Hello,
I'm currently designing a field experimental design (within a hypothetical organization) for my master thesis. I'm only supposed to design the ideal experiment but it is all only hypothetical as I will not actually implement this experiment.
It's about different goal levels as the treatment variables (CG=Challenging Goal; SG=Stretch Goal; TG=Tiered Goal) on team performance (Yi,t=2) and the interaction effect of work experience (WE). I also included the prior performance (ß9 Yi, t=1) to only measure the performance increase due to the goal treatment (as I will do clustered randomization based on locations). However, now I'm unsure if there will be a problem of multicollinearity between work experience and prior performance?!
My current idea is to state this baseline equation for my regression model:
Yi,t=2 = ß0 + ß1 CG + ß2 SG + ß3 TG + ß4 WE + ß5 WE2 + ß6 CG x WE + ß7 SG x WE + ß8 TG x WE + ß9 Yi, t=1 + ei
In my design I suggest that the treatments will be implemented at different locations of the organizations to avoid contamination but therefore there will be clustered randomization and not completely randomized. This is why I felt there is definitely a need to consider prior performance in the regression. Additionally, I would suggest to include time fixed effects and location fixed effects (if there will be more than 4 locations so not one location is equal to one treatment but several locations get each treatment) as well as further control variables such as team size, educational background, age, gender..
Do you see an issue of multicollinearity or any other problems?
I would really appreciate your input.
Thanks!
Best,
Clara
- Badges
- Science topic
- Similar topics
- Instrumentation Engineering
- Instruments