|
|
| Message/Author |
|
|
|
|
Dear all,
I want to do a CFA on the Quick Inventory of Depressive Symptoms, which has a rather peculiar scoring instruction: of several items, only the item with the highest values is considered, and then added up with other items to yield a sum score. For example: sleep = MAX(difficulty falling asleep, waking up during the night, waking up very early, sleeping too much)
QIDS sum score = SUM (sleep, appetite, interest, mood, suicidality,....)
All items have a four point Likert scale, I have 2000 cases.
This scoring instruction makes sense as a lot of the items measure opposites (you cannot sleep to little and too much). Also, I believe the developers wanted the symptom domains (sleep, appetite, psychomotor symptoms) to have the same weight as symptoms that are measured with a single item (e.g. loss of interest, low mood).
I am unsure how to convert this scoring instruction into mplus code. I could compute the symptom domains in SPSS and then do the CFA in mplus, but then I couldn't test whether this method fits the data better than treating all original items the same (which has been proposed) because the two models wouldn't be nested. I could do a second-order CFA with sleep, appetite and psychomotor symptoms as first order factors, but this doesn't accurately depict the original scoring instruction.
I would be very grateful for any ideas or hints!
Anna |
|
|
|
|
| I think this question would be good for SEMNET. |
|
|
| Back to top |
|
|
