Sage Journals: Discover world-class research

Abstract

U.S. studies have found that stratified graduate education accounts for most of the relatively strong intergenerational socioeconomic association among postgraduate degree holders. The same association has been observed, but not explained, in countries with higher education systems that differ from that of the United States. We explore the mediation role of undergraduate- and graduate-level stratification in accounting for the intergenerational occupational association among postgraduate degree holders in the United Kingdom. We find that the unequal distribution of undergraduate-level education and path dependency between undergraduate- and graduate-level stratification help to give rise to an unequal occupational outcome by social origin among postgraduate degree holders. We explain this by the tight coupling of undergraduate and graduate education in the United Kingdom. Our analysis also illustrates the need to go beyond graduate education in understanding social origin inequality among postgraduate degree holders to examine the role of undergraduate education and how it is linked to graduate education.

Keywords

class inequality higher education educational transition graduate education intergenerational mobility meritocracy

A large body of research has reported a weak (or even a null) intergenerational socioeconomic association among college graduates (Breen 2010; Breen and Jonsson 2007; Breen and Luijkx 2007; Hauser and Logan 1992; Hout 1984, 1988; Pfeffer and Hertel 2015; Torche 2011; Vallet 2004), although there is debate about how far this is an effect of college education itself or of selection into college (Karlson 2019; Zhou 2019). More recent work, however, has documented a strong intergenerational socioeconomic association among postgraduate degree¹ holders (Falcon and Bataille 2018; Oh and Kim 2020; Torche 2011, 2018; Wakeling and Laurison 2017). These findings call into question the equalizing role of higher education for intergenerational mobility, and they are particularly important because of the rapid growth in the number of postgraduate degree holders and the steep increase in economic returns to postgraduate degrees (Autor 2014; Fisher et al. 2018; Keister 2014; Lemieux 2008; Posselt and Grodsky 2017; Valletta 2015).

A few recent studies have investigated the mechanisms behind the intergenerational association among postgraduate degree holders (Oh and Kim 2020; Torche 2018). They found that differences in graduate education by socioeconomic origin account for a substantial part of the intergenerational association among this group. It is not yet clear, however, how graduate-level stratification that accounts for the intergenerational association among postgraduate degree holders relates to undergraduate-level stratification. Socioeconomic origin is associated with college destination (Davies and Guppy 1997; Goyette and Mullen 2006; Hearn 1991; Karen 2002; Oh and Kim 2020), and there are strong links between features of undergraduate and graduate education (Bowen and Bok 1998; Eide, Brewer, and Ehrenberg 1998; Ethington and Smart 1986; Goyette and Mullen 2006; Lang 1987; Tinto 1980; Zhang 2005, 2012). Nonetheless, given the current state of research, we do not know whether the graduate-level stratification that leads to unequal socioeconomic outcomes for individuals of different socioeconomic origins is driven by unequal sorting into college followed by a strong path dependency between undergraduate and graduate education or by unequal sorting into graduate education despite a more equal distribution at the undergraduate level.

We use a novel U.K. data set that contains measures of various aspects of undergraduate- and graduate-level stratification in addition to social origin and destination to answer two questions. First, how is the association between social origins and destinations among postgraduate degree holders mediated by differences in the type and nature of their graduate education? Second, how far can the mediating effect of graduate-level education be explained by prior differences in undergraduate education? To answer this second question, we need to focus on the links between undergraduate and graduate education. To date, only a few studies have investigated both undergraduate and graduate education as a vehicle for intergenerational socioeconomic persistence (e.g., Goyette and Mullen 2006; Oh and Kim 2020), in line with the paucity of studies in stratification research that explore more than one educational transition (Karlson 2011).

Background

What We Know about the Puzzle: Stratification at Graduate Education and Intergenerational Socioeconomic Association

While confirming Hout’s (1984, 1988) long-standing finding of a null intergenerational occupational association among college graduates in younger cohorts, Torche (2011) found stronger associations among postgraduate degree holders. For stratification scholars, this finding was counterintuitive. For one thing, given that attaining a postgraduate degree is more difficult than attaining a bachelor's degree, postgraduate degree holders from disadvantaged social origins are likely to be even more positively selected on abilities or aspirations than college graduates. For another, if the null or weak intergenerational associations among college graduates are attributable to a more meritocratic labor market for the highly skilled (Breen and Jonsson 2007), the labor market for postgraduate degree holders would presumably be at least as meritocratic. Nevertheless, Torche’s (2011) finding held across different measures of socioeconomic outcomes, such as class, occupational status, and earnings, and has been corroborated by studies using more recent and larger U.S. data sets and in other countries, such as France and the United Kingdom (Falcon and Bataille 2018; Oh and Kim 2020; Posselt and Grodsky 2017; Torche 2011, 2018; Wakeling and Laurison 2017).

Some studies have attempted to explain the strong intergenerational socioeconomic associations among postgraduate degree holders. Torche (2011) found that postgraduate degree holders from advantaged socioeconomic origins are more likely than their counterparts from disadvantaged origins to choose remunerative fields and degree programs in graduate education (see also Mullen, Goyette, and Soares 2003). They are also more likely to have more lucrative occupations and to earn more, even when they are in occupations similar to those held by postgraduate degree holders from disadvantaged origins. In a study of a sample of doctorate holders in the United States, Torche (2018) demonstrated that differences in graduate education (e.g., selectivity of the degree-awarding institution, fields of study, and institutional control) and occupational mediators (e.g., occupational status and employment sector) account for the association between parental education and children's earnings among White and Black men with a doctoral degree.

Oh and Kim (2020) tested similar mechanisms for holders of different types of postgraduate degrees in a large sample of U.S. male and female college graduates age 35 to 54. They found that the association between parental education and earnings of postgraduate degree holders is fully explained by stratification in graduate education and early degree completion. They further tested an explanation advanced by Torche (2011) for the contrasting patterns of intergenerational socioeconomic associations between terminal bachelor's holders and postgraduate degree holders. Torche (2011) had suggested that the role of college as a sorting machine for graduate education (Stevens, Armstrong, and Arum 2008) might explain the difference. The supposition was that children from advantaged origins are more likely to choose undergraduate fields that could increase their chances of obtaining access to more desirable kinds of graduate education. In contrast, children from disadvantaged origins are more likely to focus on maximizing their chances of securing a decent job immediately after college graduation because their chances of continuing to graduate education are slim (Goyette and Mullen 2006). As a result, among terminal bachelor's holders, children from disadvantaged origins are likely to have optimal undergraduate educational choices, whereas among postgraduate degree holders, children from advantaged origins are more likely to have made optimal choices. Oh and Kim (2020) show that heterogeneous choices of undergraduate fields and differentiated sorting into graduate fields are responsible for the contrasting patterns of intergenerational socioeconomic associations between terminal bachelor's holders and postgraduate degree holders.

These studies have contributed to our understanding of intergenerational socioeconomic associations among those at the top of the educational distribution, but the exclusive focus on the United States has left the same phenomenon unaddressed in countries with different higher-education systems. For example, in the United Kingdom, in contrast to the United States, medicine and law education takes place at the undergraduate level, and graduate-level education is not a prerequisite to practice in these fields. The relative importance of undergraduate and graduate stratification and the sorting between them will thus likely differ across countries. More importantly, these U.S. studies have paid disproportionate attention to graduate-level stratification. Despite the work of Oh and Kim (2020), it is still unclear whether undergraduate- or graduate-level stratification is more decisive in explaining unequal socioeconomic outcomes among postgraduate degree holders.

What We Know about Paths from Socioeconomic Origin to Higher Education and Socioeconomic Outcomes

One of the most well-studied stratification elements in higher-education research is institutional selectivity (Davies and Guppy 1997; Hearn 1991; Karen 2002). One mechanism linking college selectivity and differentiated postgraduate socioeconomic outcomes is the effect of college selectivity on graduate education, an important consideration given growing returns to graduate education (Autor 2014; Fisher et al. 2018; Keister 2014; Lemieux 2008; Posselt and Grodsky 2017; Valletta 2015). Studies have documented positive associations between attending a selective college and better outcomes in graduate education (Bowen and Bok 1998; Eide et al. 1998; Ethington and Smart 1986; Lang 1987; Tinto 1980; Zhang 2005, 2012). College graduates from selective institutions are not only more likely to attain a postgraduate degree, they are also more likely to attend more selective programs and institutions. Furthermore, they complete their graduate education earlier than their peers from less selective colleges. These results suggest that at least part of the explanation for the strong intergenerational socioeconomic association among postgraduate degree holders could lie in undergraduate-level stratification. Britton and colleagues (2020) suggest this may also be the case in the United Kingdom.

Along with institutional selectivity, undergraduate field of study (Gerber and Cheung 2008) and postgraduate degree type (Posselt and Grodsky 2017) have also been widely discussed as dimensions of stratification in higher education. College graduates majoring in technical or STEM (science, technology, engineering, and mathematics) fields have significantly higher earnings (Berger 1988; Grogger and Eide 1995; Grubb 1993; Kim, Tamborini, and Sakamoto 2015; Rumberger and Thomas 1993; Sullivan et al. 2018a). Recent studies have found that individuals who major in business in the United States and in LEM (law, economics, and management) in the United Kingdom also earn more than those majoring in the social sciences and humanities (Kim et al. 2015; Sullivan et al. 2018a). Earnings also vary by degree type among postgraduate degree holders, with, in the United States, the highest earnings for professional degree holders, followed by those with doctoral and master's degrees (Day and Newburger 2002; Ma, Pender, and Welch 2016).

The link between undergraduate fields and postgraduate degree types has received little attention. Studies have conjectured that some students may choose their undergraduate fields intending to pursue graduate education (Eide and Waehrer 1998). Goyette and Mullen (2006) found that college graduates who major in arts and sciences are more likely to attend graduate school than are those with vocational majors. These and other studies (Zhang 2005) suggest disadvantages induced by majoring in less lucrative undergraduate fields could be compensated for later if students proceed to graduate education.

The U.K. Context

U.K. graduate education consists of three main types: taught programs (e.g., one-year master's degrees), research programs (e.g., three-year doctorates), and other (mostly vocational) graduate programs (e.g., postgraduate certificate in education [PGCE]). These programs are markedly different in their aims and the characteristics of participating students. Taught programs account for the majority of graduate students, and they have experienced rapid growth over the past few decades. They consist of vocational and academic courses; the academic courses serve as an entry route for graduate-level research programs. Unlike in the United States, the majority of doctoral students in the United Kingdom commence their study after completing a master's course (Higher Education Funding Council for England [HEFCE] 2013).

Graduate education in the United Kingdom is characterized by unregulated fees and limited scholarships. Graduate students thus depend heavily on self-funding, which leads to high rates of part-time participation, especially in taught or PGCE programs. Previous studies suggest these financial barriers might have reduced intergenerational social mobility (Conlon and Patrignani 2011; HEFCE 2013; Lindley and Machin 2013; Wakeling and Laurison 2017; Wakeling and Mateos-González 2021; Wakeling and Savage 2015). Results from some recent studies, however, are more nuanced. Although they corroborate previous findings on substantial gaps by family background in entry to and types of graduate education, most of these gaps are explained by previous academic attainment or education received in school or university (Britton and van der Erve 2020; Mateos-González and Wakeling 2022; Scott 2020; Wakeling 2017). Similarly, the earnings premium for postgraduate degree holders is largely accounted for by previous attainments (Britton et al. 2020). These results suggest a linkage between social origin, undergraduate education, and graduate education. They also hint that postgraduate degrees in the United Kingdom do not bring much of an economic premium to graduates, and thus many students choose graduate education as “insurance” to avoid very low occupational attainments or earnings (Britton et al. 2020). This is especially the case for PGCEs, which are filled with a disproportionately large number of lower-origin students.

Analytic Approach

Previous studies of the mechanisms of unequal socioeconomic outcomes among postgraduate degree holders have paid most attention to the role of stratification in graduate education. Our approach goes beyond that to consider how much of this can be explained by undergraduate education and to assess which educational level is more significant in maintaining inequality by social origin among postgraduate degree holders. Following the literature, we characterize education along two dimensions. Undergraduate education comprises two horizontal stratification axes (institutional selectivity and field of study) and a vertical axis (undergraduate academic performance); graduate education is stratified by the same horizontal axes and a vertical axis (degree type).

We measure intergenerational association by the relationship between a person's social class origin and their class destination (unlike previous studies by Oh and Kim [2020] and Torche [2018] that examine the association between parental education and children's earnings). We chose this approach for practical and theoretical reasons. In the data we use (the UK Labour Force Survey; Office for National Statistics 2021), earnings are asked only for currently employed people, whereas social class is based on respondents’ current or past occupation, resulting in fewer missing cases. Because occupational destination precedes earnings in the temporal order of individuals’ careers, we avoid a potential problem of using earnings as a dependent variable, namely, that mediation by occupational assignment must be addressed. Our outcome of interest is a binary measure of occupational destination: being in an “elite” occupational group or not.

In our first set of analyses, we focus on the relative importance of undergraduate and graduate education in mediating the association between social origins and being in an elite occupation. Figure 1 shows the logic of this part of our analyses.² We focus on individuals who have a postgraduate qualification, and our starting point is an assessment of how much of the association between social origins, S, and attaining an elite occupation, Y, is mediated via characteristics of their graduate education, P. We then ask: How much does P mediate the association between S and Y when we also consider undergraduate education, U? We labeled the paths, for convenience, so the direct effect of social origins, S, on Y is labeled f. The total effect of S on Y, which we call T, is equal to this direct effect plus sums and products of the other paths: $T = f + a (b e + d) + ce$ . Here, a represents the association between social origins and undergraduate education, b the association between undergraduate and graduate education, e the association between graduate education and the outcome, Y, conditional on social origins and undergraduate education, and d the association between undergraduate education and Y, conditional on graduate education and social origins. Graduate education, P, meditates the $S - Y$ association through the path $ce$ and thus mediates share $\frac{ce}{T}$ of the total effect. But if we do not also condition on U, it will appear to mediate an amount equal to $\frac{(e + b d) (c + a b)}{T}$ ; failing to control for U will thus bias our estimate of how much graduate education mediates the intergenerational association by $\frac{b (e a + c d + b a d)}{T}$ . Under the assumption that all the associations shown in Figure 1 are positive, failing to control for U will lead to upward bias in our estimate of the importance of graduate education as a mediator of the relationship between social origins and access to an elite occupation. Note that if $b = 0,$ the bias will be zero because U and P would be independent (undergraduate education unrelated to graduate education). Also, if $d = 0$ and either $a = 0$ or $e = 0,$ there would be zero bias: that is, if undergraduate education, U, had no direct association with Y and either U was independent of social origins, S, or P (graduate education) had no direct association with Y, omitting U would not lead us to overestimate the degree to which P mediates the $S - Y$ association. But none of these conditions is likely to hold in reality.

Figure 1.

Graduate (P) and undergraduate (U) Education as Mediators of the Social origin (S) – destination (Y) relationship.

We find that although features of graduate education strongly mediate social origin differences in the likelihood of entering an elite occupation (“class gaps”) among postgraduate degree holders, once we control for features of their undergraduate education, the mediating role of graduate educational characteristics virtually vanishes. Our second question, then, asks how the dimensions of graduate education that seem to do most to mediate the class gaps relate to characteristics of undergraduate education.

All our models include control variables for age, age-squared, ethnicity, and country of birth. Because the models are nonlinear, we used the reformulated KHB method (Breen, Karlson, and Holm 2013; Karlson, Holm, and Breen 2012; Kohler, Karlson, and Holm 2011), which is called the linear predictor approach (Breen et al. 2021), to compare the coefficients from the nested models.

Data And Measures

Data

The U.K. Labour Force Survey (LFS; Office for National Statistics 2021) is a large, nationally representative cross-sectional survey, similar to the U.S. Current Population Survey (CPS). Since 2011, it has captured characteristics of both undergraduate and graduate education, such as degree-awarding institution, subject of degree, academic performance (for undergraduate education only), and type of postgraduate degree. Since 2014, the LFS has asked mobility-related questions in every July–September quarter survey, including one about the occupation of the main earner in the household in which the respondent lived at age 14. We use this to construct our social origin variable. Our analytic sample was constructed using seven years of LFS data (2014–2020) for men and women age 35 to 59 who held a postgraduate degree (4,207 men and 5,278 women). Observations with missing values were omitted, leaving 7,482 respondents (3,260 men and 4,222 women).

One critical limitation of these data is that degree characteristics are asked only for the highest degree. If a person's highest qualification were a doctorate, questions would be asked only about the doctorate, leaving characteristics of a bachelor's degree missing. Fortunately, around three-quarters of the postgraduate degree holders in our sample also answered questions about their undergraduate degrees in the July–September quarter survey between 2014 and 2020. Because the LFS has a five-wave rotational structure, where respondents are surveyed for five consecutive quarters, we used responses to this question in all waves to minimize missingness.

To further address the issue of residual missing data on the undergraduate institution for some postgraduate degree holders, we applied inverse probability weights (Seaman and White 2013; Wooldridge 2007). For all postgraduate degree holders, we constructed the weights from a logit model that regressed the binary measure of having answered the questions about undergraduate institution on sex, age, marital status, country of birth, ethnicity, residential region, economic status, social origin, and class destination. We then multiplied the survey weights from the LFS with the inverse of these probabilities. This procedure yields a weighted sample that is representative of the complete data.

Variables

Our main outcome variable is having an elite occupation, defined as an occupation that belongs to the U.K. National Statistics Socio-economic Classification (NS-SEC) category 1 (1.1: higher managerial, administrative occupations and 1.2: higher professional occupations) rather than having a nonelite occupation (NS-SEC 2–8). This distinction is conventional in U.K. class analyses (e.g., Sullivan et al. 2018b).

Our primary explanatory variable is the respondent's social origin based on the NS-SEC of the household's main earner's occupation when respondents were 14. The sample was restricted to respondents who, at age 14, lived with family members and at least one of whom had earnings. We grouped the eight NS-SEC codes³ into four categories to give a parsimonious representation of social origin gaps (Gugushvili, Bukodi, and Goldthorpe 2017): NS-SEC 1 (higher salariat), NS-SEC 2 (lower salariat), NS-SEC 3 to 5 (intermediate), and NS-SEC 6 to 8 (working-class) origins.

Stratification in undergraduate education is measured in three ways. For institutional selectivity, we followed previous work on U.K. higher education (Boliver 2013; Crawford et al. 2016; Sullivan and Brown 2015; Sullivan et al. 2018b; Wakeling and Savage 2015) and assigned each university to one of four categories: global top 20 (the four U.K. universities in the global top 20 are Oxford; Cambridge; Imperial College, London; and University College, London), Russell Group (24 universities), pre-1992, and post-1992 or other universities. To define subject area of degree, we followed previous studies (Sullivan et al. 2018b; Walker and Zhu 2011) and grouped 20 subjects into three categories: LEM, STEM, and OSSAH (other social sciences, arts, and humanities or combined areas). Performance at the undergraduate level was measured by the degree class awarded: first, upper second, lower second, and third/pass/other. Institutional selectivity and subject area at the graduate level were defined in the same way as at the undergraduate level. Type of postgraduate degree distinguished doctorate, master's, PGCE and others.

Results

Descriptive Statistics: Social Origin, Higher Education, and Occupational Destination

Table 1 shows the distribution of our sample of postgraduate degree holders by their social origin across each stratification axis at the undergraduate and graduate levels. Social origins are strongly linked to undergraduate institution type among men and women. The link with undergraduate subject area is also evident, although somewhat weaker, and there is, at most, only a weak relationship between origins and degree class. At the graduate level, associations with social origin are similar, although institutional differentiation is less pronounced, and the link between class origins and degree type is more marked. There are also noticeable differences between men and women: At the undergraduate level, institutional differences between class origins are generally smaller among women, but attending a selective institution for graduate study is more stratified by social origin among women.

Table 1.

Descriptive Statistics for Postgraduate Degree Holders in the United Kingdom.

	Men (%)				Women (%)
	Social Origin				Social Origin
	NS-SEC 1	NS-SEC 2	NS-SEC 3–5	NS-SEC 6–8	NS-SEC 1	NS-SEC 2	NS-SEC 3–5	NS-SEC 6–8
Undergraduate level
Institution
Selective	52.06	47.53	36.28	29.85	44.81	38.16	29.67	25.74
Pre-1992	23.97	25.00	27.04	29.33	26.03	27.42	28.85	27.13
Post-1992	23.97	27.47	36.68	40.82	29.17	34.41	41.48	47.13
Subject area
STEM	47.91	44.71	45.10	39.11	35.74	30.78	30.51	31.19
LEM	21.19	17.26	18.47	23.02	15.50	13.93	15.98	12.96
OSSAH/COMB	30.90	38.03	36.43	37.87	48.76	55.29	53.51	55.86
Degree class
First	15.56	13.99	17.06	15.14	13.48	10.73	10.04	8.91
Upper second	47.19	47.15	45.91	44.36	51.36	53.29	51.37	51.90
Lower second	24.24	25.85	23.58	26.76	24.08	25.77	27.57	27.79
Third/pass/other	13.01	13.01	13.45	13.75	11.07	10.21	11.02	11.39
Graduate level
Institution
Selective	41.69	38.48	36.56	34.11	37.56	30.51	28.25	24.66
Pre-1992	29.68	29.18	27.00	29.69	23.00	27.92	27.32	26.46
Post-1992	28.64	32.34	36.44	36.21	39.44	41.57	44.44	48.88
Subject area
STEM	46.40	42.21	42.29	36.61	32.94	28.07	28.44	29.52
LEM	21.02	17.92	18.34	22.55	15.62	13.56	14.86	13.43
OSSAH/COMB	32.58	39.87	39.37	40.84	51.44	58.37	56.70	57.06
Degree type
Doctorate	17.19	16.59	15.39	13.27	11.43	10.12	8.72	5.56
Master’s	56.51	52.42	54.30	54.24	46.36	43.75	43.66	43.49
PGCE	13.96	18.51	16.29	20.27	25.97	30.54	31.10	33.26
Other	12.34	12.48	14.02	12.22	16.24	15.60	16.51	17.69
Occupational destination
NS-SEC 1	52.35	45.78	46.93	40.04	35.00	30.07	27.57	27.64
NS-SEC 2	36.53	40.81	38.64	44.56	48.63	51.75	54.28	56.98
NS-SEC 3–5	7.57	10.08	10.96	9.46	10.59	11.31	12.26	10.43
NS-SEC 6–8	3.56	3.33	3.47	5.93	5.79	6.87	5.89	4.95
Total	1,085	814	913	448	1,265	1,145	1,272	540

Note : NS-SEC = U.K. National Statistics Socio-economic Classification; STEM = science, technology, engineering, and mathematics; LEM = law, economics, and management; OSSAH = other social sciences, arts, and humanities; COMB = combined subjects; PGCE = postgraduate certificate in education.

When we turn to postgraduate degree holders’ occupational destinations, a social origin gap is evident, mostly at the top two tiers of the NS-SEC class schema (NS-SEC 1 vs. NS-SEC 2), where over 80 percent of men and women postgraduate degree holders are found. Individuals from the highest origin were around 1.3 times more likely to have NS-SEC 1 occupations than were those from the lowest origins. In contrast, children from the lowest origins tended to enter NS-SEC 2 occupations at greater rates than their peers from the highest origins. We see no substantial social origin gap for the lower-class destinations. A considerable gender gap is also evident here: For all social origins, women are less likely to be found in the highest social class and are quite heavily clustered in Classes 2 and 3 to 5. The intrinsic association coefficient, (Bouchet-Valat 2022) fitted to the mobility tables shown in Table 1, returns a value of .10 for men and .08 for women, indicating a weaker overall origin–destination association among the latter.⁴

Mediation of the Class Gaps in Attaining an Elite Position

Our first analysis addresses the question of how class differences in elite attainment are mediated by education. Table 2 shows results from a logit model regressing our binary outcome (elite vs. nonelite occupational destination) on social origin without (Model 1) and with (Model 2) controls for graduate-level stratification. Table 3 shows the same models with graduate-level stratification replaced by undergraduate. The third columns for each gender represent indirect effects of class origins via features of education (upper row) and the percentages of the total effect mediated by them (bottom row). All indirect effects are statistically significant at the conventional significance level (p < .05).

Table 2.

Results from Logistic Regression Models (Linear Predictor Approach) Regressing Having an NS-SEC 1 Occupation on Social Origins with and without Controlling for Graduate-Level Educational Stratification.

	Men (N = 3,260)			Women (N = 4,222)
	Model 1	Model 2	Indirect Effects	Model 1	Model 2	Indirect Effects
Social origin
NS-SEC 1	Reference	Reference	Reference	Reference	Reference	Reference
NS-SEC 2	−.299*(.118)	−.158(.110)	−.141**47.20%	−.267*(.110)	−.127(.105)	−.140**52.46%
NS-SEC 3–5	−.253*(.109)	−.114(.105)	−.139**54.82%	−.428***(.110)	−.269**(.102)	−.160***37.26%
NS-SEC 6–8	−.592***(.148)	−.350*(.142)	−.242***40.82%	−.428**(.136)	−.178(.130)	−.249***58.26%
Control variables	Yes	Yes		Yes	Yes
Postgraduate degree type		Yes			Yes
Postgraduate institution		Yes			Yes
Postgraduate subject area		Yes			Yes
Postgraduate degree type
Doctorate		Reference			Reference
Master’s		−.848***(.125)			−1.025***(.123)
PGCE		−2.219***(.177)			−2.425***(.152)
Others		−.582**(.168)			−1.065***(.144)
Postgraduate institution
Global top 20		Reference			Reference
Russell Group		−.153(.155)			−.114(.165)
Pre-1992		−.270(.156)			−.390*(.170)
Post-1992		−.399*(.166)			−.307(.163)
Postgraduate subject area
STEM		Reference			Reference
LEM		.060(.111)			.387**(.120)
OSSAH/other		−.825***(.099)			−.598***(.093)
Pseudo R²	.007	.120		.006	.132

Source : U.K. Labour Force Survey (2014–2020; Office for National Statistics 2021).

Note : Bootstrapped standard errors are in parentheses. Estimates reported as log odds. The third columns for each gender represent indirect effects of class origins via features of education (upper row) and the percentages of the total effect mediated by them (bottom row). Control variables include age, age-squared, ethnicity, and country of birth. NS-SEC = U.K. National Statistics Socio-economic Classification; STEM = science, technology, engineering, and mathematics; LEM = law, economics, and management; OSSAH = other social sciences, arts, and humanities; COMB = combined subjects; PGCE = postgraduate certificate in education.

p < .05, **p < .01, ***p < .001.

Table 3.

Results from Logistic Regression Models (Linear Predictor Approach) Regressing Having an NS-SEC 1 Occupation on Social Origins with and without Controlling for Undergraduate-Level Educational Stratification.

	Men (N = 3,260)			Women (N = 4,222)
	Model 1	Model 2	Indirect Effects	Model 1	Model 2	Indirect Effects
Social origin
NS-SEC 1	Reference	Reference	Reference	Reference	Reference	Reference
NS-SEC 2	−.270*(.110)	−.156(.106)	−.114***42.36%	−.244*(.103)	−.127(.102)	−.117***48.02%
NS-SEC 3–5	−.237*(.103)	−.110(.100)	−.127***53.64%	−.394***(.103)	−.256*(.100)	−.138***34.98%
NS-SEC 6–8	−.551***(.138)	−.368**(.136)	−.183***33.23%	−.399**(.128)	−.188(.125)	−.211***52.78%
Control variables	Yes	Yes		Yes	Yes
Undergraduate institution		Yes			Yes
Undergraduate subject area		Yes			Yes
Undergraduate degree class		Yes			Yes
Pseudo R²	.007	.077		.006	.086

Source : U.K. Labour Force Survey (2014–2020; Office for National Statistics 2021).

p < .05, **p < .01, ***p < .001.

In both Tables 2 and 3 we see a social origin gradient in access to an elite occupation for men and women with a postgraduate qualification. In Table 2, the indirect effects and percentages mediated show that graduate-level stratification meditated around half of the total social origin gap. The remaining direct associations (Model 2) are still quite large but not precisely measured, so they are not statistically significant. The exceptions are for the lowest class among men and the second lowest among women, where a significant direct association persists.

Estimates of graduate-level stratification presented at the bottom of Table 2 show that for men and women, the most significant factor in reaching an elite group was the type of postgraduate degree (PhD, master's, PGCE, or other). People with all degree types had lower chances of having an elite occupation than those with a doctorate, and the chances were exceptionally slim for those with a PGCE degree. For example, the odds of having an elite occupation for male PGCE holders were .11, exp(–2.219), those of male doctorate holders. Given that a PGCE is a pathway to a job as a teacher and teachers are in NS-SEC 2, this result is unsurprising. For institutional selectivity, there is a gap between the highest (global top 20) and lowest (post-1992) tiers among men and between the mid (pre-1992) and top tiers for women. Finally, for both men and women, a postgraduate degree in OSSAH or combined areas lowers the odds of getting into an elite occupation compared to those with a STEM qualification, whereas a LEM degree increases the odds for women, net of other graduate-level educational characteristics. Overall, the factors associated with higher chances of reaching an elite occupation group correspond with stratification points where the distributional disparities by social origin were most pronounced, as shown in Table 1.

In Table 3, graduate-level stratification is replaced by undergraduate-level stratification. This changed the degree of mediation only slightly. Undergraduate-level stratification accounted for almost half of the total social origin differences, but its inclusion rendered the partial association of social origin with the outcome not statistically significant, with the exceptions of NS-SEC 6–8 among men and NS-SEC 3–5 among women. Although it explained slightly less of the gap between the highest and lowest origin among men than did graduate-level stratification, the results overall show that undergraduate-level stratification substantially explains intergenerational occupational association among postgraduate degree holders.

Test of Mediation of Intergenerational Association among the Highest Educated

Having shown that class gaps are partially mediated by both undergraduate and graduate stratification, we now examine which is the more important. Table 4 shows results from six nested logit models that regress occupational destination (attaining an elite occupation) among postgraduate degree holders on social origin with stepwise additions of educational mediators. Model 1 includes social origin and demographic controls, Models 2 to 4 add measures of undergraduate-level stratification, and Models 5 and 6 add measures of graduate-level stratification.⁵ Because the coefficients were rescaled using the linear predictor approach, they can be compared across nested models.

Table 4.

Results from Nested Logistic Regression Models (Linear Predictor Approach) Regressing Having an NS-SEC 1 Occupation on Social Origins, Controlling for Undergraduate- and Graduate-Level Educational Stratification.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Men (N = 3,260)
Social origin (reference = NS-SEC 1)
NS-SEC 2	−.296*(.119)	−.266*(.118)	−.178(.115)	−.171(.113)	−.153(.111)	−.150(.110)
NS-SEC 3–5	−.254*(.110)	−.163(.111)	−.098(.107)	−.115(.107)	−.113(.106)	−.112(.107)
NS-SEC 6–8	−.597***(.149)	−.476**(.150)	−.392**(.147)	−.398**(.147)	−.341*(.145)	−.345*(.146)
Women (N = 4,222)
Social origin (reference = NS-SEC 1)
NS-SEC 2	−.268*(.112)	−.230*(.112)	−.150(.110)	−.139(.110)	−.126(.107)	−.114(.107)
NS-SEC 3–5	−.432***(.112)	−.341**(.112)	−.294**(.109)	−.281*(.109)	−.250*(.105)	−.241*(.104)
NS-SEC 6–8	−.424**(.138)	−.295*(.139)	−.206(.135)	−.193(.135)	−.123(.132)	−.112(.132)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Undergraduate education
Institution		Yes	Yes	Yes	Yes	Yes
Subject			Yes	Yes	Yes	Yes
Degree class				Yes	Yes	Yes
Graduate education
Degree type					Yes	Yes
Institution						Yes
Pseudo R² (men)	.007	.017	.070	.077	.123	.124
Pseudo R² (women)	.006	.015	.080	.086	.140	.141

Source : U.K. Labour Force Survey (2014–2020; Office for National Statistics 2021).

Note : Bootstrapped standard errors are in parentheses. Estimates reported as log odds. Control variables include age, age-squared, ethnicity, and country of birth. NS-SEC = U.K. National Statistics Socio-economic Classification.

p < .05, **p < .01, ***p < .001.

In Model 1, as we already saw, the lower the social origin, the lower the chance of having an elite occupation. Introducing selectivity of the undergraduate institution (Model 2) leads to a substantial decrease in estimates of social origin associations for origins in NS-SEC 3 to 5 (by 36 percent for men and 21 percent for women) and NS-SEC 6 to 8 (by 20 percent for men and 30 percent for women). The estimate for men from NS-SEC 3 to 5 origins is no longer statistically significant (at the p < .05 level). The introduction of undergraduate subject area (Model 3) led to a further decline in the social origin gaps, and the estimates for NS-SEC 2 origins become nonsignificant, but it is clear that institutional selectivity is a more important mediator of class gaps than is choice of subject area.

From Model 3 onward, we see a distinct gender difference. Unlike men, whose gap was largest between the two most distant points on the origins scale, for women, the gap was largest between NS-SEC 1 and NS-SEC 3 to 5 origins. These gaps remain significant in Model 6, when both undergraduate- and graduate-level stratification are held constant, whereas gaps observed among other origins lost their statistical significance when undergraduate-level institutional selectivity and subject area are held constant. From Model 4 onward, the class origin coefficients change only slightly; the only exception is for NS-SEC 3 to 5 among women. Bootstrap estimates of the standard error of the difference in the coefficients for social origins from Models 4 (which controls for all features of undergraduate education) and 6 (which controls for both undergraduate and graduate education) show that these differences fall far short of statistical significance. In other words, conditioning on characteristics of graduate education brings about no further changes in the mediation of class gaps once features of undergraduate education are controlled.

Unequal Paths across and beyond Higher Education and Their Implications

Introducing either undergraduate- or graduate-level stratification removes much of the association between social origins and destinations, and once we condition on undergraduate education, adding graduate education does not increase the degree to which we can mediate the origin–destination association. Referring to Figure 1, this suggests that path e (the direct association between graduate education and attaining an elite occupation, conditional on social origins and undergraduate education) is weak, whereas paths a (the social origins–undergraduate association) and d (the direct association between undergraduate education and attaining an elite occupation, conditional on social origin and graduate education) are strong. Nevertheless, when we do not condition on undergraduate education, graduate education mediates the association as much as undergraduate education does. This indicates that path b (linking undergraduate and graduate education) is also strong because the mediation by postgraduate education when undergraduate education is not in the model is due to the backdoor path $P \leftarrow U \to Y$ . If this interpretation is correct, we should see very strong links between features of undergraduate and graduate education. We examine this in a series of models that relate graduate degree type, institution, and subject area, respectively, to undergraduate institution, subject area, and degree class.

Table 5 reports results from a multinomial logit model in which types of postgraduate degree (doctorate, master's, and other vs. PGCE) are regressed on the three features of undergraduate education. The links between postgraduate degree type and undergraduate education are evident, with particularly strong links with undergraduate degree class. Having attended a more prestigious university for undergraduate study, having studied a STEM subject, and having a first-class degree are all positively related to more prestigious (doctorate or master’s) graduate study.

Table 5.

Results from Multinomial Logistic Regression Models Regressing Graduate-Level Degree Type (PGCE as a Reference) on Undergraduate-Level Educational Stratification.

	Men (N = 3,260)			Women (N = 4,222)
		vs. PGCE			vs. PGCE
	Doctorate	Master’s	Other	Doctorate	Master’s	Other
Undergraduate institution
Global top 20	Reference	Reference	Reference	Reference	Reference	Reference
Russell Group	−1.434***(.295)	−1.124***(.272)	−.661(.347)	−1.562***(.281)	−1.187***(.231)	−.915**(.283)
Pre-1992	−1.532***(.303)	−1.193***(.275)	−.799*(.352)	−1.717***(.287)	−1.193***(.233)	−.824**(.279)
Post-1992	−2.841***(.309)	−1.469***(.269)	−.720*(.337)	−2.453***(.291)	−1.368***(.228)	−.840**(.272)
Undergraduate subject area
STEM	Reference	Reference	Reference	Reference	Reference	Reference
LEM	−1.188**(.405)	1.538***(.323)	1.873***(.339)	−1.787***(.361)	.620**(.219)	1.208***(.230)
OSSAH/COMB	−2.815***(.168)	−1.783***(.130)	−1.669***(.174)	−3.333***(.160)	−1.893***(.113)	−2.046***(.141)
Undergraduate degree class
First	Reference	Reference	Reference	Reference	Reference	Reference
Upper second	−1.522***(.215)	−.557**(.201)	−.214(.282)	−1.546***(.192)	−.748***(.158)	−.622**(.211)
Lower second	−3.478***(.278)	−1.076***(.211)	−.341(.291)	−3.262***(.261)	−1.271***(.169)	−.807***(.223)
Third/pass/other	−3.136***(.285)	−1.246***(.234)	−.081(.309)	−2.241***(.272)	−1.216***(.204)	−.353(.249)
Constant	2.437***(.596)	3.681***(.498)	.801(.671)	2.964***(.507)	3.289***(.378)	1.449**(.490)
Pseudo R²	.173			.150

Source : U.K. Labour Force Survey (2014–2020; Office for National Statistics 2021).

Note : Standard errors are in parentheses. Estimates reported as log odds. Control variables include age, ethnicity, and country of birth. NS-SEC = U.K. National Statistics Socio-economic Classification; STEM = science, technology, engineering, and mathematics; LEM = law, economics, and management; OSSAH = other social sciences, arts, and humanities; COMB = combined subjects; PGCE = postgraduate certificate in education.

p < .05, **p < .01, ***p < .001.

Similarly, as seen in Table 6, graduate institution is also strongly linked to all dimensions of undergraduate education. There is a clear path dependency between undergraduate and graduate institutions. Undergraduate study at a global top 20 university is strongly related to graduate study there, undergraduate education at a global top 20 or Russell Group university is strongly related to graduate study at a Russell Group university, and undergraduate study at a pre-1992 university is strongly linked to graduate study there. Both undergraduate subject area (with STEM graduates most likely to attend more prestigious universities as postgraduates) and degree class (the higher the degree class, the more prestigious the graduate institution) are also strongly linked to graduate institution.

Table 6.

Results from Multinomial Logistic Regression Models Regressing Graduate-Level Institution (Post-1992 as a Reference) on Undergraduate-Level Educational Stratification.

	Men (N = 3,260)			Women (N = 4,222)
		vs. Post-1992			vs. Post-1992
	GT20	RG	Pre-1992	GT20	RG	Pre-1992
Undergraduate institution
Global top 20	Reference	Reference	Reference	Reference	Reference	Reference
Russell Group	−3.166***(.261)	−.034(.256)	−.866**(.271)	−2.777***(.237)	.656**(.244)	−.636**(.230)
Pre-1992	−3.968***(.300)	−1.082***(.267)	.031(.267)	−3.243***(.269)	−.415(.251)	−.066(.223)
Post-1992	−5.180***(.295)	−2.291***(.260)	−1.718***(.263)	−4.373***(.270)	−1.051***(.245)	−1.505***(.223)
Undergraduate subject area
STEM	Reference	Reference	Reference	Reference	Reference	Reference
LEM	−1.106***(.245)	−1.011***(.155)	−.418**(.139)	−1.328***(.344)	−1.075***(.156)	−.566***(.140)
OSSAH/COMB	−.804***(.185)	−.479***(.117)	−.448***(.115)	−.780***(.168)	−.505***(.097)	−.298**(.101)
Undergraduate degree class
First	Reference	Reference	Reference	Reference	Reference	Reference
Upper second	−.871***(.202)	−.392*(.153)	−.056(.155)	−1.147***(.221)	−.370**(.139)	−.172(.143)
Lower second	−1.685***(.262)	−.949***(.173)	−.248(.168)	−1.326***(.266)	−.480**(.153)	−.314*(.156)
Third/pass/other	−2.018***(.317)	−1.128***(.196)	−.766***(.197)	−1.531***(.321)	−.502**(.184)	−.465*(.189)
Constant	2.524***(.570)	1.311**(.442)	.516(.446)	2.832***(.569)	.319(.375)	.092(.371)
Pseudo R²	.180			.128

Source : U.K. Labour Force Survey (2014–2020; Office for National Statistics 2021).

Note : Standard errors are in parentheses. Estimates reported as log odds. Control variables include age, ethnicity, and country of birth. GT20 = global top 20; RG = Russell Group; pre-1992 = pre-1992 university; post-1992 = post-1992 university; STEM = science, technology, engineering, and mathematics; LEM = law, economics, and management; OSSAH = other social sciences, arts, and humanities; COMB = combined subjects.

p < .05, **p < .01, ***p < .001.

By contrast, graduate subject area is not associated with either undergraduate institution or degree class, but, as one might expect, it is very strongly linked to undergraduate subject area (see Table 7). For example, among male postgraduates, the odds of moving from STEM undergraduate to STEM graduate study are 69 times greater than the odds of moving from LEM undergraduate to STEM graduate study (85 times for women). Among female postgraduates, the odds of moving from LEM undergraduate to LEM graduate study are over 400 times greater than the odds of moving from STEM to LEM (300 times for men). Given the nature of graduate education, it is not surprising that undergraduate and postgraduate subject area group are almost always the same.⁶

Table 7.

Results from Multinomial Logistic Regression Models Regressing Postgraduate-Level Subject Areas (OSSAH/COMB as a Reference) on Undergraduate-Level Educational Stratification.

	Men (N = 3,260)		Women (N = 4,222)
	STEM	LEM	STEM	LEM
	vs. OSSAH/COMB		vs. OSSAH/COMB
Undergraduate institution
Global top 20	Reference	Reference	Reference	Reference
Russell Group	.154(.320)	−.194(.428)	.163(.426)	−.002(.581)
Pre-1992	.166(.339)	.095(.455)	.297(.432)	−.156(.613)
Post-1992	−.165(.331)	−.542(.418)	.351(.440)	−.316(.595)
Undergraduate subject area
STEM	Reference	Reference	Reference	Reference
LEM	−4.241***(.480)	5.782***(.466)	−4.437***(.459)	6.030***(.508)
OSSAH/COMB	−6.591***(.252)	−1.281**(.454)	−7.424***(.283)	−1.406**(.512)
Undergraduate degree class
First	Reference	Reference	Reference	Reference
Upper second	.131(.253)	.54(.507)	.051(.264)	−.730*(.318)
Lower second	−.355(.287)	.014(.555)	−.218(.302)	−.623(.402)
Third/pass/other	−.203(.316)	.187(.563)	−.405(.338)	−.739(.545)
Constant	3.497***(.701)	−4.087***(1.120)	3.269***(.751)	−3.606***(1.043)
Pseudo R²	.776		.801

Source : U.K. Labour Force Survey (2014–2020; Office for National Statistics 2021).

Note : Standard errors are in parentheses. Estimates reported as log odds. Control variables include age, ethnicity, and country of birth. STEM = science, technology, engineering, and mathematics; LEM = law, economics, and management; OSSAH = other social sciences, arts, and humanities; COMB = combined subjects.

p < .05, **p < .01, ***p < .001.

Conclusions

We have explored the mechanisms underlying the intergenerational occupational association among men and women postgraduate degree holders in the United Kingdom. The existence of such an association among the highest educated presented a puzzle when it was first described by Torche (2011) in the United States. Recent U.S. studies have argued that graduate-level educational stratification substantially accounts for the strong intergenerational socioeconomic association among this group (Oh and Kim 2020; Torche 2018). We found that in the United Kingdom, the intergenerational occupational association among postgraduate degree holders is mediated equally strongly by undergraduate- and graduate-level education when they are considered separately. Conditioning on either of them accounted for around 40 percent of the origin–destination associations, and the remaining partial associations, although substantively large, were imprecisely estimated. However, once we control for undergraduate stratification, graduate-level stratification adds little or nothing to the mediation of social origin associations.

One contribution of this study is that we identified the paths in higher education that are associated with unequal access to elite occupations among the highest educated. The most significant aspect of graduate education associated with reaching an elite occupational group was the type of degree (Table 2), but this is very strongly related not only to undergraduate degree class but also to undergraduate institution and subject area. The other important dimension of graduate education is subject area, and we find this is very heavily determined by undergraduate subject area (Table 7).

Taken together, these results suggest that insofar as characteristics of education can account for the association between social origin and the chances of access to an elite occupation among postgraduate degree holders, it is characteristics of undergraduate, rather than graduate, education that matter, and the selection of those who continue from a primary to a secondary degree is more important than what occurs during graduate education. Graduate-level stratification, which U.S. studies have found to be important, is, in the U.K. case, epiphenomenal because it so strongly reflects patterns of undergraduate stratification among those who choose (and who qualify) to continue to graduate education.

Our study builds on previous work in two main ways. First, even though a strong intergenerational association among the highest educated has been reported in several other countries, studies that attempt to explain it have disproportionately considered the United States. Our analysis deals with the United Kingdom and so examines this question in a quite different educational system. Second, earlier studies have not examined comprehensively the paths from undergraduate to graduate education as a bridge that connects social origin and postgraduate socioeconomic outcomes, something we sought to do. Had we considered the mediation of class differences in attainment only by graduate education, we would have drawn erroneous conclusions, biased in the way we described in our discussion of Figure 1.

The differences between the U.S. and U.K. tertiary educational systems are important in understanding our results and placing them in context. Compared to the United States, graduate education in the United Kingdom features less professional education and relatively more academic education. It is also much more tightly coupled to undergraduate education, especially in field of study, and access to graduate education is, for the most part, heavily dependent on undergraduate degree class. We believe our results may be less likely to hold in more loosely coupled systems but are probably characteristic of more tightly coupled ones, found widely in Europe and in many other countries. Future work should extend this research beyond the United States or Europe to countries, such as China, where a high proportion of graduates leave to study in foreign countries, or South Korea, which has only recently introduced professional graduate schools (notably in law and medicine).

Although we focused on postgraduate degree holders, our study sheds light on a long-studied subject in stratification research: the equalizing power of college education. Almost all inequality among postgraduate degree holders in the United Kingdom has its roots in undergraduate-level education. Our findings thus cast doubt on the college-as-a-great-equalizer argument. As more college graduates pursue graduate education, it has become more problematic to draw conclusions by studying terminal bachelor's degree holders, or college graduates as a whole, without distinguishing the level of degree. This study suggests future research will need to take greater account of educational transitions in higher education in the study of intergenerational mobility and stratification among highly educated people.

Footnotes

The authors would like to thank the anonymous referees for their helpful comments. We also received insightful comments and suggestions from John Ermisch,Jan O. Jonsson,Christiaan Monden,Alice Sullivan,and participants at the ASA RC28 meeting at the University of Turku in 2021 on a previous version of this article. We thank members of the Office for National Statistics (ONS) and the UK Data Service (UKDS) for their prompt and thorough help with our data use. The use of these data does not imply the endorsement of the data owner or the UK Data Service in relation to the interpretation or analysis of the data.

Research ethics

This study uses secondary (restricted) data from the UK Office of National Statistics. It has been reviewed and approved by the Departmental Research Ethics Committee,Department of Sociology,University of Oxford.

ORCID iDs

Jung In

Richard Breen

Author Biographies

Jung In is a research fellow and postdoctoral researcher of Nuffield College,University of Oxford. Her research explores how education,family,and geographic mobility affect socioeconomic inequality and intergenerational mobility using quantitative methods. She received her DPhil from the University of Oxford.

Richard Breen is professor of sociology and fellow of Nuffield College,University of Oxford. His research interests are inequality,intergenerational mobility,and quantitative methods. His recent publications have appeared in Annual Review of Sociology,Sociological Methods and Research,and European Sociological Review . Together with Walter Müller,he edited Education and Intergenerational Social Mobility in Europe and the United States (Stanford University Press,2020). He is a fellow of the British Academy and a member of the Royal Irish Academy.

References

Autor

David H.

2014. “Skills, Education, and the Rise of Earnings Inequality among the ‘Other 99 Percent.’”Science 344(6186):843–51.

Berger

Mark C.

1988. “Predicted Future Earnings and Choice of College Major.”Industrial & Labor Relations Review 41(3):418–29.

Boliver

Vikki

. 2013. “How Fair Is Access to More Prestigious UK Universities?” The British Journal of Sociology 64(2):344–64.

Bouchet-Valat

Milan

. 2022. “General Marginal-Free Association Indices for Contingency Tables: From the Altham Index to the Intrinsic Association Coefficient.”Sociological Methods and Research 51(1):203–31.

Bowen

William G.

Bok

Derek

. 1998. The Shape of the River. Long-Term Consequences of Considering Race in College and University Admissions. Princeton, NJ: Princeton University Press.

Breen

Richard

. 2010. “Educational Expansion and Social Mobility in the 20th Century.”Social Forces 89(2):365–88.

Breen

Richard

Jonsson

Jan O.

. 2007. “Explaining Change in Social Fluidity: Educational Equalization and Educational Expansion in Twentieth-Century Sweden.”American Journal of Sociology 112(6):1775–810.

Breen

Richard

Karlson

Kristian Bernt

Holm

Anders

. 2013. “Total, Direct, and Indirect Effects in Logit and Probit Models.”Sociological Methods & Research 42(2):164–91.

Breen

Richard

Karlson

Kristian Bernt

Holm

Anders

. 2021. “A Note on a Reformulation of the KHB Method.”Sociological Methods & Research 50(2):901–12.

10.

Breen

Richard

Luijkx

Ruud

. 2007. “Social Mobility and Education: A Comparative Analysis of Period and Cohort Trends in Britain and Germany.” Pp.102–24 in From Origin to Destination, edited by Scherer

Pollak

Otte

Gangl

. Frankfurt, Germany: Campus Verlag.

11.

Britton

Jack

Buscha

Franz

Dickson

Matt

van der Erve

Laura

Vignoles

Anna

Walker

Ian

Waltmann

Ben

Zhu

. 2020. The Earnings Returns to Postgraduate Degrees in the UK. London, UK: The Institute for Fiscal Studies. https://ifs.org.uk/publications/earnings-returns-postgraduate-degrees-uk#:~:text=By%20institution%2C%20returns%20for%20masters,of%2020%25%20at%20the%20top.

12.

Britton

Jack

van der Erve

Laura

. 2020. Family Background and Access to Postgraduate Degrees. London, UK: The Institute for Fiscal Studies. https://ifs.org.uk/uploads/Family-background-and-access-to-postgraduate-degrees__.pdf.

13.

Conlon

Gavan

Patrignani

Pietro

. 2011. The Returns to Higher Education Qualifications. London, UK: Department for Business Innovation & Skills. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/32419/11-973-returns-to-higher-education-qualifications.pdf.

14.

Crawford

Claire

Gregg

Paul

Macmillan

Lindsey

Vignoles

Anna

Wyness

Gill

. 2016. “Higher Education, Career Opportunities, and Intergenerational Inequality.”Oxford Review of Economic Policy 32(4):553–75.

15.

Davies

Scott

Guppy

Neil

. 1997. “Fields of Study, College Selectivity, and Student Inequalities in Higher Education.”Social Forces 75(4):1417–38.

16.

Day

Jennifer Cheeseman

Newburger

Eric C.

. 2002. The Big Payoff: Educational Attainment and Synthetic Estimates of Work-Life Earnings. Special Studies. Current Population Reports. Washington, DC: U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau. https://www.census.gov/content/dam/Census/library/publications/2002/demo/p23-210.pdf.

17.

Eide

Eric

Brewer

Dominic J.

Ehrenberg

Ronald G.

. 1998. “Does It Pay to Attend an Elite Private College? Evidence on the Effects of Undergraduate College Quality on Graduate School Attendance.”Economics of Education Review 17(4):371–76.

18.

Eide

Eric

Waehrer

Geetha

. 1998. “The Role of the Option Value of College Attendance in College Major Choice.”Economics of Education Review 17(1):73–82.

19.

Ethington

Corinna A.

Smart

John C.

. 1986. “Persistence to Graduate Education.”Research in Higher Education 24(3):287–303.

20.

Falcon

Julie

Bataille

Pierre

. 2018. “Equalization or Reproduction? Long-Term Trends in the Intergenerational Transmission of Advantages in Higher Education in France.”European Sociological Review 34(4):335–47.

21.

Fisher

Jonathan

Johnson

David S.

Smeeding

Timothy S.

Jeffrey

Thompson

. 2018. The Demography of Inequality: Income, Wealth and Consumption, 1989-2016. Report 18-890. Ann Arbor, MI: Population Studies Center, University of Michigan, Institute for Social Research. https://www.psc.isr.umich.edu/pubs/rr18-89027dc.pdf?i=988537410178203311550019081&f=rr18-890.pdf.

22.

Gerber

Theodore P.

Cheung

Sin Yi

. 2008. “Horizontal Stratification in Postsecondary Education: Forms, Explanations, and Implications.”Annual Review of Sociology 34:299–318.

23.

Goyette

Kimberly A.

Mullen

Ann L.

. 2006. “Who Studies the Arts and Sciences? Social Background and the Choice and Consequences of Undergraduate Field of Study.”The Journal of Higher Education 77(3):497–538.

24.

Grogger

Jeff

Eide

Eric

. 1995. “Changes in College Skills and the Rise in the College Wage Premium.”The Journal of Human Resources 30(2):280–310.

25.

Grubb

W. Norton

. 1993. “The Varied Economic Returns to Postsecondary Education: New Evidence from the Class of 1972.”The Journal of Human Resources 28(2):365–82.

26.

Gugushvili

Alexi

Bukodi

Erzsébet

Goldthorpe

John H.

. 2017. “The Direct Effect of Social Origins on Social Mobility Chances: ‘Glass Floor’ and ‘Glass Ceilings’ in Britain.”European Sociological Review 33(2):305–16.

27.

Hauser

Robert M.

Logan

John Allen

. 1992. “How Not to Measure Intergenerational Occupational Persistence.”American Journal of Sociology 97(6):1689–711.

28.

Hearn

James C.

1991. “Academic and Nonacademic Influences on the College Destinations of 1980 High School Graduates.”Sociology of Education 64(3):158–71.

29.

Higher Education Funding Council for England. 2013. Postgraduate Education in England and Northern Ireland. Bristol, UK: HEFCE.

30.

Hout

Michael

. 1984. “Status, Autonomy, and Training in Occupational Mobility.”American Journal of Sociology 89(6):1379–409.

31.

Hout

Michael

. 1988. “More Universalism, Less Structural Mobility: The American Occupational Structure in the 1980s.”American Journal of Sociology 93(6):1358–400.

32.

Karen

David

. 2002. “Changes in Access to Higher Education in the United States: 1980-1992.”Sociology of Education 75(3):191–210.

33.

Karlson

Kristian Bernt

. 2011. “Multiple Paths in Educational Transitions: A Multinomial Transition Model with Unobserved Heterogeneity.”Research in Social Stratification and Mobility 29(3):323–41.

34.

Karlson

Kristian Bernt

. 2019. “College as Equalizer? Testing the Selectivity Hypothesis.”Social Science Research 80:216–29.

35.

Karlson

Kristian Bernt

Holm

Anders

Breen

Richard

. 2012. “Comparing Regression Coefficients between Same-Sample Nested Models Using Logit and Probit: A New Method.”Sociological Methodology 42(1):286–313.

36.

Keister

Lisa A.

2014. “The One Percent.”Annual Review of Sociology 40:347–67.

37.

Kim

Chang Hwan

Tamborini

Christopher R.

Sakamoto

Arthur

. 2015. “Field of Study in College and Lifetime Earnings in the United States.”Sociology of Education 88(4):320–39.

38.

Kohler

Ulrich

Karlson

Kristian Bernt

Holm

Anders

. 2011. “Comparing Coefficients of Nested Nonlinear Probability Models.”The Stata Journal 11(3):420–38.

39.

Lang

Dwight

. 1987. “Stratification and Prestige Hierarchies in Graduate and Professional Education.”Sociological Inquiry 57(1):12–31.

40.

Lemieux

Thomas

. 2008. “The Changing Nature of Wage Inequality.”Journal of Population Economics 21(1):21–48.

41.

Lindley

Joanne

Machin

Stephen

. 2013. The Postgraduate Premium: Revisiting Trends in Social Mobility and Educational Inequalities in Britain and America. London, UK: The Sutton Trust.

42.

Jennifer

Pender

Matea

Welch.

Meredith

2016. Education Pays 2016: The Benefits of Higher Education for Individuals and Society. Trends in Higher Education Series. New York, NY: College Board. https://files.eric.ed.gov/fulltext/ED572548.pdf.

43.

Mateos-González

José Luis

Wakeling

Paul

. 2022. “Exploring Socioeconomic Inequalities and Access to Elite Postgraduate Education among English Graduates.”Higher Education 83(3):673–94.

44.

Mullen

Ann L.

Goyette

Kimberly A.

Soares

Joseph A.

. 2003. “Who Goes to Graduate School? Social and Academic Correlates of Educational Continuation after College.”Sociology of Education 76(2):143–69.

45.

Office for National Statistics, Social Survey Division, Northern Ireland Statistics and Research Agency, Central Survey Unit. 2021. Quarterly Labour Force Survey, 1992-2021: Secure Access. [data collection]. 28th Edition. UK Data Service. SN: 6727, http://doi.org/10.5255/UKDA-SN-6727-29.

46.

Byeongdon

Kim

Chang Hwan

. 2020. “Broken Promise of College? New Educational Sorting Mechanisms for Intergenerational Association in the 21st Century.”Social Science Research 86(2020):102375. doi:10.1016/j.ssresearch.2019.102375.

47.

Pfeffer

Fabian T.

Hertel

Florian R.

. 2015. “How Has Educational Expansion Shaped Social Mobility Trends in the United States?” Social Forces 94(1):143–80.

48.

Posselt

Julie R.

Grodsky

Eric

. 2017. “Graduate Education and Social Stratification.”Annual Review of Sociology 43(1):353–78.

49.

Rumberger

Russell W.

Thomas

Scott L.

. 1993. “The Economic Returns to College Major, Quality and Performance: A Multilevel Analysis of Recent Graduates.”Economics of Education Review 12(1):1–19.

50.

Scott

Peter

. 2020. Discussion Paper: Access to Postgraduate Study: Representation and Destinations. Edinburgh, UK: The Scottish Government. https://www.gov.scot/binaries/content/documents/govscot/publications/independent-report/2020/01/commissioner-fair-access-discussion-paper-access-postgraduate-study-representation-destinations/documents/discussion-paper-access-postgraduate-study-representation-destinations/discussion-paper-access-postgraduate-study-representation-destinations/govscot%3Adocument/discussion-paper-access-postgraduate-study-representation-destinations.pdf.

51.

Seaman

Shaun R.

White

Ian R.

. 2013. “Review of Inverse Probability Weighting for Dealing with Missing Data.”Statistical Methods in Medical Research 22(3):278–95.

52.

Stevens

Mitchell L.

Armstrong

Elizabeth A.

Arum

Richard

. 2008. “Sieve, Incubator, Temple, Hub: Empirical and Theoretical Advances in the Sociology of Higher Education.”Annual Review of Sociology 34(1):127–51.

53.

Sullivan

Alice

Brown

Matthew

. 2015. “Vocabulary from Adolescence to Middle Age.”Longitudinal and Life Course Studies 6(2):173–89.

54.

Sullivan

Alice

Parsons

Samantha

Green

Francis

Wiggins

Richard D.

Ploubidis

George

. 2018a. “Elite Universities, Fields of Study and Top Salaries: Which Degree Will Make You Rich?” British Educational Research Journal 44(4):663–80.

55.

Sullivan

Alice

Parsons

Samantha

Green

Francis

Wiggins

Richard D.

Ploubidis

George

. 2018b. “The Path from Social Origins to Top Jobs: Social Reproduction via Education.”The British Journal of Sociology 69(3):776–98.

56.

Tinto

Vincent

. 1980. “College Origins and Patterns of Status Attainment: Schooling among Professional and Business-Managerial Occupations.”Sociology of Work and Occupations 7(4):457–86.

57.

Torche

Florencia

. 2011. “Is a College Degree Still the Great Equalizer? Intergenerational Mobility across Levels of Schooling in the United States.”American Journal of Sociology 117(3):763–807.

58.

Torche

Florencia

. 2018. “Intergenerational Mobility at the Top of the Educational Distribution.”Sociology of Education 91(4):266–89.

59.

Vallet

Louis-André

. 2004. “Change in Intergenerational Class Mobility in France from the 1970s to the 1990s and Its Explanation: An Analysis Following the CASMIN Approach.” Pp. 115–47 in Social Mobility in Europe, edited by Breen

. Oxford, UK: Oxford University Press.

60.

Valletta

Rob

. 2015. “Higher Education, Wages, and Polarization.”Federal Reserve Bank of San Francisco Economic Letter (02):1–5.

61.

Wakeling

Paul

. 2017. “A Glass Half Full? Social Class and Access to Postgraduate Study.” Pp. 167–89 in Higher Education and Social Inequalities: University Admissions, Experiences, and Outcomes, edited by Waller

Ingram

Ward

M. R. M.

. Abingdon, UK: Routledge.

62.

Wakeling

Paul

Laurison

Daniel

. 2017. “Are Postgraduate Qualifications the ‘New Frontier of Social Mobility’?” British Journal of Sociology 68(3):533–55.

63.

Wakeling

Paul

Mateos-González

José Luis

. 2021. Inequality in the Highest Degree. London, UK: The Sutton Trust.

64.

Wakeling

Paul

Savage.

Mike

2015. “Elite Universities, Elite Schooling and Reproduction in Britain.” Pp. 169–84 in World Yearbook of Education 2015: Elites, Privilege and Excellence: The National and Global Redefinition of Educational Advantage, edited by Van Zanten

Ball

S. J.

Darchy- Koechlin

. Abingdon, UK: Routledge.

65.

Walker

Ian

Zhu

. 2011. “Differences by Degree: Evidence of the Net Financial Rates of Return to Undergraduate Study for England and Wales.”Economics of Education Review 30(6):1177–86.

66.

Wooldridge

Jeffrey M.

2007. “Inverse Probability Weighted Estimation for General Missing Data Problems.”Journal of Econometrics 141(2):1281–301.

67.

Zhang

Liang

. 2005. “Advance to Graduate Education: The Effect of College Quality and Undergraduate Majors.”The Review of Higher Education 28(3):313–38.

68.

Zhang

Liang

. 2012. Does Quality Pay? Benefits of Attending a High-Cost, Prestigious College. Abingdon, UK: Routledge.

69.

Zhou

Xiang

. 2019. “Equalization or Selection? Reassessing the ‘Meritocratic Power’ of a College Degree in Intergenerational Income Mobility.”American Sociological Review 84(3):459–85.