Getting Estimation Thinking into Educational Leadership: Challenges and Next Steps

Estimation thinking is increasingly recommended across disciplines, including educational leadership, as an alternative to traditional significance testing. Unlike the binary logic of p -values and arbitrary cutoffs, such as p < .05, estimation emphasizes reporting effect sizes together with their uncertainty. This approach helps avoid common errors of significance testing, for example the mistaken belief that a non-significant result implies the true effect is zero. However, confidence intervals are often misunderstood even by experienced researchers, and poorly applied estimation methods would offer little advantage over significance testing. This article therefore focuses on correctly interpreting confidence intervals and on recent techniques for estimating sources of uncertainty beyond sampling error. Using a running example from a study of school principal leadership, the presentation is aimed primarily at applied researchers. All syntax and output files in R for this example are freely available from a public Open Science Framework (OSF) website and also presented in article supplemental materials. The goal of this presentation is to help researchers in educational leadership realize the benefits of estimation thinking.