Generalized propensity scores for multiple treatment versions can be estimated with statistical models such as multinomial logistic regression and with machine learning methods. In this video, I show how to estimate generalized propensity scores for three treatment versions using the baseline category logistic regression model, generalized boosting, and the covariate balancing propensity score (CBPS) method. The example uses data from the Schools and Staffing Survey (SASS) to estimate the effects of teachers being assigned a mentor (three versions: no method, mentor of the same area, mentor of another area) on the probability of the teacher staying in the teaching profession in the following year.
R code for propensity score analysis with multiple treatments
chapter6_part_1_estimate_generalized_propensity_score_for_multiple_treatment_versions.r | |
File Size: | 10 kb |
File Type: | r |
chapter6_part_2_obtain_iptw_and_check_balance_pairwise.r | |
File Size: | 4 kb |
File Type: | r |
chapter6_part_3_obtain_iptw_and_check_balance_with_twang_package.r | |
File Size: | 2 kb |
File Type: | r |
chapter6_part_4_obtain_mmws_and_check_balance.r | |
File Size: | 5 kb |
File Type: | r |
chapter6_part_5_estimate_the_average_treatment_effect_of_multiple_treatment_versions.r | |
File Size: | 2 kb |
File Type: | r |
Data for example propensity score analysis with multiple treatments
chapter6_example_sass_tfs_data_imputed.rdata | |
File Size: | 381 kb |
File Type: | rdata |
chapter6_generalized_propensity_score_estimation_results.rdata | |
File Size: | 68595 kb |
File Type: | rdata |
chapter6_dataset_with_mmws_weights.rdata | |
File Size: | 433 kb |
File Type: | rdata |
Related Research:
Leite, W. L., Aydin, B., & Gurel, S. (2019). A comparison of propensity score weighting methods for evaluating the effects of programs with multiple versions. Journal of Experimental Education, 87, pages 77-88. DOI: 10.1080/00220973.2017.1409179
Leite, W. L., Aydin, B., & Gurel, S. (2019). A comparison of propensity score weighting methods for evaluating the effects of programs with multiple versions. Journal of Experimental Education, 87, pages 77-88. DOI: 10.1080/00220973.2017.1409179
Proudly powered by Weebly