| Message/Author |
|
|
| ZHIYAO YI posted on Tuesday, October 15, 2019 - 7:49 pm
|
|
|
Dear Dr. Muthen,
I want to generate a dataset that have individuals repeatedly measured before and after intervention as follow and analyze it using DSEM models. I need to separate pre and post tests as two separate variables. Could you help or recommend me any reading materials that have this type of data pattern? Is there a way to generate data with no observation at particular time points.
Thank you so much in advance.
id phase obs pre post
1 1 1 Y1 no
1 1 2 Y2 no
1 1 3 Y3 no
1 2 1 no Y4
1 2 2 no Y5
1 2 3 no Y6
... |
|
|
|
|
| This is described in our Topic 12 Short Course video and handout for Parts 1-2, slide 53 and on. |
|
|
|
|
| Is it possible to use DSEM to do the interrupted time-series model (ARIMA model + intervention variable) at more than one level? For instance, compare the homicide rates 90 months before and 90 months after a social event occurred across 50 US cities and see whether this event has any effects on homicide rates? |
|
|
|
|
| It should be possible but I am not sure what the best approach would be. One simple way is to run a two-level bivariate model where the first variable represent the first 90 months and the second represents the second 90 months and then you can regress on the between level Y2 on Y1 to see how the rates changed. |
|
|
| Back to top |
