Edward F. Gocka

A Factor Analysis Of Individual Social Desirability Scale Values.

The factorial structure of individual judgments of social desirability scale value (SDSV) was studied by factor analyzing a 112 (college students) by 100 (MMPI items) data matrix of individual judgments of SDSV in two ways. One analysis was factoring by item (R technique) while the second analysis was factoring by subject (Q technique). A large general SD factor appeared when subject intercorrelations were factored, but no such general factor was identified when item intercorrelations were factored. Evidence was also presented that the large general SD factor was highly correlated with mean or normative SDSVs.

SOCIAL DESIRABILITY VALUES FOR CALIFORNIA PSYCHOLOGICAL INVENTORY ITEMS

Social desirability ratings for CPI items were obtained from 68 female and 59 male college students. Simple means and standard deviations of the ratings were compared with previously published scale values and dispersions in the case of 186 overlapping CPI-MMPI items. The results showed high comparability on scale values so that the simple means of the item ratings may be used similarly to previously published MMPI item values obtained by using scaling techniques Simple standard deviations of individual item ratings, however, tended to be generally larger than the associated discriminal dispersions published for such items.

A method of fitting non-symmetric Gompertz functions for characterising malignant growth

A basic symmetrically asymptotic Gompertz equation was extended to account for more complex growth and life-table data. This extended Gompertz or non-symmetric asymptotic regression equation was found to be suitable for fitting a large variety of data exhibiting growth and experimentally-induced regression effects expressed in the same curve. The statistical and computational properties of this fitting function are discussed in theory and example.

Factor Analysis Of Individual Social Desirability Scale Values: Second-Order Analysis.

A general social desirability factor emerged at the second-order level when a set of individual social desirability scale values were correlated between subjects (i.e., across items) and factor analysed. However, no such general factor emerged at the second-order level when the individual judgments were correlated between items (i.e., across subjects).

Analysis of Length-of-Stay Data with Incomplete Observations

Length-of-stay data of the type used as criteria for assessing various kinds of treatment programs poses a number of data analysis problems due to Ss becoming lost to follow-up, unresponsive Ss being withdrawn at the end of the study, Ss entering the study sequentially, and observations being arbitrarily truncated at some time point for purposes of analysis. Such data are often described as data containing ‘censored’ observations. Techniques for handling such data have been worked out extensively in industrial and human survival studies but are not commonly available in quantitative psychology literature. Adaptations of such methods are suggested for assessing the results of treatment programs associated with alcoholism problems, the rehabilitation of blind veterans, recidivism among neuropsychiatric patients, and the treatment of epileptic patients. The paper focuses on 5 classes of analytic procedures: actuarial, nonparametric, stochastic-Markov, randomization, and maximum likelihood regression, and discusses such procedures with respect to assumptions, sample-size requirements, characteristics of the censored observations, types of test statistics, ability to consider ancillary variables, and ease of computation.

Complex Chi-Squares: Matrix Notation and Calculation

A matrix formula available for the calculation of complex chi-squares allows several computational variations, each of which requires fewer steps than the standard formula. However, neither the matrix formula nor the associated computational algorithms have been given adequate exposure in statistical texts for the behavioral sciences. This paper reintroduces the formula, expands the notation, and shows how several computational variations can be derived.

COMMENTS ON RANDOMIZATION TESTS

Although t tests are known to be robust, randomization tests have their unique role in instances of complex test criteria, censored observations, and unknown small sample characteristics.