## Main

Professors play a unique role in the knowledge economy: they both train the next generation of thinkers and generate new scholarship, which informs national policy and advances scientific discoveries. But the professoriate has never represented the sociodemographic characteristics of the population it serves. While the diversity of the educational pipeline has been extensively studied in terms of race and ethnicity1,2,3, and the links between parental income and occupational status, and their children’s educational attainment are well documented4,5,6,7, there exist comparatively few systematic studies on the socioeconomic roots of professors or how their socioeconomic origins interact with institutional prestige. Analyses of the socioeconomic backgrounds of faculty will both improve our understanding of the social reproduction of the highest levels of academic attainment and scientific influence, and provide a quantitative basis for studies of how representational diversity influences which and what kind of discoveries are made.

From early childhood through advanced degree attainment, family socioeconomic origins shape educational outcomes8. Parental socioeconomic status (SES) impacts children’s educational attainment in numerous ways. Parents with higher SES, who tend to be college-educated and occupy high-status occupations, cultivate advantaged cultural and social capital for their children9. Parents with high SES also engage in ‘concerted cultivation’, whereby parents prioritize organized activities that facilitate educational success10, and spend an increasingly large amount of time on developmental activities that advantage their children’s attainment7. Families with parents with high SES also have more information about schooling and resourceful social networks than working-class families. Research also suggests that middle- and upper-class students request accommodations and attention in school that ultimately advantage them in their attainment11. These networks function through power and privilege across racial, ethnic and social class lines, forming unequal opportunities for academic success. SES and race also impact where families live and the schools that children attend. The dispositions and skills of students with high SES are judged differently by teachers and other school personnel who favour the cultural capital of middle- and upper-class youth12. Ultimately, parents, teachers and students behave in ways that reproduce class and racial inequality in academic achievement9. Schools in more and less advantaged neighbourhoods differ markedly in resources available to students13,14. These factors all contribute to the substantial effects of family SES on postsecondary schooling.

Research has shown notable socioeconomic differences in not only whether individuals attend and complete college, but also where they attend, with more advantaged students attending more selective institutions. Students completing degrees at highly selective institutions are more likely to come from the top 1% of the US income distribution than from the bottom 50%15. Students from more disadvantaged socioeconomic backgrounds who attend college, particularly those who attend selective colleges, also have different experiences on campus that are less conducive to academic success than their higher socioeconomic peers16,17.

Family SES also influences graduate school applications and admissions18, as well as students’ experiences once accepted19,20. In fact, graduate and professional degrees are substantially more heritable than other levels of education. Individuals with parents who have a doctorate or professional degree are increasingly overrepresented among doctorate and professional degree holders20. Moreover, research on social mobility suggests that the association between parents’ SES and their children’s status is larger among postgraduate than bachelor’s degree recipients19,20. Occupational closure plays an important role in the intergenerational transmission of status21, and an association between children and their parents’ occupations (sometimes called micro-class reproduction) has persisted for decades22. Unequal social, cultural and financial capital help reproduce micro-class advantages across generations, and the professoriate may be an occupation with a high level of closure and reproduction.

The accumulation and consequences of these tendencies can be expected to limit the socioeconomic diversity of the professoriate. However, the extent and shape of this association, and its precise interactions with gender, race, ethnicity, academic discipline and prestige are less well understood. Past work shows that faculty are increasingly representative of highly educated families23, while professors from disadvantaged socioeconomic backgrounds report hyperawareness of how those backgrounds impact their careers24, which can make them feel like outsiders in academia25. SES may also interact with academic job placement. Vague narrow definitions of merit or ‘fit’ disadvantage the work of marginalized scholars, particularly racial minorities26 or those from working-class or poor backgrounds27, due to differences in social and cultural capital between these applicants and faculty search committees21. Structural factors that influence whether and where a person earns a faculty job, as well as difficulties once employed, will influence the composition of the professoriate.

Beyond composition alone, representational diversity is known to improve academic scholarship28, and a lack of socioeconomic diversity is likely to deeply shape the type of scholarship that faculty produce and the scholars they train. Recent studies have begun quantifying how researcher identity affects research topic choices29, as well as the evaluation of that work30,31. However, it remains generally unclear what scholarship is missing as a result of limited representational diversity across the professoriate, in part because we lack a clear and quantitative view of current diversity and its roots.

Here, we develop such a view by investigating the indicators of childhood SES of US-based tenure-track professors across eight disciplines spanning science, technology, engineering and mathematics (STEM), the social sciences and the humanities. Tenure-track faculty positions are less financially precarious than contingent positions, rendering the socioeconomic representation of the faculty who obtain these privileged positions notable for social reproduction. We consider how SES is systematically related to faculty placement within academia, how the role of SES varies across broad disciplines, and how parental advanced degrees relate to the likelihood that their children become and remain as faculty. We conclude by suggesting how socioeconomic diversity and its historical trends may limit diversity in the professoriate.

## Results

To study the childhood SES of US faculty, we conducted a large survey of tenure-track faculty at Ph.D.-granting departments in the United States from eight academic disciplines. Responses include information on the education levels of parents and the zip code where faculty grew up. We augmented our survey data with national estimates of educational attainment, income and rural or urban classification by zip code from the US Census, National Science Foundation (NSF) Survey of Earned Doctorates (SED), Internal Review Service (IRS), US News and World Report (USNWR) and the National Research Council (NRC), which allowed us to contrast survey characteristics with patterns in the general US population.

We conducted the online survey between summer 2017 and autumn 2020 on a frame of 46,692 current tenure-track faculty across 1,360 Ph.D.-granting departments in Computer Science, Business, History, Psychology, Physics and Astronomy, Sociology, Anthropology and Biology. These eight academic disciplines were chosen for their diversity of scholarship and represent a broad sample of tenure-track faculty at research intensive institutions in the United States. The sample frame was constructed from the online public directories of institutions, allowing us to explicitly compare respondents with the frame.

In total, 7,204 faculty provided information on a parent’s level of highest education (15.4% of survey frame) and 4,807 provided the US zip code in which they grew up (10.3%).

### Parents’ education

Faculty tend to come from highly educated families. Nearly a quarter (22.2%) report at least one of their parents holds a Ph.D., and 3.7% of faculty report both parents hold Ph.D.s. Across all eight disciplines, over half (51.8%) of faculty have at least one parent with a masters degree or Ph.D. (29.6% and 22.2%, respectively; Table 1). In comparison, among adults in the United States aligned to when faculty were born, on average, less than 1% held a Ph.D., and just 7.4% held a graduate degree of any kind.

We calculated the relative likelihood that a faculty member has a parent with a Ph.D., compared with either the US adult population or parents of Ph.D. recipients, by estimating upper and lower bounds. These bounds correspond to differences in whether these datasets described individuals (US Census and NSF SED) versus households. Across all eight fields, we estimate that faculty are on average between 12 and 25 times more likely to have a parent with a Ph.D. than the general population, and about twice as likely as other individuals who hold a Ph.D. (Fig. 1 and Supplementary Note C).

The distributions of parents’ educational attainment are similar across the disciplines surveyed, suggesting that despite disciplinary differences in scholarship, funding and culture, having a parent with a Ph.D. is universally advantageous for becoming a professor. The rates at which parents of faculty have a college degree or higher have also slightly increased over time, which mirrors broader social trends in the US population (Fig. 2). Women are particularly more likely to have highly educated mothers: 33.3% of women versus 28.3% of men have a mother who holds a graduate degree (two-tailed test, z = 4.5, n = 7,107, 95% CI = 0.028 to 0.073, P < 0.001). Historically, rates of parents’ college completion among women faculty are higher than among men faculty; however, this gap has steadily closed over time fully reaching parity for faculty born in 1985 (Fig. 2).

We find that faculty with Ph.D. parents are far more likely to receive support and encouragement for their academic careers from their parents (on a scale of 1 to 5: 4.5 versus 3.9 for less than Ph.D., two-tailed test, t = 17.2, n = 5,725, 95% CI = 0.497 to 0.625, P < 0.001), which is consistent with prior work32. This question came from an optional later section of the survey, but responses were representative along faculty rank and departmental prestige of the frame (Supplementary Note A). In fact, the career support that faculty report receiving from their parents increases with greater parental education (Fig. 3) and does not depend on faculty gender (average rating of 4.0, two-tailed test, t = − 0.2, n = 5,695, 95% CI = −0.072 to 0.057, P = 0.827). This strong correlation between parental education and faculty career support suggests that the family resources and experiences of faculty with highly educated parents differ from those without highly educated parents and differ in ways that correlate with improving the odds that a person becomes a professor. For example, the degree of family support is known to correlate with both undergraduate and graduate student retention19,33. Given what is known about educational stratification, we expect that there are other sociocultural mechanisms through which being the child of highly educated parents increases the chances of becoming faculty, such as greater identification with academic ideals, more and earlier experiences in activities valued by academia, or simply closer role models34.

### (Proxied) parents’ income

Faculty also tend to spend their childhoods in wealthier zip codes than do the general public (Fig. 4). The median proxied household income based on zip code data for surveyed faculty when they were children is 23.6% higher than the median across all zip codes (US $73,000 versus US$59,000, Mann–Whitney U, ρ = 0.4, n = 1.2 × 108, P < 0.001). Consistent with the importance of parental education on faculty careers, proxied parental income is correlated with parental education: faculty who reported that at least one of their parents holds a college degree were associated with higher average proxied household incomes (US $78,000) than those who said their parents did not hold a college degree (US$59,000; ρ = 0.3, n = 3,916, P < 0.001). Across disciplines, median proxied parental income remains relatively high, ranging from US $67,000 (Sociology) to US$78,000 (History). Faculty are more likely to have grown up in urban areas compared with the geographic distribution of the US population around the average year faculty were born (89.6% versus 73.6%, point estimates)35. And the majority of faculty reported that their parents owned a home during the first 18 years of their life (75.7% versus 13.4% said primarily rented, and 10.9% rented and owned equally, point estimates), higher than one would expect given rates in the United States at the time (62% of homes owned by their occupants in 196036). Hence, faculty tend to come from families with relatively stable childhood financial circumstances.

### Impact of parental education on becoming faculty

Because the educational attainment of parents is strongly correlated with becoming faculty, we can use data on how many Ph.D.s are granted in a given year to forecast the changing composition of the professoriate. To better quantify this relationship, we model how the likelihood of having Ph.D. parents depends on a scholar’s characteristics.

Conditioned on having a parent with a Ph.D., the probability of becoming a faculty member is given by:

$$\Pr \left({{{\rm{faculty}}}}| {{{\rm{Ph.D.}}\,{\rm{Parent}}}}\right)=\frac{\Pr \left({{{\rm{Ph.D.}}\,{\rm{Parent}}}}| {{{\rm{faculty}}}}\right)\,\Pr \left({{{\rm{faculty}}}}\right)}{\Pr \left({{{\rm{Ph.D.}}\,{\rm{Parent}}}}\right)}$$

where our estimates of $$\Pr \left({{{\rm{Ph.D.}}\,{\rm{Parent}}}}\,| \,{{{\rm{faculty}}}}\right)$$ and $$\Pr \left({{{\rm{Ph.D.}}\,{\rm{Parent}}}}\right)$$ are given by the probability of faculty with Ph.D. parents in our survey, and the probability of an adult having a Ph.D. close to the birth year of a professor (22.2% and 0.9%, respectively, Table 1). To estimate the probability of being a tenure-track faculty $$\Pr \left({{{\rm{faculty}}}}\right)$$, we assess the proportion of the US adult workforce employed in postsecondary education recorded by the Bureau of Labor Statistics, around the year in which a professor started their job (0.4%)37. Because tenure-track faculty are just one kind of employee in postsecondary education, it is likely that this approach overestimates the percentage of the workforce employed as tenure-track faculty.

Combining these quantities, we estimate that the probability of becoming a faculty member given that one’s parents hold a Ph.D. $$\Pr \left({{{\rm{faculty}}}}\,| \,{{{\rm{Ph.D.}}\,{\rm{Parent}}}}\right)$$ is 9.5%, indicating a strong degree of both educational heritability and substantial professional advantage.

### Relationships with sociodemographic diversity

To the extent that becoming a professor is strongly influenced by having parents with doctoral degrees, our results paint a disheartening picture for efforts to build a racially diverse pipeline to the professoriate. Broad social and educational inequality within the United States indicates that Black and Hispanic adults are less likely to hold graduate degrees of any kind compared with white adults (Fig. 5). Our race/ethnicity-independent estimate of the probability of becoming faculty may overestimate the production of Black or Hispanic faculty because it fails to account for the lower probability of Ph.D. parents among Black and Hispanic children conditional on SES. Obtaining a precise race/ethnicity-conditioned estimate would require additional conditional estimates that are not currently available (for example, the proportion of Black and Hispanic people who obtain faculty positions).

However, we do find evidence of racial differences within our survey results: white professors are more likely to have a parent with a Ph.D. (23.4%, n = 5,905, point estimate) compared with Black or Hispanic faculty (17.2% and 16.9%, respectively, n = 518, point estimates). This distinction is even more pronounced among women surveyed, where 25.5% of white women have a Ph.D. parent versus 14.6% of Black women (point estimates, Supplementary Table 3). To the extent that the probability of becoming faculty depends on parental education, and specifically on having Ph.D. parents, this large racial gap in Ph.D. attainment is an intergenerational impediment to the proportion of Black and Hispanic scholars who become tenure-track faculty.

The relationship between parents’ education and the gender composition of the professoriate is complex. For instance, women in our sample are more likely to have Ph.D. parents than are men (24.8% versus 20.8%, point estimates). Identifying the reasons underlying this pattern is an interesting direction for future work. Of course, SES does not impede gender diversity in academia in the same way or to the extent that it does for racial diversity. Nevertheless, parents may differentially allocate resources to sons and daughters that influence their academic achievement38. Moreover, the intersection between race/ethnicity and gender is not neutral with respect to the likelihood of becoming faculty. For example, Black women faculty are less likely to have Ph.D. parents than are Black men (Supplementary Table 3). Prior evidence suggests that there has been a growing gender gap in college completion, with women outpacing men, among both white and, to a larger extent, Black Americans39. The overrepresentation of Black men from educationally privileged families may imply a differential selection within academia, wherein Black men need more social capital to pursue these careers. Untangling how gender, race/ethnicity and social origins interconnect to shape who pursues academic careers, and specifically why Black women professors were less likely to come from more educationally privileged families, is an important direction for future research.

### Trends in socioeconomic origins

Academia has undergone many dramatic shifts over the past 100 years, and our survey reveals several interesting and related trends. For instance, we find that the rate of faculty born from 1940 to 1960 reporting that a parent holds a Ph.D. has increased from 13% at the beginning of the period to 26% by the end (Fig. 6a), but then remains stable at above 20% across the next 50 years. This increase from the 1940s to 1960s mirrors the increasing college and graduate school enrolment rates within the United States over the same period40,41, and hence may simply reflect a general broadening of access to higher education. However, the subsequent stability of the rate at which faculty have a Ph.D. parent, in contrast to the continuing growth in and diversifying demographics of doctoral degree attainment42, suggests that the relationship between the educational attainment of faculty parents and the likelihood of becoming faculty has been consistent and strong for nearly half a century.

Parental education is so consequential that it also correlates with where in the academic hierarchy a professor lands. Across all years, we find that nearly a third of faculty at top-ranked universities across all eight fields report that one of their parents holds a Ph.D., and faculty at these elite departments are 57.4% more likely to have a parent who holds a Ph.D. than are faculty at the least prestigious departments (29.8% versus 19.0%; two-tailed test, z = 6.5, n = 2,612 95% CI = 0.076 to 0.142, P < 0.001). This concentration among elite departments is consistent with prior research documenting the ways academic hiring tends to devalue faculty of lower socioeconomic standing 26,27, and advantage faculty from more privileged backgrounds.

### Modelling faculty placement

Adjusting for faculty discipline, Ph.D. prestige, race/ethnicity and gender within a model, we find that faculty who had parents who attended college tend to be employed at significantly more prestigious universities than faculty without these childhood socioeconomic advantages (Table 2). We recognize that Ph.D. prestige may in fact be endogenous to this model of faculty placement because of Ph.D. placement’s likely relationship with parents education, but regardless include it as a control due to the strong evidence of its importance in predicting faculty placement43. Here, institutional prestige is defined by ordinal ranking, where the most prestigious universities have smaller rank (1), and less prestigious have larger rank (100). On average, faculty with a Ph.D. parent move ‘up’ in the institutional rankings by nearly 7 percentile ranks (two-tailed test, t = −3.4, n = 2,709, 95% CI = −10.536 to −2.850, P = 0.001). Faculty who grew up in wealthy neighbourhoods, who are also less racially diverse than the general population, also tended to place at more prestigious institutions. SES may constrain an individual’s search for a faculty job either by influencing which institutions they apply to, or by shaping their ability to stay on the academic track while searching for employment (Supplementary Note D). These results have direct implications for efforts to increase the socioeconomic and racial diversity of the professoriate, particularly at the most prestigious institutions that train most future professors43. The higher-prestige placement of faculty from advantaged socioeconomic backgrounds also represents a structural barrier to the visibility of the ideas of lower SES faculty because scientific discoveries made at more prestigious universities are more likely to spread throughout academia44.

Despite the significant correlation between having a parent who holds a Ph.D. and placement as faculty in the prestige hierarchy, not all faculty at elite universities (here: top 20% by USNWR or NRC) have this advantage. For elite faculty without Ph.D. parents, career support from colleagues both within and beyond their institution, and the wealth and urbanicity of their childhood zip code all become more important factors in explaining their placement at an elite institution. Faculty without Ph.D. parents who are employed at prestigious institutions report slightly higher levels of support from colleagues compared with those at lower-ranked institutions both within their institution (4.0 versus 3.8; two-tailed test, t = 3.6, n = 4,449, 95% CI = 0.078 to 0.261, P < 0.001), and outside their institution (4.0 versus 3.9; two-tailed test, t = 2.6, n = 4,448, 95% CI = 0.028 to 0.204, P = 0.009).

Most faculty earned their Ph.D.s at elite institutions43. Faculty without a Ph.D. parent, currently employed at an elite university, were more likely to come from more elite Ph.D. programmes than were similar faculty at non-elite universities (5.5 versus 14.0 median; Mann–Whitney U, ρ = 0.3, n = 3,813, P < 0.001). Furthermore, without Ph.D. parents, faculty at the top institutions are more likely to come from urban neighbourhoods (92.3% versus 87.3%) that are higher income (US $80,000 versus US$69,000 median; Mann–Whitney U, ρ = 0.4, n= 2,974, P < 0.001).

## Discussion

Using a large survey of eight academic disciplines spanning STEM, social science and the humanities, we quantify the extent to which becoming a professor is associated with indicators of SES, and in particular is most accessible to the children of doctoral recipients and those who grew up in wealthy urban neighbourhoods. Considering the intersection between race/ethnicity and class, we also conclude that this dependence on parent educational education and SES is likely to limit racial diversity within the professoriate.

To summarize, nearly a quarter (22.2%) of faculty reported that one of their parents holds a Ph.D., and over half (51.8%) had a parent who holds a graduate degree, compared with less than 10% of US adults of similar ages (Table 1). Faculty who have parents with Ph.D.s report receiving more support from them for their careers (Fig. 3) and are more likely to be employed at elite institutions. Nearly a third of faculty at top-ranked universities report that their parent holds a Ph.D. (29.8%), versus a fifth (19.0%) at lower-ranked institutions. This pattern represents a significant source of social reproduction at the highest levels of academic attainment. Moreover, given broader racial inequality in educational attainment, academia’s overrepresentation of inherited advantages represents a fundamental constraint to increasing its racial diversity (Fig. 5).

Previous studies of childhood SES of faculty have often been limited in scope, due to relatively small surveys of faculty or a lack of historical data on indicators of SES in the broader United States, both of which were necessary for this study. Given the well-established correlations between household parental income and child educational attainment and patterns of micro-class reproduction in other economic sectors, the overrepresentation of faculty with Ph.D. parents is perhaps unsurprising. Yet our results quantify just how large, and how persistent, that overrepresentation is.

In fact, the importance of having Ph.D. parents appears so great that the rate of having them nearly doubles across the transition from completing a Ph.D. to obtaining a faculty job (11.8% versus 22.2%; Table 1). This effect indicates a substantial loss of talent in the pipeline from Ph.D. to the transition to a faculty job. Doctoral students with Ph.D. parents may be better prepared for the difficulties of the academic job market, which may confer an advantage that becomes even greater during periods when academic jobs are scarce, for example, during a pandemic or a recession. Furthermore, the stability of this pattern across STEM fields, the social sciences and humanities suggests that the loss of talent at this stage in the pipeline is unlikely to be caused by the existence of attractive non-academic jobs for STEM Ph.D.s. Understanding the causes of this pattern is an important direction for future work.

But the even greater overrepresentation of faculty with Ph.D. parents at the most prestigious institutions (Fig. 6) implies that increasing sociodemographic diversity in academia is not only a pipeline problem. Indeed, our findings complement research that suggests a re-emergence of socioeconomic reproduction at the highest levels of educational attainment8,45. Progress towards broadening participation in science will remain limited if our current practices within academia favour individuals with advantages conferred by parental socioeconomic and occupational privilege21,26,27.

The interpretation of our work is limited by the granularity of the IRS and Census data we link to individual responses, and the current composition of tenure-track faculty. Several of our comparisons rely on aggregated measures used to proxy for income based on respondents’ childhood zip code. Many zip code boundaries have changed over time and may span heterogenous populations48. Each of these may limit the accuracy of our childhood income measures, especially for older faculty. In future work, better estimates of household income could be obtained from information on respondents’ childhood Census blocks, or from de-identified historical tax records49. Similarly, our analyses of race and ethnicity are limited by the use of the 2010 Census classification standard, which provides some consistency with 2020 standards50, but precludes certain more fine-grained analyses.

Furthermore, our survey was restricted to current tenure-track faculty at research intensive academic institutions between 2017 and 2020. As a result, our data say little about contingent faculty at these institutions, which represent an increasing fraction of university instructional staff51. We also do not know how gender discrimination, racism or classism impacts academics’ intentions to persist, and we do not observe faculty who left academia before our survey date, whether that was after graduate school, a postdoctoral position or a contingent or tenure-track academic position, and whether that is due to feeling pulled towards more desirable opportunities outside of academia or pushed out by discrimination in hiring or promotion within. This group may be a population biased toward faculty from lower-income backgrounds19. Such a bias would tend to lead us to underestimate the observed associations or importance of SES in our analyses if we aimed to examine associations for all faculty entering academia, but these associations do not bias our estimates for current faculty. An important and complementary direction of future work would examine SES among non-tenure-track faculty and investigate its role in recruitment and retention before and along the tenure track.

While this work highlights the advantages that SES may confer for becoming and remaining faculty, it does not explicitly speak to the social or professional difficulties that underrepresented individuals experience stemming from their gender, race, ethnicity or SES once they become faculty. For example, women are less likely to be viewed as competent52,53, less likely to be awarded for their research54, less likely to be invited for talks55 and less likely to be compensated fairly56,57. Black faculty earn less despite no measurable differences in productivity58 and may have their research evaluated less positively31, all while facing unequal service burdens and racism on and off campus59,60. While research on how SES shapes faculty careers is less comprehensive, work generally points to less supportive environments24,25 and differences in research or teaching appointments61. Individuals with disadvantaged socioeconomic origins may become faculty and still have greater difficulty navigating and finding support within academia.

Future research should consider why the importance of family background varies moderately by gender and across fields. In our survey, women are more likely than men to have a parent with a college degree or higher (Supplementary Table 2), indicating the greater barriers facing women in reaching the highest ranks of academia. The reasons for this difference among tenure-track faculty require more investigation. Percentages of faculty with a Ph.D. parent are lowest among Business and Sociology (17.3%), and highest among Computer Science and History (26.3%, χ2 = 39.1, n = 3,362, P < 0.001). Compared with the educational attainment of professors’ parents (Table 1), and generally consistent with the result presented here, a survey of lawyers found that about 12% of lawyers were the children of lawyers62, and among doctors in Sweden, 20% of physicians were the children of physicians63. Whether or not having a Ph.D. parent provides more of an advantage to academic faculty than one would expect relative to other high-status occupations is an important question for future research.

Our study’s detailed and cross-disciplinary view of the deep socioeconomic roots of current US faculty highlights the social reproduction of the highest levels of academic attainment and scientific influence. These results raise a critical question regarding what ramifications high SES overrepresentation in the academic workforce—and its skewed intersection with race/ethnicity and class in the United States—have on the type of scholarship that faculty produce and on the scholars they train. That is, what discoveries are not made and what ideas are not developed as a result of academia’s historical and current lack of socioeconomic diversity? A lack of diversity is already known to reduce research on public health disparities31, immigration, mental health and gender-based discrimination29. However, it is likely that the loss of useful research is more pervasive, because the relationship between researcher identity and scholarship is complex, depending not only on scientific social capital64 but also on how that capital can be invested. Developing a quantitative understanding of how representational diversity shapes scholarship across fields is an important direction for future work, which will rely on quantitative assessments of current diversity and its roots like the one provided here. Making academia more diverse and inclusive is a considerable challenge, given the complex interplay of SES and academic achievement from early schooling through academic appointments. Progress in this direction, however, is likely to produce substantial scientific and societal benefits.

## Methods

Our survey protocol was approved by the University of Colorado Boulder Institutional Review Board. All participants were included in drawings for cash lotteries (Supplementary Note A). Although contingent faculty represent an increasing proportion of the US academic workforce, we surveyed tenure-track faculty because their unique stability in academia makes them more likely to contribute research and mentor scholars over the course of their careers.

Overall, our response rate resembles other online surveys with email invitations conducted in the context of academia65,66. While it is not possible to know whether faculty respondents differ in their childhood SES from those who did not respond, respondents were generally representative of their fields by institutional prestige and faculty rank. Additionally, women responded at slightly higher rates than expected in several disciplines (Supplementary Table 1). This pattern may imply a slight upward bias in our analysis of parents’ education: in our sample, women are somewhat more likely to come from highly educated families (one of their parents holds a Ph.D.: 24.8% versus 20.8%; z = 3.9, n = 7,145, P < 0.001), across all disciplines (Supplementary Table 2). Roughly a quarter of respondents provided a country outside of the United States where they grew up (n = 2,007; 25.1%), consistent with prior estimates of the proportion of international faculty at US universities67.

We asked faculty to reflect on their childhood and report their parents’ highest levels of education during this period (Supplementary Note A). We compared their responses to the adult educational attainment statistics in the US population in the year the survey respondent was born68, and to educational attainment of parents of US doctoral recipients in the year they graduated from their doctoral programme69. These benchmarks, respectively, facilitate comparisons of faculty with the general public, and with academics at a career stage just prior to their current one. Comparing faculty to national estimates describe how privileged the upbringings of faculty were relative to the general public, and comparisons to doctoral recipients describes how the socioeconomic roots of faculty differ from the broader set of their doctoral peers.

Data on the educational attainment of adults are drawn from the American Community Survey of the US Census, and information on family’s educational attainment among Ph.D. recipients is drawn from the NSF SED. While historical data on undergraduate and graduate achievement is available from the NSF70, it does not report demographics as percentages of the adult population or describe any earlier education levels as we require. Before 1993, the Census recorded educational attainment as years of schooling, and completion (or not) of high school or bachelor’s degree71. More recent data recorded educational attainment as highest degree held among adults 25 years and older. In our study, we refer to both. Using the earlier Census data, we compare parents’ college completion among faculty to college completion rates in the adult US population. When we require information on degree earned, we compare respondents with the Census dataset closest to the year faculty were born that records this information.

Our estimates of US adult Ph.D. completion (from 1993 at the earliest) are likely higher than the true Ph.D. completion rate in the year a respondent was born (which was 1967, on average), due to rising educational attainment in the United States. Thus, our comparisons of faculty parents with the general public may in fact underestimate academia’s overrepresentation of highly educated families. Data on the highest degree attained by parents of Ph.D. recipients is available for select years from 1993 to 2018 from the NSF69. Details on how many respondents matched with a given year of data are provided in Supplementary Note B.

As childhood household income is also predictive of educational attainment, we construct proxy measures of childhood income levels and geographic diversity among current faculty. The ideal dataset would be individual-level household income and location during the childhoods of faculty. Because it is difficult for survey respondents’ to estimate their parents’ income retrospectively, we approximate these quantities using more easily recalled information. We link zip code responses to the closest publicly available IRS tax release (1998–2018) when faculty were children (adjusted to 2020 US \$). Proxying individual SES based on aggregate zip code-level data is a widely used approach72,73. However, the geographic proxy may suffer from some mismeasurement because zip codes often span heterogeneous populations74. To measure whether faculty come from rural or urban areas, we use US Census data from 1990 to 2010, linked to faculty responses by their year of birth75, which records how many people live in urban or rural areas in a given zip code. We labelled zip codes as either rural or urban based on whether the majority of the population in that zip code lived in a rural or urban area. As with our measure of income, this approach may suffer from mismeasurement due to the underlying heterogeneity of a zip code. These measures of childhood household income and geographic diversity, when taken together with self-reported information on parental education, speak to how faculty SES shapes academic careers.

Finally, to assess whether SES impacts job placement within academia, we compare the previously described measures with the institutional prestige of a respondent’s current faculty appointment and Ph.D.-granting institution. For most of the disciplines considered here, we refer to 2017–2020 college rankings from USNWR, which are provided for most Carnegie-classified R1 and R2 doctoral universities. Because Anthropology departments are not ranked by USNWR, we use their R rank in the 2010 NRC Anthropology rankings. To account for differences in the number of departments across disciplines, we rescale each ranking to the unit interval. Where faculty were employed at institutions not ranked by USNWR or NRC (11.1% of responses), we coded their ranking as missing and excluded them from our analysis of institutional prestige.

### Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.