I’m trying to estimate total employment by adding up PERWT (excluding youth under 16, unemployed individuals, etc.) However, when I add up the PERWTs in the 1% samples from 1950->2022, I get drastically fluctuating numbers.
How is it that there can be so much fluctuation from 1950 to 1960? It can’t be that the labor force almost quadrupled in a decade, right? Is there something I’m doing wrong?
PERWT_1950 17527479.0
PERWT_1960 57081010.0
PERWT_1970 68222600.0
PERWT_1980 87450500.0
PERWT_1990 105579940.0
PERWT_2000 118469009.0
PERWT_2010 131127964.0
PERWT_2021 144170560.0
PERWT_2022 149684432.0
Without knowing more about how you arrived at these estimates, it is hard to say anything definitive. While reviewing output or code is beyond the scope of our team, I can describe how I might go about this and share that my estimates do not match yours. Hopefully this gives you a jumping off point for troubleshooting.
I created an extract containing EMPSTAT and the pre-selected variables (including PERWT) with the 1950 1% and 1960 1% samples. I summed values of PERWT by year for respondents with EMPSTAT=1 who are at least 16 years old. The resulting estimate is that about 58.5 million people were employed in 1950 and about 65.6 million in 1960.
Alternatively, you could quickly tabulate EMPSTAT in our online data analysis tool (SDA). For example, I tabulated EMPSTAT by YEAR, using the filter feature to specify years 1950 or 1960, and to specify that age must be at least 16. I chose PERWT in the weight dropdown menu. Here is my input:
And here is my output:
