There has been an increasing focus on external validity, both in terms of replication and generalisation, in political science and, especially, in experimental political science. With this increasing focus comes the proliferation of fixed and random effects meta-analysis models to pool experiment-level estimates across contexts to answer political science questions. However, few political scientists to date have made use of the individual-level data that is now readily available for thousands of studies, to perform individual participant data (IPD) meta-analysis. IPD meta-analysis allows researchers to test theories of individual-level treatment effect moderation that most individual experiments are underpowered for, and which are of interest to political scientists. In this paper, we set out best practices for IDP meta-analysis and effect heterogeneity analysis from other fields such as medicine and epidemiology, and make two methodological contributions. First, we think about how we can efficiently estimate uncertainty in common experimental set-ups, where intervention arms share the same control group and are clustered within experiments. Second, we extend this set-up to heterogeneity analysis with one and multiple moderators. We apply these new methods to a novel individual level dataset of get-out-the-vote interventions that aim to increase turnout, testing what groups such treatments mobilize most.