Fred Heilizer

An Ipsative Factor Analysis of the Ipsative Epps

This srudy was conducted on the assumption that ipsative evaluations constitute a more correct and fruitful use of an ipsative dimension than do normative evaluations. A Q-technique factor analysis was computed from an 85-item correlation matrix of the 15 needs of the Edwards Personal Preference Schedule (EPPS). Eighty items (Ss) were contributed by 20 male Ss in each of four groups: college students, neurotics, personality disorders, schizophrenics; five items (markers) were contributed by a profile of average raw scores for each of the four groups and a scale of compatibility and incompatibility of needs (CI Scale). The correlations, then, were computed among the 85 items upon scores for each of the 15 needs of the EPPS as compared to the normative procedure of computing correlations among the 15 needs upon scores for each of the 85 items. The first eight factors, accounting for 82% of the common variance, were rotated to simple structure by means of the varimax analytic procedure. All Ss and markers were assigned to the factor on which their highest loading occurred. The firsr and second factors are the most clearly related to Ss and marker assignments by diagnostic grollp. The first factor is primarily assigned by personalicy disorder Ss ( N = 9 ) and secondarily by neurotic ( N = 6) and schizophrenic ( N = 5 ) Ss. The respective group profiles are also assigned to the first factor. Some specificity of the second factor is indicated for college student S assignments ( N = 5 ) . The college srudent and CI Scale markers are assigned to the second faccon. There are a sufficient number of S assignments on each of four factors (Factors 1, 2, 3, and 7 ) to permit meaningful profile descriptions. The high scores on the first factor (endurance, abasement, nurturance, order) represent the self description of a personality disorder which is quite opposite to the description which the clinician applies to him. This opposite description is consistent with the frequent discrepancy between the self description of a personality disorder and the description of him by others. The low scores (heterosexuality, aggression, autonomy) correspond closely to Horneys model of developing inrerpersonal relations and to the personality disorder diagnoses of inadequate personality and schizoid personality, i.e., personality trait disturbances. The second factor (high scores on heterosexuality, endurance, dominance,

Social Desirability Effects in the Epps

The EPPS and a true-false version (TF-EPPS) were compared on three S dimensions: sex (males and females), age-education (college students and adults), SD (high and low). The SD content of both tests is prominent, although it is much greater in the TF-EPPS. However, SD is not important for differential functioning of Ss, since profile differences between high and low SD Ss are smaller than between males and females or college students and adults The matching format of the EPPS produces a slight reduction in differential SD effect and an increased sensitivity to S variables other than SD. A potential, subtle, SD mechanism is identified by which Ss may respond differentially. Implications for ipsative and normative measurement and evaluation are discussed.

COMPARISON OF HOW TWO CONTENT VARIABLES ARE CHANGED BY THE SPECIAL FORMAT OF THE EPPS.

This study compares the changes of the two content variables, social desirability (SD) and compatibility-incompatibility (CI), in response to the paired comparisons format of the EPPS, a special format which was intended to minimize the importance of one variable, SD, but not the other. The simple linear SD effect is greatly and consistently reduced. However, the males and females show an increased differential curvilinear response to the reduced SD content. The CI changes are more moderate and selective, generally consisting of increased responsivity to the CI content on the nine compatible needs. The method of analysis of this study occasions a rebuttal to a previous study on SD changes.

A NOTE ON VARIANCE HETEROGENEITY IN THE ANALYSIS OF VARIANCE

This article reviews and summarizes the empirical sampling studies of the effects of variance heterogeneity in the analysis of variance. The general conclusion of these studies, that heterogeneity of variance is not an occurrence of importance, is only demonstrated for the simple randomized design; it remains to be demonstrated for the more complex designs. There is also a general inattention to E(MS) and to the effects of the sampling model upon E(MS) and F.

COMPARISON OF SOCIAL DESIRABILITY EFFECTS ON THE EPPS IN RELATION TO RATINGS OF COMPATIBILITY OR INCOMPATIBILITY OF THE NEEDS

The mutually facilitating or inhibiting property of the needs of the EPPS, as measured by compatibility-incompatibility (CI) ratings, is an important determinant of the need-profiles; for the nine compatible needs it is a more important determinant than is the SD property of the needs. A tendency toward different functioning of the CI and SD ratings is noticeable on the compatible needs and marked on the incompatible needs. For all Ss, regardless of group, and for each set of needs, the scores are a direct linear expression of SD and a simple curvilinear expression of CI. SD mostly determines profile differences between men and women, while CI mostly determines profile differences between college students and adults.