Binyamin Z. Lebovits

Prospective and retrospective psychological studies of coronary heart disease.

&NA; The Minnesota Multiphasic Personality Inventory was administered at the first and fifth annual examinations of 1990 men studied in a prospective investigation of coronary heart disease. These data were used to investigate the following questions‐‐two potential sources of error in retrospective studies: Do survivors of the illness differ, on the variables being investigated, from persons who die before they can be included in a retrospective study? Is occurrence of the illness associated with systematic changes in these variables among survivors? Both questions were answered affirmatively. The results indicate that investigation of relationships between variables of behavior or personality and the occurrence of symptomatic life‐threatening disease should proceed by the prospective method.

Cigarette smoking and personality

Abstract Many attempts have been made to discover the ‘smoking personality’. In general, these attempts at best have involved controlling for the major demographic variables, e.g. age, sex and socioeconomic status. However, the resulting groups still must reflect wide individual differences, and no clearcut relationship between smoking and personality can emerge under these conditions. To test this logically derived conclusion, the Western Electric data were subjected to a number of analyses, each based upon a different mode of analysis and a different way of interpreting the data (ANOVA, factor analysis, and X2; and MMPI t-scores, high point scales, and clinical signs). On the broadcast level—an ANOVA where the variables of classification were age, educational level, defensiveness and smoking level—the MMPI scales are typically associated with smoking. However, the complexity of the ANOVA precludes a simple interpretation of the data. When more homogeneous subgroups are examined, the differences among the subgroups vary from comparison to comparison, and the results do not support the hypothesis that there is a definite personality pattern associated with smoking. Moreover, a factor analysis of the basic data, too, fails to confirm the existence of a smoking personality. These findings suggest that meaningful research will require a classification system—and the establishment of non-smoking control groups—that goes well beyond the use of major demographic variables. Unfortunately, the exact nature and the dimensions of such a system remain to be established. In the interim, both personality theorists and those who seek to modify smoking habits may find it helpful to focus upon the stimulus conditions, external and internal, that are associated with smoking, especially as these relate to sociabilitydepression and tension release.

Smoking and personality: A methodological analysis

Abstract After a careful examination of the literature, we have concluded that few studies meet minimum standards of adequacy. The studies on smoking and personality typically suffer from the following inadequacies: 1. 1. Subject variables . Little attention is paid to the fact that important differences exist in smoking behavior as a function of age, socio-economic status, family smoking habits, occupation, and religion. 2. 2. Measurement of smoking behavior . When classified according to smoking behavior, various studies not only use different criteria but also make use of what appears to be illogical criteria. 3. 3. Independent variables . Scores on personality tests are frequently used as the independent variable. However, little or no attention—and worse yet, mere lip service—might be paid to such things as (a) defensiveness, (b) the unavailability of norms that are adequate for the various age, sex, and socio-economic groups, (c) the effects of different administrations and settings on the tests, and (d) the validity and reliability of the tests that are used. 4. 4. Statistical evaluation . Statistical analyses have been performed (typically, t -tests are used) without taking into consideration the fact that a number of the independent variables are significantly intercorrelated, and that each may actually represent a somewhat different way of measuring the very same variable. Also, large numbers of tests of significance may be made and no mention of this fact is found. Instead, the significant findings alone may be detailed. 5. 5. Derivation of hypotheses . Hypotheses are not derived from theoretical formulations, and little attention is paid to the data that are available in other areas, e.g. child and adolescent development, sociology, group dynamics. However, there are a few exceptions, notably the use of psychoanalytic formulations. A few authors have attempted to utilize psychoanalytic constructs and to explain smoking behavior as a function of them [30, 31]. Here, though, we have serious questions about the meaningfulness of an independent variable that is studied in retrospect. There is little ground to assume that the reports about the behavior of parents are accurate ones, and we have little faith in the assumption that fantasy reports are as useful as veridical ones [30]. Thus, it is clear that little experimental expertise is reflected in the work on smoking and personality. While the more recent studies [32, 33] appear to be more sophisticated, there is room for much improvement.

NOTE ON THE EDWARDS SOCIAL DESIRABILITY SCALE

A set of actual scores on the MMPI are compared with a set of SD-generated scores. However, the SD-generated scores are not obtained in the usual way, viz., from the mean SD score of Ss. Instead, the SD score of each S is taken separately and an MMPI profile is generated for him. 15 MMPI scales are examined. The differences between the actual means and the means of the SD-generated scores are evaluated using the t test and the Pearson product-moment correlation. The results show that for 14 of the 15 scales the differences between the means are significant (p ≤ .01), and that 11 of the 15 correlations are significantly different from zero. However, the correlations account for only a relatively small portion of the variance. Only in the case of 4 scales does the correlation account for as much as 45 to 65% of the variance. The authors recommend, on the basis of the data and a logical analysis, that the SD-generated scores not be used as a substitute for the actual scores on the MMPI.

ON EMPTY BATHTUBS: A REPLY TO MEEHL.

Abstract In the present state of ambiguity, negative studies of the validity of psychodiagnostic tests can easily be generated. But such studies throw little light upon the capacities of the tests and are confounded by the utilization of clinical interpretations. It would be most advisable to deal first with the central problems: the need for an integrated theory delineating tests in terms of the areas of optimum assessment and levels tapped; and, the development of realistic expectations on the part of clinicians. The studies frequently cited as throwing question upon the validity of projective tools are rejected on three grounds: the unrealistic wording of the hypotheses (i.e., are these tests valid or invalid, rather than under what circumstances and in what areas are the tests valid); the failure to consider the populations and the background experience of the clinician; and the methodological weaknesses in the designs of the studies. While the author admits that in the long run Meehl may turn out to ...