STEM in America: Data, Approaches, and Questions (Part II)

If you missed Part I of this series, find it here.

Part II: Trends

In Part I, we looked at some numbers: majors and graduates in US four-year institutions in and out of STEM fields. Those data gave us an overview of the state of STEM education in 2012 – who’s pursuing it (mostly men, to varying degrees depending on race and ethnicity) and who isn’t (mostly women across all race and ethnicity groups).

In this segment, we have two objectives. First, I’d like to provide a little more context for the discussion by looking at the value of STEM versus non-STEM degrees for the graduates who hold them. Then, we’ll look at data on higher education in the US from 2002 and 2012 to set up a longer perspective on how disparities have been changing (if indeed they have). In our efforts to make college more accessible to all Americans, have we been making progress? And when we look at STEM in particular, do we see similar patterns? Or are disparities in STEM widening when compared to higher education as a whole?

Are disparities widening when we look at STEM education v. total higher ed? Click To Tweet

The STEM Difference in $

There are plenty of excellent reports on the economic value of a STEM degree and the salaries for various fields that require them, as well as income disparities between STEM and non-STEM degree holders that persist even in jobs not requiring a STEM degree. I will briefly summarize here, and provide links for further exploration.

The mean annual salary for jobs in STEM fields is twice the mean annual salary for all jobs combined. This data comes from 2009, and includes 97 STEM occupations. Only four of those occupations yielded a salary lower than the national mean of $43k. The mean for all STEM occupations was $78k. Part of this disparity is likely due to the inclusivity of “all occupations” and the limited definition of “STEM occupations” – see the report for more details on the methodology.

The US Department of Commerce gives us a little more detailed breakdown of wage differences by including level of education in their calculations. This data is from 2010, from their report “STEM: Good Jobs Now and For the Future”:

Average Hourly Earnings Difference
STEM Non-STEM Dollars Percent
HS diploma or less $24.82 $15.55 $9.27 59.60%
some college/associate $26.63 $19.02 $7.61 40.00%
bachelor’s only $35.81 $28.27 $7.54 26.70%
graduate degree $40.69 $36.22 $4.47 12.30%

STEM fields outperform non-STEM fields at every level, but much more dramatically so with less education. This is interesting and useful information to have; though this section of the research deals primarily with higher education, we will need to return in the future to pre-college exposure to STEM when looking to determine what population segments need to be provided with more opportunities.

Since we’ll be looking specifically at engineering populations in the next section, it’s worth breaking STEM incomes down a little bit more and looking at engineering incomes v. overall STEM incomes. These figures are median annual US salaries from 2010, again from the NSF:

29 & younger All ages
All occupations $47,000 $70,000
S&E occupations $56,000 $80,000
Engineering occupations $62,000 $87,000

Just as science as engineering fields together pay much more than the median for all occupations, engineering fields alone pay much more than the median for science and engineering fields together.

Trends in Higher Education Populations

Now for our trends in student populations in higher education. All data in this section comes, again, from the NSF.

The following charts compare the population of US residents aged 18-24 with the population of students in US higher education institutions of all kinds. This is not, of course, a perfect overlap – many 22 year-olds have graduated from college, while some college students are older than 24. Furthermore, I have excluded non-residents from these statistics and matched the US Census Bureau’s “two or more races (non-Hispanic)” category to the NSF’s “other/unknown” category. These choices made the integration of the two data sets more intuitive and useful. “Hispanic/Latino” can include any racial identity, but there is no overlap – the other categories do not include individuals who identified as Hispanic, regardless of other affiliations.

These data show us something heartening: between 2002 and 2012, the US student body population became more similar to the 18-24 year-old segment of the US population. For every group except the unknowns, percentage of the higher ed population was more similar to percentage of the total population in 2012 than in 2002.

There are still disparities, though. Black students and white students are underrepresented in higher education at about the same rate in 2012, with about a 1% difference in their share of the total vs. higher ed population (in 2002 black students were twice as underrepresented, while white students were overrepresented); Hispanic students are much more underrepresented, at 4.4%, though that is still an improvement over the 2002 rate of 5.9%. Meanwhile, the somewhat mysterious category of students identifying as more than one race (non-Hispanic) or “other” has increased its representation in higher ed over 2002, and in both years was already much greater than its representation in the general population. Given my lack of a clear definition for this group, however, I won’t draw any conclusions about it from the data sets I have available.

This table displays the differences in a group’s percentage of the general population and its percentage of the student population. A negative number indicates that its student representation is smaller, as a percent of the whole, than its total population.

Identity Group 2002 2012
total 0.0% 0.0%
white 1.6% -1.1%
black -2.2% -1.0%
Asian/Pacific Islander 1.8% 0.6%
Native American/Alaska Native 0.1% 0.0%
two or more/other 4.6% 5.9%
Hispanic/Latino -5.9% -4.4%

Now that we can say with some confidence that our student population is getting more similar to our general population, let’s ask the next question: when looking at STEM degrees in particular rather than higher education as a whole, do we see the same population trends?

The data I have for the range of time we’ve been looking at (2002-2011) deals specifically with engineering fields, which we’ve already seen to have much greater disparities than all STEM fields together. That makes our engineering data both less useful – it provides a distorted picture relative to the whole of STEM – and more useful, since disparities are thrown into relief for us to see more clearly.

We know that women are underrepresented in engineering, but let’s quickly quantify that. These charts represent undergraduate women and men in all US engineering programs as percentages of the total number of students in those programs:

women v men in US engineering programs 2002, 2012 pie chartsGiven that in the total undergraduate body we know that women outnumber men, these charts should drive home the stark evidence of women’s lack of participation in engineering, and the lack of progress made on that front in a decade. There was some fluctuation between 2002 and 2011: women’s share of engineering education actually went down further before climbing again to just slightly above its 2002 level.

men and women in engineering programs 2002-2011

What about ethnic and racial affiliation groups? In the following graph, we can see each group’s share of the whole from 2002 to 2011:

EngiRacEth_linesThis graph is hard to read in terms of progress, though. The changes in numbers look small, and it’s hard to tell whether or not we’re getting closer to representations that match our overall percentages of students in higher ed.

To answer that question, we need to compare each group’s share of the total population of undergraduates to its share of the total population of undergraduates in engineering programs. Those shares are calculated as percents of the whole; the differences refer to differences in strength of representation, not to numbers of students. Data points above zero indicate that a group has a greater share of students in engineering than of undergraduates as a whole; points below zero indicate that a group has a smaller share of students in engineering than of undergraduates as a whole. If interest in and access to engineering degrees is evening out across groups, then we should see all trend lines getting closer to zero.

trends in pursuit of engineering degrees 2002-2011 by race and ethnicityThis, however, is not the case. (Note that my data sets only include figures for the category “other or two or more” in 2010 and 2011.) The engineering share for black and Hispanic groups, as of 2011, was dropping; for Asian students, it was rising slightly, and for white students it was dropping after a long rise, althoug white students still exhibit the greatest ratio of engineers to total undergraduates. Furthermore, we can see that engineering trends are the opposite of total higher education trends: while total higher ed shares are getting closer to representing shares of the general population, engineering trends are getting farther from representing shares of the total education population.

Engineering trends by race and ethnicity are the opposite of total higher education trends. Click To Tweet

Given the financial disparities between STEM and non-STEM incomes — and, in particular, between engineering incomes and non-STEM incomes — we might consider these trends to be a rather noteworthy indication of future economic inequality in spite of the progress made for higher education as a whole.

Up next: exploring possible strategies for making engineering and other STEM fields more accessible for those groups with less access. What are the roadblocks? How can we break them down?

 

Leave a Reply

Your email address will not be published. Required fields are marked *