Events

Instructor: Dr. Gary T. Henry, Duncan MacRae ’09 and Rebecca Kyle MacRae Professor of Public Policy, Department of Public Policy and Director of the Carolina Institute for Public Policy at the University of North Carolina at Chapel Hill

Description: Evaluators often face the situation where program outcomes vary across different participants and they want to explain those differences. To understand the contribution of the program to the outcomes, it is often necessary to control for the influence of other factors. In these situations, regression analysis is the most widely used statistical tool for evaluators to apply. The objective of this course is to describe and provide hands-on experience in conducting regression analysis, and to aid participants in interpreting regression results in an evaluation context. The course begins with a review of hypothesis testing (t-tests) and a non-mathematical explanation of how the regression line is computed for bivariate regression. A major focus is on accurately interpreting regression coefficients and tests of significance, including the slope of the line, the t-statistic, and the statistics that measure how well the regression line fits the data. Participants will also learn how to find outliers that may be unduly influencing the results. Participants will have opportunity to estimate multivariate regression models on cross-sectional data; diagnose the results to determine if they may be misleading; and test the effects of program participation with pretest-posttest and posttest-only data. Regression-based procedures for testing mediated and moderated effects will be covered. On the third day, students will be given the opportunity to conduct an independent analysis and write-up the findings. Both peer feedback and instructor feedback will be provided to build skills in interpreting findings and explaining them to interested audiences. Participants will use SPSS software to compute regression analyses and given opportunity to apply it on data from an actual evaluation. Students and instructor will work on interpreting the results and determining how to present them to evaluation audiences. The class will be in a lab where each person has a computer for application of content.