|
Message/Author |
|
|
I have a question about an analysis and am hoping to get your advice. I have data collected daily over the course of t=10 work days. At the end of each day (between 6-9 pm) participants reported the level of Harassment they experienced that day. The next morning they reported on anxiety (mediator) and job sat (outcome). At the within level the model is harassment --> anxiety --> job satisfaction. To this there is a between level variable related to personality measured before daily data collection. It is hypothesized to act as a cross level moderator of the relationship between harassment and anxiety. Anxiety was treated as a latent variable, all others were treated as manifest variables. We tested the model using Mplus. Here are my two questions. A reviewer suggested that we predict changes in anxiety from one time point to the other by controlling for anxiety at t-1 in predicting anxiety at t0 from harassment at t-1. Do you have any suggestions for how to accomplish this in Mplus? That reviewer also suggested we test for reciprocal effects between harassment and anxiety. Again, do you have any suggestions for how to accomplish this in Mplus? |
|
|
You can do the suggested analysis either in single-level wide format or in two-level, long format. For the wide format with 3 variables at each of the 10 time points, you would analyze 30 variables (30 columns in the data). The The controlling for can be stated as anxiety2 ON harrass1 anxiety1; etc for each time point. For the long format you would have 3 columns in the data and you can use the DSEM analyses we describe in the webinars at http://www.statmodel.com/TimeSeries.shtml |
|
Back to top |
|
|