Personal Involvement, Social Judgment, and Action
Carolyn W. Sherif, Merrilea Kelly, H. Lewis Rodgers, Gian Sarup,
and Bennett I. Tittler
Pennsylvania State University
Relationships among the latitudes of acceptance, rejection, and noncommitment were investigated to develop indicators of ego involvement, using the own-categories method and the method of ordered alternatives. Degree of involvement was differentiated across attitude positions (extreme-moderate) and among persons taking the same position. Findings from five studies were that degree of involvement is (a) inversely related to fineness of discrimination in judging beliefs about one's reference group; (b) associated with the priority of normative concerns (values) in one's reference group; (c) positively related to selectivity in attributing credibility to communicators and to contrast effects in judging communicator positions; (d) inversely related to attitude change in response to short communications; and (e) positively related to the probability of action and positive response to social pressures to action congruent with attitude. Factors affecting the indicators and possible implications for attitude research are discussed.
Theories about attitude functioning and change frequently make predictions that are contingent upon the degree of intensity, personal meaning, importance, or ego involvement of the attitude object to the person (cf. Bieri, 1967; Festinger, 1957; Harvey, 1963; Kiesler, Collins, & Miller, 1969; McGuire, 1969; Osgood, Succi, & Tannenbaum, 1957; Shaw & Constanzo, 1970; Sherif & Hovland, 1961; Smith, 1969). Yet, relatively little research has explored behavioral indicators for degrees of personal involvement.
The dilemma is not unique to the involvement variable for, as Triandis (1967) remarked, "There is a gap between those who are primarily concerned with the measurement of attitudes and those who have written theoretically about attitudes." The former, he noted, are often content to provide a "score" that signifies "that which is measurable by my attitude scale," while the latter too frequently make theoretical distinctions without "precise and standard procedures for measurement [p. 228]."
This article reports five studies in a continuing research program aimed at studying the structural properties of attitudes and differential personal (ego) involvement through social judgment. In earlier research the properties of attitude structures were used to predict systematic variations in judgment, placement of the positions represented in communications, reactions to the communications, and susceptibility to attitude change as well as its direction (cf. Sherif, Sherif, & Nebergall, 1965; Sherif & Hovland, 1961; Sherif & Sherif, 1967, 1969). The social judgment formulation has proved relatively viable in predictive power (cf. Kiesler, Collins, & Miller, 1969; Rhine & Severance, 1970; Shaw & Constanzo, 1970).
Yet, research in the program left a number of important questions unanswered and raised new issues. Each study reported here was addressed to such a question, all relating to the problem of specifying differential ego involvement and its effects. The studies concerned the relationships between involvement and keenness of discrimination when attitudes are favorable (Study I) and when they pertain to norms of varying priority in identifiable reference groups (Study II), judged difference of opinion from communicators of varying credi-
( 312) -bility (Study III), attitude change (Study IV), and the probability of actions related to attitude (Study V).
The need for indicators of personal involvement is not met by most conventional methods of attitude-test construction. For example, the more rigorous Thurstone method of item selection effectively eliminates items with self-reference, which are notably subject to systematic variations in judgment (assimilation-contrast effects) according to the person's attitude (Schulman & Tittle, 1967). Rigorous statistical assumptions about item variability can produce tests composed of abstract stereotypes with little bearing on behavior in actual situations the person faces. The calls for reform in test construction have included increased self-reference in items, for example, by soliciting personal intentions and/or commitment for action (e.g., Fishbein, 1967; Triandis, 1964), as well as capitalization on systematic item variability as a basis for inferring attitude (Sherif et al., 1965; Zavalloni & Cook, 1965).
Most previous research on degree of involvement has relied on some form of rating or ranking as to importance and upon extremity of attitude scores on conventional tests (e.g., Likert, semantic differential). A major difficulty in equating extremity and involvement is the problem of interpreting intermediate ratings. The difficulty can be illustrated through analogy with opinion surveys, in which the equating of "no opinion," "don't know," or moderate responses with neutrality has led to embarrassing errors, for example, in the 1948 presidential election and the 1969 mayoralty election in New York City.
Other difficulties are closely related to specific features of the research situation (cf. Rosenthal & Rosnow, 1969). The set of items prepared by the investigator may be loaded with social significance, but of moderate or minor importance to the persons studied. The result is high ranks for the pressing social issues which, as we have found repeatedly, are rapidly depressed when more personally salient matters are included in the set to be rated. In addition, self-ratings are particularly subject to the person's expectations about the intent of the researcher, possible personal evaluation, and what appears socially desirable in the situation (Edwards, 1957).
The present research sought methods for inferring degree of involvement from structural properties of behavior while the person engages in tasks that do not confront him directly with the issue of how salient the attitude object is for him personally. While offering no panacea for the obstacles in the path of attitude research, the development of such indicators would add to the sparse tools now in the workroom. The magnitude of attitude problems will continue to require multiple methods for collecting convergent data if predictive validity is a viable goal.
Indicators of Attitude Structure and Involvement Based on Social Judgment
The keystone of social judgment theory is a conception of the self-system (or ego), Since this conception differs from other widely used characterizations of self (Sherif, 1968), it will be specified briefly. Self is conceived as a system of attitude structures which, when aroused by ongoing events, are revealed in more characteristic and less situation-specific behaviors toward objects or classes of objects (Sherif & Cantril, 1947; Sherif & Hovland, 1961; Sherif & Sherif, 1969; Sherif et al., 1965). Behavior indicating involvement of self in psychological processing is typically emotional-motivational, as revealed, for example, in the categorization of objects into positive and negative subclasses, in approach-avoidance activities, or systematic variations in response thresholds.
The degree of self or ego involvement (Sherif, 1936) is inferred from the relative consistency and characteristic structure of behaviors with reference to their situational context. In thus inferring an order of personal involvement, we assume cognitive processing: selecting, discriminating, and comparing salient features of events with both antecedent and concurrent stimuli. Since this judgment process also becomes evaluation when events bear closely on cherished premises and beliefs, attitude may be defined operationally in terms of the person's categories for evaluating a stimulus domain when circumstances or instructions do not predetermine the categories to be used (Sherif & Sherif, 1967, p. 115).
( 313) Attitude structure, in turn, refers to the relative scope, width, or latitude of categories used in evaluation, namely the latitudes of acceptance, rejection, and noncommitment (Sherif, 1960). Latitude of acceptance is that position (or category) in a class of objects that the person finds most acceptable plus all others also acceptable, while latitude of rejection encompasses the comparable objects that are objectionable. Latitude of noncommitment consists of those objects that the person neither accepts nor rejects, or for any reason fails to evaluate, when not required to respond to the entire stimulus domain presented. In this article, "relative size" of the three latitudes always refers to the relative frequencies of items accepted, rejected, or not evaluated (noncommitment) by the person.
Relationships among the evaluative latitudes form the basis for the indicators of involvement. These relationships were largely determined by comparing the social judgments of persons opting for extreme versus moderate positions on social issues, the independent criteria for the greater involvement of those taking extreme positions being public commitment and participation in pertinent actions (Sherif et al., 1965). When free to generate their own categories, those more involved used fewer categories and distributed items bimodally, the largest mode coinciding with items rejected. The relative sizes of the latitudes of rejection, acceptance, and noncommitment differed systematically according to the extremity of the person's stand. Briefly, with increasing attitude extremity, the latitude of rejection became increasingly larger and greater than the latitude of acceptance, while the latitude of noncommitment became increasingly smaller, even vanishing.
The question remained whether variations in relative sizes of the latitudes indicated degrees of involvement when the person's attitude was not extreme. Can a person alienated from available alternatives on an issue, or undecided, or moderate be highly involved in the issue? If so, would the relative sizes of the latitudes differentiate the degree of their involvement? The questions were resolved by Beck and Nebergall (1967) in research on persons opting undecided or very moderate positions concerning candidates during the 1966 senatorial campaign in Texas. The distinctive circumstances of that campaign resulted in numerous Democrats giving wishy-washy support to one or the other candidates or remaining undecided. A sizeable proportion withdrew from political activity, while others were politically active despite their lack of enthusiasm for the candidates. Both undecided and moderate persons who were politically active exhibited large latitudes of rejection, approximating those of persons upholding extreme positions, while the politically inactive exhibited large latitudes of noncommitment and little rejection.
In the studies reported here, the indicators included comparison of the latitude of rejection to the sum of items acceptable and noncommittal (the greater preponderance of rejection indicating higher involvement), the size of the latitude of rejection (for controversial social issues on which partisans adopt bipolar stands), and the size of the latitude of noncommitment (when bipolar stands were not taken). Criteria for degree of involvement included observed behavior in a variety of life situations and selection of objects for judgment rated or ranked as more or less important by pretest samples from the same populations.
STUDY I 
Do Differences in Involvement in a Reference Group Affect Judgment When Attitudes Are Favorable?
Experience suggests that one person may favor a viewpoint without trembling at its contradiction while another, similarly positive, feels the flush of indignation at the hint of a challenge. A study of national identification by Sarup (1969) bears on the problem. Given favorable attitudes toward one's country, does the strength of one's involvement or identification affect a person's judgments?
Sarup studied Indian students at several
|Sample||Mean number of categories||Mean item frequencies for the latitude of|
n = 10 in each sample
* p < .05
** p < .001
large universities through observation, the own-categories method (Sherif et al., 1965, pp. 92-126), and a questionnaire requesting various judgments and predictions about India.
Through interviews of Indian students and pretesting at several universities, Sarup developed a set of 50 statements about India purportedly made by Indian students in the United States and ranging from 7 extremely favorable statements to 7 extremely unfavorable statements. On the average, the 36 remaining statements were rated as moderately favorable (10), intermediate (16), or moderately unfavorable (10). Intermediate items were rated with significantly greater variability than the extremes.
Criterion samples were selected to validate the method of designating students as more or less involved in their nation. These samples were observed over a period of time, and reports were collected from acquaintances of the students. Ten publicly committed Indian nationalists and 10 Indian students not active in Indian affairs were selected.
Both for the criterion samples and the subsequent research, the own-categories task was presented to students individually or in pairs so situated that they could not observe each other's activity. The investigator requested the judgments as a favor in constructing a questionnaire for his doctoral research on the "brain drain," for which competent judgments were needed. The task was to sort the statements into categories ordered according to how favorable or unfavorable they were to India using any number of categories deemed necessary for the task, such that items within any category "belong together." Items could be distributed into the categories with any frequency that appeared suitable. After completing the judgment task, each judge was asked to label the categories as to which was most acceptable and most objectionable, as well as any others that might be acceptable or objectionable.
Table 1 presents the mean number of categories used by the two criterion samples, along with the mean sizes (item frequencies) of their latitudes of acceptance, rejection, and noncommitment. As predicted, those highly involved used significantly fewer categories, rejected significantly more beliefs, and were noncommittal toward fewer statements than those less involved but also favorable toward their country.
Accordingly, Sarup established his criterion for high involvement as a latitude of rejection greater than the sum of the frequencies accepted and those not labeled (noncommitment). If the latitude of rejection equaled or was less than the sum of the latitudes of acceptance plus noncommitment, a student was labeled as less involved.
The judgments of 67 Indian students at a large state university were investigated, this sample comprising the population with the exception of 7 students who were not available on campus at the time and 2 who refused to cooperate. All but 5 were Hindu. Mean age was 28.3 years (SD = 3.74). All but 5 were students of science or engineering. All held Indian passports, and all but 1 had student or exchange visitor visas.
Would those students classified as highly involved according to the stated criteria use fewer categories than those less involved? Table 2 presents the findings. Those classified as highly involved on the basis of the relative sizes of their evaluative latitudes used significantly fewer categories than those classified as less involved. The sizes of the latitudes for the two samples (which were constituted by criteria specifying only "greater than" or "smaller than") also differed significantly in predicted directions. Those less involved rejected significantly fewer items and were significantly more noncommittal. There were no significant differences in the number of items
|Sample||n||Mean number of categories||Mean item frequencies for the latitudes of|
|* p < .001|
(315) accepted or in their favorableness toward India.
Questionnaire data revealed no significant differences between the subsamples in their predictions about their country 5 years hence. Even more important, the two subsamples did not differ significantly in their ratings on linear scales about the desirability of India as a place to live permanently or a place to pursue one's chosen profession. When asked to locate the status of India in American eyes on a list of nations ranked according to prevailing social distance norms in the United States, the highly involved and less involved students did not differ significantly in the rank they assigned to Indians, an intermediate rank actually higher than that accorded to Indians by Americans.
What, then, were the differences between the highly involved and less involved students that were associated with their different categorizations of statements about their country? First, the less involved students were significantly less certain of predictions of India's future, namely, with respect to the likelihood of a dictatorship and the maintenance of a secular state. Second, although more and less involved students did not differ in their reports of the number of their own friends who would, in all probability, return to India, they did differ significantly in the average number of their friends they reported as willing to return and in their ratings of the legitimacy of their country's right to expect overseas students to return. More involved students reported more friends willing to return and regarded India's expectation that they do so as more legitimate.
Finally, when asked to rank a variety of groups in terms of their relative priority in their own personal concerns, the students who were more involved differed from those less involved in the average rank assigned to their nation. They were permitted to rate groups as completely unimportant to them if they chose to do such and to assign tied ranks. Students in both samples ranked the family as the most important group. However, the more involved students, on the average, gave tied ranks (2.5) to "nation" and "humanity." Mann-Whitney U tests indicated that their ranks for nation and humanity were stochastically higher than those assigned by the less involved students (U = 325.5, p < .05; U = 426.5, p < .05, respectively). In short, the more involved students ranked their nation just below the family and approximately equal to their concern for humanity, while the less involved students saw the family towering above the nation in importance.
To be favorable to a reference group (one's country in the present study) is not necessarily to place it at the pinnacle of one's concerns. Behaviorally, the difference in involvement is found when the person categorizes beliefs about that reference group. The highly involved person is more prone to reject categorically statements that might impugn his reference group and sees less need to differentiate finely among beliefs, thereby using fewer categories than the less involved person.
STUDY II 
Is Degree of Involvement Related to Priorities of Reference Group Norms?
Since Newcomb's (1952) classic survey of the shifts in Bennington students' attitudes toward the liberal norms of their college reference group, it has been widely assumed that individuals incorporate the norms of their reference group as their personal attitudes. If so, it would follow that the priority of normative concerns in the reference group should be reflected in the relative involvement of individual members in relevant attitude objects.
Using the own-categories procedures, two recent studies have supported the proposition that individual members of groups that differ with respect to the salience of a norm-regulated activity are differentially involved in categorizing stimuli related to them. Sherif (1973) found significant relationships between social judgments and the salience of blackwhite interactions for different black students. Rollins (1973) reported a high and significant correlation between social distance accorded to one's national origin and number of categories used to judge statements about one's group.
The present study of social judgment and priority of norms was undertaken after prolonged observation of natural groups of adolescent girls. After identifying seven reference groups through unobtrusive observation, two areas of normative concern were selected differing in their relative salience for the different groups. The differential priority of the norms was expected to be indicated in individual judgments as follows:
1. Individual members would use fewer categories and be less noncommittal in judging stimuli related to normative concerns high in priority than those lower in priority in their groups.
2. Individual members of groups in which a norm had higher priority would use fewer categories and be less noncommittal than members of groups in which the same normative concern had lower priority.
Regular observations (at least two periods per week) were made in a large high school over a total period of 14 months.  Observation was unobtrusive, the observer functioning with the cooperation of school authorities in various nonauthoritative roles that permitted her to move freely around the building (e.g., as a distributor of information from the counseling office).
Seven informal groups of girls were identified during the period solely on the basis of their frequent and recurrent patterns of interaction within and outside of the school. Regular reports on observed interaction also revealed discriminable differences among the groups in customary practices, concerns, and activities. On this basis, the area of personal appearance was selected as high in the normative concerns of all groups, though more salient for some than for others. Since all of the girls were high school seniors as the observation drew to a close, future activities after graduating were selected as an important but less salient area of normative concern for all groups, again varying in priority from group to group, as revealed in time spent in discussing it and pressures on individuals to follow courses of action consistent with a particular future. Ranks assigned groups for observed concern were based on the observer's reports prior to the experiment.
The labels assigned by the observer (Stiffler, 1968) and brief descriptions convey observed differences in these normative concerns. Each is followed by the group's rank in observed concern with appearance and the future: (a) governors---attractive, conservatively and well dressed, highly active in school organizations, college bound (Appearance Rank 2, Future Rank 1) ; (b) wild ones---attractive, modishly dressed, keenly oriented toward boys, observed frequently outside school with college boys, not oriented toward school work (Appearance Rank 1, Future Rank 7) ; (c) pep club girls---nucleus of official booster club for athletic teams, active, college bound (Appearance Rank 4, Future Rank 2); (d) business girls---neat and proper, enrolled in business curriculum, together in and out of class (Appearance Rank 3, Future Rank 5) ; (e) longhairs---avant-garde interests in folk music, art, current affairs, associated in and out of school (Appearance Rank 7, Future Rank 3); (f) sports---girl athletes in gym classes, leisure time, and school teams (Appearance Rank 6, Future Rank 4) ; (g) rurals---cluster of girls from outlying areas, differed in appearance from "town" girls, not college bound (Appearance Rank 5, Future Rank 6).
Group members were not aware that they had been observed. Since no notes were made in their presence, the subsequent information that observations were made was a complete surprise. Therefore, at the time, the request for help in a judgment task to secure reliable information about high school students was not seen as related to group membership. A total of 48 senior girls belonging, in approximately equal numbers, to each of the seven groups participated individually with permission from parents.
Stimulus Materials and Procedure
Sets of pictures (line drawings) were prepared representing (a) clothing and hair styles, (b) activities to be engaged in after high school, and (c) geometric forms. Each set consisted of 60 drawings. The pictures of clothing and hair styles ranged from those observed as acceptable among the girls to those selected as outmoded or not observed in school. The own-categories task was to categorize this set in terms of "appropriateness" for school wear. The drawings of future activities were to be categorized according to their "desirability" for a high school graduate 2 years after graduation, and included pictures of college students in academic and social situations, vocational and business students, various occupations, marriage, child care, housework, and heterosexual activities of varying propriety.
The set of geometric designs was prepared for comparison with the norm-related activities. These nonsense designs varied systematically in symmetry, the width-height ratio, and number of convolutions in the design. The task was to categorize the de-
( 317) -signs according to which seemed to "belong together." Again, the subject was free to generate any number of categories needed and to distribute items in any way that seemed appropriate.
Each girl made judgments individually of the three sets of drawings (60 drawings each). Order of presentation of the sets was counterbalanced. After categorizing each set, the student was asked to label any categories she wished as to her own acceptance or rejection.
From the observations, it was predicted that, on the average, the girls would use fewer categories and be less noncommittal about the set of clothing and hair styles than about the pictures of future activities, in which they were less immediately involved. One-way analyses of variance for related measures showed that the mean number of categories used for styles (3.6) was significantly less than the mean for future activities (5.3), and that the percentage of items not labeled (noncommitment) was significantly smaller for styles (5.17o) than for future activities (12.7%). While the only prediction made about the categorization of geometric forms was that they would differ from that for the norm-related sets, it was interesting that the mean number of categories for geometric forms was intermediate to the other sets (4.8) as was the percentage of noncommitment (9.8%). The overall differences in mean number of categories and percentage of noncommitment for the three sets were both statistically significant (p < .01).
The rank orders of the groups as to their normative concern with appearance and with future activities that were based on observational reports were not significantly related. However, the ranks of the groups in observed concern and their ranks in number of categories used and in percentage of noncommitment were significantly related, both for appearance and for future activities. Table 3 presents the rank-order correlations (Spearman rho). In addition, the rank-order correlations are given for the groups ranked according to number of categories used and frequency of noncommitment for each of the three sets of pictures. (Since there was no observed concern with geometric forms, the third column is blank in the first two rows.)
|Ranks considered||Appearance||Future||Geometric forms|
|Observed concern and number of categories a||.93**||.96**||—|
|Observed concern and frequence of noncommitment a||.75*||.78*||—|
|Number of categories and frequency of noncommitment b||.72*||.69*||.07|
Higher rank indicates greater observed concern and fewer categories (or less
b Higher rank indicates fewer categories and less noncommitment.
| * p
** p < .01.
As predicted, the higher the group rank in concern over appearance or over the future, the fewer the categories used by its members in their individual judgments and the less their noncommitment. As the last row of Table 3 indicates, there was also a statistically significant relationship between the group ranks according to the number of categories used and to percentage of noncommitment, in support of previous findings that constriction in number of categories and low frequency of noncommitment are covariants, together indicating higher involvement. As would be expected, there was no significant relationship between number of categories used and frequency of noncommitment in judging the geometric forms.
Owing to the small numbers within each group (6-7), the problem of variability in individual judgments within the group was assessed in terms of the percentages of pictures in each of the three sets on which 70% or more of the members agreed in categorizing an item as acceptable, objectionable, or remained noncommittal. As might be expected, consensus on geometric forms was lowest (only the artistic long-hairs agreed on over half of the items); consensus on future activities was greater (the two groups ranked first and second in observed concern agreeing on over 70% of the items) ; and the greatest consensus was on appearance (with only the sixth-ranked sports in observed concern failing to agree on more than half of the items).
This combined field and laboratory study
( 318) found that the number of categories used by individuals and the size of their latitudes of noncommitment (as indicators of the degree of personal involvement) were related to the priority of normatively regulated concerns in their reference groups. Fewer categories and less noncommitment were found for attitude objects higher in the group's priorities and for individuals belonging to groups wherein the particular object of concern was ranked higher than by other groups. These relationships were not found when individuals categorized items unrelated to their group concerns (geometric forms).
STUDY III 
Are Judgments of Communicators' Credibility and Viewpoints Affected by One's Involvement in the Issues?
In research practice, communicator credibility has been treated, at times, as a property of the communicator. Likewise, it is sometimes assumed that the position taken by a communicator is constant for both researchers and subjects. On the contrary, communicator credibility refers to judgments made about a communicator in which consensus presumes an antecedent social process.
Research on judgments of the positions presented in communications on social issues suggested the hypothesis that the more involved the person is in an issue, the more selective he becomes in judging communicators credible in that issue. Further, the more involved the person, the greater the likelihood that he will contrast a position discrepant from his own on that issue, thereby widening the gap between the communication and his own position. Conversely, a person less involved is likely to assimilate more discrepant positions (Sherif et al., 1965, pp. 155-161). These general predictions were examined in a study by Rodgers (1968) of the credibility and judged positions of 10 well-known political figures on three current issues.
Cooperation was secured with organized and active groups of left-wing students (Students for a Democratic Society [SDS], n = 31) and conservative students (Young Americans for Freedom [YAF], n = 20). An unselected sample of undergraduates in, sociology courses at the same eastern university served as a comparison group (n = 76). By attending meetings of both political groups, Rodgers selected three topics that were being debated at the time, but on which the groups differed in devotion of time and energy: (a) the continued bombing of North Vietnam versus the immediate cessation of all bombing; (b) whether or not the United States is a racist society, an issue then being discussed following the release of a widely distributed report by a presidential commission; (c) whether or not the United States economy was endangered by the gold drain into foreign financial circles.
These issues were selected because all students were relatively concerned about the Vietnam issue, the SDS devoted more time to the race issue than the YAF, and the gold-drain issue was discussed by the YAF but not the SDS.
All students responded to attitude statements on each of the three issues, using the method of ordered - alternatives (Sherif & Sherif, 1967). A set of nine statements on each issue was presented, the statements having been reliably ranked by judges from one extreme to the other. For example, on the Vietnam issue, the alternatives ranged from a bombing halt to all-out military effort. The entire array of statements on each issue was presented four times in . succession: the first time requesting the one position most acceptable to the person; the second time any.' other acceptable positions; the third time the one position most objectionable; the last time any other' objectionable positions. The only requirements were the selection of one acceptable and one objectionable position, other evaluations being at the choice of the subject.
Next, they judged the credibility of 10 public figures on unmarked 9-centimeter linear scales ranging from "highly credible" to "not at all credible." The communicators rated were all familiar names to the students (W. Fulbright, H. Humphrey, L. Johnson, R. Kennedy, E. McCarthy, W. Morris, R. Nixon, R. Reagan, N. Rockefeller, G. Wallace)., Third, on 9-centimeter linear scales whose extremes corresponded to the extreme statements on each issue in the method of ordered alternatives, the student rated his own position on each of the three issues. Finally, on each of the three issues, he marked duplicate sets of the same linear scales to indicate where, in his judgment, each of the 10 public figures stood on each issue.
The most acceptable positions endorsed by the three samples differed as expected. On the Vietnam issue, 96.5% of the SDS endorsed one of the two extreme positions favoring a bombing halt while 75% of the YAF favored continued bombing. The unselected students varied widely over all nine positions. All SDS members endorsed the extreme positions that
( 319) the United States is definitely a racist society, compared to 40% of the YAF and 75% of the unselected students. On the other hand, 55% of the YAF adopted the position that this is not a racist society, as compared with 18.4% of the unselected students and none of the SDS. Virtually all students agreed that the gold drain was a danger to the economy, with the two most extreme positions taken by 64.6% of the SDS, 75% of the YAF, and 48.8% of the unselected students.
In addition to observational evidence, the size of the latitude of rejection was used as the criterion of relative involvement in the three issues since earlier research had supported its use on such bipolar issues (Sherif et al., 1965). Table 4 summarizes the mean frequency of positions rejected on each of the three issues by each sample. By this criterion, SDS members were significantly more involved in every issue than the other samples, but relatively less involved on the gold drain than the other two. In fact, on the average, these highly active students rejected all statements on the Vietnam and race issues that opposed their own stand, plus an intermediate statement indicating indecision and the most moderate statement supporting their own side. On the other hand, their mean latitude of rejection on the gold drain amounted merely to rejecting statements that minimized the problem.
The conservative YAF had latitudes of rejection significantly smaller that the SDS and not differing significantly from those of unselected students with one interesting exception, namely the gold-drain issue. On that issue, the YAF rejected significantly more
|Note. SDS = Students for a Democratic Society; YAF = Young Americans for Freedom|
|* p <
.05 for Duncan multiple-range comparison for YAF vs unselected
** p < .01 for F ratio in one way analysis of variance for each row
***P < .001 for Duncan multiple-range comparisons for SDS vs YAF
statements than the unselected students, indicating more involvement in the issue.
The ratings of communicator credibility were dichotomized at the scale midpoint to classify the communicators as credible or not credible to each student, and the distributions of the numbers credible for the three samples were compared on each issue using the Mann-Whitney U statistic. The comparisons were dependent, of course, on the particular set of communicators presented. An array including spokesmen of the "new left" might have increased the number credible for SDS members and decreased the proportion of the list credible by the YAF.
The comparisons of interest, therefore, pertain to relative selectivity in rating the particular communicators as credible on the three issues. The SDS members were significantly more selective than other samples (p < .001) in according credible ratings to communicators on the two issues in which they were most involved (Vietnam: Mdn = 2; race: Mdn = 3), while the YAF was significantly more selective than the SDS or unselected students on the gold-drain issue (Mdn = 3.0, 4.0, 5.0, respectively, p < .001). The comparative selectivity of the SDS and YAF on the gold-drain issue and the relative frequencies within each sample for the other issues are congruent with involvement indicators in the trend for greater selectivity on the more involving issues.
In accord with their selectivity, each political group rated the position upheld by their favored communicators as less extreme than those same communicators were rated by their opposite number. The pattern of the mean ratings of communicators as to their credibility and judged position on the issues revealed differences between the SDS and YAF members that would be expected if each side were assimilating the positions of credible communicators and contrasting those of communicators judged not credible. Yet, the analysis was complicated by the fact that on both the Vietnam and race issues, more SDS members than YAF took the extreme positions, while unselected students adopted positions over the entire range. Accordingly, the analysis was made in terms of the difference between each student's judgment of his own
|Note. SDS = Students for a Democratic Society; YAF = Young Americans for Freedom|
* p < .01 for F ratio in one way analysis of variance for each row
** p < .01 Duncan multiple-range comparisons for SDS vs YAF
***p < .01 Duncan multiple-range comparison for YAF vs SDS and YAF vs unselected
position and that attributed to each communicator on each issue.
The mean discrepancy between judgments of own position on the issue and the judged position of a communicator was calculated separately for communicators classified as more credible and less credible by the subject. On the whole, the average difference between own position and that judged as the position of a credible communicator was small (1 or 2 centimeters). The differences were generally larger on the Vietnam issue (> 2 centimeters) than on others, an outcome not surprising in view of the prevalent public charges at the time that information was being withheld or distorted on that issue.
The SDS members perceived greater differences than the other samples between their own positions on the race issue and those of credible communicators (X = 3.5, p < .01).
On the gold-drain issue, however, the discrepancy between own positions and those of credible communicators for SDS members averaged only 1 centimeter. YAF members placed the positions of credible communicators on the gold-drain issue at a significantly greater distance from their own position than the other samples, a finding in accord with their greater involvement in that issue (p < .01).
Table 5 shows the comparable mean differences between ratings of own position and communicator position on each issue for those public figures classified as not credible. On the Vietnam and race issues, the mean differences for the SDS occupy all but about 2 or 3 centimeters on the 9-centimeter scale. The magnitude of these differences reflects the fact that most SDS members were extreme in their own positions on these issues, thereby having leeway to indicate differences of a large order. However, another finding suggests caution in dismissing the differences in Table 5 as mere artifacts of the extremity of own position.
On the gold-drain issue, where similar proportions of SDS and YAF took extreme positions on the same side of the issue, the SDS separated their own positions from those of noncredible communicators by only 2.3 centimeters on the average. The YAF members, who were more involved in this issue, separated their own positions from those of noncredible communicators by over half of the 9-centimeter scale on the average.
In general, the findings were that the higher the involvement in the issue, the more selective the person was in attributing credibility to communicators and the more likely he was to perceive greater differences between his f own position and that of communicators, especially if they were not credible in his eyes. Less involved persons (e.g., the SDS on the gold-drain issue, the YAF on the race issue, and the unselected students on all three issues) were more generous in attributing credibility and tended to assimilate positions of communicators nearer to their own, perceiving smaller differences even from communicators judged as not credible. The findings suggest the importance of personal involvement j in studies of the credibility of communicators_' and their judged positions, as well as the k: vicissitudes of communicator status and judged political stance over time.
STUDY IV 
Does Degree of Involvement Affect Susceptibility to Attitude Change?
In general, attitude research indicates that persons upholding extreme stands are less likely to change their attitudes than persons taking moderate positions (Sherif & Sherif, 1969, chap. 21). The unsettled issue is whether these findings are simply the result of attitude extremity or of the higher involvement of many extremists or both. While this, issue may not easily be settled, owing to the
( 321) typically high correlations between extremity and involvement, it is possible to study change of attitudes that are not extreme, but moderate, and which differ in their relative involvement for the person. Tittler (1967) modeled such a study after Janis and Field's (1959) investigation of the persuasibility of males and females.
Two of the issues and communications used by Janis and Field were adapted for college students, namely, the merits of General von Hindenburg as an historical figure and the probability of success for an unknown comedian named O'Keefe. Through extensive pretesting for ranked importance, two additional topics were selected as involving for college students, one concerning the female and the other the male sex role. More attitude change was predicted on the less involving topics used by Janis and Field than on the more involving sex role topics.
Student samples from introductory psychology classes ranked 14 issues as to their personal concern or lack of concern. Most males accorded a high rank to the importance of choosing an occupation with an adequate income level, while women showed consensus in giving a high rank to the importance of developing abilities to understand and get along with others.
Selecting these two sex role topics and the Hindenburg and O'Keefe topics used by Janis and Field, Tittler (1967) developed separate orderedalternative instruments to assess attitudes tOWard each. Two communications (each 200-300 words) were standardized on each issue, one reliably rated by independent judges as second to one of the extreme positions on the method of ordered alternatives and the other as second to the opposite extreme position on the same issue. In addition, four communications of comparable length were designed that were totally irrelevant to the issues, for example, the use of flu vaccine or the design of shotgun stocks.
Two hundred and fifty-five students in introductory psychology (136 males, 119 females) volunteered for the study, receiving one credit to be included in their final course grade. Each filled in the method of ordered alternatives for each issue, the order of issues being counterbalanced. (See Study III for general procedures for the method.)
At a second session about 2 weeks later, the same students read four communications, then repeated the method of ordered alternatives on each of the issues. Assignment to read four communications relevant to the issues or four irrelevant communications was on a random basis. For those assigned to read relevant communications, the particular communications had been selected in terms of the student's most acceptable position on each issue at the first session. If a student's most acceptable position was on one side of the issue, the booklet contained the relevant communication located next to the extreme opposite to the side accepted. Since the students' most acceptable positions on the four issues were moderate, this procedure insured that each student faced a communication opposed to his most acceptable position, but moderately discrepant from it. The irrelevant communications were presented to approximately half of the subjects as a means of assessing the demand character of the before-after design, which strongly suggested that reading a communication between administrations of the same attitude instruments was related, in some way, to responses to the attitude tests.
Attitude change was defined operationally as change in the most acceptable position toward the discrepant and opposed communication that the person read on the relevant topics. ("Boomerang effects" in response to these moderately discrepant communications were neither expected nor found.) For those reading irrelevant communications, change was defined arbitrarily as shifts toward the side opposite to the position the student accepted on the first administration. The data are the percentage of students making such changes from the first to the second administration of the attitude instruments.
Since the Hindenburg and O'Keefe issues were not expected to be involving, it was also expected that the latitudes of noncommitment on these topics would be significantly greater than for the sex role topics. Latitude of noncommitment was studied since students did not take bipolar stands on these issues. Analysis of variance of the initial sizes of noncommitment (item frequencies) revealed a main effect attributable to the topics (p < .001), with greater noncommitment for the less involving topics (see Table 6, last row).
For convenience, Table 6 summarizes the main findings on attitude change, combining male and female samples, since there was no significant overall difference in percentage of males and females who changed. All comparisons between independent proportions of males or females who received relevant communications, on one hand, or irrelevant com-
|Net change (relevant – irrelevant||25.0||22.2||49.6||44.5|
|Mean frequency of noncommitment||2.8||2.6||3.2||3.5|
|Note. In the sample receiving relevant communications, n = 125. In the sample receiving irrelevant communications, n = 130.|
-munications, on the other, were significant (p < .05). By inspecting the percentage of change and net change in Table 6 for the two sex role topics and for the O'Keefe and Hindenburg topics, the effects of greater involvement in reducing the propensity to change is apparent (McNemar tests, p < .05).
Comparison of independent proportions of males and females changing after relevant communications on the issues yielded only one significant difference: Women changed significantly more frequently than men on the Hindenburg issue, the issue on which women also had a significantly greater latitude of noncommitment than men. The incidence of change was massive for both males (72.7%) and females (88.1%). If the changes by those who read irrelevant communications are interpreted as reflecting demand characteristics of the situation, the percentages of net change still reveal a sex difference on this issue (males = 31.3%; females = 57.5 %) . The only obvious explanations are that (as the latitudes of noncommitment suggest) women were less involved in von Hindenburg than men or that the communication about his military-political career carried more weight for women or both.
In view of earlier research revealing consistent sex differences in change across topics, it was of interest to compare the number of issues on which the same individual males and females changed. The Kolmogorov-Smirnov two-sample test was used to test whether the frequency distribution according to the number of topics on which persons changed was stochastically greater for women than men. There were no significant differences in the probability that men or women would change on 1, 2, 3, or 4 of the topics.
Tittler's findings indicate that degree of involvement in the issue is inversely related to susceptibility to change. Even though initial positions were not extreme, more involving attitudes were less susceptible to change. Further, the results raise the question that frequently reported sex differences in susceptibility to change may relate more to the nature of topics chosen by the researcher and to the demand character of the research situation than to generalized sex role predispositions. The masculine and feminine issues did not produce differences between men and women. Neither was there a significant general tendency for women to change more than men, the one significant difference (Hindenburg) emphasizing the issue-specific nature of sex differences in susceptibility to persuasion.
STUDY V 
Does Greater Involvement Increase Attitude-Action Consistency and the Effectiveness of Social Pressure to Become Engaged?
Kelly investigated the probability of action directly related to issues differing in ego involvement. The main hypotheses were (a) that the probability of action related to a person's attitude would be an increasing function of the degree of ego involvement in the issue and (b) that the effectiveness of successive social pressures exerted upon the per-
( 323) -son to act would increase with degree of involvement in the issue.
Selection of Issues
In pilot work, women in introductory psychology classes ranked 20 topics according to how interested and concerned they were with each, also rating each topic on a 7-point scale ranging from "extremely interested or concerned" to "extremely disinterested or unconcerned." The purpose of the pretests was to select three topics differing reliably in the concerns of women students and also amenable to devising action situations in the campus setting.
On the basis of the pretest ratings, three issues were chosen: personal-social development, achievement in education, and the social-civil rights of blacks-ranked in that order of importance. Subsequent pilot work in standardizing items and procedures for the own-categories techniques used in the research indicated, however, that this ordering of the topics based on the rating procedure might be misleading. As expected, the fewest categories and the greatest rejection occurred for the personal social-development topic. However, the largest number of categories and the least rejection were found for the education topic, with the black issue being intermediate in both respects. Accordingly, on the basis of previous research using these measures as indicators of involvement and with due respect for the limitations of self-rating procedures, the three issues to be used in the research were classified as follows: highly involving —personal-social development; involving—civil and social rights of blacks; less involving— achievement in education. This order was determined prior to the data analysis.
For each issue, a pool of 100 items was devised from statements in available attitude tests, from the California and Gordon Personality Inventories (for the personal-social topic), and from interviews conducted to obtain the prevailing extremes and midrange of positions current on the topics in the student population.
Women psychology students who did not participate in the experiment itself judged each set of statements as to their favorableness-unfavorableness on the issues, following the Thurstone equal-appearing intervals method. Order of presentation of the issues was counterbalanced.
Fifty items on each topic were selected with Thurstone scale values representing the entire range, such that for each topic approximately 15 statements represented the favorable end (scale values 0-3.9), 15, the unfavorable end (7.1-10.0), and approximately 20 items, intermediate scale values (4.0-7.0). These 50 items for each topic were used in the experiment. All were included in the own-categories procedure. Half of the subjects responded to the items in a Likert-type format, and their summated scores were based on those items that proved to discriminate between students with high and low total scores.
Two hundred and fifteen women students in introductory psychology volunteered to participate in research on social judgment, no mention being made of their attitudes or possible future actions. Subjects were restricted to women to avoid interaction effects involving sex and involvement in the issues, as well as researcher-subject interactions according to sex. Each student received one research credit for making the judgments (regardless of subsequent action or inaction). Owing to mass administration of procedures on a sequence of forms, 27 students were eliminated for failure to follow instructions or to complete any one of the three forms, leaving a total of 188 students for the subsequent call to action. Since random assignment of subjects to one of the three issues for the call to action was made at the time of the administration of forms, the eliminations affected randomization adversely, chiefly by overrepresenting those assigned to the black issue and to the Likert procedure.
The students were randomly assigned to respond to the attitude items either in a Likert format (n= 98) or using the own-categories procedure (n = 90). Each responded to the same type of procedure on each of the three topics, the topics appearing in counterbalanced order. Then a series of briefly described situations pertaining to each of the three topics was presented, the task being to indicate which of five alternatives one would most probably follow if faced by the situation. (Since a satisfactory index of reproducibility for Guttman scaling of the alternatives was not obtained, scores on this form are not reported here.) Finally, through random assignment, each student was asked for signed commitment to act on a series of alternatives pertaining to one of the three topics.
From 1 to 3 weeks after the session just described, each student was sent a letter informing her that an organizational meeting was to be held on the one topic to which she had been assigned (randomly) to secure verbal commitment. She was requested to attend at the time and place stated, or to return the letter indicating a date and time when she could meet with the researcher to learn what had transpired and what she could do to participate, or to return the letter indicating that she had no interest in the activity.
Graduate students personally interested in each of the topics conducted the meetings, providing discussion and information that students could use personally. The meeting on social-personal development concerned opportunities to participate in human-relations (sensitivity) training. The meeting on the black issue concerned organized attempts on campus to increase interracial understanding and acquaintance. The meeting on education concerned a series of small tutorials geared to learning on topics of relevance to undergraduates.
Attendance at the meetings was low. A series of social pressures by the researcher was then initiated.
( 324) Those who had not attended or had returned the letter giving alternative dates were contacted by phone and asked to make an appointment. Then, those who missed an individual appointment made either by letter or by phone were called and asked to reschedule the appointment. There were six distinct possible outcomes in response to the call to action (see Table 7 ).Results
The attitude positions of the students were calculated separately for those following owncategories procedures and those taking the Likert tests. On the own-categories procedure, the criterion for attitude position was the median Thurstone scale value for items placed within the person's latitude of acceptance. By this criterion, the ranges of attitude positions for students who received a call to action on each of the three topics were small (personal development: 1.47-6.94; black issue: 1.216.02; education: 1.42-5.76). While strict comparison between attitude position in the Likert and in the own-categories data is not possible, the distributions of Likert scores relative to the possible range were similarly restricted.
Accordingly, the student's attitude position was defined as very favorable, favorable, or moderate according to the location of her score in the upper, middle, or lower third of the distribution on the procedure she had followed. The effect of this division is seen most clearly with reference to the Thurstone scale values for items defining the thirds and the item content. For example, on the black issue, the very-favorable third had a range of median scale values (1.21-3.73) whose least favorable statement was represented by "I would make a small cash contribution to help blacks on campus" (scale value = 3.62). The second third, labeled favorable, had a range of median scale values of 3.79-4.44. The lowest third, labeled moderate, had a range of median scale values of 4.51-6.02, the least favorable being represented by the statement "Blacks on campus usually prefer the company of their own race to the company of whites" (scale value = 6.02).
The relative unimportance of the extremity of attitude position within the restricted ranges represented on the three issues was apparent when the relative frequencies of those who did take some form of action (with or without continued pressure) were compared with those who did not take action (with or without social pressure) according to whether their attitude was classified as very favorable, favorable, or moderate regardless of testing method. Favorableness of position was not" significantly related to action or nonaction (personal development: X' = 2.09, p > .30; black rights: X` = 3.98, p > .10 < .20; education: X2 = 2.403, p > .30). Owing to the restricted range, these findings should not be construed as indicating the triviality of attitude position. They do warrant presentation of the findings without regard to attitude position but according to the relative involvement of the topic, which is of primary interest. -'
Table 7 presents a summary of the frequencies of response to each of the possible alternatives to the call for action on each issue. Its significance can be viewed relative to' the general problem of the consistency or inconsistency of attitude-related behavior (cf Campbell, 1963). Of particular interest is Alternative 5 listed in the table, which shows the relatively small frequencies of individuals
|Highly involving (personal adjustment)||Involving (black rights)||Moderately involving (education)|
|1. Attend meeting||21.66||7.32||8.33|
|2. Kept appointment made by letter||10.00||8.83||6.66|
|3. Kept appointment made by phone||25.00||27.94||23.33|
|4. Kept rescheduled appointment made phone ∑=||6.6663.33||4.4248.51||0.0038.33|
|5. Missed second scheduled appointment||8.33||4.42||8.33|
|6. Not interested (letter or phone) ∑=||28.3336.66||47.0651.48||53.3361.66|
|Note. For the highly involving topic, n = 60. For the involving topic, n = 68. For the moderately involving topic, n = 60.|
(325) who were definitely not consistent by Campbell's criterion, in that they agreed on two occasions to appointments and broke both.
The proportion of those who stated by letter or on the phone that they would not attend a meeting or make an appointment (and hence were consistent) varied inversely with the involvement of the topic, being smallest for personal adjustment (28.33%) and greatest for the education topic (53.33%). Conversely, the percentages of those who acted consistently by attending meetings or keeping appointments increased regularly according to the predetermined order of involvement in the topic.
Statistical tests of frequencies in the various action-no-action categories according to the issues confirmed expectations. The differences in relative frequencies of action versus nonaction on the three issues were significant (X2= 7.59, df = 2, p < .01, one-tailed test), with the proportions who acted increasing with higher involvement in the issues and those not acting regularly decreasing. A more rigorous test compared the relative frequencies of those attending the meeting in response to the letter without further pressure (Alternative 1) with those acting after the researcher telephoned and those not attending. The chi-square was significant (X2= 12.74, df = 4, p < .01, one-tailed test), with departures from chance expectancy reflecting the predicted trends.
Thus, the findings indicate that, in addition to other variables that affect engagement in activities related to attitude (e.g., potential evaluation by one's reference groups, social constraints, competing attitudes and activities, time schedules, etc.), the degree of personal involvement affects the likelihood of action or inaction, including the likelihood that social pressure to act will be effective.
Findings from the five studies support the following conclusions:
1. The relative sizes of the latitudes of acceptance, rejection, and noncommitment indicate degree of involvement across attitude positions (extreme-moderate). Since indicators were derived from two independent methods, they are not merely artifacts of one or the other (cf. Sherif, 1972).
2. Different degrees of involvement produce predictable differences in social judgment among persons with the same attitude position, with high involvement resulting in less fine discrimination (use of fewer categories, categorical rejection of discrepant alternatives, and reduced noncommitment-Study I).
3. Degree of involvement, as revealed in social judgments, is related to the priority of observed normative concerns (values) in reference groups of which the person is an identified member (Studies II and III). This finding represents concrete linkage of the group and psychological levels of analysis.
4. Degree of involvement in different issues is related to judged credibility of communicators such that those highly involved are more selective in attributing credibility and more prone to emphasize discrepancies from communicators on issues of high priority (Study III).
5. Susceptibility to change in response to short-term exposure to communication is inversely related to degree of involvement in the issue at hand (Study IV).
6. The probability of engaging in action related to attitude and of complying to social pressures to act are positively related to degree of involvement in the attitude objects (Study V).
The research relied on a combination of methods to ascertain attitude and involvement, including unobtrusive observation of public actions and rankings of importance. Judgment indicators of involvement included the sizes of latitude of rejection (for bipolar controversial issues) and the latitude of noncommitment (for conformity-deviation gradients represented by personal appearance and sex role behaviors). However, these single latitudes were convenient shortcuts in that both theoretically and empirically the basis for involvement indicators is structure of the person's own categories as he interacts with a relevant stimulus domain and is free to categorize without restrictive instructions. The three latitudes are interrelated parts of the structure, none being "primary" in the sense of invariant determination of the others. Thus
( 326) indicators based on all three latitudes (as in Study I) are more discriminating and more veridical to theory and findings than a single latitude containing only partial information on structure. Research is currently underway to develop a quantitative indicator of involvement better reflecting structure, which would also permit the ranking of individuals as to their relative involvement in the situation.
The indicators rest on certain procedural assumptions. Both methods used here assume that the entire range of the stimulus domain be represented as nearly as possible (a possibility sometimes limited, as we found for future activities of adolescent girls, where propriety dictated the truncation of deviant extremes). Relationships among latitudes do vary with the range presented, resulting for example in finer discrimination within the acceptable latitude when objectionable extremes are absent (Sherif, 1961). The method of ordered alternatives assumes that alternatives were reliably ranked by judges upholding different positions, an assumption whose violation produces different structures (Koulack, 1970). Finally, use of the indicators for involvement assumes that the person is not confronted directly with the knowledge that personal involvement is being studied and that procedures are carefully administered without possible modeling effects from others performing the tasks.
Future basic research may profitably inquire into other variables that may affect the involvement indicators, such as background differences, information, individual differences in cognitive structure or style, and situational factors. For example, it is reasonable to contend that the greater susceptibility to change by Study IV's subjects on the less involving topics reflected lack of information and the learning of that supplied. Such change may be more aptly termed learning or attitude formation, terms equally appropriate to much "attitude change" research using unfamiliar, neutral, or trivial stimuli.
In exploring relationships between involvement and information, however, we may need to formulate information concepts and tools more differentiating than mere quantity or accuracy of recall for preselected facts. The more involved Indian students who were observed regularly seeking more information about India were, nonetheless, less discriminating in dealing with beliefs about India (Study I). Likewise, high-prestige adolescent girls, whose backgrounds and higher grade point averages are associated with more information, discriminated less keenly among future activities than their less favored peers (Study II). Probably, assessment of information-involvement relationships requires consideration of the value premises and sources governing assignment of truth value, for these clearly affect the kinds of information judged as factual or false (Sherif & Jackman, 1966).
It is possible that the indicators are affected by individual differences in cognitive structure or style, since they are based on categorizations. If so, it is significant that repeated measures of categorical structure on different stimulus arrays differ according to the person's relative involvement (Studies II, III, and IV). Some consistency has been reported, however, as indicated by low positive autocorrelations (around .30) across tasks (cf. Glixman, 1965; Sherif, 1961). Finally, situational influences may affect the indicators. The latitude of noncommitment, for. example, would seem particularly susceptible`; to variations in social pressure, constraints, threats, or procedures alienating the subjects.
The need for further research on involvement in attitude research is proportional to . its relative neglect in the past. In research on the communication situation, Sereno (1969, ,. p. 74) has even proposed that relationships among communication and source variables with attitude change may have been oversimplified or overemphasized owing to neglect of the involvement variable. Studies by Rhine and Severance (1970) and Gorn (1971) reported interaction effects for involvement and ' communicator credibility such that the wellknown source-credibility effect on attitude change virtually disappeared with sufficiently:. high involvement.
Similarly, as Study V indicates, involvement is one parameter defining the situational threshold for action (in the sense used by . Campbell, 1963) and hence may interact with other variables known to affect the probability of attitude-consistent actions, such as social constraints within and outside of the
( 327) research situation and the normative climate of institutional settings and of one's reference groups (Bowers, 1968; DeFleur & Westie, 1958; Fendrich, 1967; Linn, 1965; Warner DeFleur, 1969).
Finally, the involvement variable needs fuller consideration in extending the implications of research on games and negotiations in ad hoc groups to critical real-life conflicts. The stalemates so frequent in intergroup conflict (e.g., Blake, Shepard, & Mouton, 1964; Sherif et al., 1961) also occur in ad hoc dyads when the pairs' involvement is sufficiently high as indicated by their initial latitudes (Sereno, 1969). Such indicators also permit predictions about those alternatives where mutual agreement is potentially feasible or stalemate most likely through comparing the locations of initial latitudes of acceptance, rejection, and noncommitment of the antagonists.
Realistic research on attitude formation and change will eventually link degree of identification with reference groups, the priority of their normative concerns (values), and personal involvements with the individual's behavior over time. Such research requires both field- and laboratory-type methods as well as convergent measurements of variables in a variety of situations. One step in this direction is development of more valid indicators of personal involvement, to which this research is intended as a contribution.
BECK, D., & NEBERGALL, R. E. Relationship between attitude neutrality and involvement. Paper presented at the annual meeting of the Speech Assocation of America, Los Angeles, 1967. (Mimeo)
BIERI, J. Attitudes and arousal: Affect and cognition in personality functioning. In C. W. Sherif & M. Sherif (Eds.), Attitude, ego-involvement and change. New York: Wiley, 1967.
BLAKE, R. R., SHEPARD, H. A., & MOUTON, J. S. Managing intergroup conflict in industry. Houston: Gulf Publishing, 1964.
BOWERS, W. J. Normative constraints on deviant behavior in the college context. Sociometry, 1968, 31,370-385.
CAMPBELL, D. T. Attitudes and other acquired behavioral dispositions. In S. Koch (Ed.), Psychology: A study of a science. Vol. 6. New York: McGraw-Hill, 1963.
DEFLEUR, M. L., & WESTIE, F. R. Verbal attitudes and overt acts: An experiment on the salience of attitudes. American Sociological Review, 1958, 23, 667-673.
EDWARDS, A. L. The social desirability variable in personality assessment and research. New York: Dryden, 1957.
FENDRICH, J. M. A study of the association among verbal attitudes, commitment and overt behavior in different experimental situations. Social Forces, 1967, 45, 347-355.
FESTINGER, L. The theory of cognitive dissonance. New York: Harper, 1957.
FISHBEIN, M. Attitude and the prediction of behavior. In M. Fishbein (Ed.), Readings in attitude theory and measurement. New York: Wiley, 1967.
GLIXMAN, A. R. Categorizing behavior as a function of meaning domain. Journal of Personality and Social Psychology, 1965, 2, 370-377.
GORN, G. Effects of ego-involvement, communication discrepancy, and communicator prestige on attitude change. Unpublished doctoral dissertation. Pennsylvania State University, 1971.
HARVEY, O. J. Concluding comments on the current status of the incongruity hypothesis. In 0. J. Harvey (Ed.), Motivation and social interaction. New York: Ronald Press, 1963.
JANIS, I., & FIELD, P. B. Sex differences and personality factors related to persuasibility. In C. I. Hovland & I. L. Janis (Eds.), Personality and persuasibility. New Haven: Yale University Press, 1959.
KOULACK, D. A test of social judgment theory with the use of an interval scale. Journal of Social Psychology, 1970, 82, 275-276.
LINN, L. S. Verbal attitudes and overt behavior: A study of racial discrimination. Social Forces, 1965, 43, 353-364.
McGUIRE, W. J. The nature of attitudes and attitude change. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology. Vol. 3. Reading, Mass.: Addison-Wesley, 1969.
NEWCOMB, T. M. Attitude development as a function of reference groups: The Bennington study. In G. E. Swanson, T. M. Newcomb, & E. L. Hartley (Eds.), Readings in social psychology. New York: Holt, 1952.
OSGOOD, C. E., SUCCI, G. J., & TANNENSAUM, P. H. The measurement of meaning. Urbana, Ill.: University of Illinois Press, 1957.
RHINE, R. J., & SEVERANCE, L. J. Ego-involvement, discrepancy, source credibility, and attitude change. Journal of Personality and Social Psychology, 1970, 16, 175-190.
RODGERS, H. L., JR. Reference groups as determinants of the judged credibility of communicators and their judged position on issues. Unpublished master's thesis, Pennsylvania State University, 1968.
ROLLINS, J. H. Reference identification of youth of different ethnicity. Journal of Personality and Social Psychology, 1973, 26, 222-231.
ROSENTHAL, R., & ROSNOW, R. L. Artifact in behavioral research. New York: Academic Press, 1969.
SARUP, G. Reference groups: Some determinants and consequences. Unpublished doctoral dissertation, Pennsylvania State University, 1969.
SCHULMAN, G. I., & TITTLE, C. R. Assimilationcontrast effects and item Selection in Thurstone scaling. Social Forces, 1967, 46, 484-491.
SERENO, K. K. Ego-involvement: A neglected variable in speech-communication research. The Quarterly Journal of Speech, 1969, 60, 69-77.
SHAW, M. E., & CONSTANZO, P. R. Theories of social psychology. New York: McGraw-Hill, 1970.
SHERIF, C. W. Established reference scales and series effects in social judgment. Unpublished doctoral dissertation, University of Texas at Austin, 1961.
SHERIF, C. W. Comment on interpretation of latitude of rejection as an "artifact." Psychological Bulletin, 1972, 78, 476-478.
SHERIF, C. W. Social distance as categorization of intergroup interaction. Journal of Personality and Social Psychology, 1973, 25, 327-334.
SHERIF, C. W., & JACKMAN, N. R. Judgments of truth by participants in collective controversy. Public Opinion Quarterly, 1966, 30, 173-186.
SHERIF, C. W., SHERIF, M., & NEBERGALL, R. E. Attitude and attitude change: The social judgmentinvolvement approach. Philadelphia: Saunders, 1965.
SHERIF, M. The psychology of social norms. New York: Harper, 1936.
SHERIF, M. Some needed concepts in the study of attitudes. In J. Peatman & E. L. Hartley (Eds.), Festschrift for Gardner Murphy. New York: Harper, 1960.
SHERIF, M. Self concept. International encyclopedia of the social sciences. Vol. 14. New York: Macmillan, 1968.
SHERIF, M., & CANTRIL, H. The psychology of ego involvements. New York: Wiley, 1947.
SHERIF, M., HARVEY, O. J., WHITE, B. J., HOOD, W. R., & SHERIF, C. W. Intergroup conflict and cooperation: The Robbers Cave experiment. Norman, Okla.: University of Oklahoma, 1961.
SHERIF, M., & HOVLAND, C. I. Social judgment: Assimilation and contrast effects in communication and attitude change. New Haven: Yale University Press, 1961.
SHERIF, M., & SHERIF, C. W. Attitude as the individual's own categories: The social judgment involvement approach to attitude and attitude change. In C. W. Sherif & M. Sherif (Eds.), Attitude, ego-involvement and change. New York: Wiley, 1967.
SHERIF, M., & SHERIF, C. W. Social psychology. New York: Harper & Row, 1969.
SMITH, M. B. Social psychology and human values: Selected essays. Chicago: Aldine, 1969.
STIFFLER, M. K. The influence of peer group membership on some social attitudes of adolescent girls. Unpublished master's thesis, Pennsylvania State University, 1968.
TITTLER, B. I. The relationship between attitude change and ego involvements and its relevance to sex differences in attitude change. Unpublished master's thesis, Pennsylvania State University, 1967.
TRIANDIS, H. Exploratory factor analyses of the behavioral component of social attitudes. Journal of Abnormal and Social Psychology, 1964, 68, 420430.
TRIANDIS, H. C. Toward an analysis of the components of interpersonal attitudes. In C. W. Sherif & M. Sherif (Eds.), Attitude, ego-involvement and change. New York: Wiley, 1967.
WARNER, L. G., & DEFLEUR, M. L. Attitude as an interactional concept: Social constraint and social distance as intervening variables between attitudes and action. American Sociological Review, 1969, 34, 153-164.
ZAVALLONI, M., & COOK, S. W. Influence of judges' attitudes on ratings of favorableness of statements about a social group. Journal of Personality and Social Psychology, 1965, 1, 43-54.
(Received February 18, 1972)
- The article reports five studies in a continuing research program which has been directed from its inception by Muzafer Sherif. Research reported here was supported by grants to the program from the Rockefeller Foundation and the National Science Foundation between 1966 and 1970 when the first author was coinvestigator and the other authors were graduate students
- Requests for reprints should be sent to Carolyn «'. Sherif, Department of Psychology, Pennsylvania State University, 417 Psychology Building, UniversityPark, Pennsylvania 16802.
- Noncommitment is, therefore, not equivalent to neutrality. It refers to any stimulus not included in latitudes of acceptance or rejection. Variables associated with this lack of inclusion include ego involvement, but could also include ignorance, boredom, fear, etc.
- Study I was conducted by Sarup (1969).
- Study II was conducted by Merrilea Kelly Stiffler (1968).
- This unit in a research program on natural groups directed by Muzafer Sherif was supervised by the first author. The judgment data were analyzed by the first author from Stiffler (1968). The observational data were collected by Stiffler as an assistant in the research program, supported by the National Science Foundation. The authors appreciate the cooperation of the School Board, administration, and counseling program of the State College Area High School.
- Study III was conducted by Rodgers (1968).
- Study IV was conducted by Tittler (1967).
- Study V was conducted by Merrilea Kelly. Owing to her untimely death, her doctoral dissertation entitled "Attitude, Involvement and Social Constraint Effects on Verbal and Behavioral Response Consistency," 1970, which included the research summarized here, was not completed. This study is based on the dissertation draft and data subsequently compiled by Kelly (Table 7), which were analyzed by the first author. Extensive correlational analysis between attitude measures, behavioral intentions, signed commitment, and sources assigned for various actions is omitted here, primarily because the truncated range of attitude positions and the assumptions in scoring some measures limit its implications. The order of issues according to relative involvement was based on pretest data collected before the experiment itself, and was not a post hoc decision by the first author, as explained in the text.