A program for examining the effects of variance and slope of time-varying variables in intensive longitudinal data
Dec 2016 – Present
Data gathered using intensive longitudinal methods such as ecological momentary assessment are often messy with a lot missing values. We designed Mix-WILD – “Mixed model analysis With Intensive Longitudinal Data” – a desktop application (for Windows and MacOS) for multilevel modeling of behavior using EMA data. Mix-WILD provides a graphical user interface to add or remove regressors from the model, manipulate missing value codes, and configure other model parameters such as quadrature, convergence criteria, ridge, and the number of resamples (see figure below). MixWILD allows testing of random intercepts and slopes as predictors, mediators, and moderators of outcome variables in intensive longitudinal data.
Mix-WILD allows behavioral researchers to:
- Extend the stage 1 regressor model with the possibilities of random slope
- Feed random slope and other effects into stage 2 model
- Save output files in the users’ desired format and location
- Run the analysis on both Windows and Mac operating systems
We are live and open source now. MixWILD has been used in biostatistics workshops at the Society for Behavioral Medicine symposium in 2018 to train health behavior researchers to use robust methods to analyze behavioral data.
The screenshots presented here are from MixWILD v0.7. To download and try out the latest version of the program, please click here, and to contribute to our open-source efforts to improving MixWILD, click here.
Application Development, Interaction Design
Prof. Donald Hedeker (Prof., University of Chicago), Prof. Genevieve Dunton (Assoc. Prof., University of Southern California), Dr. Eldin Dzubur (Research Consultant, Cedars-Sinai), Rachel Nordgren (Ph.D. student, University of Illinois, Chicago), Jixin Li (Northeastern University, Boston)
The research and development of the software discussed was supported by the National Heart, Lung, and Blood Institute (NHLBI) of the National Institutes of Health (NIH) under award number 5R01HL121330.