This study addresses several methodological problems that have confronted prior research on the effect of class size on student achievement. Unlike previous studies, this analysis accounts for the hierarchical data structure of student achievement, where grades are nested within classes and students, and considers a wide range of class sizes…

In most multi-campus studies of college impact that have been conducted over the past four decades, investigators have relied on ordinary least squares (OLS) regression as the analytic method of choice. Recently, however, some investigators have advocated the use of Hierarchical Linear Modeling (HLM), a method specifically designed for analyses…

A review of the literature and of currently available computer programs for the use of log-linear models in analysis of qualitative data is presented, and two types of log-linear models and procedures for investigating both types are outlined. (MSE)

A statistical methodology applied to three years of teaching load data from 21 major public universities yields an objective, broadly applicable set of student credit hours (SCH) weight factors. A table of optimum weighting factors for upper division and graduate SCH is presented. (Author/MLW)

An analysis of student decisions to withdraw from specific courses before assignment of grades but after the "add-drop" period at the University of Minnesota is presented. In a probit model, withdrawals appear to occur randomly, with notable exceptions. Student, class, and teacher characteristics are considered as variables related to…

This study used Tobit analysis to estimate retention probabilities and credit hours at two universities. Tobit was judged as appropriate for this problem because it recognizes the lower bound of zero on credit hours and incorporates this bound into parameter estimates and forecasts. Models are estimated for credit hours in a single year and…

This study refined and applied an integrated model of undergraduate persistence (accounting for both behavioral and perceptual components) to examine first-year retention at a private, highly selective research university. Results suggest that including behaviorally based measures of involvement improves the model's explanatory power concerning…

A study compared approaches to assessing the influence of student financial aid on within-year persistence of traditional college students. Results indicated that models including tuition were better predictors of persistence, that use of multiple approaches provided more insight into policy's role, and that tuition charges had a consistent…

Two competing models of the relationship between college graduates' perceived learning and satisfaction with the college experience are examined, one representing a true effect and one treating the relationship as an artifact of the halo effect. It is recommended that researchers be cautious in interpreting self-reports of learning and…

Analysis of data on freshman-to-senior developmental gains in 722 University of Tennessee-Knoxville students provides evidence of the advantages of structural equation modeling with latent variables and suggests that the group differences identified by traditional analysis of variance and covariance techniques may be an artifact of measurement…

The study tested a structural equation model of community college enrollment patterns of White and Hispanic males and females and the invariance of parameter estimates among subgroups. High school preparation in mathematics and science (M/S), M/S attitudes, M/S enrollment patterns, parent education, encouragement, and high school grades were…