Download Full Text (728 KB)
Many professors are familiar with students who come into their first statistics course with a pronounced lack of interest (Rajecki, Appleby, Williams, Johnson, & Jeschke, 2005), or even an intense fear of math. Often, when statistics is paired with a research course, the context of using math to answer a question about human behavior helps them understand what those numbers mean, and if we are lucky, their fear turns to interest or even excitement. But is the reverse true--can understanding statistics help students understand how science works and how to do better research? Incorporating a meta-analysis unit in introductory statistics is an excellent way to reinforce basic concepts, provide a rich context for understanding how statistics and research design fit together, think critically about how to interpret statistics, and encourage students to value psychology as a science. In this chapter, I discuss the benefits and challenges of incorporating a meta-analysis unit in the undergraduate classroom and provide suggestions for activities and lectures, whether you have one day or one week to spend on the topic.
Meta-analysis is both a set of statistical techniques and a research method designed to estimate the overall strength of a relationship in the population by combining all of the existing data on that relationship in one analysis. Following the "replication crisis," concerns about publication bias, and criticisms of null hypothesis significance testing, the way we conduct and evaluate psychological science is changing, and meta-analysis is at the forefront of those changes. The field is moving toward greater emphasis on effect sizes over NHST, increased reporting of confidence intervals and statistical power, publishing interesting null findings, and focusing on cumulative methods such as replication and meta-analysis (Cummin, 2014; Eich, 2014; Stanley & Spence, 2014; Vazire, 2016). In light of these changes, several researchers have highlighted the need for meta-analysis to be more widely taught at both the graduate and undergraduate level (e.g., Funder et al., 2014). However, according to a recent national survey of psychology programs, most introductory statistics courses only cover effect size and confidence intervals for two days or less, and the overwhelming majority--between 73 and 84 percent--do not cover meta-analysis at all (Friedrich, Childress, & Cheng, 2018).
Why would we recommend that such an advanced technique be introduced at the undergraduate level? On a purely practical note, one reason we should introduce meta-analysis early on is that students will inevitably find meta-analyses in their literature searches. In my experience, most do not understand what they are, and thus discard them. But the benefits of this instruction go far beyond teaching students how to cite meta-analytic results. Besides training students in line with best practices in the field, a meta-analytic mindset prepares students to be better scientific thinkers by reframing their approach to research and data early in their academic careers. Below, I review a few of the specific ways meta-analytic instruction can benefit introductory statistics students.
Society for the Teaching of Psychology
For the Love of Teaching Undergraduate Statistics
Copyright © 2020 by Dr. Jennifer Fayard
Psychology | Scholarship of Teaching and Learning | Statistical Methodology
Fayard, Jennifer, "Meta-Analysis as a Tool for Increasing Students' Scientific Thinking" (2020). Books and Monographs. 66.