~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Session title: Latent Profile Analyses using tidyLPA
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Abstract: Do adolescents who try alcohol before age 15 have worse health in adulthood than those who try alcohol later in life? Are there different subtypes of COVID-19 symptoms? How do these subtypes predict hospitalization risk? What types of advertisements most strongly predict if someone buys a product?
All these questions can be answered using the statistical technique, latent profile analysis (LPA). LPA is a powerful statistical tool to identify hidden clusters in your data based on a set of variables. LPA is superior to other data reduction techniques like k-means clustering in that it generates probabilities of group membership. In this talk, a brief introduction to LPA using the R package tidyLPA will be covered. Using sample data on daily sleep patterns of nurses, we will walk through an example using the current gold standard 3-step approach: 1) estimate class membership, 2) extract class probabilities, 3) examine predictors of class membership. Best practices and advantages and disadvantages of LPA will be discussed.
Bio: Dr. Danica Slavish is an Assistant Professor of Psychology at the University of North Texas. Danica’s research focuses on how sleep problems and stressful life experiences interact to predict health outcomes across time. Her research team uses a variety of statistical techniques for analyzing longitudinal data in R, including multilevel modeling, time-varying effect modeling, network analyses, and Bayesian analyses. Danica currently teaches courses in health psychology, advanced statistics, and research methods. Before joining UNT, Danica completed her Ph.D. in biobehavioral health at Penn State, where she was a National Science Foundation Graduate Research Fellow. She holds an undergraduate degree in psychology from Beloit College, in Beloit, WI.
Twitter: / danicaslavish
19 сен 2024
