I'm working on a RCTs study on the impact of nutrition training on crop diversity of small farmers. We will train them on nutrition and observe the changes in crop diversity and their diet.
We do randomize at village level.
We will check how it is different across seasons: lean and abundant. So that, we will collect two rounds of baseline and end line according to the two seasons, and we train them in between.
The two seasons are only 2 months different.
Right now I find the treatment effect by running the following regression:
Yi = β0 + β1T1 + β2X+ β3S + β4S*T + β6V + µi
Yi: Outcome indicators of all households in treated villages and control villages in both end-line rounds
β1: Treatment effect
T: Treatment
X: Household characteristics
S: Seasonality dummies (S=0: lean season, S=1: superfluous season)
V: village dummies
However, I am not sure about my estimator. Does anyone have any idea on this?
Thank you so much.
I believe the best approach would be applying multilevel modeling, which can estimate random intercept and slope information. You can save output of these random variables and use them for subsequent evaluation. NLME or LME4 packages in R can handle complex multilevel analysis. You may search LEXJENSEN website for SAS codes. I think SPSS can handle mixed effect modeling (another name for multilevel modeling), but I have not applied such methods using SPSS. Good luck.
Kindly check https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6293457/
Also check https://www.luroufan777.com/sites/default/files/downloads/report/Baseline-Report-Mixed-Methods-Cluster-RCT-of-eSchool-360-Model-in-Rural-Zambia-April-2018.pdf
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training