Date of Award


Degree Type



Maxine Goodman Levin College of Urban Affairs

First Advisor

Bowen, Chieh-Chen

Subject Headings

Science -- Study and teaching -- United States, Technology -- Study and teaching -- United States, Engineering -- Study and teaching -- United States, Mathematics -- Study and teaching -- United States, Education innovations -- United States


The federal government has been spending a large amount of funds on STEM programs. It is important to examine the effectiveness of such spending. Much research has been conducted in the past 30 years for this particular purpose however, the results of such evaluations have not painted a clear picture of the effectiveness of STEM programs. The goal of this meta-analysis is to investigate whether STEM programs are successful in the outcomes they claim to achieve. Such a meta-analysis must integrate all of the empirical studies which reported their effort in evaluating the effectiveness of STEM programs, based on measures of (1) engagement, (2) capability, and (3) continuity (Jolly, Campbell, and Perlman (2004). Previous studies on the STEM program effectiveness used either a within-subject design or a between-subject design. First, each of these research designs was investigated independently to examine whether the particular design has an effect on estimates of STEM program effectiveness. In addition, other moderators were investigated to determine factors that could influence estimates of STEM program effectiveness: pedagogical types of programs, program funders, program creators, grade level of sample groups, regional locations, instrument reliabilities, publication types, time, etc. The meta-analysis covers literature published between 1980 and 2010. The total number of studies included in this meta-analysis is 91. This study finds that all three outcome variables have positive effect sizes at the moderate level with the weighted mean effect sizes of .346 for engagement, .456 for capability, and .369 for continuity measurements. Additionally, the between-subject design versus within-subject design has an effect on estimates of the engagement and capability outcomes of STEM programs. Finally, some moderator variables were statistically significant on the outcome variables with the mean effect sizes: program strategy, program creator, regional location, and educational level of sample group. These fi