4

An Overview of Broad Patterns of Inequality in PreK–12 STEM Educational Outcomes

In the previous chapters, we discuss how inequity has been a feature of U.S. education throughout history, as well as the nested nature of education policy. With that background in mind, in this chapter we now turn to address the committee’s task of providing an overview of broad patterns of inequality in preK–12 education. Specifically, this chapter captures what national and other large-scale data systems reveal about differences between student groups on what are often considered external markers of science, technology, engineering, and mathematics (STEM) success, most notably achievement test scores. The chapter also provides a snapshot of disparities in access and opportunity in STEM learning. Finally, it offers a critique of a national emphasis on such measures.

Much of the data available to characterize the state of K–12 STEM equity at the national level reflects educational policies and investments in large-scale data systems to monitor those policies. For example, in K–12 education, a major impetus of policy with respect to equity in STEM has been to reduce gaps in achievement—specifically on standardized tests—between different groups. As noted in Chapter 2, the No Child Left Behind Act of 2001 (2002; NCLB) ushered in an era in which it has become commonplace to gather and disaggregate standardized test score data by race, socioeconomic status, disability status, and language spoken at home—all in order to monitor progress toward equity goals like parity in achievement outcomes. Since the passage of NCLB, states have established longitudinal databases that enable policymakers to assess whether gaps between different groups are being reduced in accordance with policy goals. Researchers, in turn, have been able to aggregate data from state systems using

sophisticated algorithms for comparing scores to create portraits at the national level, as exemplified within the Stanford Education Data Archive (Fahle et al., 2021). These databases, in turn, have enabled new kinds of analyses about how educational “inputs” and opportunities relate to—and possibly partly explain—differences in achievement scores.

In addition to data on achievement, investments in longitudinal databases through federal grants and contracts have enabled a deeper understanding of inequities of access and opportunity. For example, one well-documented source of inequity of access is the “tracking” of students into higher- and lower-level courses in mathematics, which has consequences for students’ future opportunities to pursue postsecondary education and career opportunities (Domina et al., 2016; Gamoran & Mare, 1989; Giersch, 2018; Oakes, 1985; Stevenson et al., 1994). Databases for research such as the Education Longitudinal Study of 2002, the High School Longitudinal Study of 2009, and the National Longitudinal Study of Adolescent Health (1994–1995) include information about students’ academic course taking patterns that allows researchers to better understand how tracking occurs and the role that patterns of course taking have on educational attainment. Because these databases also include data on social and contextual factors (e.g., friendship ties, school climate), one can gain a sense at the national level for how these factors contribute to inequities in students’ educational trajectories (e.g., Frank et al., 2008).

At the same time, national datasets reveal much less about the policies and practices responsible for maintaining inequities of opportunity and outcomes, inequities in how resources are allocated to districts and schools, or about students’ experiences in STEM classrooms and how inequities emerge from the interactions that take place within classrooms. In large part, this may be due to the fact that such measures do not figure into contemporary accountability systems, even though in the past there have been calls for more data on opportunities to learn (McDonnell, 1995), and, more recently, for data on student experience (Schneider et al., 2021). Adequate data at the national level to analyze inequities in resources, such as per pupil funding, have only recently been possible through the creation of the National Education Resource Database on Schools (Shores et al., 2021).

But large-scale quantitative data sources also have significant limitations when it comes to studying teaching and learning in classrooms. Systems are only beginning to emerge that can capture, for example, differences in rates of participation in classroom discussion by gender and race (e.g., Reinholz & Shah, 2018), and these require fairly intensive data collection and analysis procedures. There are often entire groups of students whose achievement and opportunities are unaccounted for, either because they are rendered invisible by small sample sizes, such as Indigenous students (Demmert et al., 2006), or because relevant identities are not captured,

such as for students who identify as transgender or gender non-binary¹ (Leonardi et al., 2021). Moreover, the kinds of data provided about trends in achievement and attainment, and even in opportunity to learn, do not provide an account of how those patterns came to be. Statistical methods for addressing small sample sizes for intersectional groups of students are only beginning to be developed and explored in STEM education (e.g., Van Dusen et al., 2022).

In this chapter, we review trends that reveal inequities in achievement, and discuss problems with the public and policy emphasis on group differences in achievement test scores. Additionally, we consider data on differences in opportunities to learn STEM content in K–12; and data on inequities in out-of-school STEM learning. We conclude with a call for different kinds of data in the future that can help in improved assessment of national progress toward equity in STEM education at any point in time, as well as longitudinally, to address the limitations of current data systems.

DATA RELATED TO ACHIEVEMENT TEST SCORES

In this section, we review data about test score gaps in national assessments of student achievement, discuss the limitations of focusing on test score gaps, and then highlight efforts to better understand and contextualize such gaps using a new database that integrates local achievement data from across the country, the Stanford Education Data Archive (SEDA), as well as analyses of data from the U.S. Census’s American Community Survey.

The National Assessment of Educational Progress (NAEP), often referred to as the Nation’s Report Card, is the most prominent data source used to identify gaps between groups on standardized achievement tests. A snapshot of scores from the 2023 NAEP assessments in mathematics, science, technology, and engineering literacy shows gaps among different groups with respect to overall scores. Such gaps can also be seen across all STEM subjects and grades with respect to race/ethnicity, language spoken, disability status, and socioeconomic status. Consistently, white and Asian/Pacific Islander students score higher on NAEP than do Black, Latinx, and Indigenous students. Students who are emergent multilingual learners (that

___________________

¹ The committee notes that although we believe it is important to report on disparities based on sex and gender, the existing survey literature in this space is inadequate for describing the state of part of the field. First, much of this literature conflates sex and gender, where sex represented by a set of biological characteristics and gender represents an affective expression of identity. Generally, education research is more concerned with gender than with sex, but the terminology is oft conflated (Glasser & Smith, 2008). Additionally, the survey methodology typically used to describe disparities in gender relies on an understanding of gender as a binary state. The committee notes that this construction of gender ignores the complexity of gender as it is expressed by today’s student population, to the erasure of many identities.

is, those classified as English language learners) score lower than those who are English speakers. Students without identified disabilities score higher than those with such disabilities, and students from families with lower incomes, as indicated by eligibility for free or reduced-price lunches, score lower than those with higher incomes. In contrast to such persistent and consistent patterns of higher performance among other socially advantaged groups, the presence of a statistically significant male advantage in math and science test scores is variable across year, age, and subject; and even in cases where it is significant, the difference is always small in scope, at about one-tenth of a standard deviation (compared to racial gaps of more than one standard deviation in magnitude).

Data released in the fall of 2022 revealed that scores decreased and gaps among certain groups had grown since the last time NAEP was administered, prior to the pandemic. Declines in fourth grade mathematics were evident for all racial and ethnic groups; those declines were statistically significant for Black, Latinx, and white students, but not for Indigenous and Asian students or for students from two or more races. In addition, the score gaps between white students and their Black and Latinx peers were larger in 2022 than in 2019. Finally, although very small in scope, the gender gap widened during the period.

There are many possible explanations for the change in scores. Some may be quick to cite the results as evidence of “learning loss” in the wake of the COVID-19 pandemic. But the patterns in data do not show a “loss” of learning at all, since different populations of students take the test each year. The decline in scores means that fourth and eighth graders in mathematics scored more poorly in 2022 than fourth and eighth graders did in 2019; they are not the same students. Those differences in scores, moreover, are offset by an overall growth trend since 1990. Others suggested that the shift to remote learning was to blame for the widening of gaps between groups. But declines are evident both in districts that persisted longer in remote learning and in those that did not, and remote learning is correlated with other factors related to achievement scores, including poverty (Fahle et al., 2022). What appears true about the pandemic is that it made it much harder for students to learn in school, and it exacerbated existing inequities based on geography, income, and race.

These score gaps open many questions about inequities in STEM that cannot be answered by looking at scores at a single point in time or even over a longer period, and relatedly, the great emphasis on such gaps in and of itself is problematic in several regards. First, focusing on achievement test score gaps can promote a narrow conception of learning outcomes as that which can be measured easily on standardized tests (Gutiérrez, 2008). Second, focusing on gaps without considering the opportunities students have had to learn the content tested can reify ideas of innate differences

between groups. This leads to unproductive discussions about the deficits of children from different groups and lower support for the initiatives that might increase educational equity in STEM, rather than discussions and initiatives about the equitable distribution of resources or how to change instruction and other factors that may contribute to those gaps (Quinn & Desruisseaux, 2022). Third, attention to aggregate differences on average test scores between groups ignores substantial diversity within groups (e.g., the broad categories of “Latinx” and “Asian” each conflate the educational experiences of students from many different countries of origin and immigrant generational status backgrounds; e.g., Castillo & Gillborn, 2022). It also ignores variation across places (e.g., schools, districts, states; Reardon & Stuart, 2019). And finally, test score gaps between groups do very little to explain later disparities between groups in STEM degree attainment (e.g., Cimpian et al., 2020; Riegle-Crumb et al., 2019), even though promoting equity in STEM attainment is a very common refrain invoked for why the public and policymakers should care so much about students’ test scores. We unpack many of these issues in subsequent sections of this chapter, highlighting research on systematic inequality in students’ opportunities to learn STEM content in K–12.

Gaining Insight About Variability in Score Gaps

Analyses of large-scale databases can shed light on important variability in average test score differences between groups. One database that is potentially useful for gaining a deeper insight about achievement gaps is SEDA (Fahle et al., 2021), a database constructed from aggregating achievement data from schools and districts across the United States in all tested subject areas. While much of the data are not specific to STEM, analyses of SEDA datasets have been conducted on mathematics and other subjects, offering insights into some of the sources of inequity in achievement that can help expand the conversation about achievement beyond a focus on gaps among groups.

One analysis that examined inequities in mathematics highlighted the significant role of school districts in accounting for different gaps in test scores. In this analysis, Reardon, Weathers, et al. (2019) found that on average, across the country, there was no gender gap in achievement in mathematics. At the same time, they found that gaps varied among school districts, with some districts having more male-favoring gaps and others having more female-favoring gaps. The researchers also found that gaps in mathematics tend to favor males more in socioeconomically advantaged school districts and in districts with larger gender disparities in adult income, education, and occupations.

More broadly, in explaining achievement gaps between white and Black students, analyses of the SEDA dataset point to the significant role of disparity in the overall baseline poverty level of schools attended by these two groups, with higher-poverty schools where Black students are concentrated consistently posting lower scores (Reardon, 2016). The degree to which there are such racial gaps in concentrations of students from low-income households, moreover, varies considerably among states and among districts within states (Jang & Reardon, 2019). In other words, place matters, with economic, demographic, segregation, and schooling characteristics explaining 43 percent to 72 percent of the geographic variation in these gaps (Fahle et al., 2020; Reardon, Kalogrides, & Shores, 2019). However, there are few if any true “outliers” to the overarching patterns that show a strong pattern of association between school-level racial-economic segregation and achievement (Reardon, Weathers, et al., 2019).

Analyses conducted using other longitudinal databases confirm the importance of geography in mediating effects of school composition on student outcomes. For example, using data from Missouri and the U.S. Census American Community Survey data, Tate and Hogrebe (2015) analyzed relationships between poverty and algebra performance. Using advanced statistical techniques to control for spatial autocorrelation, they found that the association between poverty and student outcomes varied by regions across the state.

Analyses like these—of datasets that integrate local data across states and include other information about characteristics of schools and neighborhoods—point to important potential explanations of achievement score gaps that are not due to individual differences, but rather to larger patterns of segregation by income and race (see Box 4-1). And they point the way toward future research that might inform efforts to promote equity by pointing toward the need for system-level solutions (Reardon, 2016). At the same time, the data do not tell the story of why, for example, students are in racially and socioeconomically isolated schools. Which is to say, the historical and continuing policies that maintain segregated schools are missing from these analyses.

DIFFERENCES IN ACCESS AND OPPORTUNITIES TO LEARN STEM CONTENT IN K–12 SCHOOLS

In this section, we review evidence related to race, social class, and gender differences with respect to opportunities to learn² STEM content

___________________

² “Opportunity to learn” is an evolving concept in education scholarship that signifies the extent to which students, teachers, and schools have access to the conditions and resources necessary to support both short- and long-term student success (Marion, 2020).

in K–12 school settings. In this section, we focus on access to advanced coursework in STEM and key courses like algebra that are “gatekeeper” courses for advanced courses. The data come from a variety of national and longitudinal databases. As with achievement, there are a lot of data documenting “gaps” in representation between different groups, yet research points to factors that influence access to opportunities to learn STEM, which we will discuss below.

Enrollment in advanced courses in STEM subjects in K–12 is well established as a beneficial academic outcome. Within such courses, students typically have the opportunity to learn high-level and rigorous curricula that are typically taught by the most well-qualified and experienced teachers (Ingersoll 1999; Oakes 1990); therefore, students enrolled in such courses early in their educational careers generally learn more STEM content and are positioned to continue on to subsequent advanced courses (Xu et al., 2021). While educational researchers point out that explicit academic tracking (where students are designated as being in a college-preparatory track, regular track, or vocational track, each with their own unique set of courses) is no longer operative in most schools and districts, nevertheless, in practice, students are often clearly labeled according to whether they take advanced, regular, or remedial courses in math and science, with those students taking advanced classes early in their educational careers positioned to start accumulating educational advantages relative to their peers who are not enrolled in such courses (Lucas, 1999; Oakes & Guiton, 1995; Xu et al., 2021). Yet access to advanced courses is systematically unequal, as students of color and students from low-income backgrounds are less likely than their racially and economically privileged peers to attend schools that offer such courses, and less likely to enroll in such courses when attending schools that do offer them (Price, 2021; Xu et al., 2021). Related to the last point, small differences in early achievement test scores between groups grow larger via differences in opportunities to learn, which are then used to justify subsequent under-representation of racially marginalized and low-income youth in advanced STEM course work; such under-representation places students in a position of disadvantage to learn subsequent STEM content, including that measured on standardized tests (Condron, 2009; Long et al., 2012; Mickelson et al., 2013; Quinn, 2015). Disparities in early access to advanced course work, then, contribute to subsequent inequities of access, making it difficult to assert that students from different races and social classes are on a level playing field by the time they apply for higher education or postsecondary jobs.

Inequality in Advanced STEM Course Offerings

The Office for Civil Rights at the U.S. Department of Education collects data every two years related to educational equity from preK–12. The Civil Rights Data Collection (CRDC) supports efforts to monitor compliance to federal laws requiring that schools not discriminate on the basis of race, color, national origin, sex, and disability. The most recent CRDC in 2017–2018 yielded a dataset with more than 17,000 districts and close to 100,000 schools. Data collected include demographic information about

students, information about school discipline, course offerings, and teachers (Office for Civil Rights, 2018b).

Analyses of CDRC data show strong disparities in course offerings according to the racial composition of the school (Office for Civil Rights, 2018a). Specifically, analyses of nationwide data show that while about 50 percent of high schools nationwide offer calculus, only 38 percent of schools with a high concentration of Black or Latinx students (defined as 75% or greater) offer this course, which is a foundational course in the

pursuit of many STEM bachelor’s degrees. Large disparities exist in science as well. While 60 percent of all schools offer physics to their students, only 51 percent of predominantly Black or Latinx schools offer this course.

In another study using data from the CRDC, researchers examined associations between racial isolation and poverty concentration in schools, on the one hand, and schools’ course offerings, on the other (Leung-Gagné et al., 2020). In this study, researchers found large disparities in advanced course offerings (in both mathematics and science) between schools with high concentrations of students of color (defined in the study as those who are African American, Asian, Latinx, Native American, Pacific Islander, or of two or more races) and those with low concentrations. This same study found similar disparities in course offerings between schools with higher concentrations of students from low-income families and those with lower concentrations. Of low-poverty schools (schools enrolling 0–6% of students eligible for free or reduced-priced lunch of the total student body), 95 percent offered algebra II, while 93 percent offered advanced math. In contrast, algebra II was offered at 87 percent of high-poverty schools (schools enrolling 77–100% of students eligible for free or reduced-price lunch), and 71 percent offered advanced math. Eighty-seven percent of low-poverty schools offered calculus, almost twice the percent of high-poverty schools (45%).

The above results are consistent with the U.S. Government Accounting Office’s (2018) own analysis of federal data, which also pointed out that these disparities in offerings are consequential for students, in that they mean students in schools with high concentrations of students of color and low-income students are less likely to take courses required for admission into public four-year colleges and universities. Specifically, that study found that when there were more Latinx and Asian students, there were lower odds that a school would offer the three foundational math courses for college matriculation (algebra I, geometry, and algebra II); in science, schools with more Black, Latinx, or Indigenous students were less likely to offer the three foundational science courses (biology, chemistry, and physics). In addition, these results are consistent with analyses presented in the previous section showing disparities in achievement linked to these highly segregated schools.

The study mentioned above found some striking differences among states. For many states, the gaps in advanced course offerings between schools with high versus low concentrations of students of color (and high vs. low concentrations of students from low-income families) were more than 20 percent, with some approaching 50 percent. There were notable outliers as well, such as Washington state, where gaps related to race were low, and Florida, where gaps related to income were low. In each state, some have speculated that those differences are due to policies pursued for

several years to require districts to provide access to advanced course work (Blad, 2020). See Box 4-2 for an additional example of how lack of access has distinct consequences in STEM contexts.

These inequities in access to advanced courses are important to address because this access is related to achievement and attainment outcomes. A study by Long et al. using data from Florida (2012), for example, found that all students benefited from enrollment in advanced course work in mathematics and science, yet the related increases to the probability of

graduating from high school and attending college were even larger for Black, Latinx, and low-income students compared to their white and high-income peers. Similarly, an analysis of data from the Educational Longitudinal Study of 2002–2006 found that enrollment in more advanced courses in mathematics had a positive impact on college enrollment (Byun et al., 2015). A study by Evans (2019) found that those students who took an Advanced Placement (AP) test in a STEM field earned higher grades in their first-year college STEM courses than those who had not taken an AP test. In addition, the impact of AP classes may be especially important for the postsecondary outcomes of female-identified, first-generation, and racially minoritized students (Smith et al., 2018). Further, the beneficial postsecondary consequences (college attendance, performance, and attainment) of high school advanced math and science course taking in turn have far-reaching implications for students’ prospects beyond college, including college debt, employability, and future income (Evans, 2019).

Inequality in Enrollment in Advanced Science and Mathematics Courses

While the above patterns of inequality in opportunities to learn advanced STEM content are found across schools and are related to school segregation by race and social class, additional inequality is found within schools. Mickelson (2001, 2015) refers to this as “second-generation segregation,” whereby within schools that offer advanced courses, Black, Latinx, and low-income students are under-represented in these courses, while white, Asian, and high-income students are over-represented. In part, this inequality in high school courses is the culmination of many years of accumulating advantages in prior advanced course taking for white and Asian and high-income students; yet, as discussed later, prior disparities cannot fully account for such patterns, and factors such as bias in teacher recommendations and evaluations and the interventions of white and privileged parents also contribute. Additionally, as girls and boys nationwide typically attend the same schools, they have the same course offerings available to them. However, some gender differences do exist in course enrollment, as discussed below.

Inequality in Eighth Grade Algebra Enrollment

Enrollment in eighth grade algebra is a key entry point into later advanced math coursework, because in the prerequisite structure of most U.S. high schools, students must enroll in this course to have sufficient time to reach calculus by their senior year. (The typical sequence is algebra in eighth grade; geometry in ninth grade; algebra II in tenth grade; trigonometry

or precalculus in eleventh grade; and calculus in twelfth grade, Schiller & Hunt, 2011.) A study by McEachin et al. (2020) also indicated that enrollment in algebra in eighth grade increased the likelihood of enrollment in future advanced math coursework. Eighth grade enrollment in algebra increased the likelihood of ninth grade placement in geometry by approximately 30 percentage points. The probability of enrollment in tenth grade algebra II increased by about 20 percentage points, and enrollment in trigonometry or precalculus as an eleventh grader by 16 percentage points. The study’s results also showed girls, students of color, and multilingual learners benefited disproportionately by early algebra enrollment (e.g., increased test scores). Calculus, importantly, serves as the gatekeeper to college STEM majors and is generally a signal or credential associated with higher levels of enrollment in college in general (Moreno & Muller, 1999). This sequence clearly illustrates what the civil rights activist and educator Bob Moses meant when he referred to equity in the pivotal course of eighth grade algebra as an issue of civil rights (Moses et al., 1989).

Despite the initiatives of many states and districts in recent years to expand access, strong racial disparities remain. According to CRDC data, while white students comprise just under 50 percent of students attending schools that offer eighth grade algebra, they comprise 58 percent of students enrolled. By contrast, while 17 percent of students attending schools that offer eighth grade algebra are Black, only 11 percent of students enrolled in the course are Black; Latinx students are likewise under-represented, comprising 25 percent of students enrolled in schools offering eighth grade algebra, but only 17 percent of students enrolled in this key course. Indigenous students are also underrepresented in eighth grade algebra: in schools that offer algebra in eighth grade, enrollment of Indigenous students in the course is two times lower than their representation in the school overall (Office for Civil Rights, 2018a). Notably, across all races, there is a gender gap in favor of girls, as 25 percent of girls take eighth grade algebra compared to 22 percent of boys (InformED, 2018).

Further, research finds evidence of bias related to eighth grade algebra enrollment that is particularly problematic for Black youth. For example, a study using administrative records from a very large school district in the South revealed that Black youth remained under-represented in eighth grade algebra even after accounting for prior test scores and honors-level course taking, as well as family social class background (Morton & Riegle-Crumb, 2019). Additionally, a study using national data from the U.S. component of the Trends in International Mathematics and Science Study data revealed that net of teacher, student, and school characteristics, eighth grade algebra teachers in predominantly Black schools reported spending less time on algebra content (and more time on less-advanced math content) compared to other schools (Morton & Riegle-Crumb, 2019). This indicates that not only

are Black students more likely to attend schools that do not have advanced math course offerings, as described above, but they are also more likely to attend schools that provide less rigorous algebra content, placing them in a likely disadvantageous position as they progress to high school math.

Inequality in Enrollment in Advanced Math and Science Courses in High School

Relatedly, data from the CRDC reveal that while Black students comprise approximately 16 percent of U.S. students enrolled in high school, they represent only 13 percent of those students enrolled in advanced high school math and only 8 percent of those enrolled in calculus, as well as only 12 percent of students taking physics (Office for Civil Rights, 2018a). Patterns of underrepresentation in math are also clear for Latinx students, who comprise almost 25 percent of high school students nationwide, but only 19 percent of those in advanced math, and 16 percent of calculus students; Latinx enrollment in high school physics, however, is proportional to their representation as high school students overall. The pattern of representation in advanced mathematics is different for Asian students, who make up 5 percent of the total enrollment but 14 percent of students enrolled in calculus and 8 percent of students enrolled in physics. Data on Asian students, however, are not disaggregated by ethnicity, immigration history, home language, etc. and so the difference in access to STEM education for Hmong students versus Japanese students, for example, is invisible in these statistics.

Regarding emergent multilingual students (in the SEDA dataset, those classified as English learners), about 6 percent of high school students nationwide, only 2 percent of students with this designation are enrolled in calculus, and approximately 4 percent are enrolled in physics (Office for Civil Rights, 2018a). Finally, female students comprise half of those enrolled in calculus, a signal of a larger pattern of gender equity in math course taking that has been apparent for the last few decades (Dalton et al., 2007). Yet a small gender difference remains in high school physics, where male students are overrepresented by approximately 3 percentage points.

The CDRC data (Office for Civil Rights, 2018a) also point to disparities in STEM course enrollment by home language and identified disability. Students identified as English learners are underrepresented in algebra II, calculus, advanced mathematics, chemistry, and physics. Students with disabilities are underrepresented in high school courses except algebra I.

While the CRDC data present a compelling snapshot of inequality, its cross-sectional nature does not allow an investigation of pathways over time. However, researchers using data from nationally representative high school transcript studies have examined racial and ethnic differences in

math course-taking patterns for several cohorts. Specifically, a recent study by Irizarry (2021) using data from the High School Longitudinal Study (HSLS) found that Black and Latinx students who took eighth grade algebra were much less likely than their peers with racial privilege to begin high school in geometry (the subsequent advanced math course placing them in a position to reach calculus by their senior year); furthermore, these racial and ethnic disparities could not be explained by students’ prior academic performance (or other measures of student background). Additional analysis suggested that high school counselors might be deterring such students from continuing in advanced math pathways.

The importance of school counselors in shaping course pathways is underscored in several studies. For example, Cholewa et al. (2015) analyzed national data from the HSLS: 09 dataset and found that a higher proportion of Black students were significantly more likely than white students to identify their school counselor as influencing their thinking about postsecondary education, as were first-generation students in comparison to those whose parents went to college. Concerningly, an experimental audit study investigating potential bias against Black students (Francis et al., 2019) found that counselors are not as likely to recommend Black females for AP calculus, indicating that if indeed these counselors are differentially influential for Black students, their practice may contribute to maintaining inequality in access to advanced courses. A qualitative study of public school computer science education in a large urban district found that counselors had biases about who they believed could excel with technology—that matched stereotypes about computer scientists being “boy genius wonders” of either white or Asian descent—and that this influenced which students received information and/or encouragement to enroll in computing classes (Margolis et al., 2008/2017).

An earlier study by Kelly (2009) utilizing national data from the National Educational Longitudinal Study of 1988 (1988; NELS) dataset, a nationally representative sample of students who were in eighth grade in 1988 and who were then resurveyed four times through 2000, found that differences between Black and white students in enrollment in advanced math courses in high school were largest in racially desegregated schools, where the white advantage could not be entirely explained by prior academic performance and opportunities nor by family background. A more recent study following three cohorts of students in North Carolina shows similar findings (Francis & Darity, 2021). While these quantitative studies of Black-white gaps in course taking could not pinpoint the mechanism, a recent qualitative study of an integrated high school in an economically privileged school district by Lewis and Diamond (2015) points to the power and persistence of white, privileged parents who advocated for the

placement of their children in advanced courses even when prior student achievement was relatively low, and who opposed initiatives to broaden the representation of racially minoritized students in advanced courses by eliminating prerequisites and/or expanding the number of seats.

Gender, too, can be the basis for differences in enrollment in advanced science courses. Research suggests that the social construction of gender-specific norms and stereotypes related to physics contributes to a perception that physics is a heteronormative masculine field where females are not welcome (Cheryan, 2017). As with some racial patterns of inequality noted above, however, research has also found that gender disparities in physics course taking in high school vary across place. Specifically, using national data from the survey Add Health, Riegle-Crumb and Moore (2014) found that the percentage of girls in high school physics varied in relation to the percent of women in the community employed in STEM occupations. This pattern was related to a combination of student, school, and other community characteristics, and indicates how local gender norms and expectations can shape inequality in STEM fields. Relatedly, other research using data from Add Health has found that while girls are as or more likely to take advanced math in high school compared to boys, nevertheless their rates are even higher in local contexts where doing well in math appears to be less stereotyped as masculine (Frank et al., 2008). As with the racial patterns discussed above, research using national and other large-scale quantitative datasets therefore goes beyond establishing group differences in enrollment to further highlight how the representation of students with different racial and gender identities in these relatively elite math and science spaces in schools is shaped by social norms, stereotypes, and expectations that typically favor white and male students in these domains.

Data from the High School Longitudinal Study provide some relevant information about inequality of participation in AP courses in science and mathematics.³ Specifically, among high school graduates nationwide in 2013, 46 percent of Asian students and 17 percent of white students completed an AP math course; the corresponding percentage for Black students was 6 percent and for Latinx students, 12 percent. Regarding AP science courses, 40 percent of Asian students and 16 percent of white students completed an AP course, compared to 10 percent of Latinx students and 8 percent of Black students. Female students were also slightly less likely than their male peers to complete an AP science course but were equally likely to complete an AP math course.

___________________

³ For more information, see https://nces.ed.gov/programs/raceindicators/indicator_rce.asp

Participation in Engineering and Computer Science

Within the larger umbrella category of STEM, looking beyond long-established and traditional advanced courses in math and science, national data also reveal systemic inequality and enrollment in computer science and engineering courses. Data from the NAEP revealed that while 61 percent of white students and 62 percent of Asian students took a computer or engineering course in 2018, lower representations of other racial groups were observed (i.e., 56% of Black students, 49% of Latinx students, and 47% of Indigenous students).⁴ The data also reveal disparities by economic status, such that there is a 10-percentage-point enrollment difference between students who were eligible for free or reduced-price lunch and their peers who were not. Students classified as English learners were less likely than their peers to take an engineering or computer science course in eighth grade by a difference of approximately seven percentage points. Finally, while 61 percent of male students take engineering and computer science courses in eighth grade, this is compared to 53 percent of female students; given that the gender difference in eighth grade algebra enrollment favors girls, this pattern is noteworthy given its opposite direction.⁵

Regarding computer science course taking in high school, data from the College Board reveal underrepresentation of Black, Latinx, and female students in AP computer science (also referred to in the literature as AP computer science A, or AP CSA) for the last several decades. Among students in the high school graduating class of 2019, strong disparities remain, with Black students comprising only 4 percent of students in this course, Latinx students comprising 12 percent, and women comprising 24 percent (Wyatt et al., 2020).

To combat entrenched patterns of inequality, a new course called AP Computer Science Principles (CSP) was launched in 2009. As part of an initiative by the National Science Foundation (NSF), the intent of the course was to reach traditionally underrepresented students in computer science; a key distinction of the course is that it can be taken early in high school and has less of an emphasis on programming skills compared to the traditional CSA course. The College Board added CSP to its AP course offerings in 2017. On the one hand, in terms of equity in enrollment, AP CSP fares better than its more traditional counterpart. Specifically, among the high school graduating class of 2019, the representation of Black, Latinx, and Indigenous students was almost twice that observed in the CSA course, and the representation of female students was about ten percentage points higher (Wyatt et al., 2020). Another study found that Black and Latinx students,

___________________

⁴ For more information, see https://www.nationsreportcard.gov/tel/student-questionnaires/

⁵ For more information, see https://www2.ed.gov/datastory/stem/algebra/index.html#data-story-title

female students, and first-generation students who take AP CSP are more likely to declare a computer science major than their counterparts who did not (Sax et al., 2022). On the other hand, a study on the rigor of AP CSP showed that in comparison to AP CSA, the previously existing AP course in computer science, there was a lower minimum standard for programming skills, with potential implications for the acquisition of transferable skills for future study and employment in computer science (Havard & Howard, 2019). Additionally, participating in AP CSP alone did not predict long-term computing interest although a positive association appeared for women (Sax et al., 2022).

INEQUITIES IN INSTRUCTION AND OPPORTUNITIES FOR STUDENT AND TEACHER LEARNING

In this section, we review available data related to what happens in classrooms in order to develop a more nuanced understanding of opportunities for student learning in STEM education. We also look at data on opportunities for teachers to advance their own learning, which is one source of variation in student opportunities to learn. Such data are more limited in range than data on student achievement and opportunities to learn via access to and enrollment in advanced coursework. Nonetheless, they provide important information about opportunities to learn that are missing when one only considers the percentage of students enrolled in particular advanced courses or scores on achievement tests. Being able to look inside the classroom—even if it is only a snapshot—provides a deeper view of student opportunities to learn. A more granular view of what happens inside STEM classrooms in relation to equity is elevated in Chapter 5, which focuses on how equity (and inequity) are witnessed in the lived experiences of children and youth.

Since 1977, Horizon Research has periodically conducted a nationally representative study of mathematics and science education focused on detailed measures of student opportunities to learn (e.g., amount of instructional time, and exposure to engaging and rigorous curriculum), as well as teachers’ opportunities to learn through professional development (Banilower, 2019). The most recent National Survey of Science and Mathematics Education (NSSME) concluded in 2018, and it revealed inequities of opportunity related to socioeconomic status, community type (urban, rural, suburban), race/ethnicity, and prior achievement of students’ classrooms in both science and mathematics education. Here, we summarize patterns by socioeconomic status, community type, and race/ethnicity from this report.

Opportunities for Students

In science, there were several significant differences in opportunities to learn based on socioeconomic status, community, and race/ethnicity (Trygstad et al., 2020). While schools with high percentages of students from low-income families spent about as much time on science in the elementary grades as other schools, fewer of these schools offered advanced courses in science. The latter trend is consistent with the trends in course offerings discussed in the previous section. Instruction in high-poverty schools focused more on test-taking skills and involved more reading of textbooks, with fewer laboratory activities.

Interestingly, there were no differences by community type with respect to student engagement in the science and engineering practices. Urban schools spent more instructional minutes in science at the elementary grades. Laboratory activities were more common among suburban schools than schools in cities and rural areas. As with socioeconomic status, community type made no difference with respect to the frequency with which students engaged in the science and engineering practices. Spending on laboratory materials and use of commercially published kits and modules was higher in rural than urban areas. Box 4-3 provides further insight into the ways that inequity can play out in rural contexts.

With respect to schools with high percentages of students from a historically underrepresented race/ethnicity group (defined in the NSSME report as American Indian or Alaskan Native, Black or African American, Hispanic or Latino, Native Hawaiian or Other Pacific Islander, and multiracial students), those schools spent more time in elementary science, but offered upper-grade students fewer advanced science classes. Again, this finding is consistent with the trends on course offerings discussed in the previous section. In addition, instruction was more traditional, with teachers placing more emphasis on learning facts and vocabulary, test-taking strategies, and reading from a textbook. At the same time, teachers in schools with high percentages of racially minoritized students were more likely than other teachers to emphasize the science and engineering practices in their teaching. They had limited supplies, though, and were more likely to use locally developed lessons from the state and/or district.

In mathematics, the pattern of results was largely similar to the results in science (Malzahn et al., 2020). When comparing high-poverty schools with schools with lower percentages of students from low-income families, there were no differences in elementary instructional minutes for mathematics, a similar pattern for advanced course taking was evident. There were more opportunities in schools with fewer low-income students to take Algebra 1 prior to ninth grade, and there were more advanced classes available to them in high school. Instruction was more traditional

in high-poverty schools, with students reading more from a textbook than in other schools, spending more time learning vocabulary and test-taking strategies, and learning to perform computations with speed and accuracy (detailed vignettes provided in Chapter 5). There were similar rates of engagement with the Standards of Mathematical Practice of the Common Core. Students in high-poverty schools were more likely to be using units or lessons developed by their state, county, district, or diocese, and generally had less adequate resources as well. With respect to community type, key differences were that students in rural schools were less likely to take algebra or geometry before ninth grade, and urban schools were more likely to use materials developed by their state, county, district, or diocese.

With respect to schools with high percentages of racially and ethnically minoritized students, some key differences emerged that paralleled differences related to poverty. Interestingly, there were comparable minutes of instruction devoted to mathematics at the elementary level, but a similar pattern of fewer advanced courses being available to students at high-poverty schools. Likewise, there was evidence of more traditional instruction that focused on vocabulary, test taking, and reading from a book in schools with high concentrations of students from underrepresented groups. In those schools, students were more likely to use state, county, district, or diocese-developed units or lessons and have less adequate resources for instruction.

Opportunities for Teachers

Data from these surveys also pertain to teachers and showed significant differences among them as well. In both mathematics (Malzahn et al., 2020) and science (Trygstad et al., 2020), teachers in high-poverty schools and in schools serving high percentages of racially and ethnically minoritized students were less likely to have autonomy in their teaching decisions. Teachers were more likely to be underprepared (by their own self-report) and less experienced in high-poverty schools, urban schools, and schools serving high percentages of racially and ethnically minoritized students, but also more likely to be from a group that is underrepresented in science. Teachers in those groups were also more likely to receive professional development about how to incorporate students’ cultural backgrounds into instruction. For urban and suburban teachers, the total amount of professional development was similar, while rural teachers got less than their urban and suburban counterparts (Banilower et al., 2018).

It is important to note that these data come from self-report surveys collected in 2018, prior to the COVID-19 pandemic and the nation’s ongoing racial reckoning. These events have shaped many institutions of society, including schools. With respect to students’ opportunities to learn, recent

events have aggravated an already inequitable distribution of teachers (National Academies of Sciences, Engineering, and Medicine, 2022) and the resources those teachers need to thrive.

Equity Implications of Trends in the Teacher Workforce

There is compelling large-scale evidence that students do not have equitable access to professionally prepared teachers (NASEM, 2020). Students from lower-income families, students of color, and students who have lower achievement levels are more likely to have teachers who are less qualified than their more advantaged peers (NASEM, 2020). For example, in Massachusetts, Cowen et al. (2017) found that lower-income mathematics students were twice as likely to be assigned a teacher who was assigned a “needs improvement” or “unsatisfactory” rating and three times as likely to be assigned an inexperienced teacher than their more advantaged peers. Inequitable access to teachers occurs at all levels of the educational system—within schools, across schools in the same district, across districts, and across states (NASEM, 2020).

In addition, prior to 2020, there was a long history of significant and inequitably distributed STEM teacher shortages, particularly in schools with higher proportions of economically disadvantaged students (Cowan et al., 2016; Goldhaber et al., 2015). Recent evidence from a nationally representative sample found that vacancies are widespread in STEM. Of the K–12 schools that had teaching vacancies in 2020–2021, many schools found it difficult or were not able to fill the vacancies for physical sciences (37%), mathematics (32%), life sciences or biology (31%), computer science (31%), and for career or technical education (31%) (Taie & Lewis, 2022). These shortages are best understood at the local level because teacher labor markets tend to be local (NASEM, 2020). There is debate about the exact shape of the shortage; however, emerging evidence suggests the shortages have gotten worse since 2020 (Schmitt & deCourcy, 2022).

In addition to this national context of longstanding STEM teacher shortages and inequitable distribution of teachers, many school environments do not support teachers’ general wellbeing. While healthy school communities are a resource to teachers’ learning, thriving, and ability to connect with and support students’ growth, recent data suggest that at least two of the basic components of a thriving school community are often a challenge across institutions, and directly impact teacher wellbeing. These challenging components are (a) an adequate professionally prepared workforce as noted above (Schmitt & deCourcy, 2022; Taie & Lewis, 2022) and (b) a working environment that supports teachers’ mental health (Diliberti & Schwartz, 2022; Kush et al., 2022).

To the latter point, despite important concerted efforts to improve conditions, teachers continue to report ongoing mental health challenges. A large recent survey documented that teachers have been experiencing significant mental health challenges that are larger than office workers and health care workers (Kush et al., 2022). Specifically, this survey revealed, teachers reported experiencing anxiety symptoms more frequently than other workers, while within this group, teachers who teach remotely reported higher levels of distress than those working in person. In a second nationally representative survey that was administered in January 2021, researchers found that a greater proportion of teachers were experiencing job-related stress (78%) with higher frequency than other employed adults in the nation (40%; Diliberti & Schwartz, 2022). Teachers also experienced more depressive symptoms (28%) than the general population (18%). School administrators have noticed teachers’ experiences and are concerned. In the spring of 2022, in a nationally representative sample, 92 percent of administrators said that teachers’ mental health was a moderate or major concern (Diliberti & Schwartz, 2022, p. 5, Figure 8). Districts are taking actions—hiring more staff and cultivating more positive working conditions in schools—but challenges remain (Diliberti & Schwartz, 2022).

The structural challenges that face teachers, schools, and communities are significant, and they underscore the need to re-envision equitable STEM teaching and learning. Only by re-envisioning what STEM teaching and learning is, can schools hope to attract and retain an adequate, skilled diverse STEM workforce. We take up this new STEM vision in Chapters 7 and 8.

INEQUITES OF OPPORTUNITY IN OUT-OF-SCHOOL LEARNING

Scholarship points to the importance of out-of-school experiences for science aspirations (DeWitt et al., 2011; Tai & Maltese, 2010). Out-of-school learning can be a place to develop interest and identification with STEM (National Research Council, 2009), factors associated with later pursuit of STEM (Zhang et al., 2021). Unfortunately, there are familiar disparities in who participates in out-of-school STEM learning. For example, analyses of data from Black, Latinx, and white students in the NELS (1988), a nationally representative sample of students who were in eighth grade in 1988 and who were resurveyed four times through 2000, found that socioeconomically disadvantaged youth were less likely to participate in a wide variety of activities than wealthier peers. This 1988 study also showed that patterns of participation differed by race, with Black and Latinx youth participating less often than white students in parent-driven activities, such as scouting, but not school-based extracurricular activities (Bouffard et al.,

2006). A more recent analysis using the same dataset focused on participation in mathematics- and science-related out-of-school activities (Chan et al., 2020). That study found that Black, Latinx, and Asian students were less likely to participate in science-related out-of-school activities than white students in eighth grade. In addition, Black students were less likely to participate in math-related out-of-school-time programs in eighth grade compared with white students.

In addition, studies have documented how much more higher-income families spend on their children’s out-of-school enrichment activities than do lower-income families, and how that gap has grown since the 1970s (Kaushal et al., 2011). Neighborhood affluence is also a factor, independent of individual family income (Dearing et al., 2009). Schools with high percentages of low-income students are less likely to have out-of-school offerings related to engineering clubs and mathematics competitions, according to national survey data (Malzahn et al., 2020; Trygstad et al., 2020). Some scholars have proposed that material wealth inequity cannot explain these differences, but, rather, that there is evidence that a practice of “concerted cultivation” from more educated mothers, which explains out-of-school participation differences linked to income; that is, participation in out-of-school programs is part of an intentional parental strategy to ensure the competitiveness of their children for further opportunities in education through the development of skills that can be acquired through extracurricular activities (Weininger et al., 2015).

Studies have found that one source of inequities in opportunity in the out-of-school space, gender segregation, can negatively influence female-identified students’ future plans in STEM. Legewie and DiPrete (2014), for example, found that more gender segregation in out-of-school activities in high school predicts lower probability of intent to pursue a STEM major (Legewie & DiPrete, 2014). They argued that one likely explanation for this effect was that gender segregation produces more reinforcement of stereotypes about what boys and girls can and cannot do. This is important to note, as girls’ lower participation rates in out-of-school STEM activities like robotics clubs have likely negative implications not only for their own development of STEM identity and interest, but for the reification of gender stereotypes held by others (e.g., boys in such spaces; Meschede et al., 2022; Nguyen et al., 2016, 2022).

More recently, scholars have applied a spatial lens to mapping inequities of opportunities in communities. For example, Pinkard et al. (2016) analyzed a large database of out-of-school programs gathered as part of the Chicago City of Learning initiative (an effort to make visible to the community available opportunities for out-of-school time) to explore how program offerings varied by neighborhood. Overall, they found that there were many more opportunities concentrated in some parts of the city—particularly

downtown—than others, but that types of programming available varied widely from ZIP code to ZIP code. The equity implications of the geographical distribution of opportunities are discussed in Chapter 11. The team’s analyses also showed the variety of programs varied from neighborhood to neighborhood (Quigley et al., 2016). Such geographic differences have implications for equity, in that they highlight another dimension of accessibility to be considered—namely how close or far away an opportunity is from where a student lives.

TOWARD NEW SYSTEMS FOR ASSESSING THE STATE OF STEM EDUCATION

Much of what can be said about the state of STEM education comes from either large-scale achievement tests or databases about what courses students are taking. Such data provide a partial picture of how variable the United States is by region. Treating snapshots of performance or even longitudinal analyses of course taking as adequate for understanding inequity and its causes is likely to limit conversations about how to promote equity. In this section, the committee highlights the profound need for both a broader imagination about the kinds of information needed to understand and track progress toward equity, as well as a purposeful orientation toward student experience, the role of geography (place and lands), and systems in shaping patterns of inequity. Below, we outline some possible future directions that are grounded in evidence of what is possible to learn from adopting such an orientation. We view these as complementary to, rather than replacements for, currently available datasets regarding achievement and opportunity, which can have a place alongside these other kinds of data, but without occupying the privileged position as the only ways to speak of the state of STEM education in our society.

Data on Inequities in Resources in Schools and Communities

Consistent with an earlier National Academies report (2019) on monitoring educational equity, we see the need for more data on inequities in resources that matter for student experience of STEM classrooms (see Chapter 5 for discussion about inequities in funding and resources). Monitoring Educational Equity called for continued monitoring of inequities in access and achievement, as highlighted in this chapter, as well as for the development of indicators that “measure equitable access to resources and opportunities, including the structural aspects of school systems that may affect opportunity and exacerbate existing disparities in family and community contexts and contribute to unequal outcomes for students” (p. 3). In this category of indicators, the committee recommended monitoring

disparities in exposure to economic, racial, and ethnic segregation; level of access to early childhood learning opportunities; as well as more detailed indicators related to the quality of teaching and available school supports—including school climate and access to non-exclusionary disciplinary practices—than are currently monitored across the United States. Other data may be needed to capture availability of resources at the state and district levels, but even today, these data remains hard to find. These include availability of funding for professional learning and instructional materials that connect to the everyday lives of students, as well as the availability of supports for improving teaching, such as coaching.

More detailed information about available resources is necessary, because resource distribution practices and outcomes influence the kinds of disparities in educational opportunity and outcomes that have been identified across a range of studies as reviewed in this chapter thus far. Developing such data on a national scale could provide information at the macro and meso levels that is necessary to identify leverage points for promoting more equitable opportunities and outcomes in STEM.

Data on and Models for Analyzing and Using Student Experience

One of the hallmarks of dignity-conferring educational experiences is that they afford meaningful participation to students in learning activities (Espinoza et al., 2020). At the national and state levels, there is minimal collection of systematic data on the quality of students’ learning experiences in school. There is strong evidence from large-scale studies that student experience matters for learning outcomes (Kane & Staiger, 2012). Student experience data provide information on school quality not captured in accountability tests, and are one modest means of giving students voice in evaluating their experiences (Schneider et al., 2021). Gathering data and disaggregating it by race, gender, and other variables can help with gaining a better sense of how groups of students who are owed an education debt by society find their learning experiences in class (Martinez & Guzman, 2013).

Additionally, it is vital to expand the use of survey tools to capture students’ views of their STEM classrooms. Instruments such as the Tripod survey, developed by Ronald Ferguson and used in the Measures of Effective Teaching study, can capture students’ views regarding how much their teachers provide productive academic challenges, respect their ideas and encourage their expression, and display caring and compassion (Ferguson & Danielson, 2014). While such measures are often used to assess overall “teacher quality,” with the aim of increasing scores on standardized achievement tests (e.g., Wallace et al., 2016), one could learn more about

whether and how teachers and schools cultivate equitable STEM learning environments by examining, for example, whether girls, boys, and gender non-binary students in the same classroom report different experiences with their teachers, or whether reports of having STEM teachers that are both caring and academically challenging were more common among students in predominantly white schools compared to schools with more minoritized students.

Further, recently developed observation tools such as classroom videos hold potential to measure students’ experience of equitable STEM learning environments and could potentially be used at a large scale. The EQUIP tool (Equity QUantified in Participation) was designed to answer questions such as, “Who gets access to participation in rigorous mathematical discourse?” by disaggregating across groups or subgroups (e.g., Latinx girls), and purposely captures not only who participates in the classroom, but how they participate, and how teachers’ actions encourage or discourage equitable participation (Reinholz & Shah, 2018). For example, a recent study using EQUIP to collect data from 100 math classrooms across one large racially diverse school district found that while boys across racial groups participated more in math discussions than girls, Latinx students of all genders had the lowest overall rates of active participation (Reinholz & Wilhelm, 2022). Such patterns have likely implications not only for who has more opportunities to develop mathematical understanding, but also for who is publicly positioned in the classroom as holding mathematical expertise.

Another example of a data collection effort that could serve as a model for gathering information about student experiences and opportunities is the National Indian Education Study (NIES). Since 2005, the NIES, conducted as part of the National Assessment of Educational Progress, has provided information about the educational experiences and the academic performance of fourth- and eighth-grade Indigenous students in the United States. In the most recent survey in 2019 (National Center for Education Statistics, 2021), about 7,000 fourth graders and 6,300 eighth graders completed the NIES survey, which oversamples from schools and 15 states with high percentages of Indigenous students. The survey reports on such topics as students’ opportunities to learn about their culture and heritage language, use their own language in learning, and attend cultural events. Teachers and administrators also respond to surveys about the ways they integrate Indigenous culture, language, and history into their instruction, including in mathematics. Findings from the survey also break down responses by type of school as well—those with high concentrations (“high density”) of Indigenous students, those with low concentrations (“low density”), and Bureau of Indian Education schools. These reports allow for a portrait of Indigenous students’ opportunities to experience instruction

in which their cultural practices are represented within the United States and, as such, provide one model for eliciting information about the degree to which culturally relevant and sustaining pedagogies with respect to Indigenous students are being enacted in schools. Similar efforts can provide useful insights for other marginalized groups in STEM.

One district-level approach to gathering and using experience data in STEM has been to embed measures into curriculum materials in science, thus providing formative information teachers can use to adjust their instruction (Penuel & Watkins, 2019). As part of a long-term research-practice partnership (cf., Farrell et al., 2021), the district and researchers have collaborated to design brief “exit ticket” surveys focused on students’ affective experiences of the lesson, including whether or not they feel as though their ideas were taken seriously by others. Educators collect these exit tickets periodically during units and then can use a data system to visualize data disaggregated by race and gender to explore patterns in their classroom, supported by professional development to help them make sense of data (Raza et al., 2021, 2022). This effort illustrates how student experience data could yield usable data on the state of STEM education locally to inform ongoing equity efforts at the district level that are also useful to educators in their own classrooms.

A study by Munter and Haines (2019) used student experience data to explore students’ perceptions of opportunity to learn within mathematics classrooms. Students of color in the sample of secondary classrooms reported more bias toward giving white students more opportunities to learn. This was especially true in classrooms where teachers provided demanding academic tasks but lowered the cognitive demand over the course of lessons as they taught. This study illustrates the power of connecting student experience data to instructional practice, with potential implications for teacher preparation.

Relatedly, national survey instruments typically query students about their broad beliefs and attitudes toward math and science—for example, asking them how much they enjoy math. But such general questions obscure more nuanced beliefs and experiences, as students may feel very differently regarding their experiences, for example, in geometry than in algebra; and perhaps girls’ confidence in physics in high school show divergence from their confidence in biology, given what is known about patterns of participation in college majors. By utilizing more refined survey questions that ask students to share their thoughts about specific topics and classes, researchers could learn much more about where and when students are experiencing more positive and inclusive learning environments (and where and when they are not). Further, designing data collection systems where survey instruments are individualized to student schedules could also allow them to give feedback on what they find most engaging and relevant in the

specific STEM classes in which they are enrolled, including their participation in computer science and engineering classes, domains that are often not included in student surveys because a relatively small percentage of students take them.

Data and Analyses Toward Mapping Geographies of Inequality and Possible Futures

What national datasets underscore time and again is the importance of local variability—in schools and districts—in shaping patterns of inequality and inequity. Yet, there is not a strong body of research on mapping geographies of inequality and inequity, despite calls for a spatial approach to understanding the role of spatial injustice in education going back more than a decade (e.g., Tate, 2008; Tate & Hogrebe, 2011). Beyond understanding where education takes place, mixed-methods studies are needed to understand better how space and place matter for educational inequality and inequity. Today, there are models for analyzing geographic data that can support large-scale data analyses (e.g., Fotheringham, 2009), as well as visualization tools like Tableau that make interpreting geographic data accessible for decision makers.

Further, national datasets are generally designed to be “nationally representative” of student populations, such that different groups are represented in the dataset in proportion to their representation in the school-age population in the United States, and different school types and regions of the country are similarly represented. While certainly useful for providing a broad national picture, this breadth in design presents obstacles for more in-depth understanding of the design and experiences of (in)equality and (in)equity in STEM education, including among groups of students that are smaller in number (e.g., Indigenous students). In addition to current, prevalent approaches, new large-scale data collections could be designed with known variability across schools taken into account; for example, a design done intentionally to include many schools and districts that have been identified as being more equitable on some chosen dimension were included, as well as others that are identified as the most inequitable. This could provide an extremely rich opportunity for researchers to investigate the complex constellation of factors that converge to create more equitable or inequitable STEM learning experiences.

Relatedly, with the exception of the Add Health data (which started in 1994–1995, with students in seventh through twelfth grades, and the last wave of students in 2016), national datasets utilize small random samples of students within schools, making it difficult for researchers to study and understand school contextual factors and classroom factors in a deep way. New data collections that surveyed all students—and teachers—within a

school could be added to the repertoire of tools to allow us to better see into classrooms and schools. Measures of classroom peer beliefs and behaviors could be constructed for students’ science and math classrooms, capturing dimensions of local spaces that have been shown, for example, to shape girls’ intentions to major in STEM fields (Riegle-Crumb & Morton, 2017).

Space has been shown to be an important factor in equity and education. There are multiple scales at which space matters for educational equity, from politically defined boundaries (e.g., cities; Bullock, 2017) to those defined by natural features (e.g., watersheds; Chanse et al., 2017). The organization of space within and around school buildings matters, too, in that it shapes and is shaped by interactions that take place in different parts of the school (McGregor, 2004) and affects the wellbeing of members of the school community (Bruno, 2000). How space is organized within schools can reinforce marginalization of groups, such as students with disabilities (Holt, 2004) within school communities. At the same time, groups such as LGBTQIA+ students can reclaim spaces as safe spaces for themselves within schools (Schmidt, 2015). Ma (2016) has explored how embodied spatial relations can facilitate students in constructing geometry figures, as they move in outdoor spaces.

A spatial perspective on equity can also inform large-scale efforts to transform systems as well. Green (2015), for example, has argued for the need to map “opportunities of geography”—assets that can support the learning and wellbeing of students within a community, to counter deficit perspectives on marginalized communities. Ecosystems initiatives, for example, may benefit from the kinds of visualizations created by Pinkard et al. (2016; see above) for characterizing spatial variation in access to STEM programming (see Chapter 11 for an in-depth discussion of an ecosystems perspective on education and learning environments). Other scholars’ work points to the role of participatory involvement of youth in planning at the city level, both for building students’ STEM skills and in helping re-imagine cities (Taylor & Hall, 2013). Such work focuses on the role of placemaking in working toward more just futures, rather than the more limited project of using mapping to mark current inequalities and inequities (Derr et al., 2018). Place here refers to the social, cultural, and material aspects of spaces as experienced by people who inhabit them (Agnew, 2011; Baroutsis et al., 2017; Massey, 1994; Tuan, 2001).

Understanding the intersection of geographies with historical policies can also be instructive. For example, more than 90 percent of school-age Native youth go to school in urban, suburban, and rural places—not reservation-based schools (National Indian Education Association, 2024)—due in part to federal termination and relocation policies of the 1950s and 1960s. This has meant that Native youth systematically are in schools where they are the only or one of very few Native students.

When considering spatial inequality and inequity, it is important not just to focus on place and placemaking, but also to focus on lands and waters, with an appreciation for the importance of Indigenous sovereignty over education within lands within the United States and its territories. Sovereignty over lands and education within Indigenous lands is protected by law, even if it is not always honored. For many Indigenous nations, land is not just a place: learning about and from lands and waters is a core component in many Indigenous models of education (Brayboy & Bang, 2014).

Data on Policies and Practices at the Systems Level

Missing from this chapter’s description of inequity in opportunities to learn STEM are the activities and resources of actors at the district and state levels. Data on their actions is critical for gaining an understanding of key policies and practices that shape student outcomes, whether it is achievement, or patterns of course taking. Yet only a few school systems self-reflexively gather data on how well they function. One exception is Chicago Public Schools, which every two years collect data from parents, students, and educators about their experiences and, critically, about the organizational conditions of schools. This practice is guided by a theory—with a strong evidence base—for how to improve schools (Bryk et al., 2010). District leaders elsewhere have used that framework to guide improvement efforts, even without the same underlying structure of measures of organizational conditions (Penuel et al., 2018).

Additionally, actors might be better able to address equity concerns in their contexts if more data existed on strategies of district and community organization leaders to promote equity within their systems. While this is a growing area of research in education more broadly (e.g., Horsford et al., 2018; Rorrer et al., 2008; Schuerich & Skyrla, 2003; Skrla et al., 2000), little research in recent years has focused on STEM education. In the 1990s, evaluation studies of NSF-funded Local Systemic Initiatives explored district-level strategies and actions to promote change (e.g., Banilower et al., 2006; Weiss & Pasley, 2006), but equity was not a central focus within these studies, and most analyses focused only on professional development as a policy lever for building capacity. There are other levers—like policies pertaining to resource allocations and policies relating to teacher evaluation and assessment, as well as integration—that are likely relevant for understanding district change (Finnigan et al., 2016; Penuel, 2019). Studies of the implementation of district-level policies and practices could be important for understanding how, if at all, they can promote equity (NASEM, 2022).

Gathering data on the actions of state leaders to promote equity is another area of potential value to understanding the state of STEM. As part

of the Advancing Coherent and Equitable Systems of Science Education Project, an NSF-funded partnership between the Council of State Science Supervisors and researchers at the University of Washington and University of Colorado, researchers have partnered with state leaders to gather data on the equity initiatives being pursued in each state via surveys and interviews. Through case studies that integrate these data sources, the team has identified conditions that are supportive of making progress in enacting different equity-oriented initiatives (Wingert et al., 2020). Survey data have identified where leaders at different levels within each state—within districts, among professional development providers, and community organizations, for example—are engaged in different “equity projects” such as promoting racial justice and disrupting ableism in science education (Bell et al., 2022). In addition, survey analyses identified systems-level policies and processes that leaders judged to be aligned to equity goals for science education, as well as those working in opposition to such goals (Rhinehart et al., 2022). Still other studies have examined how state leaders in this collaboration use and share research, such as National Academies reports, to inform their standards implementation support work (Hopkins et al., 2019). These kinds of data are rarely collected about state-level system actors, but they provide information on a layer of systemic change that is especially important in a post-NCLB world, where state education agencies play a significant role in implementing federal policies (Smarick & Squire, 2014).

CONCLUSIONS

In this chapter, we examine the data currently available to make sense of what is known about how students fare in STEM-related courses as well as the availability of these courses throughout their academic careers. Throughout, we capture what national and other large-scale data systems reveal about gaps between student groups on what are often considered as external markers of STEM success, most notably achievement test scores. This is also a discussion of whether the current system has the capacity to pinpoint if current programs or methods for teaching STEM education are exacerbating or reducing inequities for students. It also provides a snapshot of disparities in access and opportunity in STEM learning.

Finally, the chapter provides a critique of a national emphasis on using achievement data to assess students’ experiences in STEM education: even though the achievement gap dominates the evidence on equity in STEM education, other domains are pertinent to equity (e.g., resource distribution, opportunity to learn, quality of experience). Attending to these domains aids in developing a more expansive understanding of equity and implementing more comprehensive approaches to achieve greater equity.

Conclusion 4-1: Results from national and state-level assessments of

performance in STEM subjects document persistent achievement gaps across groups, despite accountability-based reform efforts that try to address gaps in achievement.

Conclusion 4-2: There is systemic inequity in opportunity to learn science, technology, engineering, and mathematics in both preK–12 and informal education settings that benefits predominantly white, male individuals and well-resourced communities.

Conclusion 4-3: The current system for documenting the state of STEM education focuses primarily on student achievement on standardized test scores only. Current data are inadequate for documenting how policies and practices contribute to inequities and do not provide sufficient information to guide systemic changes that can address gaps in opportunity, access, and quality of experience.

Conclusion 4-4: Access to science, technology, engineering, and mathematics disciplines is uneven across preK–12 education. While mathematics is available in some form throughout preK–12, tracking policies across the country limit access to advanced mathematics coursework. On average, students in elementary grades rarely touch on science, computer science, or engineering coursework, and advanced coursework in secondary grades is unevenly distributed.

Conclusion 4-5: There are overarching patterns in achievement that show a strong association between school-level racial-economic segregation and achievement. However, within this overall pattern, there are regional and district-level variations in the size of the gaps. This variation in gap size is partially explained by differences in economic and demographic factors, and patterns of school and neighborhood segregation.

REFERENCES