Research

The SiGMA lab develops, evaluates, and applies quantitative methods in two primary areas: scoring, and mediation.

Scoring

Choosing how to score a measure is often a quick decision a researcher makes, but the choice of the type of score can affect the conclusions a researcher will make about their analyses. In some cases, the choice of score can dramatically affect the accuracy and precision of estimates. In the SiGMA lab, we develop and evaluate ways of creating scores from psychological measures (e.g., sum scores, factor scores). We ask when simple sum scores perform adequately enough and when they materially affect reliability, validity, and interpretation, particularly when those scores are used in models such as regression models and structural equation models.

Mediation

Statistical mediation analysis is a method that is used to understand mechanisms. Mediation is best used with longitudinal data, but researchers do not always have access to such data. When mediation analysis is applied to cross-sectional data, or when baseline measures are not included, bias can arise in the mediated effect estimates. To help researchers have more confidence that their mediated effect estimates are correct, we developed a sensitivity analysis for cross-sectional mediation.