Building the Future, Alone: The Effect of Working in Science, Technology, Engineering and Mathematics Occupations on Civic Engagement in the United States

Itay Weiss

Abstract:  Breakthroughs in science, technology, engineering and mathematics (STEM) change the way we live our lives. While these breakthroughs can find the cure for the most lethal diseases, every change carries heavy social implications. Unlike previous revolutions, this one is mainly carried out by the scientists, technologists, engineers and mathematicians. Using logistic regressions on data from the U.S. Census Bureau’s Current Population Survey, September 2017: Volunteering and Civic Life Supplement, I examine the hypothesis that workers in STEM occupations are less civically engaged. The findings of this research support the hypothesis and show that people who work in a STEM occupation are less likely to participate in activities that define civic engagement such as volunteering, voting in a local election, attending a public meeting, and engaging with someone from a different race or ethnicity . In addition, I show that workers in STEM occupations with academic education tend to be even less civically engaged. This analysis aims to improve understanding of exogenous factors facilitating harmful social effects of innovation and to help decision-makers in the public and private sectors design programs and policies to address the problem. 

I. Introduction

Unprecedented progress in the STEM field has changed every aspect of contemporary society. Social media platforms have altered the ways people consume news and communicate with friends. Similarly, Wikipedia has changed the way pupils learn and determine facts about the world. In some firms, even the traditional boss has been replaced by an algorithm that allocates workload based on consumers reviews. In the future, emerging technologies, such as artificial intelligence (AI), genome editing, or interconnected devices (the Internet of Things, or IoT), will bring further breakthroughs: in health, finding cures for serious illness; in markets, automating many occupations; in transportation, saving countless lives with self-driven cars; in education, allowing information to be spread around the globe with minimal barriers to transaction.

However, these breakthroughs disrupt the ways we live our lives, both individually and within communities (MIT 2018). In Myanmar, for example, military officials were behind a systematic social media disinformation campaign targeting the Muslim Rohingya minority, which has indisputably led to the minority’s victimization through murder, rape, and forced migration (Mozur 2018). Most Americans—72% of U.S. internet users above the age of 15—have seen someone being harassed online. Internet users who are between the ages of 15 and 29, black, and/or identify as lesbian, gay, or bisexual are all more likely to suffer online harassment (Lenhart 2016). In the online workforce, many platform workers (e.g. Uber drivers) fear the consequences of not completing a job, causing some to put themselves in harm’s way. Some refrain from filing complaints on inappropriate customers’ behavior to avoid penalization through a lower rating and loss of future work (Ticona 2018). In the U.S. criminal justice system, an algorithm with harmful implicit bias is used to make decisions about pretrial release and parole. This algorithm has been found to contain a strong racial bias; in Wisconsin, the algorithm was nearly twice as likely to categorize black defendants as high risk than their white counterparts (Angwin, Larson and Mattu 2016).

These problems do not result from a technically poor structure, flawed code,  or lack of innovation. Instead, the negative externalities of technological advancements are linked by a lack of the three ingredients of civic engagement: education (history, ethics, social, and politics), diversity, and the civic involvement of STEM professionals. Implicit in this critique is that developers should themselves be of diverse backgrounds, roundly educated, and involved with those affected by their work. However, these priorities are not usually central to STEM professionals’ work. This critique raises two important policy questions:

• First, are people in the STEM field really less civically engaged?

• Second, what is the effect of education on civic engagement in the STEM field?

In reference to these two questions, I test the following hypotheses. First, I hypothesize that workers in STEM occupations are less likely to be civically engaged. Second, I posit that workers in STEM who hold at least a bachelor’s degree are even less likely to be civically engaged. This novel analysis investigating the relationship between working in STEM occupations and various measurements of civic engagement sheds light on a topic almost unexplored by the academic literature.

II. Background

In 1959, the British physicist and novelist C.P. Snow delivered a famously controversial lecture at Cambridge University. Describing a dangerous postwar schism between two groups—scientists and the literary world—Snow identified an emerging divide, across which each party was more than happy to sneer at the other (Dizikes 2009). Those divisions within the academy now seem more deeply entrenched than ever. Ultimately, the world needs engineers, programmers, and scientists who are able to think as holistically, as well as practically, and who see social answers with equivalent importance as technical answers. The world needs STEM leaders who are as conversant with Aristotle and the rise of fascism in the 20th century as they are with Pythagoras and the second law of thermodynamics.

The technology sector, containing the vast majority of STEM workers, is the most influential sector in the economy (Evans 2001). It is estimated to be worth $11.5 trillion globally, equivalent to 15.5% of global GDP, and it has grown 2.5 times faster than global GDP over the past 15 years (Huawei & Oxford Economics 2017). A 2009 Intel-Newsweek Study found that  78 percent of Americans believe that technological innovation is more important than ever in driving U.S. economic success in the next 30 years.  (McGinn 2009). The individuals who develop technology have directly or indirectly directed critical decisions, such as which news stories we read, the nature of personal data being collected, and who gets parole, insurance, or housing loans.

In recent history, the contract between the public and the STEM sector has been that technology companies will make life better, easier, cheaper, and more enjoyable while protecting consumers against malware. However, recent incidents, including personal data breaches, anti-competitive behavior, and massive disinformation campaigns online (such as the foreign interference into the United States 2016 presidential election), have raised concerns that technology companies have been neglectful in oversight of the potentially adverse implications of their innovations. Recently, a growing body of practitioners (Skirpan et al. 2018) and leading research institutions (e.g., Data & Society, Stanford, MIT, Oxford and University of California) have begun exploring the effect of technology on society. On Capitol Hill, Congress is realizing the risks of irresponsible innovation and is working to make sure technology companies will be held accountable.  It is clear that socially responsible innovation will become an imperative priority moving forward. Central to that goal is a STEM workforce that is civically educated, informed, and engaged in the community it serves. 

What is STEM?

The Standard Occupational Classification Manual classifies STEM occupations into six groups, including computer and mathematics, architecture and engineering, and life, physical, and social sciences (OMB 2018). More generally, STEM workers use their knowledge of science, technology, engineering, and math to try to understand how the world works and solve problems (Vilorio 2014). In 2015, there were nearly 8.6 million STEM jobs in the U.S., representing 6.2% of the country’s total employment. Computer occupations comprised nearly 45% of STEM employment, and engineers made up an additional 19% (Fayer et al. 2017) .

Figure 1: STEM Employment by Type of STEM Occupation

In May 2015, seven of the 10 largest STEM occupations were related to computers and information systems. Application software development was the most common STEM occupation, employing nearly 750,000 American individuals (BLS 2015).

Figure 2: Employment for the Largest and Smallest STEM Occupations, May 2015

Wages for STEM occupations vary widely. In 2015, the average annual salary for all STEM occupations was $87,570, nearly double the average salary for non-STEM occupations ($45,700) (Fayer et al. 2017) . Ninety-three percent of STEM occupations had higher salaries than the national average salary for all occupations ($48,320) (Fayer et al. 2017) . Over 99% of STEM employment is in occupations that typically require some type of postsecondary education for entry. In comparison, only 36% of overall employment in the U.S. requires a postsecondary degree. Occupations that typically require at least a bachelor’s degree for entry, like software development and engineering, comprise 73% of STEM employment but only 21% of overall employment (Fayer et al. 2017) (Figure 3). Over half of the remaining STEM employment is in occupations that typically require at least an associate’s degree for entry, like web development and being an engineering technician.

Figure 3: Employment Distribution by Typical Entry-level Educational Requirement

The next chapter will review the existent literature and show how this thesis will contribute to it.

III. What do we know about civic engagement in the STEM labor force?

Civic engagement is a broad concept describing one’s level of, or involvement in, empowerment and political action, groups and networks, trust and solidarity, information and communication, and social cohesion and inclusion (Huyser 2017).

Civic Engagement

The academic literature often use the American Psychological Association’s (APA) definition of civic engagement: “individual and collective actions designed to identify and address issues of public concern” (Carpini 2020). The APA definition includes volunteerism, voting, and organizational memberships. The Pew Research Center’s Internet & American Life Project divides these volunteer and organizational membership activities into different categories to allow for ranking of the levels of political participation (Smith 2013).

To better understand civic engagement, it is important to explore the concept of social capital and the features of social capital that enable the creation of groups to work toward a social goal. Putnam (2000) claims that U.S. social capital, as a result civic engagement, has suffered from a steep decline since the late 20th century. Putnam defines social capital as “features of social life –networks, norms, and trust – that enable participants to act together more effectively to pursue shared objectives.” (Putnam 2000). The theory widely accepted in the literature is that volunteering, attending public gatherings, voting, protesting, and engaging within communities fosters more trust and confidence in others, both for activists and among community members. This in turn strengthens community members’ abilities to take collective action (Jennings and Stoker 2004). Holbert et al. (2005) also explore the relationship between civic associations and social trust and find a positive association between holding trusting attitudes toward others and the likelihood of becoming civically engaged.

Civic Engagement and Education

Traditionally, the education system has introduced civics and ethics to American children with the goal of shaping their moral and democratic development as active citizens. The literature shows that “engaging young children in civic activities from an early age is a positive predictor of their participation in later civic life” (Bers 2008). Currently, high school-level civic education is limited both in scope and long-term effect, multiple studies demonstrate a correlation between education and civic engagement (Branson 1998; Tarsem et al. 2007; Brown 2018; Skirpan et al. 2018), and traditionally, people with higher levels of education tend to be more civically engaged (Campbell 2006). Hillygus (2005) identifies SAT scores and college curriculum as the factors most strongly associated with increases in civic engagement. She also finds a positive correlation between taking social science courses and being civically engaged in the years after college (Hillygus 2005).

As demonstrated in the literature described above, students’ exposure to civics, social education, and sustained engagement with information about the political system tends to be associated with higher levels of future involvement. These findings imply that STEM education that fails to expose students to these subjects may affect their civic engagement outcomes later in life (Hillygus 2005; Steele et al. 2012).

Civic Engagement and STEM Education

As an educational field, STEM differs significantly from many other academic disciplines. STEM  training is largely based on quantitative teaching methods. Traditionally, the field lacks diversity and gender equality, and it does not incentivize community engagement (Harkavy et al. 2015; Tarsem et al. 2007). In recent decades, the number of women, underrepresented minorities, and persons with disabilities earning STEM degrees and entering STEM careers has increased (NSF 2013); however, significant gaps remain. In their 2011-2012 biennial report to Congress, the Committee on Equal Opportunities in Science and Engineering (CEOSE) acknowledged the progress in diversity efforts within STEM but asserted that the achievements were insufficient in overcoming the historical patterns of marginalization (NSF 2013). For example, women, minorities, and persons with disabilities remain underrepresented in STEM higher education professions even though they comprise an increasing portion of the overall workforce. The share of women has remained disproportionately low in graduate programs in engineering (23%), computer sciences (25%), physical sciences (33%), and economics (38%) (Marcus 2013; Mason et al. 2010). In 2008, women made up only 29% of tenure-track faculty in STEM fields, and in 2010, women accounted for less than one-third of all individuals employed in sciences and engineering (Marcus 2013; Mason et al. 2010). STEM educational pathways also fail to produce representative portions of black, Hispanic, American Indian, and Alaska Native individuals completing bachelor’s degrees; this leads to underrepresentation at the highest levels of STEM employment. Together, these traditionally marginalized groups comprise just 10% of workers in science and engineering occupations (Beede 2011).

As a result of the lack of diversity and low participation of underrepresented groups, graduates of STEM fields are assumed to be less civically aware, less connected with their communities, and less aware of the unintended social consequences of their creativity than their non-STEM counterparts. The Organization for Economic Cooperation and Development (OECD) aptly frame the problem as one concerning social cohesion, which they describe as "the extent of connectedness and solidarity among groups in society” (OECD 2011). Solidarity and connectedness are not advanced when historically underrepresented groups cannot be full participants in key societal activities. As with other forms of inequality, the lack of representation in STEM serves as a barrier to social cohesion (OECD 2011).

IV. Are people who work in STEM likely to be less civically engaged?

As noted in the literature, traditionally, high levels of education are positively correlated with civic engagement. The vast majority of STEM workers have high levels of education. However, I hypothesize that STEM workers with academic education may be less civically engaged than their similarly educated peers.

My conceptual model focuses on civic engagement as the dependent variable, but the definition of civic engagement is complex. This study examines four different formulations of the term:

1. Vote in Local Elections: the likelihood that an individual voted in local elections, such as mayoral or school board elections, in the last 12 months.

2. Volunteer: the likelihood that an individual volunteered for any organization or association in the last 12 months.

3. Attend a Public Meeting: the likelihood that an individual attended a public meeting, such as a zoning or school board meeting, to discuss a local issue in the last 12 months.

4. Engage with Others: the likelihood that an individual talked or spent time with individuals from racial, ethnic, or cultural backgrounds that differ from their own in the last 12 months. This may have been in person, over the phone, or through the internet/social media.

 Whether one works in a STEM occupation is the key independent variable. To allow for another dimension of analysis, the model includes two additional second-tier key independent variables: whether one lives in a STEM metropolitan area and whether one works in a STEM industry. The first variable references eight U.S. metropolitan areas with the highest ratio of STEM employees as a share of total employment.[1] The second variable references the ten U.S. industries with the highest STEM employment as a percentage of total employment.[2] Also considered are a number of demographic characteristics that are assumed to be correlated with civic engagement: educational attainment, employment sector, number of hours worked in main occupation, household income quartile, service in the U.S. armed forces, age, marital status, sex, geographical-regional area, and race.

V. Data and Methods

I use a logistic regression to examine the likelihood of civic engagement. The dataset used for the empirical analysis is the 2017 U.S. Census Current Population Survey: Volunteering and Civic Life Supplement, which includes employed, working-age individuals.[3] Data are provided on civic participation during a one-year period from September 1, 2016 to the date of the interview (September 2017). In order to test for a relationship between working in a STEM occupation and holding a bachelor’s degree or higher, a second model includes interaction terms for the two highest levels of education. The decision to use logistic regression was based on a comparison of various methods (logit, probit, and OLS). Logistic regression produced the most accurate, meaningful, and significant results. Because the outcome of logistic regression ranges between zero and one, it produces probability measurements that fall within a meaningful and interpretable range without negative numbers or percentages exceeding 100.

Dependent Variables

Civic engagement refers to broad and malleably defined concepts that take on different meanings depending on context and that are not amenable to direct statistical measurement. However, specific dimensions of these broad constructs—behaviors, attitudes, social ties, and experiences—can be more narrowly and tangibly defined and are thus more feasibly measured (Prewitt et al. 2014). Prewitt et al. (2014), who led the panel on Measuring Social and Civic Engagement and Social Cohesion in Surveys of the Committee on National Statistics for the National Academy of Sciences, were asked to define research needs and goals related to civic engagement and social cohesion. Their report defined measures of civic engagement and categorized them by units of analysis, data collecting modes, and the nature of the phenomena. Based on their extensive groundwork, this model uses measures that exist in available data and are relevant to the unit of analysis. The relevant measures are voting (political act), volunteering (nonpolitical act), attending a public meeting, and engaging with others. While Prewitt et al. (2014) do not mention the last measure specifically, it is consistent with their analysis. The original CPS dataset defines each of these metrics as binary variables with the exception of “engaging with others,” which is a categorical variable with 6 categories (every day, a few times a week, a few times a month, once a month, less than once a month, and not at all).

Key Independent Variables

The key independent variable is a binary variable that aggregates the individuals who work in STEM occupations. As defined the CPS dataset, STEM occupations consist of 51 specific careers in computer and mathematical sciences, architecture and engineering, and life and physical sciences. Notably, the survey employs a conservative definition of STEM occupations. STEM teachers, post-secondary education teachers, and specific kinds of STEM sales representatives are not included in this analysis.

As previously mentioned, in addition to the key independent variable, two second-tier independent variables are included: living in a predominantly STEM metropolitan area and working in a STEM industry. Living in a STEM metropolitan area is defined as residing in one of the eight U.S. metro areas where STEM occupations comprise over one-fifth of employment. Although STEM occupations account for only 6.2% of national employment in 2015, they account for nearly 23% of employment in California-Lexington Park, Maryland and over 22% of employment in San Jose-Sunnyvale-Santa Clara, California and Huntsville, Alabama. Boulder, Colorado and Corvallis, Oregon also had among the highest employment shares of STEM occupations.

Working in a STEM industry is defined as being employed in one of the 10 industries where STEM occupations account for between one- and two-thirds of jobs. Over 66% of STEM occupations are in the computer systems design and related services industry. Over 60% of total employment in architectural, engineering, and related services and software publishers, nearly 60% in the computer and peripheral equipment manufacturing industry, Scientific Research and Development Services, and over 40% in Data Processing, Hosting, and Related Services and in 4 other high technology manufacturing industries are STEM occupations.

Controls

The first control variable is education, which is coded as 16 ordered categories.  For simplicity, I create separate dummies for each level of relevant educational attainment: less than high school diploma (base category), high school diploma, some college education, associate’s degree, bachelor’s degree and postgraduate studies. The second control is the class of worker or sector of work. In the original dataset, the metric is organized into eight levels. This model reorganizes these levels as four dummies: public sector (all levels of government), private sector, nonprofit sector, and self-employed. The third control is household income, which is broken down into 16 categories in the original dataset. For ease of comparison, however, this model encodes the income variable into income quartiles. The fourth control reflects the effect of being a veteran using a binary variable indicating whether an individual ever served in the U.S. Armed Forces.

The model also employs an age control, divided into three ranges of 16 to 35, 36 to 56, and 56 to 66. Because the survey collected data from both citizens and noncitizens, I also include a binary control for being a U.S. citizen. I also incorporate five race dummies (Black, White, Asian, Native, and Other), two binary variables for sex and marital status, and four geographical variables (West, South, Midwest, and Northeast) to control for all fixed effects across regions.

Model 1

Civic Engagement[4] = 𝛽0 + 𝛽1stem_occupation + 𝛽2stem_metro + 𝛽3stem_industry + 𝛽4highschool + 𝛽5somecollege +𝛽6associatedegree + 𝛽7collegegrad + 𝛽8morethanba + 𝛽9publicsector+ 𝛽10nonprofitector + 𝛽11selfemployed + 𝛽12incomeq2 + 𝛽13incomeq3 + 𝛽14incomeq4 + 𝛽15veteran + 𝛽16citizenship + 𝛽17age16-35 + 𝛽18age36-56 + 𝛽19age56-66 + 𝛽20married + 𝛽21male + 𝛽22black + 𝛽23asian + 𝛽24native + 𝛽25other + 𝛽26hispanic

 Model 2

Civic Engagement  = 𝛽0 + 𝛽1stem_occupation + 𝛽2 stem_occupation*collegegrad + 𝛽3stem_occupation*morethanba +𝛽4stem_metro + 𝛽5stem_industry + 𝛽6highschool + 𝛽7somecollege +𝛽8associatedegree + 𝛽9collegegrad + 𝛽10morethanba + 𝛽11publicsector+ 𝛽12nonprofitector + 𝛽13selfemployed + 𝛽14incomeq2 + 𝛽15incomeq3 +  𝛽16incomeq4 + 𝛽17veteran + 𝛽18citizenship + 𝛽19age16-35 + 𝛽20age36-56 + 𝛽21age56-66 +  𝛽22married + 𝛽23male + 𝛽24black + 𝛽25asian + 𝛽26native + 𝛽27other + 𝛽28hispanic

VI. Descriptive Statistics

Tables 1, 2, and 3 present descriptive statistics for the outcomes, key independent, and relevant control variables. Because all of the variables are binary, their means represent percentages in the population. The statistics for the dependent variables show that, in the last 12 months, 55% of the survey population voted in local elections, 32% volunteered, 12% attended a public meeting and 52% engaged with others. In addition, 30% have post-secondary education, 68% work for the private sector, 94% are U.S. citizens, 52% are married, 80% are white, and 37% live in the southern U.S.

Table 1: Descriptive Statistics for Outcome Variables

Table 2: Descriptive Statistics for Key Independent Variables

Table 3: Descriptive Statistics for Independent Variables

VII. Results

Tables 4 and 5 summarize the findings of the logistic regression. The results show evidence of meaningful relationships between working in STEM and different measures of civic engagement. Most of the relationships are negative and, in that sense, support the hypotheses that workers in STEM occupations are less likely to be civically engaged and that STEM workers who hold advanced degrees (bachelor’s or more advanced degree) are even less likely to be civically engaged than their similarly educated non-STEM counterparts. However, several estimates lack statistical significance, and some results have a substantively insignificant magnitude. Table 4 shows the logistic regression estimates in the context of civic engagement measurements: voting in local elections, volunteering, attending a public meeting, and engaging with someone from a different race or ethnicity more than several times a month. Table 5 shows the logistic regression results when incorporating two additional interaction terms for each of the regressions in Table 4.

The interaction terms evaluate the relationship between individuals in STEM occupations who have at least a bachelor’s degree and the civic engagement measures. The results presented in Table 5 further support the hypothesis and demonstrate the need for further research on the effect of STEM education on civic engagement.

Table 4: Logit Regression Results for Model 1

Table 5: Logit Regression Results for Model 2

Vote in Local Elections

The first logistic regression, shown in column one of Table 4, shows a negative relationship that is compatible with the hypothesis, but it is not statistically significant at a 10% level (p-value of 0.16). People who work in a STEM job were 2 percentage point less likely to vote in local elections.

Interestingly, the coefficients for living in a STEM metro area and working in a STEM industry show a positive, statistically significant relationship. Some demographic characteristics that are associated with the average STEM worker provide additional information. For example, the coefficient on having a bachelor’s degree or a postgraduate degree and having a household income level in the highest quartile, both of which are common characteristics of STEM workers, is statistically significant and indicates a strong, positive association with voting in local elections (e.g. having a bachelor’s degree is associated with a 28 percentage point increase in the likelihood of voting in comparison to people with less than high school diploma). Table 5 shows that testing for the interaction between working in STEM occupations and having high levels of educational attainment and voting in local elections produces better statistical results: a p-value of 0.11 for working in STEM occupations and having a bachelor’s degree and a p-value of 0.16 for having post-graduate education and working in STEM occupations. Overall, the results doubled: individuals working in STEM occupations with a bachelor’s degree or with post-graduate education are 4 percentage point less likely to vote in local elections (in contrast with only 2 percentage point decrease for the general workers in STEM occupations).

Volunteer

The second logistic regression (Table 4) considers the effect of working in a STEM occupation on the likelihood of volunteering and yields no statistically significant results at a 10% level. Further, the results for the second-tier variables are mixed. Though the relationship between living in a STEM metro area and volunteerism was not statistically significant at a 10% level, the relationship between working in a STEM industry and volunteerism was statistically significant at that level. The results suggest that working in a STEM industry is associated with 2 percentage point increase in the likelihood that an individual volunteers. Similar to voting in a local election, having high levels of education and income, which are common traits among STEM workers, is also positively associated with volunteerism. STEM employees with bachelor’s degrees or post-graduate education were 20 and 25 percentage points more likely to volunteer, respectively, than STEM workers with less than a high school diploma.

However, Table 5 also shows that the interaction between the terms for the highest levels of education (holding a bachelor’s or postgraduate degree) yield a statistically significant, negative correlation. This relationship is compatible with my hypothesis. Bachelor’s degree-holders in STEM occupations are four percentage points less likely to volunteer than those not working in STEM. Workers in a STEM occupation that hold a postgraduate degree are six percentage points less likely to volunteer than those not working in STEM. More than 45% of the STEM workers in the survey sample hold bachelor’s degrees, and more than 25% hold postgraduate degrees. As noted above, occupations that require a bachelor’s or more advanced degree comprise 73% of the STEM job market. Therefore, it is reasonable to conclude that the majority of the people who work in STEM occupations are less likely to volunteer than non-STEM employees.

Attend a Public Meeting

Results of the third logistic regression are shown in the third column of Table 4. The regression tests the relationship between working in a STEM occupation and attending a public meeting without yielding statistically significant results at a 10% level. As above, results for the second-tier variables are mixed. The model indicates that residents of a STEM metro area are one percentage point less likely to attend a public meeting than nonresidents, but individuals working in the STEM industry are two percentage points more likely to attend a public meeting.

Coefficients on other demographic variables, such as higher education levels and the highest income quartile are statistically significant, but their magnitude is low. Therefore, the usefulness of their interpretations remain vague. There is, however, one exception: postgraduate degree-holders are 12 percentage points more likely to attend a public meeting than people without high school diplomas. Nevertheless, the interactions between the educational variables and my key independent variable yield no statistically significant results at a 10% level.

Engage with Diverse Populations

The fourth logistic regression, the results of which appear in the fourth column of Table 4, yields statistically significant results at a 1% level. In accordance with my hypothesis, STEM employees are three percentage points less likely to talk and/or meet with people from a different culture, race and/or ethnicity more than few times a month than were their counterparts employed in other professions. Again, the second-tier variables show mixed results.  The coefficient on living in a STEM metro area is statistically significant and shows that residents are seven percentage points more likely to engage with people whose demographic backgrounds differ from their own. The result suggests that people working in a STEM industry are two percentage points less likely to engage with diverse populations than non-STEM employees but the coefficient on the term is not statistically significant. In terms of the demographic variables, similar to the previous regressions, high levels of education are strongly associated with increases in the probability that individuals will engage with people whose backgrounds are dissimilar from their own.

The interaction between working in a STEM occupation and having a bachelor’s or postgraduate degree is statistically significant, which indicates that people who work in STEM occupations and hold the most common levels of education within these occupations are less likely to engage with someone from a dissimilar background. STEM workers with a bachelor’s degree are 5 percentage points less likely than their non-STEM counterparts, and STEM workers with postgraduate education are 9 percentage points less likely than their non-STEM counterparts to do so more than a few times a month. These findings support my second hypothesis that STEM workers with higher levels of education are even less likely to be civically engaged.

A salient characteristic of the survey sample relates to the Asians in the survey. While Asians comprise only 5% of the survey population, they account for a disproportionately high 20% of STEM workers in the survey. While Asians working in STEM are more likely to volunteer and engage with others than most races working in STEM, they are less likely to vote in local elections or attend public meetings.

Overall, these findings support the hypothesis that there is a negative relationship between working in STEM and being civically engaged, which is expressed as volunteering, attending a public meeting, voting in a local election, and engaging with people from a different race and/or ethnicity. Moreover, my results suggest that people who work in STEM occupations and hold advanced degrees are even less civically engaged than STEM employees at lower education levels.

VIII. Policy Recommendations

The analysis presented here suggests that people working in STEM occupations, especially those with advanced degrees, tend to be less civically engaged. In today’s tech-focused culture, people working in STEM occupations are often at the forefront of the innovations that will change the way individuals in society communicate, learn, work, and trade. Decisions they make have a tremendous impact on society, and it is essential that they possess the tools and insights required to think holistically about the effects of their creations. Unfortunately, my results indicate that people who work in STEM may be less engaged and informed about the needs and concerns of their communities than the average population. Therefore, there is a risk that the decisions STEM innovators make will be ill-informed.

The results of this study show a strong relationship across all measures between high levels of education and civic engagement. Thus, one would expect that individuals employed in STEM, more than 70% of whom have a bachelor’s or more advanced degree, would be more likely to participate in their communities than the average population. Among STEM workers, however, educational attainment appears to have an opposite relationship, and these individuals become less civically engaged. To correct this, I propose that STEM higher education programs integrate ethics, civics, and humanities into their curricula. Currently in the U.S. and around the world, a small number of STEM intensive schools offer electives in these subjects and only few require them as mandatory classes. (O’Neil 2017). Civic engagement, especially, should be formally taught alongside hands-on experiences that expose students to the challenges of their future technological constituency.

Another policy implication of this analysis is to encourage the government to incentivize volunteerism in private companies. Either through financial incentives or in terms of government contracts, private companies should be encouraged to grow a corporate culture of volunteering and engaging with communities around them. Looking specifically in the information technology sector, the dominant policy agenda has emphasized low regulation or self-governance for years (B. Smith 2019). However, this policy has led to some challenges and to the erosion in the public trust in technology. To inspire renewed trust in the new technologies of the future, it is important to ensure that technologists engage with the public and use civic means to do so. It should be an inherent part of the STEM organizational culture.

A final recommendation concerns diversity both in schools and in the labor force. The results of this study show that working in STEM occupations and industries makes people less likely to talk or spend time with individuals from different races, cultures, and/or ethnicities. The issue is exacerbated at higher levels of education. The prominence of STEM in the economy and in designing the future makes this a long-term risk for the field and for society. Government, academia, and the private sector should address this issue by creating more opportunities for diversity and adequate representation. Government, both at the local and national level, should invest in efforts to increase the participation of minorities and women in the STEM field. In addition, college admissions for STEM programs should prioritize diversity and apply affirmative action policies when needed. It is clear that diversity in companies improves performance and increases innovation, collaboration, and creativity (Gompers and Kovvali 2018). Therefore, STEM companies would be wise to adopt policies that encourage diversity in recruitment and create special programs for outreach to minorities and women.

IX. Further Research

As noted above, measuring civic engagement is a complex task. Limitations on the design of the survey and the collection of the data used in this study lead to suggestions for further research. Organizations, such as the Census Bureau, BLS, and nongovernmental organizations, should launch new surveys to map the ever-changing technological world. These surveys should include questions about internet usage, social media engagement, and other social-technological activities.

Many advocacy activities take a new form in cyberspace, and advocates no longer require the physical resources they traditionally needed to be effective. Dominant social movements, such as “#MeToo”, “Black Lives Matter,” and “March for our lives against gun violence” all started and grew online, only taking an in-person form after their online success. Thus, while many respondents may say they did not volunteer, they might still have engaged with a social cause and shared and amplified its voice over the web. Future surveys should attempt to capture this kind of online engagement. Collecting data on other activities that are not measured in the CPS survey, such as attending a protest, attending a religious institution (church), and belonging to a homeowner’s association could also improve the model. Accounting for additional details and even creating indexes of civic engagement to weigh the activities on a scale would allow for a further in-depth analysis of civic engagement.

This study aims to lay foundations for further examination of the social and civic effects of working in the STEM field. Nonetheless, it analyzes only four measures of civic engagement. To make a more insightful analysis, it will be necessary to examine all measures in the CPS Civic Engagement and Volunteer survey and in complementary CPS surveys, such as the Voting and Registration Survey and the Computer and Internet Survey.

A related area for improvement is in data collection on STEM as a field. Ideally, the Census Bureau and BLS will conduct more in-depth analyses of the demographic and geographic characteristics of people who work in STEM, including their educational backgrounds, and provide accurate measurements across all the surveys they conduct. For example, although BLS differentiates between types of sales representatives (e.g., STEM-related sales representatives and retail sales representative) in many of its economic analyses and datasets, most CPS surveys group these occupations together as “sales representatives.”

Further research is also needed to clarify the connection between STEM education and its effects on civic engagement. Current surveys do not collect detailed data on the different educational backgrounds of STEM workers, such as college majors or training. At the same time, more and more voices in academia—and especially in social and political science departments—have been calling for the incorporation of more ethics, civics, and humanities materials in STEM curricula. It is important to support those voices with empirical evidence about STEM education and civic engagement outcomes. 

Civic engagement is not just hard to measure; it is hard to encourage. Encouraging such engagement among STEM workers, however, is essential. As this study demonstrates, the STEM workers building our future may be less engaged in their communities than citizens would hope. On a daily basis, individuals working in STEM make decisions that impact our daily lives. It is vital that they do so with the input and consideration of the communities around them. Such civic engagement will equip them with a better understanding of the social consequences and possible unintended results of their creations, allowing them to build a more just and sustainable future for all.