Thursday, October 20, 2016

STEM, computer workers and their degrees

Credit: Census Bureau
Someone asked a really good question about chemists and computer occupations over at the Chemistry Reddit:
It seems to me that a disproportionate number of chemists end up taking up programming to some extent, and sometimes transfer to that field entirely. Does anybody else feel the same way? And why do you think this would be?
I thought it would be interesting to take a look at the Census Bureau's data about this, where they correlate the various STEM degrees and the occupational fields of their degree holders through the American Community Survey. As you can see, the green/teal line is the number of B.S. physical science degree holders who go into occupations classified as "computer workers." As you can see, it's large, but not especially large compared to those who get engineering degrees or computer science degrees.

To get further into the weeds, I calculated the percentages of computer workers for degree holders for the "STEM" fields:

Computers, mathematics and statistics degrees: 43% computer workers
Engineering degrees: 15% computer workers
Physical science degrees: 7% computer workers
Biological, environmental and agricultural sciences degrees: 3% computer workers
Psychology degrees: 3% computer workers
Social sciences degrees: 4% computer workers

Looking at the data, it seems to me to be equivocal. If you compare to "TEM" workers, no, chemists do not end up disproportionately as programmers. However, the data does suggest that, of the "S" fields, the physical sciences disproportionately end up as computer workers. 


  1. I wonder if these percentages would be very different if comparing PhD degrees?

    1. I would presume so. Sadly, I don't have that data.

  2. I can add to that - I have a BS in Chemistry and now work in Data Analytics.

  3. This comment has been removed by the author.

    1. Needed an edit...

      I find programming to be an essential tool for developing customized data analyses, modelling the phenomena I measure as well as the back end data reduction and report generation. I find a lot of common ground with the statistics and data science communities with their emphasis on reproducible research and verification of algorithms and output. As one of my coworkers put it, these tools make our lives better by helping us to automate repetitive analyses, test out various models, and better visualize our results. I even learned to love version control and automated software testing because it has saved my bacon on many occasions ... Programs like Software Carpentry help scientists learn helpful skills quickly.