Hi,
I'm doing a cross-country comparison. My data for each country has a match pair design i.e. each company within a country is match with a control firm from that country. The data is cross-sectional. Now I want to compare the two countries.
Could anyone please guide me as to which statistical analysis should be used for such a comparison?
Thanks and regards,
Juhee
We used SEM and a GLM for cross-country analysis, where we used the combined sample as well as individual samples. We made the different countries independent variable, so that we could do an comparison for the combined sample. Refer to my ResearchGate profile and refer to the three YouTube South Africa and Romanian articles as examples. I am not sure if this will help, but nevertheless, is a starting point!
You say that you are studying two countries and for each country you have two matched companies ; so that is data on 4 companies ; is that correct?
THis site has a lot of material on matching and subsequent analysis of the matched data
https://gking.harvard.edu/publications/term/1585
In matched pairs design you usually need to look at differences. The first step is visual analysis and then you can use panel data models.
For visual analysis of panel data let me recommend this work:
Poster A Visual Framework for Longitudinal and Panel Studies (with ...
Hi Kelvyn,
Thanks for your message. I have two countries. For country A there are ~900 firms in the experiment group and the same number of companies (~900) in the control group.
For country B, there are ~400 companies each in experiment group and control group.
The control, independent and dependent variables are the same for both the countries.
I have done an analysis for both the countries separately and now I want to do a comparative analysis of these countries together. What would be the best way to analyse them is the question.
Further, the data is cross-sectional and not panel. Given the the design is matched pairs, I have used conditional logistic regression in analysis of the countries individually.
Best regards,
Juhee
I have never done this but here is a possibility
Article Conditions for consistent estimation in mixed models for bin...
and then additionally include two dummies (and no intercept) to get separate estimates of the mean for each country.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis