Recently, we have begun to focus on research that may improve the implementation of quantitative methods in psychology.
We recently wrote an overview of evidence for a research to practice gap in quantitative methods, and how leveraging the approach of implementation science might help to close that gap.
First, we have written reviews for applied audiences to guide the application of new methods, such as longitudinal models, to their data.
We have also developed a graphical tool to improve interaction plots for linear regression, Interactive, and are working to extend this tool to non-linear GLMs.
We're also interested in improving inferences in simpler models. We recently showed why stepwise methods for testing cross-level interactions is underpowered.
We have several studies focused on the detection and influence of random responses in survey data. Our research has shown that zero-tolerance methods for screening random responses select too many conscientious respondents unless the screening items function perfectly.
Moreover, the influence of random responses can be to inflate any covariance based statistical parameter (correlation, regression coefficient, factor loading, etc.) when the data are skewed, as they often are in psychopathology data.
Privacy Policy: https://www.washington.edu/online/privacy
Terms of Use: https://www.washington.edu/online/terms