How to obtain population proportion in MRP model

Modeling
1

i am studying multilevel regression and poststratification(MRP) and want to apply this model in the real data in China. however, IPUMS International (IPUMS International) does not give the data on 2010 and 2020 population census. And the yearbook on population census does not give the combination of stateicp, sex, age, education, but the subset of this combination. for example, the combination of stateicp, age and gender, the combination of stateicp, age and education. so, my question is : can I use the combination available to infer the full combination of stateicp, gender, age, education?

Any suggestion is appreciated.

1 Like
2

One thing you may look into would be Multi-Level Regression with synthetic poststratification (MrsP) which allows you to use the marginal distributions rather than require joint distributions to construct your post-stratification frame. Here is a paper on it:

https://onlinelibrary.wiley.com/doi/abs/10.1111/ajps.12319

1 Like
3

Thank you for your reply. Is there any package or codes for this model?

1 Like
4

This chapter has some good R code to set up your synthetic post-stratification frame that might be helpful: Getting the Most Out of Surveys: Multilevel Regression and Poststratification | SpringerLink

5

Thank you very much!

1 Like