2018 Session I: Suzanne M. Scharer

Nivedita Bhaktha
Educational Studies, QREM
Ann O’Connell, Advisor

Predicted versus Actual GPA: Do Large Differences Signal an Early Warning for Student Attrition?

This study examines whether a measure of under-/over-performance given a student’s first-term grade point average (GPA) contributes to identifying students with greater chances of leaving the institution in year two. Over-/under-performance is calculated as the difference between the actual and predicted GPA. Results suggest that under- and over-performanceĀ help to identify students who may be more likely to leave the institution and where transfer students may go.

Meng-Ting Lo, Yixi Wang & James Uanhoro
lo.194@osu.edu, wang.7636@osu.edu, uanhoro.1@osu.edu
Educational Studies, QREM
Dorinda Gallant & Ann O’Connell, Advisors

Is An Alternative Method Always the Better One? A Comparison between ANOVA, Welch’s F Test and Kruskal-Wallis Test

The current study compares the performance of traditional F test, Welch’s F test and Kruskal-Wallis test by examining the nominal error rates under varying degrees of non-normality, and heterogeneity of variance using a variety of sample sizes and a balanced design. Recommendations are provided based on the study results.

James Uanhoro
Educational Studies, QREM
Ann O’Connell, Advisor

Analyzing binary outcomes, going beyond logistic regression

The standard regression for analyzing binary outcomes is logistic regression. We often obtain odds ratios (OR) and interpret them using phrases like, “… times more likely”. However, this interpretation of the OR is often misleading. I suggest an alternative regression model that allows us correctly make such statements, and demonstrate it in SPSS