Scott L. Hershberger

The Growth of Structural Equation Modeling: 1994-2001.

This study examines the growth and development of structural equation modeling (SEM) from the years 1994 to 2001. The synchronous development and growth of the Structural Equation Modeling journal was also examined. Abstracts located on PsycINFO were used as the primary source of data. The major results of this investigation were clear: (a) The number of journal articles concerned with SEM increased; (b) the number of journals publishing these articles increased; (c) SEM acquired hegemony among multivariate techniques; and (d) Structural Equation Modeling became the primary source of publication for technical developments in SEM.

The Relation Between Sexual Orientation and Penile Size

The relation between sexual orientation andpenile dimensions in a large sample of men was studied.Subjects were 5122 men interviewed by the KinseyInstitute for Research in Sex, Gender, and Reproduction from 1938 to 1963. They were dichotomouslyclassified as either homosexual (n = 935) orheterosexual (n = 4187). Penile dimensions were assessedusing five measures of penile length and circumferencefrom Kinseys original protocol. On all fivemeasures, homosexual men reported larger penises thandid heterosexual men. Explanations for these differencesare discussed, including the possibility that these findings provide additional evidence thatvariations in prenatal hormonal levels (or otherbiological mechanisms affecting reproductive structures)affect sexual orientation development.

Cooperation and Competition Between Twins: Findings from a Prisoner’s Dilemma Game

Abstract Attention to factors influencing cooperation and competition during human social interaction has increased within recent years. This study tested the hypothesis that higher levels of cooperation would be associated with increased genetic relatedness between interactants, and explored questions concerning the expression of cooperative behavior over time. A Prisoner’s Dilemma game, in which participants elect to display cooperative, competitive, or exploitative behaviors relative to a partner, was administered to 59 monozygotic and 37 dizygotic twin pairs, between 10.92 and 82.67 years of age. Results from multivariate analysis of variance procedures, cross-lag sequential analyses, and hierarchical linear modeling supported associations between zygosity, and frequency and continuity of cooperation. Mechanisms by which zygosity may affect cooperation were explored by examining relationships between response combinations, and twins’ IQ similarity and social closeness. The findings are considered with reference to an evolutionary perspective on behavior that offers a theoretical basis for considering how the relative genetic relatedness of social partners affects their social-interactional processes and outcomes. This report is the first in a series of studies designed to address mechanisms underlying differences in cooperation among pairs who vary in average genetic commonality.

Genetic and environmental influences on individual differences in masculinity, femininity, and gender diagnosticity: analyzing data from a classic twin study.

Analyzing data from Loehlin and Nicholss (1976) classic twin study, we computed measures of Masculine Instrumentality (M), Feminine Expressiveness (F), and Gender Diagnosticity (GD). Quantitative genetic modeling analyses of within-sex individual differences in M, F, and GD indicated that: (1) Additive genetic factors contribute significantly to individual differences in M, F, and GD. (2) The environmental effects on M, F, and GD tend to be nonshared. (3) The genetic and environmental components of individual differences in M, F, and GD tend not to show gender differences. Finally, (4) the estimated within-sex heritability of GD (.53) is significantly greater than the estimated within-sex heritabilities of either M (.36) or F (.38).

Meeting One's Twin: Perceived Social Closeness and Familiarity:

Perceptions of social closeness and familiarity were assessed among 44 monozygotic (MZA) and 33 dizygotic (DZA) reunited twin pairs, and several individual twins and triplets. Significantly greater MZA than DZA closeness and familiarity were found. Closeness and familiarity ratings for co-twins exceeded those for nonbiological siblings with whom twins were raised. Correlations between perceptions of physical resemblance and social closeness and familiarity were positive and statistically significant. However, most correlations between social relatedness and contact time were non-significant. Associations between social relatedness and similarities in selected behavioral traits were also examined. The findings support various theoretical perspectives anticipating greater affiliation among close relatives than distant relatives.

Assessing Content Validity and Content Equivalence Using Structural Equation Modeling

Content validity is rarely evaluated based on empirical data, independent of that from a panel of content experts. In this article, procedures are used to construct parallel test forms based on statistical equivalence rather than content equivalence. This article describes an alternative approach to assessing content validity and content equivalence in terms of item-content structures and content area constructs. Structural equation modeling is applied to item-response data from 2 Regents College examinations to empirically verify content constructs developed by a panel of content experts and to examine content equivalence across-parallel test forms. The results suggest the different degrees of inconsistency and bias of content experts in assigning items to their corresponding content areas. The results also show that content equivalence across-parallel test forms is disputable based on randomly splitting methods. The implication and importance of the study is discussed in terms of test development.

A Note on Determining the Number of Imputations for Missing Data

A desirable method for imputing missing data when the missing data mechanism is ignorable is random imputation. Imputation generally involves regressing each variable with missing data onto all the other variables. Single random imputation involves creating one data set in which each of the missing values is imputed by adding a random error term from the residual distribution of each imputed variable. Adding a random error term reduces the bias in standard errors of parameter estimates ordinarily observed under traditional methods of imputation (Schafer, 1997). However, even single random imputation produces standard errors that are generally biased downward and parameter estimates that are relatively inefficient. Improvements in bias and efficiency occur with multiple random imputation in which the imputation process is done more than once, producing multiple data sets whose parameter estimates and standard errors are averaged (Rubin, 1987). Many structural equation modeling programs (e.g., AMOS; Arbuckle, 1997) are now able to compute multiple random imputations for missing data. Relatively little guidance has been given as to the number of imputations that should be conducted. Is two enough? Are the five imputations given as the default in PROC MI of Statistical Analysis System (SAS, 2002) enough? Or, is some greater value required? Based on the results of a Monte Carlo study, Rubin and Schenker (1986; see also Li, Raghunathan, & Rubin, 1991) suggested that the number of imputations per missing value should be roughly two when 10% STRUCTURAL EQUATION MODELING, 10(4), 648–650 Copyright

Dating and commitment choices as a function of ethnicity among American college students in California

The incidence of interracial and interethnic dating and marriage in the United States has increased. This investigation examined dating and commitment choices as a function of ethnicity and sex among groups of Euro-American, Hispanic-American, Asian-American, and African-American college students. A convenience sample of college students comprising 329 heterosexual subjects (134 men, 195 women) was surveyed regarding their partner preferences for dating, visiting parents, marriage, and bearing children. It was hypothesized that subjects would consider dating partners from different ethnic groups, but when making a commitment to marriage and children would prefer members of their own group. This hypothesis was supported in half of the groups: Euro-American men, African-American men, Asian-American women, and African-American women. A discussion of dating and commitment choices among ethnic and sex groups is presented and discussed.

The Overparameterized Analysis of Variance Model

Analyses of variance (ANOVA) with the general linear model (GLM) in many standard statistical packages use an Overparameter ized model, a model unfamiliar to most behavioral science researchers. Estimates and significance tests with GLM procedures are calculated by computing generalized inverses and estimates of estimable functions. Using simple examples, the authors discuss the concepts that underlie the solutions for 1-way and 2-way ANOVAs with Overparameterized models and illustrate how these models allow one to evaluate the research hypotheses. The authors also extend the discussion of Overparameterized models to a more general modeling approach than GLM, the general linear mixed model. Many students and researchers in the behavioral -ciences routinely conduct analyses of variance <ANOVAs) using the general linear model procedure in SAS (1997) or SPSS (1998; GLM, as named in both packages). Unbeknownst to many of these students and researchers, both GLM procedures use Overparameterized models to compute ANOVAs. These GLM users may occasionally suffer some minor discomfort when the procedure produces mysterious messages about matrices being singular, gener

Drug Use, Personality and Partner Violence: A Model of Separate, Additive, Contributions in an Active Drug User Sample.

Drug use is considered a main contributing factor to crime and violence. This research examined the evidence regarding the relationship between drug abuse and the occurrence of intimate partner violence. Current drug using men were assessed on aggression related personality variables, their drug use, and the occurrence of violence in their close relationships. A latent aggression factor and recent amphetamine use were the only variables found to be significantly associated with violence. No other drug use variables were found to be associated with violence by the participant and the overall drug use factor was not found to be associated with violence or aggressive personality. The widely accepted notion that increased substance use directly leads to increases in violent behavior was only partially supported, at least within this drug using population. The assessment of aggressive personality, rather than of drug use, is suggested for correctional as well as clinical settings in which drug users are prevalent when determining susceptibility to violence.