Social Mobility and STEM education: its a complicated story.

Americans like to think that their country is a fair and just meritocracy. Central to these meritocratic principals, is the idea that with dedication and a strict work-ethic any individual — regardless of ethnicity, place of birth, race, gender, or any other personal characteristic — can have a successful life. Not only that, but baked into the ethos of this country is the belief that parents can expect their offspring to live richer lives than they did. At least, the expectation goes, their children are guaranteed better prospects so long they access a good education. So is the “american dream” really true? Can future generations expect higher income than their ancestors? Does a good education guarantee a successful and affluent life?

Figure 1

These questions are not easily answered. There exists a wealth of literature trying to understand social mobility, education and their interaction (see bibliography). Yet, research into the role played by science, technology, engineering and mathematics (STEM) education is lacking. By using datasets from the National Science Foundation on STEM educational outcomes (at various levels) and data on social mobility from the Opportunity Insights Lab, we can begin to understand the relationship between STEM educational outcomes and social mobility in the United States.

Specifically, we can visualize how the percent of public-school students who score at or above the national standard, in both science and mathematics, varies across states and across time. Figure 1, shows the percent of public school students in fourth and eighth grade deemed proficient in both domains. It is immediately clear that there exists high variance in proficiency across states for both disciplines (science and mathematics). Some states such as Mississippi, Louisiana and Alabama, have rates of around 30% or less proficiency for both age groups (fourth and eighth grade students). Others such as, Vermont and New Jersey, achieve closer to 50% proficiency. Yet, one thing is consistent across states, the vast majority have seen improvements in proficiency, for both disciplines and grade-levels, in the past decade or two. However despite these improvements, most states, with few exceptions, continue to have 50% or less proficiency in both disciplines. So while we have seen improvements across the country in both primary and secondary school attainment in both science and mathematics, the country is still a far cry from achieving consistently adequate proficiency in primary and secondary STEM education

Furthermore, by grouping between regions, we can see how unequal science and mathematics proficiency is across state lines. In figure 2, each dot shows the average proficiency (averaged across years) for an individual state. States in both the Northeast and Midwest are more tightly grouped together, with the worst performing states in these regions achieving just below 30% proficiency and the best performing states achieving close to 50% proficiency. The South and West of the country, however, shows a completely different picture. While there are high performers in both regions, achieving almost 50% proficiency, there are also states with 20% or less proficiency. The ranges of proficiency in the South and West regions illustrate just how unequal and varied STEM proficiency is across the United States for both primary and secondary schooling.

Figure 2

We now move to understand a different indicator of STEM education achievement: the number of science and engineering (S&E) degrees conferred per 1000 individuals aged 18–24 years old. In figure 3, we can see how the number of bachelors conferred has changed in the past two decades.The number of S&E degrees conferred for each state per year is encoded by color. The darker color means more degrees were conferred and vice-versa for the brighter colors. We can also visualize the change from year-to-year by looking at the direction of the triangles, with triangles pointing down signifying there was a fall in degrees conferred from the previous year and vice-versa for triangles pointing up.

Figure 3

There is a lot of information to digest from figure 3, but two findings are worth dissecting. First, D.C. has seen a consistently high number of students graduate with S&E bachelors degrees, and has seen continual improvement over the past two decades. Second, Nevada and Alaska have seen a low number of STEM bachelors since the turn of the century with minimal improvements across the years. As with fourth and eighth grade achievement, it is clear there is high variance across states in S&E bachelors’ attainment.

There exists high variance in STEM educational outcomes across states, regions, and education levels, yet some trends are consistent regardless of region. Figure 4 allows us to, albeit superficially, understand the relationship among different STEM indicators. Each dot in figure 4 represents one state, they are colored according to their region. The gray lines, which shows the best fit line, can be interpreted as the relationship between the variable plotted on each axis.

Figure 4

Some clear patterns emerge from this visualization. First, there is a strong, almost perfect, linear relationship between the same discipline but different education levels (i.e., fourth grade math is positively almost perfectly linearly correlated with eight grade math). Furthermore, the same relationship, but a tad weaker, is present when comparing across disciplines (i.e., science proficiency has a strong linear relationship with math proficiency, regardless of grade level). Most interesting, however, is the relationship between S&E bachelors conferred and public-school proficiency levels. The top row of figure 4 shows how, while still positive, the relationship between bachelor attainment and fourth/eighth grade attainment is much weaker.

The Opportunity Insights Lab, a pioneering research group with the mission “to identify barriers to economic opportunity and develop scalable solutions that will empower people throughout the United States to rise out of poverty and achieve better life outcomes”, makes the data from their research publicly available. Using their work, we can show the very stark reality of social mobility in the United States.

Figure 5

Figure 5 illustrates the massive decline in social mobility that occurred between the end of the XX century and the beginning of the XXI century. All people born in 1940, regardless what state they lived in, had at least an 85% chance of earning more than their parents by age 25–34. For the cohort born in 1980, the picture looks much grimmer. Residents of every single state, without exclusion, have seen the prospects of earning more than their parents, fall. For some states such as Alaska, that fall has been astounding. Alaskan residents born in 1940 had almost a 90% chance of earning more than their parents, those born 40 years later, have less than a 40% chance.

The picture for the “best performing” state hardly look any rosier. Residents from D.C. born in 1980 had around a 65% chance of earning more than their parents, those born 40 years before had almost a 90% chance. It is thus clear that the promise of giving your children a better life is becoming unlikelier with each passing generation.

The question then becomes, is STEM educational attainment a good predictor for social mobility? And if so which level of STEM education attainment (fourth grade, eight grade or bachelors) is the best predictor? To understand these relationships, we make use of a correlation matrix.

Figure 6

Figure 6 shows a couple of things we had already made clear in previous figures, mainly that fourth grade and eighth grade attainment are strongly related, in both domains. Yet, what now becomes clear, is the relationship between social mobility and STEM educational outcomes, shown in the bottom row of figure 6. All education variables have a positive relationship with social mobility, yet it is S&E bachelors’ attainment that has the strongest relationship with social mobility. From all STEM education variables, S&E bachelors per 1000 residents (age 18–24) might be the best predictor for social mobility. Below, we will see how this relationship plays out state by state.

Figure 7

Figure 7 shows in more detail how the relationship between our chosen social mobility index (probability of earning more than your parents between ages 25–35 for the 1980 cohort) and S&E bachelors attainment varies throughout different states. Certain states, New Hampshire D.C. and Maryland for example, have relatively high social mobility as well as high bachelors S&E attainment. Yet, this trend is hardly consistent across the country. For example, states such as Arkansas (AR) and Kentucky (KY) enjoy high social mobility, yet graduate very few S&E bachelors per 1000 residents. In contrast, some states such as Vermont (VT) graduate many S&E bachelors per 1000 residents, yet less than 50% of Vermonters born in 1980 can expect to earn more than their parents.

Figure 8

The inconsistent relationship between bachelor’s attainment and social mobility carries out in a similar fashion when comparing fourth and eighth grade proficiency to social mobility. Figure 8 shows how each state performs relative to the national average. States in green performed worse than the national average while states in blue performed better. It seems that while there is a positive relationship between social mobility and stem educational outcomes, this relationship is hardly consistent across all states. Thus, while STEM education is important it is not sufficient in predicting increased social mobility.

Concluding remarks, the goal of this piece was not to describe a specific research project or to make an exhaustive review of the literature. Rather, the intention of this piece is to visualize a set of relevant datasets constructed by governmental agencies and academic research institutions. It is clear that the relationship between social mobility and STEM education outcomes is complex, yet there are three key take-aways that I hope all viewers keep from this analysis.

• The viewer will recognize that STEM education proficiency is highly unequal across states and regions.
• The viewer will remember that between the end of the XX century and the beginning of the XXI century the United States saw massive declines in social mobility.
• The viewer will understand that STEM education is important but not sufficient in predicting increased social mobility.

Bibliography
Data Sources:

• NSF Science and Engineering Education Indicators:
  https://ncses.nsf.gov/indicators/states/
• Opportunity Insights Project:
  http://www.equality-of-opportunity.org/data/

Bibliography:

1. Chetty, R., & Hendren, N. (2016). The Impacts of Neighborhoods on Intergenerational Mobility I: Childhood Exposure Effects. doi:10.3386/w23001
2. Mitnik, P. A., & Grusky, D. B. (2020). The Intergenerational Elasticity of What? The Case for Redefining the Workhorse Measure of Economic Mobility. Sociological Methodology, 50(1), 47–95. doi:10.1177/0081175019886613
3. Bell, A., Chetty, R., Jaravel, X., Petkova, N., & Reenen, J. V. (2018). Who Becomes an Inventor in America? The Importance of Exposure to Innovation*. The Quarterly Journal of Economics, 134(2), 647–713. doi:10.1093/qje/qjy028

Further resources

• The Brookings Institute: Thirteen Economic Facts about Social Mobility and the Role of education.
  https://www.brookings.edu/wp-content/uploads/2016/06/thp_13econfacts_final.pdf
• PEW Charitable Trust: Economic Mobility in the United States
  https://www.pewtrusts.org/~/media/assets/2015/07/fsm-irs-report_artfinal.pdf