I have collected data longitudinally for the same people four times to test the relationship between team turnover rate and team effectiveness outcomes at the unit level of analysis. do i have to use only SPSS or SPSS with AMOS? can i use multiple regression or panel regression? or there are other tests? i am using a single item for employee turnover rate? how can i test one item?
Hi Ebtesam
Alvin's suggestion is very good. Should you decide to go with latent growth modelling using AMOS, link below (webzoom) gives you step-by-step guide - it's short and precise. Exactly what is needed at introductory stage. You can, of course, go deeper after that.
Alternatively, if only variations at different point of measurement taken you need to proceed with, one-way repeated measure ANOVA is a good way to start with.
Best wishes
Saiyidi
http://webzoom.freewebs.com/danielboduszek/%601Latent%20Growth%20Analysis.pdf
As others pointed out a latent growth model can be used.
A good alternative would be to use a multilevel regression to fit the growth curve which in your case may offer two advantages.
1. If there are any cases for which you do not have all the data (e.g. you do not have data for each team for ALL four times) you will have to remove them before doing the model with SEM. WIth a multilevel model you can test the growth curve without removing any teams with incomplete data.
2. The second advantage depends on how you actually measured turnover rate. I would assume the turnover rate is a count variable measured within a specific timeframe. If that is the case then its unlikely that it is normally distributed as count variables usually follow a poisson distribution. As such you will need to use a multilevel poisson regression (which is a multilevel generalised linear model). With such a model you will be able to use the actual turnover frequencies (instead of the rate) as the dependent variable and treat the timeframe as the offset in your model.
I don't use SPSS but as far as I know it can fit multilevel generalised linear models - and therefore it should be able to do a multilevel poisson regression.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS