Loglinear models: A way to study main effects and interactions for multidimensional contingency tables with categorical data.

Until recently, behavioral researchers failed to examine categorical data based on multidimensional data in a unified whole because of the complex arithmetic such procedures entail. With the appearance of the loglinear model and user-friendly computer software, the problems associated with the analysis of multifaceted contingency tables no longer exist. In this article, we describe the loglinear model and apply it to categorical data cross-tabulated on four dimensions. We define contrasts similar to those of the analysis of variance and describe post hoc and planned comparison strategies. We illustrate hypothesis testing and model building for categorical data, and provide guidelines for performing an analysis on frequency data by using the loglinear model. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Post hoc analysis of repeated measures interactions and gain scores: Whither the inconsistency?

Presents the solution to a problem posed in 1975 by S. W. Huck and R. A. McLean (see record 1975-26619-001). Specifically, it is pointed out that if the "differential effect" contrast procedure of L. A. Marascuilo and J. R. Levin (1970) is employed following the detection of an interaction in a pre-posttest repeated measures design, the conclusions will be identical to those based on post hoc comparisons of gain scores. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Nonparametric Post Hoc Comparisons for Trend.

Post hoc procedures using orthogonal polynomials are illustrated for 3 nonparametric tests that are frequently employed by behavioral scientists in the analysis of empirical data. These examples involve null hypotheses which have been rejected by the Kruskal-Wallis 1-way analysis-of-variance test for rank data, by the Friedman 2-way analysis-of-variance test for rank data, and by the Cochran extension of the McNemar test of change for dichotomous data. Finally, the proposed post hoc procedures are compared to planned-comparison methods. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Planned and post hoc methods for multiple-sample McNemar (1947) tests with missing data.

Pre-post experimental designs with dichotomous dependent variables are encountered frequently in behavioral research. If there is only one group, the McNemar (1947) test can be used to test the hypothesis of no change in parameters. The McNemar test has been extended to cover multiple groups. The problem with these tests is that complete data must be available for all subjects. If post- or pretest data are missing, subjects must be discarded. Ekbohm (1982) provided a solution for the one-group model that has good statistical properties. We extend this method to the multiple-group case. We describe and illustrate procedures for planned and post hoc contrasts. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Type IV errors and Games.

Replies to P. A. Games's (see PA, Vol. 51:Issue 3) criticisms of the authors' conception of Type IV errors in analyses of variance. It is argued that contrasts must be congruent with the mathematical model underlying the structure of the variable, while Games appears to believe that the latent structure is relatively unimportant since the interaction and simple effects models can be combined to form a cell means model. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Selection of pairwise multiple comparison procedures for parametric and nonparametric analysis of variance models.

Discusses 2 issues that are often overlooked in the evaluation of pairwise multiple comparison procedures (MCPs): Comparisons of powers for competing MCPs are not useful unless the procedures have identical nominal experimentwise error rates; and "protected" MCPs do not control the experimentwise Type I error rate at the nominal alpha except in the complete null case. Also, procedures based on the ranks of all K groups do not test the same hypothesis as MCPs that involve reranking the data for each pairwise comparison. (41 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Interactions for dichotomous variables in repeated measures designs.

Describes a method for analyzing interactions in a repeated measures design for a dichotomous dependent variable. The method is based upon the Cochran Q model but is restricted to the case in which a planned analysis is adopted. The model is illustrated for ordered and unordered independent variables. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Comment on Barnard: Another look at strength and direction of attitude using contrasts.

G. A. Barnard (see record 1980-06661-001) proposed the partitioning of the chi-square statistic obtained from attitude data into 2 components, 1 representing strength of attitude and 1 corresponding to direction of attitude. His method is based on a single-sample test of independence. The present authors contend that one problem with Barnard's presentation is his failure to display the tables that correspond to the direction of view component. Furthermore, although it is possible to generate these tables, the hypotheses associated with them are conditional in nature and are difficult to interpret. If the model is reconceptualized as a test of homogeneity, the dimensions of strength and direction of attitude can be investigated by defining contrasts that correspond to the specific hypotheses that are likely to be of most interest. This alternative analysis is statistically sound and yields results that are easily comprehensible. (9 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Nonparametric alternatives for testing main effects hypotheses: A model for combining data across independent studies.

Describes G. A. Mack and J. H. Skillings's (1980) and A. Benard and P. Van Elteren's (1953) forms of the combined Kruskal-Wallis test, 2 nonparametric tests that can be used to make comparisons across K groups in a design with B blocks or in B independent studies. It is suggested that the dependent variables across studies may be alternative measures of the same underlying construct, and an application of the models to B. G. Davis's (1972, 1974) data on the clinical skills of 3 groups of nurses is described. (12 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)