Comment on Interpretation of latitude of rejection as an 'artifact.'
Carolyn W. Sherif
Pennsylvania State University
The use of latitude of rejection as an indicator of involvement is based on findings from two dissimilar procedures for attitude research. Interpreting this indicator as artifactual owing to its correlation with own position, Markley neglected one procedure, assumed equal intervals (untenable for both methods), ignored variability of the rejected latitude across positions, and constructed a threshold measure unrelated to the threshold concept in the research under review. Available research not reviewed supports the indicator of involvement for persons upholding positions other than the extremes, although as previously proposed the relative sizes of acceptable, objectionable, and noncommittal latitudes should serve as a more adequate indicator.
Several errors of interpretation and fact in Markley's (1971) article on latitude of rejection as an indicator of involvement call for correction.
Markley's article discussed the latitude of rejection as an indicator of personal involvement in an attitude object, confining his speculation to the method of ordered alternatives. The method of ordered alternatives is one technique for research on the frequency of judgments in evaluative categories as a function of the person's most acceptable position (own position) and relative involvement in the object of judgment (Sherif & Hovland, 1961; Sherif & Sherif, 1967; Sherif, Sherif, & Nebergall, 1965). Markley's critique assumed, for "didactic" purposes, that items used in this method had equal intervals for acceptance, rejection, and noncommitment, and attempted to demonstrate that the size of the latitude of rejection is "artifactually contaminated" owing to its correlation with extremity of own position.
The basic error is the author's persistence in assuming a model of scaling (equal intervals) that was never assumed in construction or use of the method of ordered alternatives, as he recognized (Markley, 1971, p. 359). This method was, in fact, an attempt to study attitude without certain arbitrary assumptions of conventional scaling models,
particularly equal-appearing intervals, an assumption which, in Markley's words, "is not achieved in most applications and can never be perfectly validated [p. 359]."
Relationships between the latitude of rejection, own position, and ego-involvement , (including those cited by Markley, 1971, p. 359) were based on data from the method of ordered alternatives and the own-categories procedure (Sherif et al., 1965). In both procedures, the size of the objectionable latitude is determined simply by the relative fre-quency of items rejected. In the own-categories procedure, only the extreme items are unambiguously ordered (low variability of judged position). The subject is free to use 3 any number of categories and to distribute the items into them as he sees fit in order to group items whose position he sees as belonging together. Thus, Markley's didactic exercise of assuming equal item intervals for ordered alternatives (which did not make the assumption) would be equally erroneous for the own-categories procedure, which he regards as a "viable alternative [p. 359].">
The sources cited by Markley repeatedly emphasize relationships among the relative 1 sizes of acceptable, objectionable, and non- j committal latitudes, rather than the absolute frequency within any single latitude. Nevertheless, the latitude of rejection was used as an indicator of involvement on the basis of findings from both methods, which were in agreement. Markley cited only one of the conclusions supporting latitude of rejection
( 477) as an indicator of involvement, that is, the correlation between extremity of position and latitude of rejection. The inference that this 'correlation was related to greater involvement of extreme subjects was, in turn, based on (a) the well-known U curve relating extremity and intensity ratings (e.g., Cantril, 1946) and (b) the large latitudes of rejection typically obtained from extreme subjects actively committed to upholding their stands (Sherif et al., 1965, p. 29).
The second finding supporting latitude of rejection as an indicator of involvement was its variability across attitude positions from ,extremes through moderate own positions (Sherif et al., 1965, p. 156). Thus, another error in Markley's comments lies in constructing a "didactic" model of three response sets for different extremities in own position :(Table 1, p. 358) without considering variability among persons upholding the same position. In a study of reactions to communications on a national election, there were sufficient numbers of subjects adopting each of wine positions (extreme through moderate) who had latitudes of rejection composed of give or more positions (as did only the extreme response set in Markley's Table 1) to compare their ratings of communications with subjects adopting each of nine positions who :ejected four or fewer items (Sherif et al., ;1965, pp. 156 ff.). Designating subjects across the extreme and moderate positions as high n involvement if they rejected five or more positions, the research found significant differences between their judgments of communications and those by less involved subjects (latitude of rejection of four or fewer items).
The fact that subjects adopting moderate positions do, in substantial numbers, exhibit attitudes of rejection as large as the typical extreme subjects is a blow to the main thrust of Markley's critique based on the unwaranted assumption of equal item intervals. His key proposition is that for the person adopting a position at one extreme "the latitude of rejection ... is spuriously lengthened due to the increased distance to the most rejected item [p. 358]." The proposition implies that persons taking extreme positions reject more items simply because there is more "room" to do so between their own position and the most rejected item. Markley seems to forget that there is no logical or procedural requirement in the ordered alternatives method that the extreme subject do so. Instructions require only that he indicate his most acceptable and objectionable positions, leaving him free to reject or accept others if he wishes or not to respond to other items (remain noncommittal).
Many highly involved persons taking extreme positions do respond as in Markley's Response Set 1 (Markley, 1971, Table 1), rejecting all opposing positions and the position E (see Markley's Table 1), which is a noncommittal statement. If they do so because of an "artifact," such highly involved persons should respond differently when using the own-categories procedure. Yet with that procedure, such highly involved subjects typically employ very few categories and construct an extremely broad objectionable category, lumping all of the items opposing their own stand together. The proposition that the wide latitude of rejection thus obtained is "artifactual" is all the more difficult to sustain, as an equal-interval assumption is equally untenable, and the items are arranged in random order so that the subject may encounter the most objectionable item in any serial position.
It is not entirely clear why Markley coupled his proposition about distance to the most objectionable position with the possibility of a ceiling effect for the subject adopting the most extreme position. It is difficult to envision even a Republican zealot endorsing a more extreme position than Position A in Markley's Table 1 (p. 358), namely: "The election of the Republican presidential and vice presidential candidates in November is absolutely essential from all angles in the country's interests." In fact, even this position was so extreme that comparatively few of the subjects who were avowed and active supporters of the Republican candidates adopted it as most acceptable. The bulk of such committed Republican supporters adopted the adjacent Position B as most acceptable, with more than 50% of them fail-
(478) -ing to endorse the extreme Position A, and nearly 20% rejecting this adjacent, more extreme position (Sherif et al., 1965, p. 37). Therefore, the addition of a still more extreme position in the ordered alternatives might well have increased the latitude of rejection by providing an alternative "too extreme" even for the extremists.
Markley's interpretation of "threshold" of rejection, in terms of number of positions between the most acceptable position and the first rejected position, is not only "novel." It has nothing to do with the use of the threshold concept in the references cited, which refers to the probability that a stimulus will be placed in a given category rather than an adjacent or other category. This is the usual definition of the difference threshold in psychophysical research. Far from being "essentially an interval measure [p. 359]," the difference threshold is a means of defining limits of psychological categories in terms of stimulus values. True, the stimulus values in most psychophysical research are arrayed in equal intervals, but application of the threshold concept to problems of evaluative categories (acceptance, rejection, noncommitment) need not make this assumption in view of the fact that evaluations are necessarily human judgments and need not follow the regularities of physical dimensions.
If Markley's basic exercise, assuming equal intervals, and his conclusion of an "artifactual cortamination",oflatitudes of rejection for extreme subjects re valid, by the same token moderate sub;` _ should be subject to the converse artifact, .preventing them from rejecting as many: positions because their own position 1G e to both extremes. As noted above, substantial numbers of moderate subjects do reject five or more nut of nine positions. In work published before Markley's (1971) article was received for publication, there is further evidence that a moderate or noncommittal subject can be highly involved and reject items as frequently as extreme subjects (Sherif & Sherif, 1969). For example, subjects noncommittal on the choice of available candidates in an election (adopting Position E in Markley's Table 1, p. 358), but active and highly involved in the outcome, display latitudes of rejection comparable in size to those upholding extreme positions (adopting Position A in Table 1, p. 358). Politically inert subjects upholding the same position (Position E) typically evaluate the alternatives as in Markley's Response Set 3, reject-1 ing fewer positions. Similarly, those siding moderately with one candidate or another, whether Republican or Democratic, reject l significantly more items if they are politically active than if politically inactive (Sherif & Sherif, 1969, p. 365). These, and several other more recent investigations, permit the conclusion that subjects upholding positions other than the extremes can be highly involved in them and that when they are, they exhibit latitudes of rejection comparable to' those of the typical subject upholding an extreme position. Hopefully, these comments may indicate some dangers in a selective review based on criteria (psychometric scaling and equal-appearing intervals) that the research under, review explicitly rejects as stifling progress in the important but muddied problem area of attitude and attitude change.
CANTRIL, H. The intensity of an attitude. Journal o ,Abnormal and Social Psychology, 1946, 41, 129-1, 1.35.
MARKLEY, 0. W. Latitude of rejection: An artifact of own position. Psychological Bulletin, 1971, 75, 357-359.
SHERIF, M., & HOVLAND, C. 1. 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.
SHERIF, C. W., SHERIF, M., & NEBERGALL, R. Attitude and attitude change: The social judgment-involvement approach. Philadelphia: Saunders.1965.
(Received June 1, 1971)
Requests for reprints should be sent to Carolyn W. Sherif. Pennsylvania State University, 417 Psychology Building, University Park, Pennsylvania 16802.