No abstract

No abstract

Finding reliable and reproducible effects is crucial for the progress of any scientific field. Key determinants in this pursuit are the quality of the methodology, the number of participants in a study (i.e., sample size), and the magnitude of the studied effects (i.e., effect size). In a recent publication, Marek et al. provide important insights into the magnitude of effect sizes in neuroimaging MRI research and related sample sizes: brain-wide association studies (BWAS) linking neuroimaging features with behavioural phenotypes in the general population are characterised by (very) weak effects. Consequently, large sample sizes of over 3,000 participants are required to lead to reproducible effects. The largest effect sizes reported by Marek et al. concern correlations |r| as small as 0.06 to 0.16 across multiple imaging modalities and behavioural measures tested. Besides examining brain correlates of normative behaviour, another central goal in neuroimaging is to study brain structure and function under disease conditions, with the aim to shed light on pathophysiological mechanisms and identify potential markers of clinically relevant features. Brain patterns related to disease conditions are likely more pronounced than general brain-behaviour associations; other factors remaining equal, smaller samples should be sufficient in these situations to obtain reliable and reproducible findings. This is important because clinical neuroimaging studies must balance reproducibility with cost-effectiveness and efficiency. Here, we show by means of power calculations and empirical analysis that neuroimaging studies in clinical populations need hundreds -and not necessarily thousands- of participants to lead to reproducible findings.