VALUES, ATTETUDES, AND PERCEWED INSTREEMENTAU'EY Thesis [‘00 {The Dogma 05 pk. D. MICHIGAN STATE UNIVERSITY Robert John Homant E970 fHF‘uR‘“ LIBRARY Michigan State University —-— This is to certify that the thesis entitled 9 VHL 1/15.; Pfl (’fL-tcsflts. I (we!) flee/Mela? (Q LSTfZ’L‘flQ/Vfi 7! presented by {Bo/Jeflf jot”; HO'qu—f/ ‘ has been accepted towards fulfillment of the requirements for FA 0‘ fiegreein fits/Vt 40/527; .V W 1 flung/5‘“) 1‘4»an cue 41v Major professor Date QM Q /;7/ 0-7639 ABSTRACT VALUES, ATTITUDES, AND PERCEIVED INSTRUMENTALITY BY Robert John Homant The purpose of this research was to investigate the relationship between attitudes and values. It was hypothesized that attitude change would result from a manipulation of perceived instrumentality, where "perceived instrumentality" is defined as the importance of the value to which an attitude object is linked. Three studies were carried out in order to test this hypothesis. Studies I and II were carried out in a field setting and attempted to change voters' attitudes toward a graduated income tax. Study 111 was a laboratory study which attempted to change college students' attitudes toward sliding scale tuition. In each study a pretest measured subjects' attitudes as well as the relative importance of several values to each subject. The manipulation.which took place some weeks later, consisted of a communi- cation which associated a particular attitude to either a relatively important or a relatively unimportant value. In each study it was hypothesized that more attitude change would occur with subjects who had had an.important (rather than an unimportant) value linked to the attitude. Robert John Homant Study I was conducted at the time of an actual vote on a graduated income tax referendum. The results, based on the returns of 163 registered voters, failed to show any evidence supporting the per— ceived instrumentality hypothesis. use only was there no difference between subjects receiving an important value communication and subjects receiving an unimportant value communication, but also there was no difference between subjects receiving any communication and a control group receiving no communications. It was felt that Study I failed for two reasons: the wording of the ballot was very confusing to many subjects, and subjects paid too little attention to the communications. Study 11, therefore, attempted to clarify the graduated income tax issue. It was conducted a few months after the election. This time, 136 registered voters participated in the study. The results of Study 11 showed that the value communications could at least produce attitude change. But, once again, they failed to show any difference as a result of important versus unimportant value communica- tions. Finally, Study 111 was designed to insure close attention to the communications by the subjects. Seventy-eight college students par- ticipated in the study. The posttest, which took place immediately after subjects had read the communications, showed that a communica- tion linking an attitude object to a relatively important value produced significantly more attitude change than did a communication linking that object to a relatively unimportant value. Thus we were able to conclude that attitude change could be considered a function of changes in perceived instrumentality. Robert John Homant In addition, certain other topics were discussed: the differences between field and laboratory studies, possible sleeper effects in Study II, and the relevance of consistency theories for explaining our results. ’/1] J I I Approved: ’éZLZQt;4~/<€Z¢é;ap.80). Hypothesis Ib Hypothesis Ib predicted that relatively more people would change 55 Table 6. Changes in voting intention for groups A & B compared to group C. Voting Change Voting-intensity Change Percentage number Hot Group Changing Changing Changing Number Mean score8 A & B 51.6 32 30 62 1.15 C 51.4 l8 17 35 1.23 Chi-square - 0.0003 5" -0.271 b p greater than .975b p greater than .75 a) A mean score of 1.0 would represent an average change of one unit, or from "probably yes" on the pretest to "vote no--uncertain" on the posttest. b) Two-tailed probability. 56 their voting intentions in group C than in the combined groups D and E. This hypothesis was tested in exactly the same manner as hypothesis Ia had been, and the results are shown in Table 7 (page 57). Although these data were in the predicted direction, there was little difference either in the percentage of subjects changing (chi-square = .329, p ).25) or in the amount of change (5.8 +0.28, p;>.40) in the two groupings. Thus, neither part of hypothesis I was supported. Hypothesis II The second hypothesis was based on the perceived instrumentality concept. It predicted that the voting change scores as well as the voting intensity change scores would be ordered from lowest to highest for groups A through E, respectively. The difference in the number of subjects changing in the five groups (the voting change scores) was tested with a chi-square developed by Cochran (1954), which takes into account the predicted direction of the scores in a multi-celled chi-square table by assigning weights that reflect the predicted magnitude of the scores. This test has no special name, and will be referred to simply as "Chi-square with a test for trend." The weights that were pre-selected for the test of the data relevant to hypothesis II were +2, +1, 0, -1, and -2, for groups A through E, respectively.27 27These weights are, of course, arbitrary. One could, for example, argue that weights of +2, +0.5, 0, -0.5, and -2 would be more appropriate, since the low value appeals (to groups B and D) might have been expected to have very little effect. Cochran (1954) discusses this point as follows: Such scores (weights) are to some extent subjective and arbitrary, and some scientist may feel that the assignment of scores is slightly unscrupulous, or at least they are uncomfortable about it. Actually, any set of scores gives a valid test, provided that they are constructed without consulting the results of the experiment. If the set of scores is poor, in that it badly distorts a numerical scale that really does underlie the ordered classification, the test will not be sensitive. The scores should therefore embody the best insight available about the way in which the classification was constructed and used. 57 Table 7. Changes in voting intention for groups D & E compared to group C. Voting Change Voting-intensity Change Percentage number NOt Group Changing Changing Changing number Mean score8 C 51.4 18 17 35 1.23 D & E 45.5 30 36 66 1.15 Chi-square 8 0.329 E,- +0.279 p greater than .25b p less than .40b a) A mean score of 1.0 would represent an average change of one unit, or from "probably no" on the pretest to "vote yes--uncertain" on the posttest. b) One-tailed probability. 58 The five sets of voting-intensity change scores were tested by an analysis of variance with a test for trend (see Hayes, 1963, pages 555 to 558). This test is based on the same principle as the test for trend with the chi-square test mentioned above. The linear weights +2, +1, 0, -l, and -2 were once again assigned to groups A through E. Table 8 (page 59) summarizes the data for groups A through E, and reports the relevant chi-square and analysis of variance data. The very low value of chi-square (0.69) reflects the fact that there was virtually no difference in the voting change scores for four of the five groups. The extremely low value of F (0.01, highly non-significant) is due to the fact that group A, which was hypothesized to have the highest mean voting- intensity change score, in fact had the lowest mean score. Hypothesis II, therefore, was not supported. By inspection, the mean voting-intensity change scores had a somewhat bow-shaped appearance; for this reason Table 8 also gives the F test for a curvilinear trend. This too was highly non-significant. Pretest Attitude and Voting Behavior The data from this study were also relevant to several other questions (see pages 22-24). The first of these questions concerned the correlation between subjects' pretest attitudes toward a graduate income tax and their subsequent voting behavior. Table 9 (page 60) shows this correlation for each of the five groups, as well as for all subjects combined. Although the correlations for the combined groups leave little doubt (p less than .005) that pretest verbal attitudes did correlate with be- havior in this case, the correlations in Table 9 are uniformly low. 59 Table 8. Changes in voting intention for groups A through E. Voting Change Group A B C D E Number of subjects 31 31 35 32 34 Number Changing 16 16 l8 l6 14 Percentage Changing 51.6 51.6 51.4 50.0 38.9 Chi-squareE—e 0.686, df - 1, p - .40 Voting-intensity Change GrOup A B C D E Mean Changeb 1.06 1.25 1.23 1.22 1.09 Summary of analysis of variance testing for linear trend: Sum of Degrees of Mean Source Squares Freedom Square F Between 0.890 4 - Linearc 0.008 1 .008 .01 n.s. Quadraticc 0.875 1 .875 .48 n.s. Other 0.007 2 .0035 - Error 285.660 158 1.808 Total 287.440 162 a) This chi-square statistic used a test for trend (Cochran, 1954) which assigned weights of +2, +1, 0, -1, and -2 to groups A through E, respectively. b) A mean voting-intensity change score of 1.0 would represent an average change of one unit (e.g., a change from "definitely no" on the pretest to "vote no--uncertain" on the posttest). c) The hypothesis being tested was for a linear trend; the test for the quadratic, or curvilinear trend was included because the direction of the means appeared to be bow-shaped. 60 Table 9. Correlations between pretest attitude and voting on the graduated income tax amendment. Group N Votinga Voting-intensityb A 31 .205 .310* B 31 .245 .312* C 35 .209 .186 D 32 .233 .167 E 34 .397** .229 All 88 163 .2625“: .233*** *p less than .05 **p less than .01 ***p less than .005 (all significance levels are one-tailed) a) point-biserial correlation b) product-moment correlation 61 Possible reasons for the low correlations are discussed in the next chapter (pages 71 ~72) . Reasons for Voting A second question concerned the reasons which subjects gave for their vote on the graduated income tax amendment. We had intended (see page 23) on seeing whether these reasons had any relation to the value letters which subjects were sent by the experimenter. Neither of our hypotheses were supported, however, and there was very little evidence that the subjects paid any attention to the letters. Therefore, the reasons which a subject gave for his voting behavior were compared with his values (rather than with the particular letter he received). It should be noted, however, that there is no theoretical reason why a person should feel that his more important values are necessarily relevant to a particular issue. Therefore, we shall treat these data simply as providing an answer to the empirical question: Do subjects tend to perceive this issue (the graduated income tax amend- ment) in terms of their more important values? If, in the course of the telephone interview, a subject had not already told why he was for or against the graduated income tax amend- ment, he was asked, "Wes there any particular reason you were against (or 'favored') the graduated income tax?" Seven subjects (out of the 147 contacted by telephone) preferred not to answer the question. Although many other subjects (41) simply said that there was no reason, or gave an unintelligible reply, 99 of the subjects did give a reason. For the most part, subjects' replies could be classified into a few basic responses (e.g., simply being against any sort of tax; or trying to vote according to what was fair to everyone). Table 10 (page 62) presents 62 Table 10. Frequency of various reasons being given for voting. Subjects voting for Subjects voting against Reasons graduated amendment graduated amendment Taxationa 0 42 No reasonb ll 30 Fairnessc 21 16 Safe nod 0 9 Incentivese O 8 Peoplef 2 1 a) Subjects in this category all expressed fear that the tax would get out of hand--either because counties, cities, etc. would im- pose it, or because there was no limit set on it. While their general concern was economic, there was no objection to the graduated aspect of the amendment. b) Many of the subjects in this category simply said that they had e) d) f) no reason for being against the amendment. Others gave an am- biguous or unintelligible response. Subjects in this category all expressed a desire to do what was right or fair--for both the yes and no voters this was thought to involve paying according to one's ability, but the no voters felt that this was achieved by a constant rather than a graduated rate . Subjects in this category said that they did not understand the issue, and felt that it was safer to vote no on something which one did not understand. Subjects in this category all felt that a graduated tax was un- fair and discriminatory to those who worked hard to get ahead, and that simply because a person earned more money was no reason for taxing him at a higher rate. Subjects in this category all said that they voted solely in accordance with someone else's wish. (The extent of personal influence was, of course, much higher than 3 out of 147 subjects; many others pointed out that one of the above reasons was arrived at in conversation with others.) 63 a summary of the reasons given by subjects for their votes, as well as the number of subjects giving each reason. Relation of reasons and values. After the subjects had been classified according to the reason which they had given for their vote, median tests were used to determine whether these subjects could be distinguished on the basis of their values. These data, presented in Table 11 (page 64), fail to reveal any significant differences between the subgroups on the basis of their terminal values. In Table 11 no distinction was made between yes and no voters. It might be, however, that a person voting yes and giving as his reason "wanting to be fair" could be quite different from a person who voted no for that reason. Therefore, we selected the two reasons which seemed to be most logically related to particular values; namely, incentive (with a comfortable_;ife) and fair (with eguality).28 Then, using the Mann-Whitney test, we compared subjects voting no and giving an incentive reason with the other subjects voting no. This comparison was made only on the value a comfortable life. Likewise, we compared subjects giving a "fair" reason with the remaining subjects, testing yes and no voters separately on the value eguality. Once again, the results (see Table 12, page 65) gave no support to the possibility that subjects' reasons were correlated with their values. In fact, in all cases the subjects who gave the particular reason ranked the hypothetically associated value lower than did subjects who did not give that reason. Thus, from our 28The reasons classified as "incentive" were virtually identical to the reasoning contained in the comfortable life letter (see Appendix B). Likewise, those reasons classified as "fair" were usually similar to the reasoning in the equality 1etter--especially with those subjects who voted yes. 64 .anOumou cameo mo mouumauomov Hasm a you oH wanes mom Am «m. e~.~ oo.m an.» oo.n ma.6 we.» mm.o sounds as. mm.o oo.a oo.oa oo.a m~.¢ oo.m ca.m aaeanaoaum «any aw. mm.a oo.oa om.na co.ea so.ea mn.na oa.¢~ coauaamoouu aqaoom em. ~m.a n~.o oo.oa ma.o m~.6 me.a aa.~ ausaauu-mamm an. ea.s oo.~a oo.a m~.w n~.n ma.o om.a aoaua>asm mm. um.m oo.na om.ea ww.na ~s.ea as.ma oa.ea «newsman no. me.m oo.ms oo.oa oo.a an.» m~.m ma.w assuauoa Haaoaumz mm. oa.m oo.a om.~a nh.aa oo.~a ms.~a om.aa o>o~ «snug: we. ~6.s oo.aa oo.s na.m so.m a6.m oa.m assays; “mesa ma. Hm.~ oo.aa om.a oo.a mo.m mo.a om.m asaaaaaqm cs. «o.m n~.m on.e m~.n ow.m e~.¢ ha.n seasons 65. «a.~ oo.~ oo.n nh.u oe.~ mm.m oo.e guesses. sesame an. ne.m oo.n oo.a mm.e co.» Nm.n mm.a guidance as. om.o oo.ea on.ea n~.ma «4.4H mm.ea mm.ee assume we canes < om. mm.m oo.m om.o oo.e m~.e oo.n mw.m momma an vanes < unmanned «m. am.n oo.~a- oo.aa so.oa no.a oo.m mm.oa -aaouoa «6 «anon < «N. mm.e oo.a om.w n~.sa nm.ma 6m.ma m~.¢H «use assumes» a< mm. ~a.~ oc.e~ om.~a m~.~a oo.ma. oo.Ma ha.aa «was manuuuoyaou < a saunas-aeo m m a an as «a uz nosau> Illlmqmmlmmwmmml mamomm 0>wusooea on swam unashamm cannon oz noaumxsa 11mmmnmuuaqmmmmmll .ouo> unonu you msomsou msowum> mmw>uw sausages now mwcwxcmu o=Hm> swaps: .dH manna 65 Table 12. Median ranks for selected values according to the reason given for voting in a particular direction on the graduated income tax. A comfortable life Subjects Median Giving "incentive" reason 8 12.50 (all of.whom voted no) Others voting no 98 11.59 Mann-Whitney U . -O.3l; n.s. Equality Subjects N Median Giving "fair" reason 21 7.13 and voting yes Others voting yes 13 6.60 Mann-Whitney U - -O.54; n.s. Equality Subjects N Median Giving "fair" reason 16 10.00 and voting no Others voting no 90 7.88 Mann-Whitney U - -0.37; n.s. 66 data, there was no evidence of any relation between the reasons subjects gave for their vote and the importance of subjects' values. Values, Demographip Variables, and Voting The relationships between values and reasons, which we examined in the previous section, were strictly empirical. That is, they did not result from any experimental manipulation, and no particular theoretical issue was involved. This is also true of the relation to be reported in this section, between values and voting, and between demographic variables and voting. Table 13 (page 67), shows the correlations between each of the values and subjects' voting behavior. Table 14 (page 68) shows the correlations between each of the demographic variables and subjects' voting behavior. Considering both tables, and both measures of the dependent variable, there are only 6 (out of 50) correlations that are significant at the .05 level or beyond, the highest of which was only .29 (political party with voting-intensity). Thus-~with the possible exception of political identification--voting behavior on the graduated income tax amendment was not related to any of our variables to any meaningful degree. In the next chapter we shall pr0pose that this lack of psychologi- cally meaningful relationships, as well as the virtually complete failure to obtain positive results for our various hypotheses can be explained on the basis of factors that were accidental, both to the graduated income tax issue and to the design employed by this study. In addition, we shall propose two studies which could provide a fairer test of our hypotheses. 67 Table 13. Correlations between values and voting behavior on the graduated income tax amendment. Voting Voting-intensity Value (point biserial r)a (product-moment r) A comfortable life -.020 +.006 An exciting life -.042 -.012 A sense of accomplishment +.l40 +.l66* A world at peace '+.052 +.022 A world of beauty +.119 +.110 Equality 1+.069 +.025 Family security 1+.l60* +.132 Freedom -.080 -.085 Happiness -.106 -.102 Inner harmony -.014 +.004 Mature love -.l84* -.201** National security -.034 -.023 Pleasure -.021 -+.027 Salvation '+.013 -.006 Self-respect -.027 1+.002 Social recognition +.006 +3001 True friendship 1+.022 +.059 Wisdom -.031 -.056 a) A positive correlation means that those high on that particular value favored the graduated income tax amendment. *p less than .05 (two-tailed) **p less than .01 (two-tailed) N-l63 for each value. 68 Table 14. Correlations between demographic variables and voting behavior on the graduated income tax amendment. Variable N Voting Voting-intensity Sexa 153 +.049 +.080 Ageb 155 -.125 -.157 Childrenc 158 -.053 -.059 Incomed 112 -.052 -.079 Educatione 154 +.054 -.018 2611:1ca1 partyf 149 +.231** +.287** Religion3 143 -.072 -.l48 The correlations with voting are point biserial except for sex and religion which are phi coefficients. The correlations with voting- intensity are product-moment, except for sex and religion, which are point biserial. **p less than .01, two-tailed. a) A positive correlation indicates that males favor the amendment. b) A positive correlation indicates that younger people favor the amendment. c) A positive correlation indicates that people with more children favor the amnhent. d) A positive correlation indicates that people with higher income favor the amendment. e) A positive correlation indicates that people with more education favor the amendment. f) A positive correlation indicates that democrats favor the amendment. This variable includes the "independents," who were scored as falling between republicans and democrats on the political continuum. g) A positive correlation indicates that Catholics favor the amendment. (Only Catholics and Protestants were included in this variable.) CHAPTER IV DISCUSSION In this discussion we will be concerned with explaining the null results, and with proposing what needs to be done in order to substantiate our explanations. Hypotheses I and II Our hypotheses depended on the effectiveness of the manipulation, which can be considered under two aspects: (1) were subjects influenced by the letter which they received, and (2) was the persuasiveness of the letters a function of the subjects' value systems. Since the second aspect was the major point of the study, we will begin by considering evidence on this point. First of all, there was some evidence to support our hypothesis that groups A and E (who received a high value letter) were more in- fluenced than their respective counterparts, groups B and D. In Table 9, for example, it can be seen that the highest correlation between pretest and voting (+.397) was obtained for group B, the high value group who were urged to vote in the same direction as their pretest attitudes had been. Furthermore, the lowest correlation shown in Table 9 (+.205), was obtained with group A, the high value group who were urged to vote in the opposite direction from their pretest attitudes. In fact, if we place the correlations between pretest attitudes and voting in the 69 70 order of their expected magnitude, only the correlation with group B is out of order. These data are reflected in a different way in Table 8. Once again, looking at the voting change data, the "percentage changing" in each group orders in the predicted direction (with the exception of a tie between groups A and B). Of course this evidence is not sufficient for accepting hypothesis II. The differences between these groups, as we have reported, are far from significant. Furthermore, the voting-intensity data generally contradicted the trend of the voting data. The null results with respect to this aspect of the manipulation are not too surprising. One could argue, for example, that the difference in importance between the high and the low value was simply not great enough to produce a statistically noticeable difference in the effectiveness of the letters (see Table 4). These results were surprising, however, with respect to the other aspect of the manipulation--the simple question of whether the letters, irrespective of whether a high or low value was appealed to, had an effect. This point was tested specifically by hypotheses Ia and Ib. Table 6, which shows the results testing hypothesis Ia, presents an extremely small chi-square and comes about as close as one possibly can to "proving the null hypothesis" that the letters to groups A and B had no effect. The test of hypothesis Ib (see Table 7) is not much more encouraging. (Perhaps an optimist would infer a trend.) These null results were surprising because the pilot study (Table 2) had demonstrated, rather conclusively, that the letters could produce 71 attitude change. Thus the null results on this point have to be explained in terms of whatever differences there were between the main study and the pilot study. We will return to this point on page 72. The Pretest Votipggporrelation Our study has produced only one positive result: the correlation between pretest attitudes and voting (Table 9). Even here, however, the obtained correlations (+.26 and +.23) were much lower than what we had expected (from +.40 to +.60). Therefore, instead of being satisfied with the fact that the pretest did correlate with voting behavior, we will try to explain why the correlation was so low. One possible explanation for the low correlation between pretest attitudes on the graduated income tax and voting behavior on that issue would be that subjects' perceived the issue differently on the two occasions. At the time the pretest questionnaire was sent out, the official ballot for the NOvember 5 election--which was to contain the specific wording of the preposed graduated income tax amendment-~had not yet been printed. Consequently it was impossible to word the pre- test attitude measure exactly as the ballot was to be worded. After the election, subjects reported several aspects of the amendment which, in retrospect, can account for the fact that while 67% of the subjects had favored the idea of a graduated income tax at the time of the pretest, only 23% of them actually reported that they had voted for the amendment which would have allowed that type of tax. Once again, the amendment was worded as follows: Shall section 7 of article IX be amended to read as follows: ’An income tax at flat rate or gradu- ated as to rate or base may be imposed by the state or any of its subdivisions.' 72 Subjects emphasized the following points in explaining why they voted no. (1) Because the ballot read, "Flat rate or graduated," some subjects felt that the graduated income tax would be in addition to the flat-rate tax. (2) Other subjects indicated that they had expected the ballot to specify what the amount of the graduated tax would be. They felt that the legislature was being given too much power. (3) Finally, because of the phrase, "...or any of its subdivisions," many subjects felt that cities and counties were being given the power to impose an income tax, whether graduated or not. (When in fact the amendment would merely have removed a constitutional restriction on this power, rather than granting the power.) In summary very few subjects voted against the amendment because of anything to do with the concept of a graduated tax. Referring to Table 10 (page 62), the eight subjects whose reasons were classified as "incentive" and the 16 subjects who voted no and whose reasons were classified as "fairness" were the only subjects that we can reasonably be sure voted no because of the graduated aspect of the tax. The fact that so many subjects voted on the basis of factors that were not necessarily related to their pretest attitude on the graduated income tax, is in itself sufficient to provide an explanation for the low correlation between the pretest and voting. The Ineffectiveness of the Value Letters If most subjects' votes were determined by factors other than the graduated aspect of the income tax amendment, this also provides a possible explanation for the failure of the letters. The letters focused completely on the graduated aspect of the amendment, while for most 73 subjects other considerations seemed to be more important. Over and above this, there is another explanation for the failure of the letters--a supplementary rather than a competing explanation. As we pointed out earlier (see page 52), only 12 out of the 147 subjects contacted by telephone mentioned the letter which they had been sent. Although it is possible that with a little more probing on the part of the experimenter many more subjetts might have mentioned the letter, still the letter certainly could not have been very salient to the subjects. This is in contrast to the pilot study with these same letters (see page 35). At that time, although no effort was made to solicit information about the letter, about one out of every four subjects mentioned the letter. This usually took the form of a statement like, 'Why, as a matter or fact I just received something in the mail about that the other day.‘ What accounts for this difference between the pilot study and the main study? The only difference in the two studies, as far as the letters were concerned, was the obvious difference that the first set of letters was sent out on July 29, while the second set was sent out on October 31 (cf. footnote 20). Could this be enough to account for the difference? There is some evidence to support this possibility. Several of the subjects who did not mention receiving the letter mentioned that they were "sick of" or "fed up with" the amount of campaigning that had gone on. They referred to the high amount of television exposure, as well as to the high number of fliers that had been left at their door. It is possible that the value letters, even though they were individually typed and sent as first class mail, were quickly classified along with the other political 74 propaganda. Many subjects expressed pride in the fact that they never let themselves be influenced by such political campaigning. Proposed Field Follow-up The purpose of our speculation was not to explain away the study's null results, but rather the purpose was to pinpoint likely flaws in the total procedure which, hopefully, could be taken into account in a sub- sequent study. We have suggested two major causes for the failure of our manipu- lation: (l) the subjects' votes were determined by factors other than the graduated aspect of the proposed amendment, and (2) the value letters were sent at a time when they would be ignored because of the high level of political activity that was going on at that time. In the next chapter we will report the results of a follow-up study designed to check on these possibilities. Proposed Laboratory Follow-up In addition to the direct follow-up to the first study, it was also thought desirable to conduct a study under more controlled, albeit artificial conditions. It is possible for example that no matter when the value letters might be sent out, one could never be sure that subjects would pay sufficient attention to them for the difference between the high and the low value letters to affect subjects' behavior. The purpose of a third study, then, would be to determine whether the basic hypothesis of this dissertation is valid, if sufficient control can be achieved over extraneous variables. Chapter 6, then, will present the results of a laboratory study which used a procedure ana10gous to that used in the study which we 75 have just reported. The purpose of this study was to show that under controlled conditions the persuasiveness of an appeal is a function of the importance of the value appealed to. llullllli 711111!" II CHAPTER V FIELD STUDY FOLLOW-UP (STUDY 11) Introduction In the previous chapter we suggested that we failed to produce positive results for two reasons: (1) subjects' votes on the gradu- ated income tax amendment were determined by factors which were irrele- vant to the concept of a graduated income tax, and (2) subjects received the value letters at a time when they were most likely to ignore them. The purpose of the study to be reported in this chapter was to repli- cate the first study, while taking into account these two reasons. Although positive results will not pppyg our reasoning correct; they would offer the best circumstantial evidence which we can think of to support our contentions. The study to be reported below will be referred to as "Study II," and the study which was reported in chapters one to four will be referred to as "Study I." Because of the changed circumstances some changes were required in the procedure. First, subjects could only be asked how they would vote if the issue came up again. This meant, in effect, that we were measuring attitude change rather than behavior change. Secondly, there was no longer a plausible rationale for urging subjects to be against the graduated income tax amendment. Therefore, only the pro letters (based on equality and fgpily_§ppppipy) were used in this second study. This meant that instead of the five conditions in Study I, 76 77 subjects in Study II were assigned to one of three conditions: a high- effective letter condition (group A'), a low-effective letter condition (group B'), or a no letter (control) condition (group C'). Hypotheses There were two hypotheses which were tested by Study 11. The rationale for these hypotheses was the same as for the corresponding hypotheses in Study I (cf. pages 21-22). ‘ Hypothesis I. After receiving the value letters, subjects in groups A' and B' will have a more favorable attitude toward the graduated income tax amendment, than will subjects in group 0' who receive no letter. Hypothesis II. Subjects in group A', who receive a high value letter, will have a more favorable attitude toward the graduated income tax issue than will subjects in group 3' who receive a low value letter. Method Subjects The subjects for Study 11 were the 169 subjects whom we had been able to contact for the posttest of Study 1 (see page 47). This had the advantage, of course, of eliminating costly and time-consuming pretesting. It had the concomitant disadvantage of limiting the generality of the results--this point will be discussed at the con- clusion of this chapter. Value Letters Only the eguality and family security letters were used for Study II. These letters had to be modified considerably, first to provide a 78 rationale for the letter when ostensibly the issue was dead, and secondly, to take into account the several factors that had influenced subjects' voting in November (see page 72). The two letters, given in Appendix I, expressed concern that the amendment had been defeated, because the amendment would have aided a particular goal (fppily gecurity or eguality). Then the letters each pointed out that many people had voted against the amendment because of certain.misunderstandings. The rationale for a letter being sent out at this particular time (February 20--over three months after the election) was to "correct some false impressions...while...still fresh" and to urge the recipients "to support the graduated income tax the next time it is voted on." Experimental Tgeppgegt Since pretest information was already available on the subjects, the only thing that was required was to assign subjects to one of the three conditions (A', B' or C') and send out the letters. The 169 subjects were randomly assigned to three groups of 56 Ss each. (The .)29 remaining subject was eliminated from the smmple Half of the subjects in groups A' and B' were to receive the egpality‘ 291a Study I it was necessary to match the groups on pretest attitudes because subjects were to receive pro or con letters depending on that attitude. In Study II this was unnecessary, since all subjects were urged to change their attitude in the same direction (in favor of the graduated tax). Technically, there was no reason why the groups in Study II had to be matched in number of subjects; this was done simply for the sake of clarity. Also, it may have been ”neater" to match the groups on pretest attitudes, rather than to trust to randomisation, but since the pretest attitudes barely correlated with voting behavior (see Table 9) this seemed unnecessary. The various permutations of pretest attitude and voting-intensity would have presented a hopelessly confusing variable to match groups on. 79 letter and half the family segurity letter. In group A', the 56 letters were assigned so that, hypothetically, they would be most effective (according to the subjects' pretest rankings of family security and eguality). In group B', the 56 letters were assigned so that they would be least effective. The procedure for assigning the letters was identical to that used in Study I (see page 40 and Appendix E). Once again, as in Study 1, there was a question as to whether the procedure for assigning value letters was successful in creating a difference in the importance of the value appealed to in the high and low value conditions. Table 15 (page 80) shows the differences between the mean ranks of subjects' values in groups A' and B'. These differences, which are based on subjects' pretest value ranks, were highly signifi- cant. Of course, a more important question is whether these differences in subjects' values remained six months later. Unfortunately, we did not get a retest of subjects' values at that time. On February 20, 1969 the letters were sent out to the designated subjects in groups A' and B'. As in Study I, they were addressed to the subjects personally. The date was intended to be one when other distractions (i.e., Christmas and income tax) were at a minimum. Posttest Four days later, on February 24, the 168 subjects were sent a brief questionnaire, along with a stamped return enveloped addressed to the experimenter. (The complete questionnaire is given in Appendix J.) After a brief introduction which explained the reason for contact- ing the recipient again, the subject was asked to answer the following two questions: 80 Table 15. Mean pretest value ranks for subjects receiving the various letters in the high and low value conditions. (The means represent the mean score for the value contained in the letter which each subject received.) Value gopditiop High (A') Low (3') Letter N Mean N’ Mean .p Equality 28 6.34 28 12.51 7.11 Family security 28 2.15 28 6.01 4.45 Combined 56 4.25 56 9.26 7.34 81 If you were deciding on the issue at_this time, would you be in favor of an amendment to allow Michigan to have a graduated income tax? Definitely yes Probably yes Probably no Definitely no and Have you received any information on the graduated income tax since the election last NOvember? Yes No In addition, subjects who answered "yes" to the second question were asked to give a brief description of what they had received. The reason for contacting subjects by questionnaire rather than by telephone should be explained. Our concern was with the effec- tiveness of the letters approximately three days after they had been received. In Study I, this could be obtained by telephone--over a two week period--because subjects remembered how they had voted. In Study II, however, it was necessary to measure all of the subjects attitudes within a one or two day period--a requirement that would have been virtually impossible to meet by telephone interview. For this same reason, we did not follow-up those few subjects who did not return our questionnaire. Presumably, most of those subjects who did return it filled in their replies within a few days after receiving the question- naire. Results Return Rate One hundred and thirty six of the 168 subjects filled out and returned the questionnaire. An additional seven questionnaires were returned unanswered. Fifty-five per cent of the questionnaires were 82 returned within 10 days after they had been sent out. The remainder of the questionnaires were returned over the following two weeks. Effectiveness of Manipulation In Study II the subjects were asked specifically if they had received any information on the graduated income tax amendment. Thirty-six of the 92 subjects in groups A' and B' who returned the questionnaire specifically mentioned that they had received a letter. While this is still little over one-third of the subjects, it is a marked improvement over Study I (which had a percentage of 10.3). Data Analysig The data analysis for Study II is simpler than it was for Study I. The dependent variable (attitude toward graduated income tax) was measured by the subjects' responses to the first question given on page 81. In order to make the measure comparable to a voting decision, subjects who responded "definitely yes" or "probably yes" were recorded as favoring the graduated income tax, and subjects who responded "probably no" or "definitely no" were recorded as being against the graduated income tax. Differences between letter conditions could then be tested by chi-square. As a supplement to this, responses to the four categories were coded from 1 (definitely yes) to 4 (definitely no), respectively, and differences between the letter conditions were tested by §_ratios. Hypothesis I Hypothesis I predicted that subjects in groups A' and B' would be more in favor of the graduated income tax than would subjects in 83 Table 16. Posttest attitude toward graduated income tax for groups A' & 3' compared to group 0'. Percentage Subjepts Group N favoring Favoring Opposing Mean Score A' & B' 92 41 38 54 2.76 C' 44 25 11 33 3.23 Chi-square I 3.43; p less than .05 (one-tailed) p.ratio - 2.21; p less than .025 (one-tailed) *Refers to number returning the questionnaire. 84 group C'. Table 16 (page 83) presents the relevant data. In groups A' and B' 41% favored the graduated income tax, vs. 25% of group C'. This difference was significant at beyond the .05 level (chi-square = 3.43). The.£ ratio for the difference in mean attitude scores is slightly more significant. Thus hypothesis 1 was confirmed. Hypothesis II Hypothesis II predicted that subjects in group A' would be more in favor of the graduated income tax than would subjects in group B'. Table 17 (page 85) presents the data testing this hypothesis. Although in the predicted direction (43% of group A' in favor vs. 40% of group B'), the difference was minimal, and neither the chi-square value nor the‘p ratio approached an acceptable level of significance. Thus hypothesis II was not supported. Effect of Recalling Letter As we mentioned above, 36 subjects in groups A' and B' reported receiving the value letters. In order to see whether these subjects were more sensitive to the manipulation, the subjects in groups A' and B' that did ppp_report receiving the letter were eliminated from the analysis and hypotheses I and II were tested again. Tables 18 (page 86) and 19 (page 87) report these data for hypotheses I and II respectively. Once again hypothesis I is supported, with the chi-square value increasing slightly and the §.ratio decreasing slightly. And, once again, hypothesis II is not supported, with both the chi-square value and the 5 ratio being slight in the reverse direction. 85 Table 17. Posttest attitude toward graduated income tax, high vs. low value conditions. Percentage Spbjects Group N favoring Favoring Opposing Mean Score A' 44 43 19 25 2.75 B' 48 4O 19 29 2.77 Chi-square - 0.12; p less than .40 (one-tailed) £_ratio - +0.07; p less than .48 (one-tailed) 86 Table 18. Posttest attitude toward graduated income tax, for subjects who reported receiving value letters, for groups A' & 3' compared to group C'. Percentage Subjects Group N favoring Favoring Opposing Mean Score A' & B' 36 47 l7 19 2.72 C' 44 25 ll 33 3.23 Chi-square . 4.30; p less than .025 (one-tailed) §.ratio - +1.88; p less than .05 (one-tailed) 87 Table 19. Posttest attitude toward graduated income tax for those who reported receiving value letters, high vs. low value conditions. Percentage Subjectg Group N favoring Favoring_v Opposing_p Mean Score A' 16 44 7 9 2.81 B' 20 50 10 10 2.65 Chi-square - 0.14, n.s. §.ratio = -O.38, n.s. 88 Discussion The primary purpose of Study II was to substantiate the reason- ing given earlier (cf. pages 72-74) for the discrepancy between the results of Study I and the pilot study (cf. pages 35-38). Once again (in Study 11) letters arguing in terms of values were shown to be effective stimuli for attitude change, when sent at a time when there would not be too much interference from other sources. We have already pointed out that such results do not prove anything about the relation between attitudes and values, since there were many aspects of the letters which did not concern values. Still, the significance of the pilot study and Study II, taken together, should not be minimized. First of all, they constitute a field replication of Carlson's (1956) study (see pages 7-9). As we pointed out in the first chapter, attitude change has not usually been easy to demonstrate in a field situation--especially with non-trivial matters (Hovland, 1959). We do not expect that this type of attitude change will be easy to replicate. There were some special circumstances pertaining to the two studies (the pilot study and Study II) which were favorable to obtaining attitude change and which limit the generalizability of the results. With respect to the pilot study, we have already pointed out that subjects at that time (July, 1968) knew very little about the graduated income tax issue. The communications which they received had the advantages of both primacy and recency with respect to an issue of relatively low familiarity to the subjects. (See Hovland, Janis, and Kelley, 1953, for experimental studies on this point.) 89 With respect to Study 11, something almost the opposite may have occurred. The posttest of Study I was, in effect, a pretest for Study II. This in itself (not to mention the pretest and the value letters for Study I) may have been enough to sensitize subjects on this particular issue, and to make them especially susceptible to the type of persuasive appeal used in Study II. (See Hovland, Lumsdaine, and Sheffield, 1949, for the sensitization effect of a pretest.) In other words, one should be cautious in generalizing from the series of studies which we have reported to the possible effective- ness of the same type of letter on other issues and in different cir- cumstances. One further limitation should be pointed out. The two studies which did find attitude change (the pilot study and Study II) were the studies which did not measure behavior. Thus, even though the attitude question in Study II was expressed in terms of how subjects thought that they would behave, and even though subjects had already had the experience of voting on this issue, we still cannot generalize from this to how the subjects actually would have behaved. The third study, to be reported in the next chapter, also measured attitudes rather than behavior.30 Finally, of course, we failed once again to verify hypothesis II. Thus we still have no convincing evidence that the importance of the value appealed to will determine the amount of attitude change. One could suggest some possible reasons for our failure. (1) The letters 30While this limitation is important with respect to practical appli- cations of the persuasive technique used in the three studies, it does not affect the theoretical interpretation with respect to theories of attitude change and the importance of perceived instru- mentality as the link between attitudes and values. 90 may not have had enough of an effect to allow us to discriminate a high value from a low value letter. Or, (2) since subjects' values were measured almost six months before they received the value letters (in Study II), there may simply have been too much unreliability to allow us to differentiate which value e ualit or family security) was more important to a given subject. (3) The fact that even in Study II only a little over a third of the subjects reported noticing the letters, may indicate that too few subjects paid sufficient attention to the letters for a noticeable difference between groups A' and B' to have been produced. And, lastly, (4) the appeal to the values may not have been the cause for the attitude change (or may have contributed only a small part to the letters' effectiveness). In Study III we attempted to construct a situation as analogous as possible to the graduated income tax studies, in which we could achieve a high degree on control over subjects' attention to the per- suasive communications. It was hoped that under these "laboratory" conditions we would be able to observe a difference between the effect- iveness of the high and low value communications. Further consideration of the significance of Study II will be put off until Study III is reported. CHAPTER VI LABORATORY STUDY FOLLOW-UP (STUDY III) Introduction Work on the laboratory follow-up, to be referred to as Study III, began while Study II was still being conducted. The purpose of Study III was to determine whether the perceived instrumentality relationship could be demonstrated at least under the relatively controlled conditions of a laboratory situation. In Study III, then, subjects were seen individually by the experimenter, knew that they were in an experiment, and received a communication which they were virtually forced to attend to. This last point, the guarantee that subjects would pay attention to the communication, constituted the major difference between Study 111 and the first two studies. On pages 13 to 15 we pointed out that this attention factor was one of the chief differences between field and laboratory studies. We have also summarized evidence which indicated that the failure of the manipulation in Study I was due, at least in part, to the failure of the subjects to attend to the value letter. Just how this attention factor was controlled in the laboratory study will be taken up in the method section below. Other important differences in Study III center around the use of a college population, a somewhat different attitude issue, and an immediate, rather than a three-day posttest. 91 92 W In order to make Study III relevant to a college population, Michigan State University's "sliding scale tuition" plan was selected for the attitude issue. According to this plan, which was in effect during the 1968-69 academic year, students from low income families were eligible for a tuition reduction. According to the Academip Handbook (1968): Students paying Michigan resident fees and not receiving scholarships and grants of $95 or more eligible to apply for fee reduction when the total parental income for 1968 is less than $18,400 for undergraduates ($19,400 for graduate students). In such cases, fees will be 1% of total parental income for 1968, but not less than $123 ($148 for graduate students). Some students who are completely independent and self-supporting and who are in no way supported by their parents may apply on the basis of their own income. Married students, veterans, and older students often qualify on this basis. Muchigan families with two or more MSU students may apply for the maximum reduction for all students after the first, regardless of parental income. It should be noticed that the fee reduction is analogous to the concept of a graduated income tax. That is, the amount of money to be paid varies with one's income, with wealthier people paying more, and poorer people paying less. Thus the selection of this issue had the ad- vantage of allowing us to use communications that were comparable to those used in Studies I and II, since the same four values (freedom, a comfor- table 1ife, eguality and family security) were relevant to the issue.31 31If some validation is needed for our assertion that the sliding scale tuition is similar to the graduated income tax, we could quote from one of our subjects from Study I (who should be given some credit for giving us the idea in the first place). In discussing her problem in deciding how to vote on the graduated income tax she said: I think that if you make more you Cshoulcy pay moreunaturally, because you probably do use the resources of other people, and the country more. But at the same time, I don't think that you should tax initiative, and this is what I figured this graduated thing really boils down to. Just like I'm against your bit out at Michigan State where you are charging more tuition-'now this isn't--is more or less the same thing--for people who can pay more. 93 Pilotygata Before beginning Study III, data were obtained on student attitudes toward sliding scale tuition. It was found than an over- whelming majority of Michigan State students agreed with the question, "Are you in favor of the sliding scale tuition plan presently employed at MSU?" Because of this, it seemed that it would be quite difficult to change attitudes if the issue were worded in this way. Therefore, a group of subjects were asked: "Would you be in favor of doubling the fee reductions currently allowed under MSU's sliding scale tuition plan?" Even worded in this way, most students answered yes to the question-- although not nearly to the extent that others had to the original question. Therefore, it was decided to word the issue in this latter way for Study III. It was also decided only to use communications that argued against the sliding scale tuition (using the values p_ comfgpppbleilife and freedom). Thus all experimental group subjects would be hypothesized to change in the same direction (as subjects had been in Study 11). Overviep of Procedure As in Study 1, it was necessary to pretest a group of subjects, divide them into treatment groups, have them receive an experimental treatment, and then posttest the subjects. In Study III, subjects were divided into three treatment conditions: group A", in which subjects received a high value communication,32 arguing against the sliding 32Throughout Study III we use the term "value communication" rather than "value letter," since no mailing was used. Rather, subjects received the communication in the form of a booklet. They were instructed to view the communication as one of a series of "brief editorials" or else simply as any type of "persuasive argument." 94 scale tuition, group B", in which subjects received a low value communication arguing against the sliding scale tuition, and group C", a control condition in which subjects received no communication on the sliding scale tuition. Hypotheses The hypotheses for Study III were virtually identical to those for Study II. The major difference was due to the communication arguing against, rather than in favor of the issue. Once again, the reader is referred to pages 21 to 23 for the rationale for these hypo- theses. Hypothesis 1. After receiving the value communkzations, subjects in groups A" and B" will have a less favorable attitude toward the sliding scale tuition than will subjects in group C", who receive no communication on this issue. Hypothesis 11. Subjects in group A", who receive a high value communication, will have a less favorable attitude toward the sliding scale tuition than will subjects in group B", who receive a low value communication. 14.2.0129 Subjects Subjects for Study III were 78 Michigan State University under- graduates in introductory psychology courses who received credit toward their class grade for participating. Subjects were recruited by a notice on the class bulletin board, informing them of a study concerned with "student opinions." The notice also informed the sub- jects that they would have to report twice for the study in order to 95 receive credit for their participation. Subjects reported individually to the experimenter's office, where they were immediately pretested (see below). Fifty males and 28 females (mean age of 19.4 years) participated in Study 111. Mpterialg Two sets of materials had to be constructed for Study 111: an attitude questionnaire (to be used for the pretest and the posttest), and two value communications. The attitude questionnaire. As we mentioned above, the attitude on which Study III focused concerned doublipg the sliding scale fee reductions. This attitude was measured by the following question: Would you be in favor of doubling the fee reductions currently allowed under MSU's sliding scale tuition plan? Subjects circled one of five answers: definitely yes, probably yes, no opinion, probably no, definitely no. Then subjects were asked: How important is the above issue to you? Once again, subjects circled one of five answers: very important, important, slightly important, unimportant, very unimportant. In addition, the subjects responded to six other campus issues. All seven attitude questions are given in Appendix H. The complete questionnaire also included Rokeach's terminal value scale (see Appendix A), which followed the attitude questionnaire. The question concerning the sliding scale tuition took the same form as the question on the graduated income tax that had been used in Studies I and II. One difference was the addition of a "neutral" response (since there was no longer any reason to force the subjects 96 into a voting-like dichotomy). The other difference was the addition of a question about the importance of the attitude item. This question was added to provide additional information about the attitude change process. To the extent that our reasoning concerning perceived instrumen- tality is correct (attitude change is a function of the importance of the value to which the new attitude is linked), we would expect an increase in the importance of the issue to accompany our hypothesized attitude change. No specific hypotheses could be formulated concerning the importance scores, however, because attitudinal position and judged importance are usually confounded (see, for example, Sherif and Hovland, l960--this point will be discussed more fully later). The importance scores were intended to be looked at only in the event that hypothesis II was accepted, in which case it was hoped that they would corroborate our reasoning. Value communications. The experimental manipulation required the construction of two communications arguing against the sliding scale tuition plan at Michigan State, one in terms of pjcggfortable_li§p, and the other in terms of freedom. These communications are given on the fifth and sixth page of Appendix L. Their construction was guided by the same general logic on which the pppfppppplp_lifp_and fpgpgpm_ letters were based in Study I (see pages 33 to 35). Other factors (differing from Study I) were also taken into account: the presumed higher intelligence of the subjects, the fact that no behavior was being urged, and the fact that subjects would be likely to pay relatively close attention to the communication. 97 Beneath each of the communications were four scales (see Appendix L) on which subjects were instructed to judge the communication's clarity, fairness, persuasiveness, and the extent to which it influenced them. These scales were not meant to provide any data for the experi- ment. Their only purpose was to motivate the subjects to pay attention to the passages. As far as the subjects were concerned, the reason why they were reading the passages was to rate them on the four scales. After the four scales, subjects were also asked a question, to determine which of them directly benefited from the sliding scale tuition. It was felt that these subjects might be less likely to show any attitude change-~in either value condition--and that additional information about the attitude change process might be gained if these subjects were looked at separately. In addition to one of the two value communications against the sliding scale tuition, it was felt necessary to give the subjects some irrelevant communications. Therefore, three more communications were constructed concerning three other campus issues on which subjects' attitudes had also been measured. These communications (see the second, third, and fourth page of Appendix L) were similar in format to the value communications, except that none of Rokeach's terminal values were explicitly referred to. Using the five communications (two value and three irrelevant) three sets of booklets were assembled. These booklets present the experimental manipulation as well as the immediate posttest. The booklets contained six parts: (1) A set of instructions orienting the subjects to the communications (Appendix L, first page); 98 (2) The three irrelevant communications (Appendix L, second, third, and fourth page); (3) either: i) the freedom communication (Appendix L, fifth page), ii) the comfortable life communication (Appendix L, sixth page), or iii) no communication; (Thus part 3 constitutes the difference between the three sets of booklets.) (4) A brief set of instructions orienting the subjects to the posttest (Appendix L, seventh page); (5) The seven attitude questions (Appendix K); (6) The terminal scale (Appendix A).33 PretepgyProcedure When a subject reported for the experiment, the experimenter told him that he was conducting a study of some student opinions. The experimenter handed the subject the pretest questionnaire (which included the value scales), and told him to fill out and return the forms as soon as possible. The subject was also asked to fill out alone without discussing any part of them with anyone else until after he had completed them. Seventy-eight subjects were seen in this manner between March 31 and May 5, 1969. Subjects were instructed to return for the second part of the study some time after May 11, but before June 7 (the end of the school year). Thus there was a period of from one week to ten weeks between the pretest and the experimental treatment. For the majority of the subjects this period was between three and four weeks. 33The value scale was included in the posttest booklet only to obtain an estimate of reliability. 99 Experimental Treaggent As in Studies I and II, the first step in the experimental treat- ment was to assign the subjects to treatment conditions. Seventy- eight subjects were randomly assigned to one of three groups (A", B", and C") with the limitation that groups be matched on pretest attitudes toward the sliding scale tuition.34 The second step was to determine whether the individual subjects in groups A" and B" would receive the freedom or the comfortable life communication. In group A", the communications were assigned so as to be most effective, according to the subjects' pretest values. In group B", the communications were assigned so as to be least effective. Thirteen subjects in each group were to receive the freedom communica- tion and thirteen the comfortable life communication. The exact pro- cedure for assigning the communications was the same as had been used for the value letters (see page 40 and also Appendix E). As in Studies I and II (see pages 41 and 79), a check was made on the effectiveness of the procedure for creating high and low value groups. Table 20 (page 100) summarizes the mean value differences for groups A" and B". Once again, the procedure was effective, and once again the more important question concerned the stability of these mean differences. This time, in Study III, retests were available on all subjects so that the question could be answered directly. 34The reason that groups were matched in Study III but not in Study 11, was because the pretest for Study III was considered reliable, whereas no reliable pretest existed for Study II. (See page 106 for reliability results.) Since hypothesis II was the major hypothesis for Study III, it was decided to match groups A" and B" exactly, and then match C" as closely as possible to those groups. The breakdown of the three groups by pretest attitude is given in Table 21 (page 101). In each group, the mean attitude score was 2.42 (or just to the "no opinion" side of "probably yes"). 100 Table 20. Mean pretest value ranks for subjects receiving the various communications in the high and low value conditions. (The means represent the mean rank for the value contained in that particular communication.) o tio Value in High (A") Low (3") Comunication N Mean N Mean 3 A comfortable life 13 9.53 13 14.76 3.56 Freedom. 13 4.38 13 8.23 3.04 Combined 26 6.96 26 11.50 3.60 101 Table 21. Pretest attitude toward doubling sliding scale fee reductions according to treatment group. fippjegts Respopgipg Definitely Probably No Probably Definitely Group N yes yes Opinion no no Mean A" 26 8 7 6 2 3 2.423 B" 26 8 7 6 2 3 2.423 C" 26 7 8 6 3 2 2.423 Total 78 23 22 18 7 8 2.423 102 As we pointed out in the "materials" section above, each of the value communications was placed in a booklet where it was preceded by other communications and followed by a posttest. Once it had been de- termined which communication a subject was to receive, his name was written on the first page of the appropriate booklet. (Of course the names of each of the subjects in group C" were written on one of the control group booklets.) The final step in the experimental manipulation was simply to wait for the subjects to return for the second part of the study. When a subject reported back to the experimenter he was handed the booklet with his name on it. Then the experimenter briefly went over the in- structions on the first page of the booklet, and told the subject to work through the booklet page by page. The subject was also told to fill out the booklet alone, without discussing the contents with anyone until after he had completed the booklet. Posttest The subject filled out the posttest immediately after he had finished the experimental manipulation. The reason for not separating the posttest from the experimental manipulation (which would have paralleled Studies I and II) was simply that it was quite difficult to recruit subjects who were willing to return even for a second time. To have required subjects to return for a third time would have resulted in too high an attrition rate.35 Possible biases which this immediate posttest might have caused will be discussed in the next section. 35About 20 subjects had to be telephoned in order to remind them that they were expected to return for the second part of the study. Most of them had simply forgotten about it, but about six subjects had to be persuaded to come for the second part of the study. (Since the results weren't in yet, we appealed to both high and low values: for the sake of science as well as their grade.) 103 Results Effeggiyeness of Manipulation One of the dangers in a laboratory, as opposed to a field study, is that subjects will "see through" the design, realize intuitively what they are expected to do, and react to this expectation rather than to the variable supposedly being manipulated. This reaction can take the form of compliance to the experimenter's hypothesis (see Orne, 1962, for a discussion of "demand characteristics"), or of resis- tance to the attempt to change the subject (see Rosenberg, 1965, for a discussion of "evaluation apprehension"). The design of Study III, with its immediate posttest, may have seemed especially susceptible to such effects. For example, to the extent that subjects in groups A" and B" felt that the value communica- tion was intended to change their attitudes on sliding scale tuition, they may have increased or decreased their tendency to change as a result of the manipulation. In order to determine how subjects did perceive the experimental situation, twelve of the subjects were debriefed by the experimenter just after they had completed the posttest. All of them felt that the experiment was concerned mainly with the stability of student opinion (over time, rather than following a persuasion attempt). All of them felt that their names had been written on a specific booklet just so the experimenter would keep track of the subjects. None were aware that others had received a somewhat different booklet. With some probing, five of the subjects also reported that they felt that the posttest was to determine whether the four communications had had any influence on them. None of the subjects, however, were aware that the 104 experimenter was concerned only with the sliding scale tuition. And, more importantly, none of the subjects showed any awareness that the value communications contained a reference to a value which they had rank ordered. All of the subjects felt that the value scales had been in- cluded simply as a measure of another type of student opinion. All of the subjects expressed surprise when the logic of the design and the reason for the pretest with the values were explained. (This is, after all, not too surprising; had any one subject seen both the high and low value communications, he might have had a better idea of what was being tested.) From the results of this debriefing, it would seem that there could be a possible bias with respect to hypothesis I--subjects may have altered their tendency to change to the extent that they believed the experiment was designed to measure attitude change as a result of the communications. With respect to hypothesis 11, however, there seems to be no possible way that subjects' reactions to a high or low value communication could have been differentially affected. There was also some "internal" evidence which showed that most subjects tended to take all of the communications quite seriously. When subjects disagreed with the content of a communication, they tended to become quite involved in writing a reply containing a counter- argument. This was especially the case with the sliding scale tuition communications, where many subjects were directly affected by the issue. For example, in response to the question, "Do you have any particular interest in this issue?" one subject, who disagreed with the communication, wrote: 105 Yes. The small amount I get ($50) out of my dad's $10,000 a year salary is hardly going to hurt some- one who makes $20,000 a year and probably pays less taxes anyway since‘pp can make use of loop-holes, etc. Another subject, who also disagreed with the communication, wrote: Low income families are not taking higher income families' money. They get grants, loans, and scholar- ships. That's what's the matter with the world-- everyone feels that he or she is being taken by someone. The logic of these responses aside, such answers were typical of subjects who disagreed with the communications. Many of these responses were written in, if not a violent, at least a highly excited style.36 A few other subjects, who agreed with the communication, took time to express (in writing) their pleasure that someone had finally echoed their own point-of-view. Thus there is little doubt that many of the subjects took the value communications seriously. The fact that some subjects were also aware of the attitude change aspect of the design should qualify our acceptance of the results pertaining to hypothesis I, but not hypothesis II. The Pretept Begultg In Table 21 (page 101) we reported the pretest results for attitudes toward sliding scale tuition. That attitude measure also contained an importance score. We mentioned briefly above that this 36Two days before the close of the experiment, the board of trustees unexpectedly abandoned the sliding scale tuition for the coming school year. While this only affected about six subjects, two of them were highly upset and replied that they might have to drop out of school for financial reasons as a result of the board of trustees' action. These subjects were simply included in the analysis, since there was no way that they should have biased the hypotheses. 106 importance score has been found to be confounded with attitude position (Sherif and Hovland, 1960). People with extreme attitude scores (e.g., "definitely yes" and "definitely no" responses) have been found to consider the issue more important than do people with neutral ("no opinion") attitude scores. Table 22 (page 107) presents the relation between pretest attitude and judged importance. The expected relationship is very evident. An inspection of Table 22 shows the difficulty in predicting the changes in importance scores that should be expected to accompany the hypothesized changes in attitude position. Since each group's pretest score put it in the "probably yes" region of the table,37 and since the communication urged subjects to shift toward the "definitely no" side of the continuum, there should be a natural tendency for the issue to decline in importance until attitude scores move beyond the neutral region. In general, however, we can say that if groups A" and B" shift to a new mean attitude position, then their mean importance scores should be higher than the baseline established for that position by the pretest data (e.g., if group A" were to shift their mean attitude score from 2.42 to 3.0, we would expect their mean importance score to at least be greater than 2.33). Reliapility pf Mpgpurep The test-retest reliability of our attitude measure was +.65, 37From a group's mean attitude score we cannot predict what its mean importance score should be (even before any manipulation) without taking into account the distribution of the attitude scores. For example, if a group had a bimodal distribution of attitude scores, its mean attitude score could be in the neutral region and its mean importance score could still be extremely high. But in the case of unimodal distributions (such as we had for groups A", B', and C") we can expect the mean importance score to be related to the mean attitude score. 107 Table 22. Importance of issue for different attitudinal positions toward doubling sliding scale fee reductions. Attitudinal Position Definitely Probably No Probably Definitely yes yes Opinion no no N s 23 22 18 7 8 Mean Importance 4.33 3.49 2.33 3.30 4.00 Score Eta - .63, df = 4.73, p less than .01 High scores indicate a high degree of importance (see Appendix H for the verbal anchors of these scales). 108 computed on the control group. (It was even higher for the experimental groups: .70 and .86 for groups A" and B" respectively. The difference between groups B" and C" is significantly greater than chance; we will discuss this result later.) Besides a reliability score for the attitude scale position, there is also a reliability for the importance scores. The test-retest reliability for the importance scores was +.71, computed on the control group. (For the experimental groups it was .46 and .61 for groups A" and B" respectively.) The terminal value scale had a test-retest reliability of .786 (median Spearman rho correlation) for the entire sample. This relia- bility estimate was approximately the same for each group: .796, .801, and .771 for groups A", B", and C” respectively. Since the design of Study 111 allowed for a retest of all 78 subjects on the value scale, we were able to determine the differences in value ranks for the high and low value groups at the time of the manipulation. These differences are shown in Table 23 (page 109). Table 23 should be compared to Table 20 (see page 100). If one compares Table 23 with Table 20, it can be seen that the difference between the mean rank of the value appealed to in groups A" and B" declined from 4.54 on the pretest to 3.06 on the posttest. Once again, this decline had been expected because of regression effects. This regression is comparable to what had been found in Study I (cf. page 47). And, as in Study 1, the question that remained to be answered was whether a mean difference of 3.06 ranks between the high and low value conditions was sufficient to produce a significant difference in the dependent variable. 109 Table 23. Mean posttest value ranks for subjects assigned to groups A" and B". (The means represent the mean rank for the value contained in that particular communication.) Value Condition Value in High (A") ’ Low (3") Communication N Mean N Mean .p A comfortable life 13 9.84 13 13.29 1.94* Freedom 13 5.07 13 7.75 1.73* Combined 26 7.46 26 10.52 2.24** *p less than .05 (one-tailed) **p less than .025 (one-tailed) 110 Data Applysis In Study III, the attitude scale responses were scored from 1 to 5, with "definitely yes" scored as a 1, "probably yes" as a 2, etc. Unlike the groups in Studies I and II, the three groups in Study III were matched on pretest attitude position (see Table 21).38 Therefore, in Study III, differences in posttest attitude positions between treatment groups were tested by an analysis of variance with proportional cell frequencies for pretest attitude position (see Myers, 1966, pages 102- 104). In addition to the attitude scale scores, the answers to the question, "How important is the above issue to you?" were also scored from 1 to 5. "Very important" was scored as a 5, "important" as a 4, etc. Since the three treatment groups were not matched with respect to their pretest importance scores, the pretest scores were subtracted from the posttest importance scores, producing a set of importance change scores (exactly analogous to the voting-intensity change data of Study I--see page 54). Hypoghesis I Hypothesis I predicted that groups A" and B" (combined) would have a less favorable attitude toward sliding scale tuition than would group C". The data relevant to this hypothesis are reported in Table 24 (page 111). As can be seen from Table 24, there were virtually no differences in the amount of change as a result of the communications. In contrast to Study 11, therefore, hypothesis I was not supported. 38As can be seen from Table 21, matching was not perfect. For purposes of statistical analysis the pretest category for two of the subjects in group C" was changed; one subject was switched from the "probably yes" to the "definitely yes" category, and one was switched from the "probably no" to the "definitely no" category. 111 Table 24. Posttest attitude toward sliding scale fee reductions: communication vs. control conditions. Pretest Posttest Mean Group N Mean* Mean* Change A" & B" 52 2.42 2.80 +.38 C" 26 2.42 2.82 +.40 Summary of analysis of variance for difference in posttest attitude scores: Source Sum of sgpares df Mean gguare F p Treatment .02 l .02 .025 n.s. Pretest 72.47 4 18.12** Interaction 1.69 4 .42 .523 n.s. Error 54.65 68 .80 Total 128.63 77 *High numbers represent opposition to the sliding scale tuition. The trend here is slightly in the reverse of the predicted direction, since the control group (C") became more opposed than the experimental group. **The F test for the pretest main effect would be highly significant. This simply reflects the fact that the posttest correlated with the pretest. 112 Hypothesip I; Hypothesis II predicted that group A" would have a less favorable attitude toward sliding scale tuition than would group B". The data relevant to this hypothesis are reported in Table 25, page 113. As can be seen from.Tab1e 25, the mean change as a result of the high value communication was much greater than for the low value communication. Group A" became significantly less (p (.01) favorable toward sliding scale tuition than did group B", and hypothesis II is supported. e o in H at eses I a d II At this point the results may seem a bit confusing. We argued in Studies I and II that an effective communication was a prerequisite for obtaining positive results for hypothesis 11. Of course, we had pointed out that the low value communication could be expected to be of minimal effectiveness, but still the only way to reconcile Tables 24 and 25 would be to say that the low value communication led to less attitude change than would be expected by chance. Could this have happened? Table 26 (page 114) shows that there can be little doubt that there was differential attitude change in the three treatment groups. The change in group A" is of very high significance, the change in group C" of borderline significance, and the change in group B" is far from significant. It would seem, then, that something about the low value comunication in Study III (as opposed to Study II) caused it to have either no effect or a negative effect. We will con- sider possible reasons for this in the discussion section. e f I e Our anticipated difficulty in testing the importance score was 113 Table 25. Posttest attitude toward sliding scale fee reductions: high vs. low value conditions. Pretest Posttest Mean Group N Mean* Mean* Change A" (high value) 26 2.42 3.11 1+.69 B" (low value) 26 2.42 2.50 +.08 Summary of analysis of variance for differences in posttest attitude scores: Source Sum of pgpppep gr Mpap sguare F 2 Treatment 4.92 l 4.92 6.34 .01*** Pretest 50.02 4 12.50** Interaction 3.24 4 .81 1.04 n.s. Error 32.59 42 .78 Total 90.76 51 *High numbers indicate opposition to the sliding scale tuition. **The F test for the pretest main effect would have been highly signi- ficant. This simply reflects the fact that the posttest correlated with the pretest. ***one-tailed 114 Table 26. Posttest attitudes for three treatment conditions. Pretest Posttest Mean p Group N Mean* Mean* Change .5 (one-tailed) A" (high value) 26 2.42 l 3.11 +.69 3.25 .001 B" (low value) 26 2.42 2.50 +.08 0.57 n.s. C" (control) 26 2.42 2.82 +.40 1.68 .10 Summary of analysis of variance for differences in posttest attitude scores for all three treatment groups: Source - Sum of ggugpes g; Mgap ggpare F_ 9 Treatment 4.95 2 2.42 3.35 .05 (two- tailed) Pretest 72.47 4 18.12** Interaction 4.91 8 .61 v 0.83 n.s. Error 46.50 63 .74 Total 128.83 77 *High numbers indicate opposition to sliding scale tuition. **The F test for the pretest main effect would have been highly signifi- cant. This simply reflects the fact that the posttest correlated with the pretest. 115 well founded. Group A", the high value group, changaisignificantly in the predicted direction, which would lead us to expect an increase in issue importance. But group A" also moved almost directly into the neutral region of the attitude scale (see Table 22), which would lead us to expect a decrease in issue importance.39 About all that we can conclude, therefore, is that the mean importance score should be greater than the 2.33 found for the neutral region on the pretest. Our expectancies for groups B" and C" are more clear-cut. Since group C" also moved toward the neutral region, it should simply show a decline in importance scores. And if, as we concluded at the end of the last section, the low value communication had no effect, then the importance scores for group B" should have shown no change. Table 27 (page 116) summarizes the shifts in importance for the three groups. Although the three groups did not differ signifi- cantly from each other, the data are fairly close to what we predicted above. The decline in importance for group C" is highly significant, while for groups A" and B" there is virtually no change. More importantly, if we consider only groups A" and C", the difference in mean importance change is significant Q; = 2.10, p<.025; not shown in Table 27). Thus there is evidence that the high value communication did affect the importance of the attitude to the subjects, as well as the attitude position. We must give one word of caution concerning the importance 39With reference to footnote 37, the distribution of scores for group A" remained unimodal, with the predominant response on the posttest being "no opinion." This is important, since had the shift in mean scores of 2.42 to 3.11 been a result of (say) subjects changing from "no opinion" on the pretest to "probably no" on the posttest, then we should expect an increase in the mean importance score even because of the so-called shift to the neutral region. 116 Table 27. Changes in importance of the sliding scale tuition issue by treatment conditions. Pretest Posttest Mean Group N Mean Mean Change t A" (high value) 26 3.25 3.33 1+.08 +0.10 B" (low value) 26 3.35 3.12 -.23 -0.35 C" (control) 26 3.60 3.05 -.55 -3.16* Summary of analysis of variance for differences in importance change scores for all three groups: Source Sum of squares df Mean square F Between 5.16 - 2 2.58 2.11 Within 88.63 75 1.18 Total 93.79 77 *p less than .005 (one-tailed) 117 scores. The mean posttest importance scores for groups A" and C" differed very little; the significant difference in the change scores may have resulted from the pretest differences. However, since subjects were randomly assigned to treatment conditions, there is no reason to expect any sort of regression effect. There is also some further supporting evidence for our contention that the attitude change in group A" was accompanied by an increase in issue importance relative to group C". In group A", subjects' attitude change scores correlated +.245 with their importance change scores. In other words, subjects who changed in the predicted direction tended to increase the importance of the attitude. In group C", on the other hand, these scores correlated -.284. While neither correlation is significant, the difference between the two correlations is signifi- cant (j_: - 1.80, df - 46, p (.05). This difference in correlations corroborates our argument that the process of attitude change differed in the two groups. The Effect of Perpoppl Interegt in the Issue After they had finished reading the communication on the sliding scale tuition, subjects were asked, "Do you have any personal interest in this issue? (e.g., Do you receive a fee reduction?)" As we mentioned earlier, we suspected that the communication may not be effective for subjects who actually benefited from the sliding scale tuition. In group A", there were seven subjects who received a fee reduction, while in group B" there were eight. Table 28 (page 118) summarizes the effectiveness of the communications for these subgroups. As can be seen from this table, it appears that receiving a fee reduction had very little effect on the effectiveness of the communications. In group 118 Table 28. Posttest attitude toward sliding scale fee reductions as a function of receiving tuition reductions. Subjects receiving Group Tuition reduction No tuition reduction ‘5 N a cha e N Mean change A" 7 +.83 l9 +.63 +0 .45 B" 8 -.12 18 1+.l6 -0.36 A" & B" 15 +.33 37 +.41 -0.18 119 A", there was even a slight tendency for these subjects to be more influenced by the communications. Discupsion At this point we will discuss only the results of Study III. Results from the previous two studies will be brought in only to the extent that they are relevant to understanding Study III. In the next chapter we shall try to integrate the results, implications, etc. of all three studies. The Apgumept for Pergeiyed Inspgumentality In Study III we have found evidence that attitude change as a result of a communication is a function of the importance of the value to which the new attitude is linked in that communication. When a subject read a communication which asserted that being-against- a-sliding-scale-tuition-plan was consistent with valuing p pomfortable lips, the extent to which he adopted this attitude was shown to be a function of his own value for a comfortable life. (This was shown to be the case with respect to freedom.also.) Of course, I have presumed here that the subject accepted the assertion of the communication. Some subjects, no doubt, did not agree that the issue was in fact related to the particular value. But there is no reason to expect that this tendency, to accept or not accept the reasoning of the communication, should differ in the two value conditions. This point deserves further consideration. It is important that both the high and low value groups accepted the communication to the same extent. Our perceived instrumentality hypothesis is based on the assumption that the strength of the association between the attitude 120 and the value be constant in the high and low value conditions (see pages 22 and 23). If we should find that the subjects in group B" had simply rejected the message, the results would take on a slightly different interpretation (e.g., the function of the value aspect of the communication was simply to get subjects' attention; the more important the value, the more attention and hence the more attitude change). I do not believe that we can completely rule out such alternative interpretations, but we can point to some evidence which corroborates our viewpoint. Subjects were required to judge the communications of four ll-point scales-~clarity, fairness, persuasiveness, and influence. If subjects in the low value condition simply did not accept the communication, then these scales might be expected to detect the difference between the two value conditions. Table 29 (page 121) summarizes the differences between groups A" and B" with respect to the four scales. While none of the differ- ences are significant, in all four cases group B" judged the communica- tion in the more favorable direction. With respect to the scale which we were most interested in, measuring judgments of clarity, group B" was almost significantly in the more favorable direction (i.e., judging the communication as more clear).40 Thus these scales fail to show any evidence against the perceived instrumentality hypothesis. We have already pointed out, in the results section, that the shifts in the importance scores (Table 27) supported our reasoning that the attitude change in group A" was caused by linking the new 401a the next chapter we will consider the possibility that this may be a replicable phenomenon. 121 Table 29. Mean judgments8 of communication by value condition. Scale Group Clear Fair Persuasive Influence A" 2.96 6.21 5.88 6.63 B" 2.25 6.12 5.79 5.96 1'. ratio - -1.85* -0.11 -0.09 -0.78 *p less than .10, two-tailed. a) Low scores indicate a high value on that scale; see Appendix L for the verbal anchors of these scales. 122 attitude to an important value, and this evidence will not be discussed further. Before concluding, however, that Study III has completely supported the perceived instrumentality hypothesis, we should try to explain what happened to group B". The Ineffegtiyenesg of the Low Value Communigatioh I do not know whether the low value communication simply had no effect, or (as the trend would suggest) a negative effect. Because of this I do not wish to speculate to any great length on the reason for its ineffectiveness. Rather I will confine myself to suggesting a few simple hypotheses. First of all, a problem is involved because Study II showed the low value communication to be effective. Study II, however, took place under quite different circumstances. For example, we pointed out (page 90) that there was a six month gap between the pretest of subjects' values and the sending of the value letters. This is in contrast to an average gap of less than four weeks in Study III. Thus we could argue that we had a much more reliable measure of high and low values in Study III compared to Study II. However, this can only be a partial explanation, since in Study II groups A' and 3' combined showed significantly more attitude change than group C', while in Study III group C" showed slightly more attitude change than the combined groups A" and B". It would seem that there would have to be some factor Operating in Study III to inhibit attitude change. I can think of two possible factors (which complement rather than compete with each other). We 123 have already mentioned the concept of "evaluation apprehension." According to Rosenberg (1965), the more obvious it is to a subject that he is being invited to change his attitudes, the more resistance he will show to that change. This is the explanation for our results which I tend to favor most. I would hypothesize that this factor was equally present in both groups, A" and B", and that in group A" the effect of the high value communication was sufficient to overcome it. To support this, there is some evidence that group B" was actively resisting change, in comparison to group C". A We mentioned on page 108 that for group B" the attitude measure had a pretest-posttest correlation of +.86, while for group C" this correlation was only +.65. This difference is statistically significant (p - 1.79, df - 46, p<.05). If one accepts the premise that +.65 represents the degree to which a group would be expected to change (non- systematically) by chance, then one is led to conclude that group B" was caused not to change.41 There is a second possible explanation for the group B" result which should not go unmentioned. We were unable to measure attitude change over time (e.g., from the immediate posttest to a three day posttest as was used in Study II). It is possible that the effective- ness of a low value communication may increase over time, just as some- times happens with a communication from a low prestige source (see Hovland and Weiss, 1951, for a discussion of this "sleeper effect"). This could happen if after a few days subjects recalled the basic tone of the low value communication, but dissociated the relatively unimportant 41In and of itself a high test-retest correlation does not indicate the absence of change. In this case, however, we have the additional fact that the mean attitude score for group B" did not change. 124 value to which the attitude had been linked. This possibility should, of course, be investigated. If true, it would not negate our perceived instrumentality hypothesis, so much as refine it (just as the "sleeper effect" studies refined hypotheses concerning source credibility). Limitatiopp of Study II; As a matter of formality, we should point out some limitations to ,the generalizability of Study III. One should be careful in generaliz- ing to non-college populations, other types of issues, other procedures for presenting the communications, and more delayed posttests. And, of course, we did not measure behavior and cannot confidently generalize with respect to the "depth" of the attitude change. With the end of sliding scale tuition at Michigan State, it may even be that Study III itself is not replicable. All of the limitations just mentioned should not be attributed to oversight on the part of the experimenter. I did not wish to induce permanent changes in the subjects in Study III. Sliding scale tuition is, I think, an issue where valid arguments can be made both for and against. Although no deception had been used, I felt that an immediate debriefing was necessary because subjects seemed to react to all the communications which they read as if they were more authoritative than they actually were. (Besides the intensive debriefing which twelve subjects received--see page lO3--all subjects were told of the nature of the experiment.) Im rta e of tud I On the other hand, the limitations listed in the previous section should not be taken too seriously. They are the same limitations that 125 may be applied to virtually all attitude change studies. Evidence on the attitude change process must, of necessity, be built up from a series of studies. Study III has shown that it is valid to think of attitudes as being a function of a person's perception of how the object of the attitude relates to his own value system. It has done this by showing that to the extent that a communication links an attitude to a relatively important value for a certain individual, as opposed to a relatively unimportant value, the individual will tend to adopt that attitude. This is not to say that we have shown all attitudes to be linked to a person's values, nor that all attitude change is a function of changed perceived instrumentality. Rather we have shown that this is a valid way to view some attitudes. The generality of the phenomenon awaits to be established. CHAPTER VII CONCLUSION m In the preceding chapters we have reported the results of three studies. Studies I and II were field studies which concerned a specific election issue: the graduated income tax. Study III was a laboratory study which concerned the attitude of college students toward a sliding scale tuition. In each of these studies we attempted to demonstrate a functional relation between values and attitudes. Mere specifically, we hypothesized that a person's attitude on a given topic was determined by the perceived instrumentality of the attitude object with respect to that person's value system. Each of the three studies was designed to show that by manipulating perceived instrumen- tality, attitude change would follow. Study I was conducted at the time of the actual vote on the graduated income tax. It failed to produce any evidence supporting the hypothesis. Study II was conducted a few months after the vote. It confirmed the fact that communications which were focused on values could be effective stimuli for attitude change. But there was no evidence directly supporting the perceived instrumentality hypothesis. In other words, the communications could have produced attitude change for reasons independent of the values they referred to. 126 127 In Study III, the laboratory study, strong evidence was found in support of the perceived instrumentality hypothesis. It was found that a communication linking an attitude object to a relatively impor- tant value produced significantly more attitude change than did a communi- cation linking that object to a relatively unimportant value. Since the only difference between subjects receiving the high value communication and subjects receiving the low value communication was the importance of the value referred to, we were able to conclude that attitude change could be considered a function of changes in perceived instrumentality. The three studies produced other results and raised other issues. For the most part, these have already been discussed throughout the previous six chapters (see pages 69 to 75, 88 to 90, and 119 to 125). In the remainder of this chapter we will consider those issues which we have put off until the data of all three studies were available. Also, we will briefly point out some questions that have been raised or left unanswered by the studies. Finally, we shall sum up*what we believe to be the major significance of the studies. Labopptory Vergpp gield Studies If nothing else, this dissertation has demonstrated again the differences between field and laboratory research. On page 16, we stated that one of the two main purposes of these studies was to deter- mine whether the functional relation between values and attitudes (and behavior) could be found in a field situation. Despite the promising results of the pilot study (see pages 35-38) and of Study II, we have not been able to show this functional relationship. At best we have shown that attitude change can be experimentally produced in a field 128 situation. It will have to remain for future research to see if this attitude change can definitely be tied to values. There are several ways to explain why the perceived instrumentality hypothesis should have worked in the laboratory situation but not in the field. We have discussed most of these reasons in Chapter VI. The explanation which I favor most, is that subjects paid much greater attention to the communications in the laboratory situation.42 (This could also be expressed in terms of motivation to attend to the communi- cation.) In other words, I do not believe that the effect which was found in Study II was due to the values directly appealed to in the communications. It seems more likely to me that the more general argument of the letters, rather than their specific value references, accounted for the attitude change in Study II. This is not to say that "values," broadly speaking, might not explain the attitude change that did occur; the general information in the high and low value letters could still have been evaluated in terms of subjects' value systems. What could be done to increase subjects' attention in a field situation? It seems to me, in retrospect, that it would require at least personal contact between subject and communicator, so that the communicator could lay proper emphasis on the relevant value. There were enough differences, however, between Study II and Study III that one could reconcile the results and still maintain that the 42In effect, I am hypothesizing levels of attention in the three studies. Subjects paid least attention in Study I, resulting in no effect for the communication. Subjects paid an intermediate amount of attention in Study II, allowing the communications themselves to have an effect, but not allowing the importance of the value to have an effect. Sub- jects' attention was greatest in Study III, resulting in a difference between high and low value communications. 129 value letters (in Study II) were effective because of the specific value appealed to. To me, the most interesting possibility is that the low value communication in Study II was subject to a "sleeper effect," which caused it to increase in effectiveness over time (see page 123). A second possibility is that the difference in the importance of the value appealed to in the high and low value groups had attenuated too much for a differential effect to be produced (in Study II). Pragtical Applications Especially after the conclusion of a field study, there is a temptation to look for practical applications for the results. We pointed out earlier that our major concern was theoretical, rather than practical. For example, we had no desire to construct some sort of ideal paradigm which could be used to produce attitude change. Still, the fact that the pilot study and Study II were effective in producing attitude change, while Study I was not, does suggest a few generalizations. Nemely, it would seems-according to the pilot study-- that the earlier a communication reaches its recipient during a political campaign, the more effective it is likely to be. This may be because people do not yet have well-formed attitudes, or because they are not being subjected to several other messages that are competing for their attention. In Study II, of course, subjects had relatively well-formed attitudes. In this case the effectiveness may have been due simply to the lack of competing messages, or to the fact that Study I sensitized the subjects to the issue. Of course, if one is concerned with influencing voting it will do little good to change attitudes a few months before or after the election. 130 The strategy that is suggested, then, would be to establish the desired attitude as soon as possible, and then try to maintain the attitude change by subsequent communications. Also, the data on voting in Study I suggest that one had better anticipate the exact form which the issue will take (see pages 71-72). The above considerations concern mainly the timing of a communi- cation. Cen we say anything about the value content-~again, with a view toward practical applications? Even if there had been a statistically significant difference in attitude or behavior change between the high and low value groups, the information which we have does not suggest that multiple communications would be effective. The practical diffi- culty of measuring each subject's values, coupled with the relatively minor difference in importance between high and low values (see Table 4, page 49) would seem to make multiple communications virtually prohibi- tive. (There would be the added practical difficulty that if this was attempted on a large scale, people would eventually compare communica- tions.) A much more practical approach would be to ppmpTg,people's values within a target area, and then construct a single communication directed toward the more important values of the entire population. Pragtipal Applipatiops and the Qpestion of gehavior One final point should be mentioned with respect to practical applications. No behavior change was observed in any of the studies (only Study I attempted to measure behavior change). Since Study I found that subjects' pretest attitudes were relatively poor predictors of their subsequent voting behavior-~for whatever reason--it would be 131 quite unjustified to assume, for example, that the attitude change in Study II would have led to behavior change. On the other hand, at the time of Study II subjects had experienced the voting situation, and most of the reasons that we suggested for the low pretest-voting correlation (pages 71-72) no longer applied. Although I myself feel relatively con- fident that the attitude change would have been accompanied by behavior change, this feeling lacks any empirical support. I sues ised° The e evance of onsiste Theor We have already discussed several of the issues raised by the three studies: possible sleeper effects, evaluation apprehension in Study III, behavior change, etc. One issue, which we discussed briefly in Chapter VI, concerns other (non-attitudinal) effects of the high and low value communications. (See also footnote 5, page 6). Even though the design of our study does not allow us to adequately test for these effects, they are still worth considering. The effects which I am referring to are best approached from a consistency theory point-of-view. Osgood, Suci, and Tannenbaum (1957) have presented one of the more popular models for analyzing consistency effects. According to their congruity principle, when one links two objects or concepts which differ in their evaluation by a given subject, the subject will tend to change his evaluation of those concepts in order to make them congruent with each other. Thus, for most of the experimental group subjects in Study III, our design linked a negatively evaluated concept (elimination of sliding scale tuition) with a 132 positively evaluated one (either fpeedpp or p gomfortablg life).43 What should happen as a result? First of all, because of the difference in value importance, the disparity between the concepts was greater for group A" than for group B". Therefore, there should have been more attitude change in group A“--as we did, in fact, find. However, the congruity principle also points to another possible effect. The disparity between attitude object and value should also affect the rank of the value--theoretically, by causing it to decline in importance (in order to balance the relationship). And, once again, this effect should be greater in group A". We were able to make an indirect check for this by subdividing groups A" and B" on pretest attitude. The logic of this check was as follows: subjects whose pretest attitudes were not in favor of sliding scale tuition should not have had their values affected by the communication (i.e., they should have experienced less incongruity) nearly as much as subjects who supported sliding scale tuition. The results of this check showed that there was virtually no difference between the subgroups. In each case the trend was slightly in the reverse direction; in group A", for example, subjects who were initially against the issue lowered the relevant value slightly more than did subjects who initially supported the issue. This finding, however, does not necessarily contradict the work of Osgood et. a1. Tannenbaum (1956) found that with highly polarized 43Technically the communications linked two positive concepts via a negative assertion. The same effect is supposed to result. It is not our intention to give a thorough exposition of the congruity principle. Rather, we are adapting it to describe a problem that arose in interpreting the data of Study III. 133 concepts (which values usually are) it was necessary to add a "correction for incredulity." In our case, this same p§.hpp,correc- tion factor simply means that group A" should have negatively evaluated the communication (relative to group B"). On page 120 we have already summarized the differences between groups A" and B" with respect to their reaction to the communication itself. The data in Table 29 (page 121) provide fair support for the hypothesis that group A" would react more negatively to the communication than would group B".44 lIn some cases it may even be that a high value communication.would be negatively evaluated to such an extent that it would result in less change than a "low" value communication. (Sherif and Hovland, 1960, point to a similar phenomenon with "assimilation" and "contrast" effects in attitude change.) I To summarize this section, our data do not necessarily rule out a congruity principle interpretation of our effects. It will have to remain for future research to determine whether the two approaches can be distinguished on the basis of some differential prediction. Sigpifigapce of Stpdieg Conflicting and tentative results aside, the three studies pro- duced two results of importance. The first result, which is mainly of methodological interest, is the demonstration of experimentally produced attitude change in a field situation with a non-trivial issue. This attitude change would be difficult to account for on the basis of demand characteristics, evaluation apprehension, or any of the artifacts that 44We refer to this as an "hypothesis" somewhat hesitatingly. While this result is clearly predicted by most balance theories, the relevance of balance theory was realized only after the once puzzling results of Table 29 were known. 134 often plague laboratory research. The second important result is of theoretical interest. We have shown, in Study III, that there is a causal relationship between values and attitudes. More specifically, we have shown that a person's attitudes can be viewed as a function of how he views the attitude object as a help or hindrance in attaining his values. I would say that there are two steps which should logically be taken next. Both relate to the extensiveness of the perceived instru- mentality relationship. It must first be detemmined if the perceived instrumentality relationship can be demonstrated in a field situation. The other step that should be taken, is to try to determine the extent to which a person's attitudes are a function of perceived instrumentality, rather than of some other process (e.g., a more basic conditioning process). REFERENCES Academip handbook (full title: Schedule of courses and academip handbook), 1968, 9;, East Lansing: Michigan State University. Barton, A. Measuring the values of individuals. Religious Education, 1962, Q}, 62-97. Burnstein, E., & McRae, A. V. Some effects of shared threat on prejudice in threatened groups. Journal of Abnormal and Social Psychology, 1962, p4, 257-263. Buros, O. K. The i th me tal meas ements ea book. Highland Park, New Jersey: Gryphon Press, 1959. Burt, C. The psycho10gy of value. British gougnal of Statiptigal Egychology, 1963, Th, 59-104. Carlson, E. R. Attitude change through modification of attitude structure. Joupppl of Abnormal and §ppia1 ngghology, 1956, QT, 256-261. Campbell, A., Converse, P. E., Miller, W. E. & Stokes, D. E. The american voter. New York: Wiley, 1960. Campbell, D. T. Social attitudes and other acquired behavioral dispositions. In S. Roch (ed.), Psychology; A study 0: a cience, Vol, 9. New York: McGraw-Hill, 1963. Chein, I., Deutsch, M., Hyman, H. & Jahoda, M. (eds.) Consistency and inconsistency in intergroup relations. Jourpal og fiogial Issuep, 1949,,2, 1-63. Cochran, W. G. Some methods for strengthening the common chi-square tests. Tiometrics, 1954, 19, 417-451. Cohen, A. R. Attitude change apd social influence. New York: Basic Books , 1964 a De Fleur, M. L. G‘Westie, F. R. Verbal attitudes and overt acts: An experiment on the salience of attitudes. Amepipan §ociological geyieg, 1958, T;, 667-673. Doob, L. W. The behavior of attitudes. Psychologipal Revieg, 1947, §_4_, 135-156. 135 ' p . O ' 5 . e C ' m c ( _ 1 m. ' ' I Q . if. .- . _ 1 , . , s - O '5' - ’ -1 , w _ -_. . . i T I I e s . . . Q ,7 ‘ >7 _. _ . , - . . 1-- _. O I v I C . . . . C . \ e . s m I a ' ' ‘ s I f ‘ 1 A ‘ a . s ' s ,., . . e a . C 1 . ’7 o g C O a . o I ' , V -’—‘7 . V V O . a _ . . I u . s o 136 Dukes, W. F. Psychological studies of values. Psychological gulletip, 1955, QT, 24-50. Evans, R. Personal values as factors in anti-Semitism. Japgppl OT Abnormal and Social Psychology, 1952, 41, 749-756. -Festinger, L. Behavioral support for opinion change. Publip Opipion Qparteply, 1964, gs, 404-417. Harding, J., Kutner, B., Proshansky, H., & Chein, I. Prejudice and ethnic relations. In G. Lindzey (ed.), Handbook of social psych010gy. Reading, Massachusetts: Addison-Wesley, 1954. Hays, W. L. Statispics for pgyphologigts. New York: Holt, Rinehart, and Winston, 1963. Hill, W. F. Learning theory and the acquisition of values. Ps cho o ical Review, 1960, 91, 317-331. Homant, R. The meaning and ranking of values. Unpublished M. A. thesis, Michigan State University, 1967. Homant, R. &.Rokeach, M. Value for honesty and cheating behavior. PerpOpality, 1970, T(2), 153-162. Hovland, C. I. Reconciling conflicting results derived from experi- mental and survey studies of attitude change. Americpn Psychologist, 1959, T4, 8-17. Hovland, C. I. (ed.) The order of ppeseppatioh ip persuasiop, Vol. 1. New Haven, Connecticut: Yale University Press, 1957. Hovland, C. I. & Janis, I. L. Perponality and perpuasibility. New Haven, Connecticut: Yale University Press, 1959. Hovland, C. I., Janis, I. L. & Kelly, H. H. Qommgnicagioh and Persuagioh; Psycholongal gtudies of opipiop phange. New Haven, Connecticut: Yale University Press, 1953. Hovland, C. I., Lumsdaine, A. A. & Sheffield, F. D. Expemeentg on mass cpmmpnication. Princeton, New Jersey: Princeton University Press, 1949. Hovland, C. I. & Weiss, W. The influence of source credibility on communication effectiveness. Public Qpinion Qparterly, 1951, Li, 635-650. _ Krech, D., Crutchfield, R. S. & Ballachey, E. L. Individual in society. New York: McGraw-Hill, 1962. La Piere, R. T. Attitudes vs. actions. Social Forces, 1934, T;, 230-237. 137 Lazarsfeld, P. F., Berelson, B. & Gaudet, H. ,Ih§_pggplgL§_ghgigg. New York: Columbia University Press, 1944. Logan, F. A., Olmstead, D. L., Rosner, B. S., Schartz, R. D. & Stevens, C. M. Behavior theory and sogial science. New Haven, Connecticut: Yale University Press, 1955. Martin, W. E. Learning and identification: III. The development of values in children. Journal of Genetic Ps cholo , 1954, 53, 211-217. Myers, J. L. Fundamentals of experimental design. Boston: Allyn and Bacon, 1966. Orne, M. T. On the social psychology of the psychological experiment: With particular reference to demand characteristics and their implications. American Psychologist, 1962, ll, 776-783. Osgood, C. E., Suci, G. J. & Tannenbaum, P. H. The measurement of meaning. Urbana: University of Illinois Press, 1957. Penner, L., Homant, R. & Rokeach, M. Comparison of rank-order and paired-comparison methods for measuring value systems. Perceptual and Motor Skills, 1968, Tl, 417-418. Piaget, J. Will and action. Bulletin of the Menninger Clinic, 1962, 2'6", 138- 145 e Rapoport, A. Fights, games and debates. Ann Arbor: University of Michigan Press, 1960. Rokeach, M. Beliefs,_attitudes,,and values. San Francisco: Jossey-Bass, 1968. Rosenberg, M. J. Cognitive structure and attitudinal affect. our of Abnormal and Social Psychology, 1956, 2;, 367-372. Rosenberg, M. J. An analysis of affective-cognitive consistency. In M. J. Rosenberg and C. I. Hovland (eds.), Attitude orgapizatiop and change. New Haven, Connecticut: Yale University Press, 1960. Rosenberg, M. J. When dissonance fails: On eliminating evaluation apprehension from attitude measurement. Journal of Personality and Social Psychology, 1965, T, 28-42. Sanai, M. The relation between social attitudes and characteristics of personality. Journal of Social Psychology, 1952, Th, 3-13. Sherif, M. & Hovland, C. 1. Social judgpent; Assimilation and contrast effects ip communication and attitude change. New Haven, Connecticut: Yale University Press, 1961. A o .. o q s . a a 0 ~ . d a o a o , a a , . . - 138 Smith, M. B. Personal values as determinants of political attitude. Journal of Pa cholo , 1949, Th, 477-486. Tannenbaum, P. H. Initial attitude toward source and concept as factors in attitude change through communication. Public Opinion Qparterly, 1956, 29, 413-425. Taylor, D. W. Toward an information processing theory of motivation. In M. R. Jones (ed.), Nebraska symposium in motivation. Lincoln: University of Nebraska Press, 1960. Woodruff, A. D. & Divesta, F. J. The relationship between values, concepts, and attitudes. Educational and Psychological Mpasurement, 1948, 8, 645-650. APPENDIX A PRETEST QUESTIONNAIRE Name Address Below is a list of questions. Below each question is a list of answers to those questions. Please put an "X" in front of the answer that best expresses your opinion. I have tried to ask the following questions as simply as possible. In doing this I may have over-simplified a few of the complex issues. Please feel free to write in any comments which you may feel are necessary to clarify your answers. 1. In general, do you feel that taxes should be reduced? Definitely yes Probably yes No opinion Probably no Definitely no 2. Do you feel that the sincerity of the various candidates is a major issue in the coming elections? Definitely yes Probably yes No opinion Probably no Definitely no 139 140 3. WOuld you be in favor of making Michigan's income tax into a graduated income tax? (A graduated income tax is one where people with higher incomes pay a higher rate of tax than do people with lower incomes.) Definitely yes Probably yes No opinion Probably no Definitely no b . Are you in favor of daylight savings time? Definitely yes Probably yes No opinion Probably no Definitely no U'l . Do you think that the economy of the country is in good condition? Definitely yes Probably yes No opinion Probably no Definitely no O‘ . Do you think that the United States has been too active in foreign affairs? Definitely yes Probably;yes No opinion Probably no Definitely no 7. Do one or two issues stand out as being quite important to you in this election year? Yes No 141 If you answered "yes", please check below those issues which seem particularly important to you: Riots in the cities Civil rights for minority groups War in Vietnam Inflation, high prices Increasing taxes Other (write in) 142 Below is a list of 18 values arranged in alphabetical order. Ybur task is to arrange them in order of their importance to YOU, as guiding principles in YOUR life. Study the list carefully. Then place a T_next to the value which is most important to ypp; place a T.next to the value which is second most important to you. Continue to number the values until you have placed an T§Din front of the least important value. Work slowly and think carefully. If you change your mind, feel free to change your answers. The end result should truly show how you really feel. A COMFORTABLE LIFE (a prosperous life) _____AN EXCITING LIFE (a stimulating, active life) _____A SENSE OF ACCOMPLISHMENT (lasting contribution) _____A WORLD AT PEACE (free of war and conflict) _____A WORLD OF BEAUTY (beauty of nature and the arts) _____EQUALITY (brotherhood, equal opportunity for all) _____FAMILY SECURITY (taking care of loved ones) ‘____;FREEDOM (independence, free choice) _____HAPP INESS (contentedness) INNER HARMONY (freedom from inner conflict) MATURE LOVE (sexual and spiritual intimacy) NATIONAL SECURITY (protection from attack) _____PLEASURE (an enjoyable, leisurely life) _____§ALVATION (saved, eternal life) .____JSELF-RESPECT (self-esteem) _____SOCIAL RECOGNITION (respect, admiration) _____TRUE FRIENDSHIP (close companionship) WISDOM (a mature understanding of life) 143 It will be helpful to me if I have the following information about my respondents. However, I DO NOT HAVE TO HAVE this information. If any, or all of the following questions seem to be too personal, do not answer them. Your sex Age Race or nationality Check one: Number of children Married Widowed Divorced Single Your occupation Your husband's (or wife's) occupation Approximate family income $ Circle the highest grade of school you have completed: (grade school) 1 2 3 4 5 6 7 8 (high school) 9 10 11 12 (college) 1 year, 2 years, 3 years, 4 years, more than 4 years Political party preference: Democrat Republican Independent (no preference) Religious preference APPENDIX B VALUE LETTERS USED FOR MAIN STUDY (STUDY I) 1616 Cambria Drive East Lansing, Michigan October 30, 1968 We are writing to you because we are concerned about everyone having an equal opportunity in our society. This coming Tuesday, November 5, you will vote on an issue that could very well be of life-long importance to you. Because this issue will be overshadowed by the presidential election, we feel that there is a need to call your attention to it, and to remind you of some facts which you may have overlooked. We are speaking about a proposed constitutional amendment which-- if it is passed--will allow the state of Michigan to adopt a graduated income tax. We urge you to vote Ln_£ayp;_gf,this amendment. The tax law which we have at present, where everyone pays the same percentage of his income, tends to deny the poor man his right to an equal opportunity to participate in our expensive society. With a graduated income tax the state of Michigan will be able to get more of its income from those who can best afford to mm W ;__o ‘ta 0: _g_e 1 :d_:_ ‘0 >10u‘ :J I o 11‘: _. {1' ‘s .1 ‘_: ‘- ucc -0- e -. 096 :- tli, e - o-e ._1 e_-i - 2. --_; o-.- ,. Wien- Help guarantee equal opportunity for all. Vote in favor of the graduated income tax amendment next November. Sincerely yours, Douglas Robinette Ad hoc committee for fair tax laws 144 145 1616 Cambria Drive East Lansing, Michigan October 30, 1968 We are writing to you because we are concerned about security for every individual and his family. This coming Tuesday, Nbvember 5, you will vote on an issue that could very well be of life-long importance to you. Because this issue will be overshadowed by the presidential election, we feel that there is a need to call your attention to it, and to remind you of some facts which you may have overlooked. We are speaking about a proposed constitutional amendment which-- if it is passed--will allow the state of Michigan to adopt a graduated income tax. We urge you to vote 1n_£§ypz_gf_this amendment. A graduated income tax taxes wealthier people at a higher rate. What this means to you and to society as a whole, is that tax revenue for the state of Michigan will be supplied by money that would have been spent for luxuries rather than necessities. We do not feel that luxuries are bad, but we do feel that the security of the taxpayer and his family comes first. WWW security is exactly what a graduated income tax would provide for. Help guarantee security for the taxpayer and his family. Vote in favor of the graduated income tax amendment next Nevember. Sincerely yours, Douglas Robinette Ad hoc committee for fair tax laws 146 1616 Cambria Drive East Lansing, Michigan October 30, 1968 We are writing to you because we are concerned about the right to pursue a comfortable and prosperous life. This coming Tuesday, November 5, you will vote on an issue that could very well be of life-long importance to you. Because this issue will be overshadowed by the presidential election, we feel that there is a need to call your attention to it, and to remind you of some facts which you may have overlooked. We are speaking about a proposed constitutional amendment which-- if it is passed--will allow the state of Michigan to adopt a graduated income tax . We urge you to vote againgt this amendment. With a graduated income tax, the more a person earns, the higher the percentage of taxes he must pay. ev e i i ls I‘ "I: " ' '.'. .Oili’ ' 3 'Ll' 3" 1!! '9-‘"‘~ ' £9;_;hgmsglygg. ‘Why should someone earning $10,000 a year have to pay out 5% of his income, while someone earning $5,000 a year pays only 2%? We do believe, of course, that the person earning more should pay more money, but he should not have to pay an even greater percen- tage of his income. Do not penalize someone who legitimately seeks a comfortable and prosperous life for himself. Vote against the graduated income tax next November. Sincerely yours, Douglas Robinette Ad hoc committee for fair tax laws 147 1616 Cambria Drive East Lansing, Michigan October 30, 1968 We are writing to you because we are concerned about the freedom and independence of the individual. This coming Tuesday, November 5, you will vote on an issue that could very well be of life-long importance to you. Because this issue will be overshadowed by the presidential election, we feel that there is a need to call your attention to it, and to remind you of some facts which you may have overlooked. We are speaking about a proposed constitutional amendment which-- if it is passed-~will allow the state of Mflchigan to adopt a graduated income tax. We urge you to vote against this amendment. With a graduated income tax the government imposes a higher rate of tax on people as their incomes increase. The issue does not concern whether wealthier people can afford to pay more. They can and should pay more. But they pay more under the flat-rate tax that we now have, where everyone pays the same percentage of his income. The reason that we are against the graduated income tax is that it gives the government the power to place a penalty on increasing one's income. It is jupt apother pase of the governmept limiting the freedom of the individual to pugsue his owp goals. Do not increase the power of the government of Muchigan to limit the freedom of the individual. Vote against the graduated income tax next November. Sincerely yours, Douglas Robinette Ad hoc committee for fair tax laws APPENDIX C LETTER SENT TO ACCOMPANY PRETEST QUESTIONNAIRE* MICHIGAN STATE UNIVERSITY East Lansing . Michigan 48823 Department of Psychology . Olds Hall Dear I am a doctoral candidate at Michigan State University. To fulfill the requirements of my studies, I am conducting a scientific research project on the 1968 elections. By filling out the enclosed questionnaire, you will help me in my attempt to find out what the public really thinks about election year issues. In order for my research to be successful, it is very important that everyone return the questionnaire. NO one is expected to have an informed Opinion on all of these questions. Even if you have no idea how to answer any question, your answers are still very important to me. Your name and address were taken from Lansing's voter registration lists, which are open to the public. Your answers will be kept strictTy ponfidential; I use your name only for the purpose of identi- fication. I realize that I am asking for about a half an hour of your time, plus the trouble of mailing back the questionnaire (a stamped return envelope is provided). The return benefit to you is indirect, but if my research is successful, I feel that it will lead to a better informed public--something that will benefit all of us. If you have any questions about the purpose of this research you may call me at anytime. My Office phone is 355-3441 and my home phone is 351-7326. Sincerely yours, Robert Homant *Letter was sent out on August 14, 1968. 148 APPENDIX D LETTER SENT TO REMIND ORIGINAL SAMPLE OF 500 SUBJECTS TO RETURN THE PRETEST QUESTIONNAIRE The following letter was sent out on August 23, 1968 to remind the original sample of 500 subjects to return the pretest question- naire. The letter was mimeographed on Michigan State University Psychology Department stationery (5% x 8%). Dear Mr(s). I am the Michigan State University graduate student who sent you a questionnaire last week. I realize that you are very busy and may have forgotten about it. This note is just a brief reminder that it would be very helpful to me if you return the questionnaire. Sincerely yours, Robert Homant 149 APPENDIX E HYPOTHETICAL EXAMPLE OF THE ASSIGNMENT OF VALUE LETTERS TO INDIVIDUAL SUBJECTS Suppose that in group A (to receive a high effective letter) there were six subjects who were to receive either the e t or family segupity letter (determined by a pretest attitude against the graduated income tax).* Suppose further that the ranks which these subjects had assigned to equality and family security were as follows: Subject Equality Family Security 1 12 9 2 6 9 3 1 2 4 5 4 5 l 17 6 10 12 Our first step would be to pick one of the letters at random, say the equality letter. Then we would assign it to the subject with the highest rank Of equality, either subject 3 or subject 5. We would assign it to subject 5, since of the two subjects he had the lower rank Of family security. Our second step would be to assign the TmeTy_ security letter to the one of the remaining five subjects who had the highest rank of family security--subject 3. Our third step would be to assign an equality letter to subject 4. The fourth step would be to assign the family security letter to subject 1, since he had *In Studies II and III, pretest attitude did not have tO be taken into account. 150 151 ranked gqpp1T§y_lower than subject 2 (who had just as high a rank of W). The fifth step would be to assign an EMU-LY. letter to subject 2, and finally we would assign the TmeTy_§pgh;15y, letter to subject 6. If this had been a group to receive a low effective appeal (group B or D), the Opposite letter would have been assigned in each C888. APPENDIX F POSTTEST QUESTIONNAIRE SENT TO SUBJECTS WHO COULD NOT BE REACHED BY TELEPHONE MICHIGAN STATE UNIVERSITY East Lansing - Michigan 48823 Department of Psychology . Olds Hall NOvember 6, 1968 Dear Mr(s). I am the Michigan State graduate student to whom you recently returned a four-page questionnaire. I appreciate your response, but in order for me to complete my study I must ask a few more questions. I hope that you will check the appropriate places below and return this to me. I had hoped to reach you by telephone to save you the trouble of having to mail this back to me, but I was unable to. The questions that I wish to ask are: Were you able to vote in Tuesday's election (Nevember 5)? Yes No How did you vote on the graduated income tax amendment? (If you did not vote, please check how you think you would have voted). I voted in favor of it (or, I would have voted in favor of it). I voted against it (or, I would have voted against it). Was your decision on the graduated income tax amendment rather definite, or were you doubtful? _____Definite Doubtful Thank you very much, this completes my study. Sincerely, Robert Homant 152 APPENDIX C TRANSCRIPTS OF THREE OF THE 147 POSTTEST TELEPHONE CONVERSATIONS* Note: All three transcripts began identically--see page 44. The transcripts reported in this appendix begin with the question, "First of all, were you able to vote Tuesday?" *It was impossible, Of course, to find "typical" conversations. The ones reported here were selected because they also exemplified other points. Transcript #1. Experimenter: First of all, were you able to vote Tuesday? Subject: Ya, uh hum. Experimenter: Then would you mind telling me how you voted on the graduated income tax amendment? Subject: We voted against it. (Subject was a housewife.) Experimenter: At the time you voted were you fairly sure of your decision or-- Subject: Yes. Experimenter: Then was there any particular reason why you were against the graduated income tax? Subject: Well, we just thought it was going to let every little township or--you know, too many governments--suburbs and everything--it would allow them to put on an income tax. I don't think that's a good way to do it. Experimenter: I'm particularly interested in what influenced your decision. Did you see or hear anything at all-- Subject: Well, we read the newspapers quite a bit--and we don't just go by what they say, but we try to get information from it. TV, a, the people themselves who are campaigning I don't pay much attention to because they promise every- thing. Pay more attention to editorials. 153 154 Transcript #2. Experimenter: First of all, were you able to vote Tuesday? Subject: I did. And I did not lose my vote. Experimenter: Then would you mind telling me how you voted on the graduated income tax amendment? Subject: I voted no. Experimenter: At the time you voted were you fairly sure Of-- Subject: Well, I didn't understand it too well, and a, I don't know, it seems like we're taxed tO death. Experimenter: I'm particularly interested in.what influenced your decision. Did you see or hear anything at all about this issue-~such as television programs, newspaper editorials, letters and so forth? Subject: Well, this last election I got so tired of it I could have thrown television, radio, newspapers, and everything right out the door. SO if you want to know how I came to my decision, I thought they'd carried it too far. There was too much repeating of everything. I know I'm an Old fuddyduddy, but I think they carried it too far. 155 Transcript #3. Experimenter: First of all, were you able to vote Tuesday? Subject: Yes. Experimenter: Then would you mind telling me how you voted on the graduated income tax amendment? Subject: Let's see--I voted no. Experimenter: At the time you voted were you fairly sure of your decision, or were you doubtful? Subject: Yes I was--I couldn't decide what to do about it. Experimenter: Then was there any particular reason why you did vote no? Subject: I couldn't make up my mind and I voted no. Experimenter: DO you know if there was anything that might have influenced this decision? Such as television, newspapers, letters-- Subject: Well, mostly by reading in the paper. Mainly because I don't watch television--very seldom. APPENDIX H LETTER USED IN OBTAINING RELIABILITY ESTIMATE FOR THE VALUE SCALE MICHIGAN STATE UNIVERSITY East Lansing - Michigan 48823 Department Of Psychology - Olds Hall November 8, 1968 I am the Michigan State graduate student to whom you returned a four-page questionnaire last summer, and to whom you talked on the phone about the recent election. I am, of course, very apprecia- tive of your cooperation with my study, but there is one more thing I would like to ask you to do for me. In the questionnaire that I sent you last summer, you ranked a list of valuesfor me, from 1 to 18. However, several people complained that these values were very difficult to choose between, and felt that they might rank the values very differently if they were to try it a second time. In order to find out if this is generally true, I am asking a small group Of the people who replied last summer to rank these values again, just as they feel about them now. It will be very helpful to me if you would do this. (I realize, of course, that this is "above and beyond the call Of duty" to social scientific research.). If you are interested in finding out how much your opinions about these values changed from last summer, I would be more than happy to send you a copy of your earlier value choices, along with your second rankings. , Thank you very much for your time and trouble to date. Sincerely, Robert Homant 156 APPENDIX I VALUE LETTERS USED FOR STUDY II Family Security Letter February 20, 1969 1616 Cambria Drive East Lansing, MHchigan Dear Mr. and Mrs. One of the basic concerns Of the supporters of a graduated income tax amendment (including such groups as the League of Women Voters) has been with a tax structure which would provide adequate revenue for the state without placing tOO heavy a burden on low income families. We feel that family segurity is a worthwhile goal which a graduated income tax would help to achieve. Therefore we were very disappointed to see the graduated income tax amendment go down to defeat last November 5th. We felt that taxing according to ability to pay would mean that every family could provide for its necessities before seeing its income dis- appear in taxes. We are contacting people at this time because we feel that the graduated income tax amendment was defeated because of misunderstandings and we want to correct some false impressions which were shared by many people while these false impressions are still fresh. Many people voted against the graduated income tax amendment because they felt it would increase taxes. This was not true. The graduated income tax was meant to take the place Of the flat-rate income tax which we now have in Michigan. It would mean higher taxes for some people, but lower taxes for others. Many other people voted against the graduated income tax amendment because they were afraid that cities and counties would have been given the power to tax people. This also was not true. NO city or county could have imposed a graduated income tax without the permission of the state legislature. This is the same situation that now exists with re- spect to the one per-cent city income taxes. We feel that a misunderstanding of these points led to the defeat of the graduated income tax amendment. We feel that now is a good time for us to try to correct this misinformation, rather than waiting until the issue is once again before the voters. We thank you very much for your time in reading this letter. In the interest of family security we urge you to support the graduated income tax the next time it is voted on. Sincerely yours, Douglas Robinette Ad hoc committee for fair tax laws 157 158 Equality Letter February 20, 1969 1616 Cambria Drive East Lansing, Michigan Dear Mr. and Mrs. One Of the basic concerns of the supporters of a graduated income tax amendment (including such groups as the League of Women Voters) has been with an equal opportunity for everyone in our society. Since we felt that the graduated income tax amendment would have helped achieve this goal, we were very disappointed to see it go down to defeat last NOvember 5th. We felt that taxing according to ability tO pay would help achieve the goal of equal Opportunity in our society. We are contacting people at this time because we feel that the graduated income tax amendment was defeated because of misunderstand- ings and we want to correct some false impressions which were shared by many people while these false impressions are still fresh. Many people voted against the graduated income tax amendment be- cause they felt it would increase taxes. This was not true. The graduated income tax was meant tO take the place of the flat-rate income tax which we now have in Michigan. It would mean higher taxes for some people, but lower taxes for others. Many other people voted against the graduated income tax amendment because they were afraid that cities and counties would have been given the power to tax people. This also was not true. NO city or county could have imposed a graduated income tax without the pemmission of the state legislature. This is the same situation that now exists with respect to the one per-cent city income taxes. We feel that a misunderstanding of these points led to the defeat of the graduated income tax amendment. We feel that now is a good time for us to try to correct this misinformation, rather than waiting until the issue is once again before the voters. We thank you very much for your time in reading this letter. In the interest of equal opportunity for all we urge you to support the graduated income tax the next time it is voted on. Sincerely yours, Douglas Robinette Ad hoc committee for fair tax laws APPENDIX J POSTTEST QUESTIONNAIRE FOR STUDY II MICHIGAN STATE UNIVERSITY East Lansing - Michigan 48823 Department of Psychology . Olds Hall Let me begin by saying that I am extremely sorry to have to trouble you again. I am the Michigan State University graduate student who contacted you last November concerning the graduated income tax. I had fully expected that I would have completed my study by now. However, I have run into two unexpected complications. The first was that many people seemed unsure of their opinion on the issue. The second was that some supporters of the graduated income tax have continued to distribute information on the issue. ' Therefore, I would be extremely grateful if you would answer my final two questions and return this letter to me. Thank you very much for your patience. Sincerely yours, Robert Homant If you were deciding on the issue at this time, would you be in favor of an amendment to allow Michigan to have a graduated income tax? Definitely yes Probably yes Probably no Definitely no Have you received any information on the graduated income tax since the election last NOvember? Yes NO (If you answered yes, I would appreciate a brief description.) 159 APPENDIX x ATTITUDE QUESTIONNAIRE (Used for both pretest and posttest on Study III) Name Age Sex Student Opinions (please circle the response that most closely re- flects your opinions) 1) DO you think that the proposed closing of the graduate library to undergraduate students is justified? definitely yes : probably yes : no Opinion : probably no . definitely no How important is the above issue to you? very very important : important : slightly important : unimportant : unimportant 2) Do you think that there should be a tougher policy toward campus demonstrations, such as those staged by SDS? definitely yes : probably yes : no Opinion : probably no definitely no How important is the above issue to you? very very important : important : slightly important : unimportant : unimportant 3) Do you think that the 9-point grading system adopted this year is a good idea? definitely yes : probably yes : no opinion : probably no : definitely no How important is the above issue to you? very very important : important : slightly important : unimportant : unimportant 160 161 4) Do you favor the complete elimination of hours for women, irrespec- tive of their year in school? definitely yes : probably yes : no Opinion : probably no : definitely no How important is the above issue to you? very very important : important : slightly important : unimportant : unimportant 5) Are you in favor Of the Administration having the power to exercise censorship over the State News? definitely yes : probably yes : no opinion : probably no : definitely no How important is the above issue to you? very very important : important : slightly important : unimportant : unimportant 6) Would you be in favor Of doubling the fee reductions currently allowed under MSU's sliding scale tuition plan? definitely yes : probably yes : no opinion : probably no : definitely no How important is the above issue to you? very very important : important : slightly important : unimportant : unimportant 7) Do you think that the use of marijuana should be legalized? definitely yes : probably yes : no Opinion : probably no : definitely no How important is the above issue to you? very very important : important : slightly important : unimportant : unimportant APPENDIX L VALUE COMMUNICATIONS FOR STUDY III* Name Instructions for Part One: On the following four pages you will read four passages, each about four paragraphs long. YOu could consider these passages to be brief editorials (or as simply any type Of persuasive argument). Your task is to read them carefully, and then judge their persuasibility. I judge these passages to represent reasonable positions on four campus issues, although I do not necessarily agree with them in their entirety. Admittedly, your standard for how persuasive something is must be subjective. Still, I am more interested in the relative persuasive- ness of the arguments (for example: is passage #1 more persuasive than passage #3?) than I am in the absolute persuasiveness Of the argument. After each passage you will find four scales like the following: How persuasive did you find the above passage? 1234;61:394ng very persuasive not persuasive at all *Appendix L contains the first six pages of the booklet which subjects received when they reported for the second part of Study III. (The last four pages of that booklet are contained in Appendices R and A respectively.) Subjects in groups A" and B" received either page 167 or 168 (which are the value communications).but not both. For subjects in group C", both pages 167 and 168 were omitted--and the instructions on this page were altered accordingly. 162 163 These scales are meant to aid you in judging the passage. Circle the number that seems tO best indicate your feeling about it. You may read the passage over as often as you wish. YOu may also read ahead in the following passages and then return to change your answer. Once you begin Part 2, however, you may not go back to Part 1. Finally, I do not expect a long answer to the fifth question on each page. I am mainly interested in anything which you think pppTg. posgibly bias your Opinion on the issue referred to in the passage (i.e., might someone else suspect that your opinion is biased). 164 Passage #1: A defense of student demonstrations at MSU There are two fundamental values of our society which have re- peatedly come into conflict in recent months. On the one hand, social order requires authority, and Obedience to that authority. On the other hand, the freedom of the individual requires the right to protest against that authority wherever it is abused. The college campus is a case in point. There is no direct, democratic method by which students can challenge those decisions which directly affect them. Consequently, the basic means by which students can check and balance the Administration is the student demon- stration. It seems Obvious to us that students should have the right to demonstrate in order to make known and dramatize grievances. Of course, this implies a peaceful demonstration which does not interfere with the rights of non-demonstrators. On this campus, SDS has generally taken the lead in organizing and conducting demonstrations. When these demonstrations have been met with understanding and a willingness to bargain in good faith, they have been relatively peaceful and orderly (especially considering the dangers inherent in any large group of emotionally aroused people). Threats of force, on the other hand, have led only to increased paranoia and resistance on the part of the demonstrators. Indeed the threat of force only makes the "confrontation" desired by many for left elements that much easier to achieve. What is needed most at this time, then, is a university policy toward demonstrations which recognizes the peaceful and non-violent intent Of the vast majority of the demonstrators. 1. How clear do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very clear not clear at all 2. How fair a presentation do you feel it is? l 2 3 4 5 6 7 ’8 9 10 11 very fair not fair at all 3. How persuasive do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very persuasive not persuasive at all 4. Does the above passage influence you at all? l 2 3 4 5 6 7 8 9 10 11 YES, quite a bit no, not at all (negatively) 5. Do you have any personal interest in this issue? (i.e., are you a member of SDS, etc.?) 165 Passage #2: A view against MSU's present grading system A grading system is intended to serve two functions. It's main function should be to give a rough index of the student's knowledge which he acquired from some specific course, or tO give a rough average of his overall performance (the GPA). A second functian is to motivate students to perform well. Ideally, of course, the pursuit of wisdom should be motivation enough. But most people feel that this is not enough, and we have no quarrel with using grades as incentives--so long as it is recognized that the grade is not intended to be the primary reason for scholastic achievement. The usual five-point marking system fulfilled these two functions adequately, and student needs would have been.met especially well if it had been supplemented by credit, no-credit courses. We Object to the use of the nine-point marking system on the following two grounds: 1) First of all, the nine-point grading system does not increase the extent to which grades fulfill their main function: Most professors will admdt that even five levels Of discrimination were too many for them to make in assigning grades. The nine-way split gives the false impression that grades have suddenly become a more accurate index of student performance. 2) Secondly, it would be difficult to assert that students are more motivated to perform because there are more levels of achievement to work for. And even if this is the case, one could raise a serious question as to whether an increase im.motivation of this type is desirable. 1. How clear do you find the above passage? 1 2_____§ 4 5. 9e 7 .115 .2. 3EL____JUL. very clear not clear at all 2. How fair a presentation do you feel it is? l 2 3 4 5 6 7 8 9 10 11 very fair not fair at all 3. How persuasive do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very persuasive not persuasive at all 4. Does the above passage influence you at all? 1 2 3 4 5 6 7 8 9 10 11 YES, quite a bit no, not at all (negatively) 5. Do you have any personal interest in this issue (other than the fact that you get grades)? 166 Passage #3: Administration control of the State News Ordinarily, we would maintain that a newspaper should be free to print anything short of libel or deliberate distortions of the truth. The State News, however, is a different matter, for two reasons: ownership Of the paper is public rather than private, and the State News is intended to represent the MSU student body. With a regular newspaper, for example, an editor is responsible to the paper's owner, and also the paper must be run on a profit basis. In contrast, final responsibility for all news and editorial content in the State News rests with the student who is editor-in-chief. This, we feel, is a potentially dangerous situation. The editor is in a position not only to shape, but also to appear to represent student opinion at MSU. We feel that it would be a better situation if the Board of Trustees (or their appointee) had a veto power over the State News--not so much to be able to censor the editor, as to guarantee that editorial comment is balanced and representative. With respect to the printing of "Obscene" words, there should also be some control (or at least the power tO control). This is not because we are in favor of prohibiting the printing Of "Obscenities," but because the State News is intended to serve all the MSU students, and the student editors should not have the unchecked discretion to print whatever they wish. The principal objection to our position would probably concern the fear that the Administration could somehow manage the news. Even if there were a real danger that this would be attempted, we feel that it could not succeed as long as the continued publication Of alternatives such as The Paper is guaranteed. (The Paper, since it does not claim to speak for all the students, should be absolutely free Of Administration interference.) 1. How clear do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very clear not clear at all 2. How fair a presentation do you feel it is? l 2 3 4 5 6 7 8 9 10 11 very fair not fair at all 3. How persuasive do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very persuasive not persuasive at all 4. Does the above passage influence you at all? 1 2 3 4 5 6 7 8 9 10 11 YES, quite a bit no, not at all (negatively) 5. Do you have any personal interest in this issue? (e.g., work for the State News) 167 Passage #4: An argument against sliding scale tuition. The sliding scale tuition plan seems to be one of those programs which gets introduced as an experiment, and then never becomes critically evaluated. We contend that this plan is directly Opposed to the freedom Of the individual, a fundamental right which should not be infringed upon unless there is no simple alternative to fulfilling a real need. The individual's freedom is infringed upon in two ways. First of all, a person should be able to attend college without having to supply a lot of confidential information about his parent's income. Secondly, and more important, there is no reason why one student (or his family) should be forced to pay for another's education, just be- cause the other's parents make less money. We do reOOgnize, of course, the right of every individual to receive a college education, but this can be done by low interest loans (whose repayment begins after leaving shcool), rather than by what is really a form of double taxation. Remember, those with a higher income have already paid a higher amount of taxes to the state Of Michigan for the support of the state's colleges. Education can be made available tO all without infringing on the freedom of the individual to the extent that the MSU sliding scale tuition plan does. 1. How clear do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very clear not clear at all 2. How fair a presentation do you feel it is? 1 2 3 4 5 6 7 8 9 10 11 very fair not fair at all 3. How persuasive do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very persuasive not persuasive at all 4. Does the above passage influence you at all? 1 2 3 4 5 6 7 8 9 10 11 YES, quite a bit no, not at all (negatively) 5. Do you have any personal interest in this issue? (e.g., do you receive a fee reduction?) 168 Passage #4: An argument against sliding scale tuition. The sliding scale tuition plan seems to be one of those programs which gets introduced as an experiment, and never becomes critically evaluated. We contend that this plan is directly Opposed to a fundamental right Of the individual-~the right to pursue a comfortable life without seeing one's economic gains taken away by double taxation. We use the term "double taxation" for the following reason. Those with higher incomes pay a higher amount in taxes to the state in the first place. Tax money pays for two-thirds of the cost Of educating a student. To turn around and then require those who have already paid more for the upkeep of the school to pay more in order to send their children to that school hardly seems fair. It almost removes the incentives for getting ahead in the first place. Of course we do not want anyone to miss a college education for financial reasons. But the necessary money can be provided by low interest loans (which must be paid back after completing one's education), rather than by penalizing those who, for the most part, have worked hard to get ahead. It is not necessary to penalize those who have worked for a comfortable life in order to make education available to everyone. 1. How clear do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very clear not clear at all 2. How fair a presentation do you feel it is? 1 2 3 4 5 6 7 8 9 10 11 very fair not fair at all 3. How persuasive do you find the above passage? 1 2 3 4 5 6 7 8 9 10 11 very persuasive not persuasive at all 4. Does the above passage influence you at all? l 2 3 4 5 6 7 8 9 10 11 YES, quite a bit no, not at all (negatively) 5. Do you have any personal interest in this issue? (e.g., do you receive a fee reduction?) 169 PART TWO You have now completed Part One. Please do not refer back to part one now that you have begun part two. Part two is the same as the forms that you filled out the first time you reported to the experiment. I am.repeating these forms because I need to know how stable student opinion is. Please do not try to remember how you answered these questions the first time. Just answer according to how you feel at this time. (I do not need your name, birthday, etc. this time.) "Tillilflfljllllflflflllfllflr