First step: Create a CPS data extract that contains the undocumented 
identifier for each person in the sample. 

Do this by following the instructions in the README file in the IPUMS subdirectory.

The code for the figures/tables reported in the paper:

pregwork.do and pregwork.log 
Create Figures 1, 2, A1 and Tables 1 and 2 using the Pew files.
The Pew CPS data files must be requested directly from the Pew Research Center.

gregwork.do and gregwork.log
Create Figures 1, 2, A1 and Tables 1 and 2 using my reconstructed CPS data.

regtrend.do and regtrend.log
Creates data for Figures 3 and 4.

regtrend2.do and regtrend2.log
Runs the regressions in Table 3.

gregls.do and greglos.log
Runs the regressions (first 3 columns) in Table 4 using my reconstructed CPS data.

gregls2.do and gregls2.log
Runs the regression in column 4 of Table 4 using my reconstructed CPS data.

gregls_figure.do and gregls_figure.log
Creates the data for Figure 5.

pregls.do and pregls.log
Runs the regressions in Appendix Table using the Pew data.

gregls_2012.do and gregls_2012.log
Runs the regressions in Appendix Table using the reconstructed CPS data.



