Samantha Anderson

Several of my research interests pertain to the often overlooked design elements of psychological studies, particularly regarding study design, statistical power, and sample size. In this area, I have developed an approach to sample size planning that uses information from a prior published or pilot study to plan the sample size for a future study, adjusting the prior study effect size estimate for publication bias and/or uncertainty. This method, "BUCSS", is accompanied by open-source software, both as an R package and series of web applications, that allow researchers to implement the approach for multiple regression and ANOVA designs with continuous outcomes. 

I also am interested in meta-science topics that investigate how methodological issues can affect the scientific literature. In this area, I study methodology pertaining to replication studies, which have recently become more common in psychology. My research has centered on taxonomies for defining what it means to successfully replicate prior work, along with appropriate analyses to achieve various replication goals. I am also interested in the influence of various "researcher degrees of freedom", such as multiple testing, which have conseuquences for replicability.

Another area of active interest involves methods for handling missing data and dropout in studies aiming to detect a causal treatment effect. Two-wave randomized pretest posttest studies can offer unique challenges for estimating treatment effects under different missing data conditions, particularly when the data violate statistical assumptions and the model is misspecified. Finally, I conduct quantitatively-informed substantive research in the area of stress and depression, and I enjoy working with intensive longitudinal data.