Latitude of Rejection: An Artifact of Own Position[1]
O. W. MARKLEY [2]
Educational Policy Research Center, Stanford Research Institute, Menlo Park,
California
It is shown that the latitude of rejection of attitude statements as measured by the method of ordered alternatives is artifactually contaminated by the extremity of own position. A more direct measure of the threshold of rejection was tried. Differential "ego" or attitudinal involvement across varying own positions cannot be validly inferred by use of either the latitude of rejection (as has been previously done) or by the newly introduced measure of the threshold of rejection since the latter ideally requires the property of equal intervals, which is not achieved in most applications. It is therefore recommended that the use of the latitude measures for comparison of differential involvement between respondents be limited to responses of equal extremity of own position.
Method of Ordered Alternatives
An attitude scaling method which has generated a sizable amount of research is the method of ordered alternatives (Sherif & Sherif, 1967, p. 116; Sherif, Sherif, & Nebergall, 1965, Chapter 2). In this method, a number (typically nine) of similarly worded attitudinal statements ranging from extremely "pro," through neutral, to extremely "anti" an issue are presented in rank-ordered sequence. The complete set of nine statements is presented on each of four sheets assembled into a booklet. On the first sheet the subject is asked simply to indicate the statement most acceptable to him; on the second, to indicate any other statement or statements which are also acceptable or not objectionable; on the third to select the statement most objectionable; and on the fourth to indicate any other statement or statements also objectionable.
The method of ordered alternatives yields four measures useful in attitude scaling: (a) "Own" position: the statement found most acceptable to the respondent; (b) latitude of acceptance: the number of statements designated acceptable or not objectionable; (c) latitude of rejection: the number of statements designated objectionable; (d) latitude of noncommitment: the number of statements neither designated acceptable nor objectionable (Sherif et al., 1965, p. 30).
The development of the method of ordered alternatives was based on results of studies of social judgment which indicated (a) that "strong commitment to a position involves a lowered threshold of rejection; that is, the subject tends to see positions different from his own as objectionable and to reject them categorically [Sherif et al., p. 30] (cf. Hovland & Sherif, 1952; Sherif & Hovland, 1961) ; and (b) that the "degree of personal involvement with a stand varies with extremity of the stand [Sherif et al., 1965, p. 26]."
Although involvement and extremity of an attitude are not identical constructs, they are typically correlated and are difficult to separate operationally (Ward, 1966). Thus, finding that the results of a number of studies supported the above hypotheses, the size of the latitude of rejection was adopted as an operational index of involvement (Sherif & Sherif, 1967, p. 120).
Artifactual Contamination of Latitude Scorings
If the size of the latitude of rejection is to be an operational index of involvement, and if involvement and extremity are highly correlated, it is important that the latitude of rejection be free of any artifact associated with extremity, per se. That is, any association of the latitude of rejection with extremity due
| TABLE 1 | |||||||||||||
| DIDACTIC RESPONSES TO THE METHOD OF ORDERED ALTERNATIVES ASSUMING EQUAL ITEM INTERVALS FOR ACCEPTANCE, NONCOMMITMENT, AND REJECTION WITH DIFFERENT EXTREMITIES OF OWN POSITION | |||||||||||||
| Response set | Scale item | Measure | |||||||||||
| A | B | C | D | E | F | G | H | I | LA | LNC | LR | TR | |
| 1 | √√ | √ | — | — | X | X | X | X | XX | 2 | 2 | 5 | 4 |
| 2 | — | √ | √√ | √ | — | — | X | X | XX | 3 | 3 | 3 | 4 |
| 3 | X | — | — | √ | √√ | √ | — | — | XX | 3 | 4 | 2 | 4 |
| Note. – √√ = most acceptable item, the subject's own position; √ = also acceptable item; — = (blank) noncommital regarding; XX = most rejected item; X = also rejected or unacceptable item; LA = latitude of acceptance; LNC = latitude of noncommitment; LR = latitude of rejection; TR = A novel measure of the threshold of rejection, being the number of items between the subject's own position and the first rejected item. | |||||||||||||
solely to properties of the scaling device would threaten the validity of any inferred involvement from the size of the latitude of rejection.[3]
One way to test the method of ordered alternatives for such contamination is to construct didactic sets of responses with differing degrees of extremeness of "own position," but with equal item intervals for acceptance, noncommitment, and rejection. Table 1 shows three such sets of responses. If extremity had no effect on the latitude of rejection, it should remain constant under the assumption of equal item intervals.
As Table 1 shows, the latitude of rejection is artifactually associated with the extremity of own position. Those subjects who endorse the extreme Item A as their own position find no items of still greater extremity. Hence, that portion of their latitudes of acceptance and noncommitment which is based on items to the left of their own position is curtailed by ceiling effects. At the same time, the latitude of rejection based on items on the right is spuriously lengthened due to the increased distance to the most rejected item. Hence latitude differences between differing extremities of own position under the assumption of equal item intervals in Table 1 are artifactually produced.
The finding of this artifact makes suspect results of previous studies in which conclusions were based on the existence of a valid relationship between involvement and the latitude scores, irrespective of differences in extremity. For example, Powell (1966) found that dogmatism was positively correlated with extremity of own position (r = +.46), and with latitude of rejection ( r = +.33). Although there is no reason to question the positive correlation between dogmatism and extremity, the present analysis suggests that the correlation between dogmatism and the latitude of rejection may be due to the extremity artifact per se, and not to any differences in involvement which are associated with differences in dogmatism.
More Direct Measure of the Threshold of Rejection
For whatever worth it may have as a "negative result," described below is an unsuccessful attempt to devise a scoring method for the method of ordered alternatives would yield an uncontaminated measure of the threshold of rejection. A number of scoring methods were tried. The most promising method was to note the number of items from the subject's own position to the first rejected item. The psychometric advantage of measure as compared to the latitude of rejection can be seen by the fact that it is invariant under the assumption of equal item intervals Table 1.
The new measure was used on data previously published by Sherif et al. (1965, Figures 2.1 and 2.3, pp. 32-52) regarding presidential candidates in the 1960 elections and on data produced by subjects who simulated having varying involvement in an attitude, but with a fixed extremity (Markley, 1967). Analysis
( 359) of these data indicated that the new measure could not validly be used to compare varying involvement across different own positions unless the scale had the property of equal intervals. This is true because a threshold measure is essentially an interval measure. The method of ordered alternatives as typically constructed, however, neither has nor was intended to have this property (Sherif et al., 1965, p. 25). While method-of-ordered-alternative-type scales could perhaps be constructed which would approach the property of equal intervals, this is not achieved in most applications and can never be perfectly validated (Green, 1954).
Conclusion
Although it is not valid to compare either thresholds or latitudes of rejection which are associated with different extremities, their comparison within a given own position is apparently permissible. Where comparisons between attitudes differing in extremity is desired, the "Own Categories Procedure" (Sherif & Sherif, 1967; Sherif et al., 1965) may provide a viable alternative.
REFERENCES
GREEN, B. Attitude measurement. In G. Lindzey (Ed.), Handbook of social psychology. Vol. I. Cambridge, Mass.: Addison-Wesley, 1954.
HOVLAND, C. I., & SHERIF, M. Judgmental phenomena and scales of attitude measurement : Item displacement in Thurstone scales. Journal of Abnormal and Social Psychology, 1952, 47, 822-832.
MARKLEY, O. W. The role playing of "ego-involved" attitudes. Unpublished master's thesis, Northwestern University, 1967.
POWELL, F. A. Latitudes of acceptance and rejection and the belief-disbelief dimension: A correlational comparison. Journal of Personality and Social Psychology, 1966, 4, 453-457.
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, C. W., SHERIF, M., & NEBERGALL, R. E. Attitude and attitude change: The social judgmentinvolvement approach. Philadelphia: Saunders, 1965.
WARD, C. D. Attitude and involvement in the absolute judgment of attitude statements. Journal of Personality and Social Psychology, 1966, 4, 465-476.
(Received December 1, 1969)
