|
|
Syntax for modeling "holiday effects"... |
|
Message/Author |
|
Jen posted on Wednesday, June 01, 2011 - 7:10 am
|
|
|
Hi! I have read several of the papers that model non-linear "holiday effects" in longitudinal data with many waves. This makes sense to me conceptually, but I wanted to make sure I understood the proper syntax. I am 8 waves of drinking data, and there seems to be a large dip in drinking for two months in the midst of winter, meaning the data is not fit well by a linear or quadratic model alone. If I want to model this drop, is the following syntax correct? ibinge sbinge qbinge | binge2@0 binge3@1 binge4@2 binge5@3 binge6@4 binge7@5 binge8@6 binge9@7; ibinge winter | binge2@0 binge3@0 binge4@0 binge5@0 binge6@0 binge7@1 binge8@1 binge9@0; (7 & 8 are the months with the drop) This model seems to fit the data well, and there is variation in all parameters that could then be predicted by covariates... Thanks! |
|
|
I would do this using an intercept only model to take the drop into account, for example, winter | binge7 binge8; |
|
Jen posted on Wednesday, June 01, 2011 - 12:32 pm
|
|
|
Thank you! That seems to work well. |
|
Jen posted on Thursday, June 09, 2011 - 11:47 am
|
|
|
One more quick question related to this -- if I am interested in modeling an effect like this that occurs at only one time point (e.g., just binge7) or at multiple, non-consecutive time points (e.g., binge3 and binge7), what would be the best syntax for that? I've noticed that the intercept only model causes some issues, even if I hold the variance at 0 for winter (and there is a lot of variance in winter when 7 & 8 are included, so the 0 variance model probably isn't the best)... Thanks for any advice! |
|
|
Perhaps the following paper that is on the website will help: Greenbaum, P.E., Del Boca, F.K., Darkes, J., Wang, C. & Goldman, M.S. (2005). Variation in the drinking trajectories of freshman college students. Journal of Consulting and Clinical Psychology, 73, 229-238 |
|
Back to top |
|
|