Hello - I am trying to make a table similar to one I saw from the Census Bureau that looks at STEM occupations (STEM Occupations by Sex, Race, and Hispanic Origin: 2012-2016 ACS). To do this, I created the variable “occSTEM” in IPUMS with the ACS codes related to STEM occupations. I am wondering if I did this correctly because the % seem too low to be accurate (0.2% in STEM occupations with both the codes below). Any feedback?
I’m using the ACS 2013-2017 estimates and the two codes I used a:
are:
Code 1
if (occ eq 110 OR occ eq 360 OR occ eq 1005-1965) occSTEM = 1
else occSTEM = 2
Code 2
if (occ eq 110) occSTEM = 1
else if (occ eq 360) occSTEM = 1
else if (occ eq 1005-1965) occSTEM = 1
else occSTEM = 2
It looks like you are correctly identifying the “STEM related” occupations according to the Census Bureau table you site above. The issue may be that you are not conditioning your calculation on employed individuals who are 25 years old or older (as the Census Bureau table does). I do not see that you are conditioning on this when calculating your percentages. I just looked into this myself, using the IPUMS online analysis tool, and was able to closely replicate the figures in the table. Alternatively, I’m not familiar with the specifics of your syntax here, but I wonder if specifying a range in values as you do here (e.g. 1005-1965) correctly identifies all values in this range.
Thank you Jeff. Are you able to share a screen shot how you listed the variables in the IPUMS online analysis tool? When I run for 25+, I still get a very small %. Appreciate you looking into this.
I was also wondering if specifying a range might be presenting a challenge. I will continue to explore that.
In the online analysis tool, I inter the following.
Row: occ
Selection Filter(s): occ(110, 360, 1005-1965) age(25-**) empstat(1)
In the 2013-2017 data, this returns an estimate of about 8,417,818 employed individuals, 25 years old or older, currently working in STEM related fields. This is relatively close total estimate reported in the Census Bureau table cited above (which uses 2012-2016 data).