it appears that Goldberg makes the same mistake. In order to demonstrate the presence of such "in» ternal conformity," anonymity of the subjects must be absolute- 1y established. Therefore, the only general conclusion.which can be surmised from the studies which have been discussed is the following: Greater unanimity of opposing consensus increases the probability that individuals will conform to some group norm with which they' would ordinarily disagree. This was established very early in the literature, and is supported throughout. In conflict here is the conclusion of Goldberg with that of Walker and Heyns. The logical synthesis of their arguments is one possible basis for the internal conformity 1O theory presented below. In this view, conformity has been established as an internalized norm in most individuals, possibly by the above process, and need not be enforced by group pressure or other rewards which are external to the individual. Thus, we have two ways of looking at conformity which are relevant to the studies discussed here. The one which has been focused upon primarily in the literature is the € group pressure hypothesis. The individual is considered to be concerned with his/her standing in the group, and views ' a that standing as partially dependent upon his/her reaction to various objective and subjective issues. This concern is seen to exist even in groups which have no history, i.e. which have been created for the sole purpose of the experiment. When asked to respond to various stimuli, the subject has several priorities. If he/she alone is responding, the first priority is always to be right. If, however, others are responding, his/her perception of correctness may become distorted in order to allow agreement with the group. The individuars concern about group standing often overrides the concern for correctness. Consequently, when an individual reaches a decision which is different from the group's decision, he/she is less likely to maintain that he/she is correct. Rather than face possible ridicule and loss of standing, the individual abandons a correct decision and a- gress with the group's wrong decision. As such, the reason for conforming is external to the subject. 11 The other explanation is one which the experiments have not really considered. This view posits the existence of a second level of conformity. The first is external, and cor- responds to the above explanation, i.e. conformity for the sake of appearance to others. The second is internal-conp formity for the sake of personal satisfaction. The latter type requires that the individual have an internalized _ _ A_. re: ~......9 _.. “A“..-‘fl r _1 value for conformity, i.e. that disagreement with others is objectionable or uncomfortable. Therefore, the individual seeks to conform, not for appearance's sake, but for inter- nal satisfaction. Clearly, the internal level requires many t that the individual be informed of the group's decision or the norm, and not that he/she be exposed to the possibility of ridicule. We would not, however, expect a person to internally cone form with a group with which he/she does not identify. That is, in order for a person to change an Opinion or judgment, he/she must be motivated by the knowledge that his/her judg- ment does not concur with that of a reference group. There- fore, a necessary pre-condition for the internal conformity theory is identification with the particular reference group. Previous studies of conformiy have the following in com- mon: When tested, the critical subject has always been in a position of being observed, either by the experimenter or by the other subjects, or both. This observation effect may tend to force the respondent into a position of agreeing in order to appear to be in agreement with the group, rather 12 than for conformity's sake alone. The studies have not attempted to discover to what extent conformity is sought for its own sake. Of interest here is this failure to consider the inter- nal conformity issue. Having the subject under observation of any kind creates a situation in which the subject may still feel compelled to agree with the consensus, above and beyond any internal desire for conformity, in order to be viewed as "normal" by the experimenter or the group. Therefore, the study being done here will eliminate both.the group observation.and the experimenter observation in order to prevent the possibility that the subject's re- sponse is a result of anything other than internal conformity. The subject will respond in.sueh a way that he/she is cone vinced that the experimenter has no way of knowing the sub— ject's response, i.e. in total secrecy. If the results of this study show that conformity is not produced, then there is strong reason to believe that the social pressure hypothesis is the crucial factor in explaining conformity. This would show that individuals have to be subjected to scrutiny before they retreat from their own judgments. If, however, these results show that conformity is still produced, even under conditions of secrecy, then some substantial change in the focus of the conformity question comes about; that is, the "social pressure” which accounts for the conformity does not require the presence of the group 13 whose pressure is being used. Furthermore, it is not even necessary that the subject be observed, i.e. that he/she be concerned with how he/she appears to others. If this is the case, then one of the mechanisms which produces conformity is internal. This, of course, does not rule out the possi- bility that external pressure is also a cause of conformity. The theoretical questions being considered in this study also include the possibility that other factors relate to conformity. One such factor is the importance of the issue to the individual. It might be expected that issues which individuals consider flmportant will be less subject to social pressure than unimportant ones. Flexibility on an issue may be a function of its importance, but it may also be a function of expertise, which, on an issue like the economy, may affect one's ability to make a firm decision, while still considering the issue important. While this variable is not a central issue in this thesis, it is one which has not been given much attention in the previous studies and which the following design will attempt to in- corporate. In sum, then, this thesis asks if an artificial consen- sus will have the effect of producing conformity, even when the subject's response is totally secret. It assumes that conformity may be internalized in part, rather than wholly externally enforced. In essence, it tests the accepted idea that fear of being ostracized from the group is the only mechanism which produces conformity. .- fl] ‘7 :TA‘Tl—“gi', _ THE SURVEY Design The thesis was tested by administering four different questionnaire forms to groups of 15 students, each of whom was told that the questionnaire was part of a national opin~ ion survey. Three of the forms gave an ostensible distribu- tion of responses from a recent national sample of college students. The fourth was a control group and gave no such information. The completed form was placed in a locked ballot box. Each questionnaire consisted of four parts. On the front there was a letter of introduction.and an ”informational" questionnaire. The letter contained an introduction to the surveyor and it explained the survey and the method being used. The purpose of the letter was threefold. First, it served to help build credibility into the questionnaire by establishing, in writing, a name and address for the polling organization. Third, the explanatory nature of the letter reinforced and justified the secret nature of the poll, i.e. that the subject would respond on a secret ballot. The other half of the front page consisted of an infor- mational series of questions, in which respondents were asked to give their sex, race, and religion. In addition, the subject was asked to evaluate eight U.S. sub—populations, only one of which, college students, was of interest. 14 15 This particular question was asked in order to obtain a measure of identification with college students. As dis- cussed earlier, it is not reasonable to expect internal con- formity with a group with which the subject does not identify. Here, we attempt to verify the presence of identification with college students, as a pre-condition. The reverse side of the questionnaire consisted of the experimental treatments, and a questionnaire designed to detect suspicion about the treatments. The treatments were different on each of the four groups. This part con- sisted of six statements about public issues. Respondents were asked to respond on a scale of 1 (strongly agree) to 5 (strongly disagree), depending on how they felt about each of the six statements. They were also asked to indicate, on a scale of 1 (very important), 2 (mildly important), and 3 (not very important), how important the particular issues were to them. This was done in order to test a possible relationship between conformity and importance. As suggested earlier, it is possible that subjects who view the issues as impormnt may be less likely to conform to a majority view. On three of the different forms, the "results" of a recent poll of college students were presented. On each form, results were presented in three ranges: low agreement (15-25%) , medium agreement (40-50%) , and high agreement (75--85%).1 Each of the six statements on these three forms was represented with each level of "artificial agreement” (i.e. the recent poll results were fake, and represent the -.r arr—ma“ (.4 16 treatments) an equal number of times. Hence, statement 1 was represented as having high agreement in the "recent poll” in one-third of the treatment forms, low agreement in one-third, and medium agreement in one-third; and so on for statements 2 through 6. Therefore, each person in the three treatment conditions received a questionnaire which contained all six statements: two with low artificial agreement, two with medium artifi- cial agreement, and two with high artificial agreement. As stated earlier, there was a control group which received no information. The arrangements of treatments were as follows: mam; High Artificial Agreement Statements 5 and 6 Medium Artificial Agreement Statements 1 and 4 Low Artificial Agreement Statements 2 and 3 Group 2 High Artificial Agreement Statements 1 and 2 Medium Artificial Agreement Statements 3 and 6 Low Artificial Agreement Statements 4 and 5 Group 3 High Artificial Agreement Statements 3 and 4 Medium Artificial Agreement Statements 2 and 5 Low Artificial Agreement Statements 1 and 6 1? Grou Control Group: No Information Given Thus, each critical subject received.all three treatments. It was decided that a single treatment per form would not be acceptable for two reasons. First, that practice would pro- vide only one-third as many observations per condition. In order to get as many observations as were obtained in my design, it would be necessary to triple the number of sub- jects. Second, having the same level of agreement for each statement might produce suspicion, especially with varied issues. It was decided that such false information would stand a much better chance of being believed if it were varied, that is, having some statements receiving strong agreement and others not. Of the parts of the questionnaire mentioned so far, the “informational” questionnaire was Part I, the treatment questionnaire was Part II, and the final part was Part III. Part III was designed to detect any suspicion. The subject was asked if the interviewer seemed trustworthy, or if the interviewer influenced any answer in any way, and to state the purpose of the survey. In addition, there were two filler items. To sum up, each questionnaire consisted of the same introduction letter, the same Parts I and III, and one of four different Part II's. 18 The Subjects Each questionnaire form was administered to fifteen subjects. Subjects were randomly selected from underu graduates who reside on the Michigan State University cam- pus. Dormitories, rooms, and floors were randomly selected. All of the names of dormitories from M. S. U. were put into a hat, and ten were drawn at random. Six [ rooms per dormitory were selected in the following manner. £ :9. Six.numbers were randomly selected (with replacement) for each i "L dormitory, representing floor numbers: and sixty numbers were then drawn representing the rooms which would be selected on each of the floors. When the chosen floor in each par- ticular dormitory was reached, I then counted from the first door on the right, beginning on the south or east hallway, depending on how the building was situated. If more than one person happened to be home when they were called upon, then the final selection was made by reference to a pre- determined list of random 1's and 2's (determined by a coin toss), wherein.a 1 would indicate the person whose name came first in the alphabet, or a 2 for the name which came last. It might be objected that this procedure results in a nonprandom selection of subjects, in that it was selective of only those who happened to be home at a specific time of day. However, return visits were made in cases in which no one happened to be home. Additionally, subjects were 19 interviewed at three distinct times of day-dmorning, after- noon, and eveningb-consequently sampling a wide variety of people, with respect to time. Forms were arranged in a pre-determined random order so that I was not aware of which form each individual was given. This should have substantially eliminated the pos- sibility of any effect from experimenter expectancy. The subjects were informed that they had been randomly selected to participate in a national poll of college stup dents, and were then.asked if they would do so. If they said yes, I went inside and gave a rehearsed spiel about the poll.2 They were told that the usual method of polling involves the pollster orally administering the questionnaire, but that many peOple object to being put on the spot by an unfamiliar person. It was then explained that "we" had developed a different and less objectionable polling method- the ballot box. They were then given the questionnaire which they were told to fill out on their own and then to insert it into the ballot box. One potential problem was that the respondents might not notice the percentages on the treatment forms. If no questions were asked within a few minutes after the subject began to fill out the reverse side, then the subject's attention was called to the percentages. The explanation used for all respondents was that a 1974 Polling Disclosures Act requires that participants in opinion polls be given full information regarding the purpose of the particular poll and the results 20 of the most recent version of the same poll being taken. Virtually all of the subjects asked about the percentages and how they were supposed to respond to the questions, thus inviting the "explanation." It was only necessary to spontaneously offer the "explanation" to two subjects, since the rest asked about the items. When subsequently questioned, none of the respondents thought the results which were presented were false. However, one subject did think that the percentages were inserted in order to influence her responses. That person's question- naire was subsequently removed from the sample. Furthermore, the subjects were inconspicuously observed in their’motions in answering the questionnaires, and none of the sixty subjects violated the intended answering sequence (i.e. 523$ letter, answer Part I, 55533; Part II, answer Part III). Thus, it was ascertained that no subject based his or her response to the rating of college students on the artificial poll results.3 RESULTS ~lanipulations As mentioned above, in this survey there was no evidence in any of the cases which are used in the final data analysis that the manipulations were not believed or not noticed by the subjects. In every case, the subjects were aware that the poll results were present. Additionally, three separate pre-tests were administered in order to ascertain that such manipulations of poll re- sults would not foster suspicion. In the first and second pre-tests, questions regarding smoking were used, and the stated source of the poll results was the Gallup Poll. In the third pro-test, the form used for this survey was em- ployed. None of 33 students, a mixture of graduates and undergraduates, found the poll results to be questionable, although some extreme results for each question or statement were presented (90 to 95% agreeing and disagreeing). Finally, the identification pre-condition was met in 39 out of 45 cases. The othesis The major hypothesis can be looked at in two ways. The first separates it into two hypotheses: (1) a positive consensus (high artificial agreement) will raise the 21 22 percentage agreeing to some point above the percentage that would be obtained from a control group, and (2) a negative consensus (low artificial agreement) will decrease the percentage agreeing to some point below the percentage that would be obtained from a control group. That is, (1) Ah>A1, and (2) Ac>Al where Ah, Ac, and A1 are the amounts of agreement where artificial agreement is high, not given, and low, respec- tively.4 Another way of looking at the hypothesis combines the two aspects of the first by saying that if (1) and (2) are true, then: (3) Ah>Ac>A1, and therefore (4) Ah>Al. Therefore, a finding that Ah is less than or equal to A1 automatically disproves (3), and consequently renders (1) and (2) superfluous. Hence, the null hypothesis is: Mean response when national college response is in high agreement (positive con- sensus) a Mean response when national college response is in low agreement (negative consensus). 23 Testing the Hypothesis The experiment was set up in such a way that it may be treated as two Latin Squares. The entries in Figure 1 are L, M, and H, which represent low, medium, and high artificial agreement, consequently giving each treatment in each level of the design exactly once. Thus, the two Latin Squares separately treat Statements 1, 3, and 5 and Statements 2, 4, and 6. Form Form 1 2 3 1 2 3 ‘5 *5 o1 M H L o2 L H M 5 5 i’3 L M H “4 M L H 3 3 r”'5 H L M m6 H M L FIGURE1 Latin Squares Design The resultant analysis of variance tables are shown in Table 1. We note that in each ANOVA there is a significant item effect. This is not surprising. Some items would be expected to have a higher mean agreement than others, as in most public opinion polls. With respect to forms, it is reasonable to expect that the form which was given.would not significantly affect the results, since all questions were asked on each form with treatments equally distributed. As expected, there were no main effects associated with forms, in either ANOVA. fl .mg 4-1 ‘EJKEIT:W 24 TABLE 1 Analysis of Variance Statements 1, 3, and 5 Source of variation df SS MS F Question (Statement) 2 19.20 9.60 5.39” Form (Group) 2 .59 .30 .17 Treatment 2 3.39 1.70 .96 Residual 2 25.07 12.54 7.05” Within Cells 134 238.27 1.78 - Statements 2, 4, and 6 Source of variation df SS MS F Question (Statement) 2 47.57 23.78 17.40” Form (Group) 2 2.79 1.40 1.01 Treatment 2 7.26 3.68 2.65 Residual 2 50.72 25.36 18.40” Within Cells 134 188.47 1.38 - 1” F2’134( e05) = 3e07 25 The thesis being put forward in this paper calls for a significant treatment effect. According to Table 1, however, the treatment main effects could easily be attributed to chance alone. Therefore, the null hypothesis of no treatment effects can not be rejected. In both.ANOVA's, there are significant between cell (re- siduals) variations, indicating the presence of some inter- action effects. Such.interaction effects can increase the apparent main effects, thus making it seem as if they exist where they might not. In this case, however, any possible contribution from residuals has still not produced an apparent treatment effect. Hence, we can be reasonably confident that none exists, and need not be concerned with significant residuals. The interaction effect does make it difficult to inter- pret the question effect, since the two effects are statisti- cally confused. The fact that the question effect is sig- nificant brings the residuals problem into the picture. However, since the issue of whether or not questions had dif- fering amounts of agreement is neither important nor relevant to this study, the technical implications of such a problem need not be discussed here.5 In addition to the ANOVA, a set of t tests were done to further test the null hypothesis that there were no differences between the high and low support treatments (positive and negative consensus). This was done in two ways. First, the individual items were tested to determine if support for the null hypothesis would be found on separate items. 26 The mean response on each item under high treatment is compared with that under low teeatment (positive versus negative consensus). Since a high mean (X) is disagreement (58strongly disagree), a confirmation of the research hypothesis occurs when x5ausMos so some mousse» .AebaaeMos has> I .Hssases In .eewuueoa IN .sewoweoa aao> I.V anon o>«»«eoa son oueoausa oasoso essences use» «ecu so» hueeuaeues so .veusounnd eases use so «choose canon» usasoauou on» no some ovum .e Anuaooanv servo .e osos .o season .o oaaogeeo .a asuooooeua .u uoaeanuu .n oneseu .m ones .s wen .u III. .5833. .u 385 .. esssuse sens-an .e eauej\sse«uess ohujxxoeup .A ssdqsu ssouuese .c ssaesesso\ev«ns .e seal .— .as«»seo«nsoo e« sesame-a use» case seasons aeeu use .eesusoos use masses on ensoaa .vee: amass ea macho sous-one saga he she“ «hose one .eoseusessoo Lao» mom .eesoaeeu goes saws wean“ on on neaueuss sssowuuosp was useoaussuousu sus» ea manna-ow ea eaasssouusesc ea» we when sash H audfi «users .533 .u 8:3. Mfimahzafin .seaueueaooo use» sou so» sauna .aeuusl one sans used his es none on euaasso«eeosc ease no use: Hanan one :4 sconce-snouno enu eueowuna oesoaa menu .e>«m so» houses one use .ouuus 133.33 .5 5 sexes 33 .3333 on» so guess osoo 2:3 eooo uncommon can .uo>oson .MH .onsoaeoa use» menocosausa no access cocoa on» new: acoauneso Has assess o» sunfish“ soon as: aosoa> Issac“ era .uu panacea on canvases «on on .uo>oson .aao: use: no» oasosm .uesewbhousa on» menu ado: esosaas aw eveflanoo on sans en canons so» use hoovessaaeruHoe ea snow shanssouaeoso era .eaasse esuueusenosaek s sesame es «one usaanmsa uo e>ng as has haso on» on same .eusoosoaooa on» huausosa an: on ea or ocean .oo«e» sedans- on no: Haas so» «use on herew>hoosu use» shows“ canons .ohos son an .vbm— no Hana one a“ souooaoo once so» nonu.oapsooaa «o: «no .oanueooa as ea .coe: osvnssoou ”sadness scones one we ewssoon .vpm. mo nopsouaom an canon gowns hoses s uo sowusssausoo s nu sunk .osov mswoo ow song: has». causes omoaaoo unsound: s go when aw oneness on sexes mason one so» scans ma harass era .sooea use soausewnsmuo nonsense genus“. on «on» usage use hauasausoowusoo no cosmos s new eeo«>oaa so». ease «as» o>eanoo om .oowuuo .o.a covenannsz use cu cuss e ca echoes hausouosa on news: has one news ousoao on mass sec souhhso uosoa>housa one» soars won «oansn ens .oesoaeeu . use» no nuunewvsoouusoo on» oases“ o» osoo meson ea uses“: one vacancy on ooxos mason one no» nouns cu ho>use one an .heoahdoo one unannousd no seamen was s so» 9 «>oua on concedes haasueuso no son e: . so A voucher ea nosououousu such .so o» aesewpueusw use» eunuchunn ov ea heaven sash nus-uneasem ween new. .a access» 8»? 18... e. Tsunamis... at... s nueeueonne chanson sensuous. so . The above was side 1 of the questionnaire. (See page 14.) 46 The Questionnaire (Continued) “I! 1! low strongly do you scree or disagree with the telloeilc stetoseuts? Drsft seeders and deserters should be given couplets and unconditional unesty. The 0.8. should have senator: secs end price controls. lohsooo smoking should be prohibited in all public places. a. 11.3. should hsve psoline rstioniu. Fooeeeoion of smell ssomte of ssrijuens should be illogsl. It is wrong for married people to here oexuel relations. (maze eseem) we; "'— Jeane sun go seem em equesep mom ‘epxoe use no: u: 0‘ annual. m oz Jeeeus no: eoumtm he on u: .xeeowezu: em "a '9 lotionsl College Response (14) “-9 0) i 15‘ «i 84‘ 1“ SE 78$?! 45¢ 11‘ 13$ a 13$ 3! us 17$ 19$” we; (mtdxo more) we “'— -oe m use ‘se; mum-m Im- emu-mu: In no '1 Indicnto your egroosont or dissareomont by eiroling your response. eons m we; we none (as; ‘ssx (Imu- acute) °ox ‘— imemoo newness: on: so; u topmos- E 4 J 4 3 Q 3 4 . 3 ‘ 3 4 3:: E -5 3:; E. E E mop see ens/en use sous oi sees easement on no '1 How important isthisissue tom? 5 a 5 * Q i E g ‘8 n .. E 1 I J 1 a 3 1 2 3 1 l 3 1 2 3 1 2 J 'umwnw 0%! 30 3o tomes on «sunny» nuetdsoo e; eeuodeu not on; useswezu; pus newest“ uses game m 1001 WI ’W n eseeu esonsenb es: s am oz suopuodoe: nous mace no es 'uoze no zenmtmmzsoemmuupuuz “tweumwehnflmedeptmm twepuodeeu Jeec menu I MIA-mu The above was side 2 of the questionnaire. (See page 15.) 47 BIBLIOGRAPHY BIBLIOGRAPHY Asch, in Guetzkow, groups, leadership and Men, "Effects of Group Pressure upon the Modification and Distortion of Judgment," Carnegie Press, 1951. Crutchfield, "A New Technique for Measuring Individual Differences in Conformity to Group Judgment," Proceedings: Invitational Conference On Testing_ Problems, Educational Testing Service, Princeton, N.J., 1954. Goldberg, Solomon, "Three Situational Determinants of Conformity to Social Norms," Journal of Abnormal and Social Ps chalo , V. 49. 325-329. Heyns, R.W. and Walker, E.L., An Anatomi for Conformity Wadsworth Publishing, 1967. Luchins and Luchins, ”Motivation to Tell the Truth Vs. Social Influence," Journal of Social Ps cholo , 1968, 76, 91-95. Tuddenham, "The Influence upon Judgment of the Apparent Discrepancy between Self and Others," Journal of Social Ps cholo , 1961, 53, 69-79. 48