MrP - Multilevel Regression and Poststratification
MrP is an interesting technique used to analyse non-probability samples in a survey setting. The idea is broadly that we risk non-representativeness in our sample and by weighting the data appropriately we hope to recover better estimates than found in the simple sample data.
I think the technique is very common in survey data analysis and well covered in the Stan ecosystem. Would be good to see a PyMC example in the docs.
MrP - Multilevel Regression and Poststratification
MrP is an interesting technique used to analyse non-probability samples in a survey setting. The idea is broadly that we risk non-representativeness in our sample and by weighting the data appropriately we hope to recover better estimates than found in the simple sample data.
I think the technique is very common in survey data analysis and well covered in the Stan ecosystem. Would be good to see a PyMC example in the docs.
Suggested categories:
Related notebooks
Related to all hierarchical notebooks but distinct due to the survey topic.
References
https://tellingstorieswithdata.com/15-mrp.html