Introduction

From the 1980s, the previous strong interest of social scientists in the formation and functioning of elites tended to fall away. But, of late, research into elites, political elites included, has shown a notable revival. In the British case, much new work on political elites has focused—as indeed did work in the last century—on questions of what has become known as descriptive representation (Pitkin 1967). That is, on questions of how far members of the political elite are representative, in their social characteristics and affiliations, of the population at large or, more specifically, of the electorate. However, differences show up between earlier and more recent work both in the way in which the political elite is understood and in the particular issues of representation that are of main concern.

In the earlier work, the political elite was, for the most part, equated with the members of cabinets (see, e.g., Laski 1928; Guttsman 1963; Johnson 1973; Burch and Moran 1985). Interest then focused on how far in their social composition, and especially in their social class composition, cabinets were representative of the electorate. And, given that this was not the case, the issues that were raised were primarily ones of political principle—of how far such a situation was compatible with ideas of representative democracy. In mostly later work, the political elite has been understood more widely, as members of parliament, and it is on their social composition and representativeness that interest then centres (see, e.g., Mellors 1978; Borthwick et al. 1991; Baker et al. 1995; Hill 2013; Lamprinakou et al. 2017; O’Grady 2019). Further, this interest has extended beyond the question of how far representativeness is lacking in an objective sense and the issues of principle that may arise. What has become of greater concern are questions of how far representativeness is felt to be lacking in different groups within the electorate and of what follows for individuals’ views on politics and their political participation.

In the present paper, we attempt to combine to some extent these two perspectives. For reasons we explain, we maintain the focus on the social composition of cabinets, rather than of MPs, although we do widen our conception of the political elite to include members of shadow cabinets. In the first section of the paper, we briefly review the results of the main twentieth-century studies of the social composition of UK cabinets. In the second section, we describe the research that we have undertaken with the aim of updating, and in certain respects extending, these earlier findings. In the third section, we present the findings of our research, with reference to two main questions. First, how far have there been changes in the class and educational composition of the UK political elite, as we would define it, and with what effects on elite representativeness? And, second, how far have such changes been associated with convergence in the class backgrounds and the educational careers of the Conservative and Labour components of the political elite? In the fourth section of the paper, we then go on to consider the implications of our findings for questions of representation. How do they relate to current discussion of the effects of the representativeness of political elites, or of its lack, on political attitudes and behaviour within the population at large?

Twentieth century studies

Laski’s (1928) pioneering study of the social composition of British cabinets was based on the biographies of the 306 persons who had held cabinet positions between 1801 and 1924. Laski sought to ‘measure the way in which the changes in the social structure of English [sic] social life are reflected in the choice of those responsible for the nation’s effective governance.’ More specifically, he was concerned with the extent to which the progressive extension of the franchise from 1832 onwards had led to ‘a widening of the area from which cabinet members are chosen’ (1928, pp. 12–13). His conclusion was that movement in this direction had been slight. Only by the fifth of six periods that he distinguishes, 1906–1916, had the number of aristocrats in cabinets become fewer than the number of commoners, and only in the sixth, 1917–1924, with the first Labour government, did any working-class representation appear. Conservative and Liberal cabinets differed only marginally in their composition—the former tending to have somewhat more aristocrats and Etonians and Harrovians. In short, political democracy had ‘developed very imperfectly’ (1928, p. 30).

The next study of importance was that of Guttsman (1963), which covered the 294 persons who were members of cabinets between the second Reform Act of 1868 and 1955. Guttsman’s central concern was the same as Laski’s. That is, with how far the creation of an electoral democracy had been matched by the democratisation of the political elite. He focuses on the social origins of cabinet members, using a threefold classification of aristocratic, middle-class and working-class.Footnote 1 The conclusion he reaches is much the same as Laski’s. He observes (1963, p. 97) that while ‘the gradual broadening of democracy after 1868 … did not stop short of the composition of the political elite’, change in this regard was still very limited. Liberal cabinets came to contain fewer aristocrats and more men of middle-class, entrepreneurial or professional, origins who had not attended Eton or Harrow. But the number of cabinet members who were of working-class origins, even in Labour cabinets, remained disappointingly low. ‘The aspiring masses achieved political power if conceived electorally. They achieved political maturity, as manifested in their organisations, but given the weight and strength of these factors, they failed, on the whole, to launch a significant number of their representatives into the political elite’ (1963, p. 98).

A third study that can be regarded as falling in the same series is that of Johnson (1973), who extends Guttsman’s analyses up to 1972. Johnson accepts (1973, pp. 35–36) that the social composition of cabinets has never approached a situation in which, in terms of their social origins, members ‘were proportionately representative of the social classes making up the electorate’ and that ‘The “lag” in the translation of electoral power into elite representation is very striking’. Moreover, he observes that some actual deterioration in this regard occurred in that the cabinets of the Labour governments of 1945–1951 and 1964–1970 became ‘almost homogenously middle class’ (following Guttsman’s threefold division). If this trend were to continue, Johnson concludes (1974, p. 69) ‘we shall be again in the situation which last obtained in the late nineteenth century when the two major parties, led by homogenously middle and upper class elites, vied for the votes of a working class which was itself effectively excluded from political participation or representation at elite level.’

A related question that Johnson introduces is that of how far this growing homogeneity of the political elite implies a ‘convergence’ in the social composition of Labour and Conservative cabinets, brought about by selection becoming, in both cases alike, increasingly ‘meritocratic’: that is, based on education and formal qualifications. Such selection, Johnson maintains, is clearly becoming more important on the Labour side, as seen in the growing number of cabinet members who have been upwardly mobile, through education, from relatively disadvantaged class origins. Certain conspicuous examples of such mobility—Edward Heath and Margaret Thatcher—can also be found on the Conservative side, but Johnson thinks it still questionable how typical they might be.

One further study of note is that of Burch and Moran (1985), chiefly in that these authors take up Johnson’s question of the extent of party convergence in the social composition of cabinets. Analysing the social backgrounds of 153 cabinet members over the period 1955–1984, they reach the conclusion that little such convergence in fact occurred. In Conservative cabinets, more members came from public—i.e. private—schools in 1955–1984 than did in 1916–1955, and there was no advance of state-school educated meritocrats—Heath and Thatcher were not typical—nor of self-made men or women of humble origins. Labour cabinets remained very different in composition. Although there was an increase in members with a university education, including at Oxford and Cambridge, there was a decline in the proportion privately educated and the proportion from working-class origins remained fairly stable. Working-class access to the political elite became, in other words, a matter of inter- rather than of intragenerational mobility. Even if not closely representative of the electorate, Labour cabinets could still be regarded as more representative than Conservative cabinets.

With Burch and Moran, what could be regarded as the tradition of studies of the social composition of UK cabinets would appear to peter out. What was, in sum, established was that over the nineteenth and twentieth centuries the lack of descriptive representation of Cabinets diminished only very slowly and to a quite limited extent—the main change being a decline in the aristocratic element in their composition, especially on the Conservative side. What was left undecided was how far this change, together with, on the Labour side, a decline in working-class representation, was leading to a growing similarity across the party divide in the social composition of the political elite, in that this was becoming increasingly made up of ‘middle class’ university graduates.

Our research

We follow the authors of the studies reviewed above in focusing on the social composition of cabinets rather than of the body of MPs. We believe that MPs are better regarded not as constituting the political elite itself but rather as the ‘pool’ from which the elite is drawn (on the distinction between elite and pool, see Bukodi and Goldthorpe 2021). We do, however, widen our conception of the political elite to include members of shadow cabinets. The cabinet is the prime decision-making body within the British political system, but its members can also be regarded as constituting the leadership group of the party in power. Correspondingly, the members of shadow cabinets can be regarded as constituting the leadership group of the main party in opposition. In other words, our conception of the political elite in the UK is one that brings together those individuals who at any one time form the leadership groups of the two main parties, the Conservatives and Labour.

Our aim then is to extend the historical continuity of earlier work by collecting biographical information of interest on all individuals who were members of cabinets or of shadow cabinets from the Attlee administration of 1945 through to the Johnson administration as of September 2021. From Table 1, it can be seen that these individuals—our target population—totalled 556, of whom 260 were Conservatives and 288 Labour. The remaining eight were Liberals who were cabinet members during the 2010–2015 Conservative–Liberal coalition government. They are left out of our subsequent analyses. Table 1 also shows the distribution of the 556 individuals over six periods in which, with the exception of 1964–1979, either the Conservatives or Labour were politically dominant. Many individuals were of course, at different times, cabinet or shadow cabinet members, so that we have in total 372 instances of the former membership and 373 or the latter, with cabinet membership being the more common on the Conservative side and shadow cabinet membership on the Labour side.

Table 1 Size (N) of political elite by period and party
Full size table

Table 2 shows the extent to which we were successful in collecting the information we sought on our target population. Overall, it can be seen that we achieved an 86% coverage. We do slightly better with the Conservative leadership than with the Labour, and rather worse for the elite in general for the most recent period, 2010–2021, than for earlier periods—this reflecting the well-known fact from elite studies that it is easier to obtain information on the dead than on the living.Footnote 2 Information is most often lacking on parents’ occupation but still in less than 10% of all cases.

Table 2 Extent of coverage of our target population of political elite on different characteristics, by period and party
Full size table

As regards the sources from which our information comes, full details are given in Appendix 1. Standard works of reference, in particular the Dictionary of National Biography and Mortimer and Blick (2018) are heavily drawn on, but newspaper profiles and obituaries and web searches are also important, especially as regards parental occupation and, for those who had attended university, their degree subject. All living elite members for whom full information could not otherwise be obtained, numbering 144, were sent a relevant email questionnaire, which was completed by 44—a 31% response rate.

Two questions might be raised. First, it could be asked whether the source of information could affect its content. In particular, it has been suggested that leading politicians may seek to play down their social origins and their own occupational level before entering parliament, and especially so in the case of Labour members, anxious to demonstrate their working-class credentials. Results from statistical modelling given in Appendix 2 show only limited support for this possibility. If information gained from works of reference is set in comparison with that more likely to have come directly from elite members, some tendency is suggested for Labour members if anything to upgrade their class origins, although perhaps downgrading the class of their employment before they became MPs.

Second, arising from our decision to include shadow cabinet members in the political elite, it could be asked whether those individuals who have served in cabinets, as well perhaps as in shadow cabinets, differ from those who have been only shadow cabinet members. The latter number 196, or 35%, of all but the proportion is higher on the Labour side (48%) than on the Conservative (22%). In Appendix 3, a series of odds ratios are reported which show that differences between the two groupings are in general rather slight and fall into no very obvious pattern. Those who have been members of Conservative cabinets are somewhat more likely than those who have been included only in shadow cabinets to have been in higher managerial and professional positions, rather than others, before entering parliament and—the most striking difference to emerge—are more likely to have studied Philosophy, Politics and Economics (PPE) at Oxford, rather than taking any other degree course.

In the light of the foregoing, we feel able to take those elite members for whom we do have the bulk of the information that we have sought as being representative of our target population in regard to the attributes in question, with the possibility of only very minor, if any, biases occurring. We do not therefore apply tests of significance in our statistical analyses.

Results

We present our results by addressing two sets of different, though obviously related, questions.

  1. (1)

    How far, over the period we cover, have there been changes in the social class composition of the political elite, specifically in terms of the class origins of its members and of their own class positions before becoming MPs? How far are there differences in these respects between the Conservative and Labour components of the elite? And how far have any changes in the composition of the elite or in its party components led to greater representativeness in regard to the electorate?

  2. (2)

    How far over this period have the Conservative and Labour components of the elite tended to converge in terms of the class backgrounds of their members,Footnote 3 and how far is any convergence related to their representativeness? How far does any convergence that is apparent in class backgrounds also show up at a more detailed level when various divisions within classes are made? And, since education has been regarded as a main source of convergence, is any convergence evident in the educational careers of Conservative and Labour elite members?Footnote 4

Changes in the social class composition of the political elite and representativeness

In all of our empirical analyses, we use a modified version of the National Statistics Socio-Economic Classification (Office of National Statistics 2005) as shown in Table 3. For some purposes, we extend the division that is made within NS-SEC Class 1 between its managerial and professional components also to Class 2 but Classes 3 and 5 and Classes 6 and 7 are collapsed.

Table 3 Modified version of National Statistics Socio-Economic Classification (NS-SEC)
Full size table

In earlier research, the distinction was not always clearly or consistently made between the composition of the elite in terms of the class origins of its members and in terms of the class positions that they themselves held before becoming MPs. We consider both in turn.

In Fig. 1 we show distributions of members of the political elite, as we have defined it, by their class origins—i.e. by the social class of their parents at the time of the elite member’s adolescence, ideally around age 14–15Footnote 5—and also by period and party. We characterise these distributions by the index of homogeneity (IH) of Blau and Duncan (1967, pp. 52–53). The percentage distributions shown in the figure are treated as the set of probabilities of individuals coming from each class of origin, and the squares of these probabilities are then summed to give the probability that any two individuals chosen at random will come from the same class. The maximum value the IH can take is 1, where—to take the present case—all individuals have the same class of origin; and the minimum value, indicating the greatest possible degree of heterogeneity given the number of classes distinguished, is the reciprocal of this number, in the present case, 1/5 = 0.20.

Fig. 1
figure 1

Distribution (%) of political elite by parental class, and index of homogenity (IH), separately by period and party

Full size image

From the upper panel of the figure, it can be seen that, considering the political elite as a whole, homogeneity in class origins peaked in the 1951–1964 period of Conservative dominance and thereafter declines, especially so from the 1997–2010 period. The proportion of the elite coming from Class 1—i.e. higher managerial and professional—origins falls away while the proportion coming from other class origins tends to increase. Given the shift in the IH from 50 in the 1951–1964 period to 28 in the 2010–2021 period, one could say that if two members of the elite were to meet by chance at the earlier time, it would be an even bet that they were of similar class origins, while by the later time the odds against this would be more than 3 to 1.

However, these findings for the political elite as a whole are rather misleading, as becomes apparent from the two lower panels of the figure in which the Conservative and Labour components of the elite are treated separately. What can be seen is that the Labour leadership is always less homogeneous in its class origins than the Conservative leadership, and further that its IH changes little over time. In particular, it may be noted that after a decline between the 1945–1951 and 1951–1964 periods, the proportion of individuals of Classes 6 and 7—or working-class—origins is fairly stable (26–28%), in line with Burch and Moran’s (1985) earlier analysis. That is to say, as regards class origins, no overall decline in working class representation has occurred in the Labour leadership.

It is then among the Conservative leadership that the decline in homogeneity of class origins very largely occurs. Onwards from the 1979–1997 period of Conservative political dominance, the proportion on the Conservative side of the elite of individuals of Class 1 origins, though still high, steadily falls. Correspondingly, there is some increase in the proportion of those of Class 2—i.e. lower managerial and professional—origins and, more importantly, of those Class 4 origins: that is, of individuals coming from the families of small employers and the self-employed. This latter change is of particular interest in the light of the discussion in earlier research of whether the cases of Heath and Thatcher—and, it could be added, of John Major—are indicative of some more general trend of upward mobility from such origins into the Conservative leadership. In this case, contrary to the earlier findings of Burch and Moran (1985), our evidence does in fact point to a trend of this kind, such that by the 2010–2021 period of Conservative rule, a fifth of the leadership were of what might be described as ‘petty bourgeois’ descent. Detailed examination reveals that the individuals in question are most frequently the sons—far less often the daughters—of men owning small manufacturing or construction businesses, farms, garages, dairies, shops, hotels and restaurants. With the Labour leadership, in contrast, it is in the proportion of individuals of Class 4 origins—being in this case mostly the sons of self-employed tradesmen—that the sharpest decline occurs.

We move on to examine how far changes in the social composition of the political elite in terms of the class origins of its members has affected its descriptive representativeness. For this purpose, we switch to a birth cohort perspective and in Fig. 2 set elite distributions of class origins against our best estimates for such distributions for the British population at large (see Appendix 5). In most previous discussion of descriptive representation, change in the class composition of the population, though obviously relevant, has not been taken into account. We measure differences between the pairs of distributions by the index of dissimilarity (ID), which shows the percentage of individuals that would need to move from one distribution to the other in order to make the two distributions identical, and thus ranges from 0 to 100.

Fig. 2
figure 2

Comparing distributions of parental class (%) between population at large(a) and political elite, in three birth cohorts

Full size image

It can be seen that over the three birth cohorts, the political elite, taken as a whole, does show a tendency to become more representative, to a greater degree than was found in earlier research (though now with shadow cabinet members being included). This change is especially marked as between the two most recent birth cohorts. Nonetheless, the ID of 34% for those born from 1950 onwards does still indicate a substantial lack of representativeness, which, as can be seen, primarily results from the strong over-representation in the elite of individuals of Class 1 origins and the under-representation of those of Classes 6 and 7 origins.

It is, though, once again the case that a more revealing picture emerges when we consider the Conservative and Labour leaderships separately. The relevant IDs indicate that, across the birth cohorts, the Labour leadership, in being less homogenous than the Conservative leadership, is always the more representative. But it is change on the Conservative side—the previously noted decline in the proportion of its members who are of Class 1 origins and the increase in those of Class 2 origins—that primarily drives the increase in elite representativeness overall. And the increase in those of Class 4 origins on the Conservative side means that this class is now in fact over-represented in the party leadership.

Turning to the composition of the political elite in terms of its members’ own class positions before entering parliament, we show relevant distributions in Fig. 3, although with Classes 3–7 now being collapsed in view of the small numbers involved. Overall, the IHs indicate no trend and are always high on account of the large majority of the elite having been previously employed in Class 1, higher managerial or professional, positions.

Fig. 3
figure 3

Distribution (%) of parental elite by own class before entry into parliament, and index of homogenity (IH), separately by period and party

Full size image

Significant differences do, however, still show up between the Conservative and Labour components of the elite. On the Conservative side, previous employment in Class 1 positions is, over all the periods covered, preponderant—the most frequently reported occupations being those of barrister, managing director or chief executive officer, and also for the later periods, entrepreneur or financier. On the Labour side, social homogeneity is quite consistently lower, although showing a tendency to increase, and especially from the 1997–2010, New Labour, period, with the proportion of those previously employed in Class 1 positions reaching a new high while the proportion previously employed in less advantaged, Classes 3–7, positions more or less dwindles away. The previous occupations most frequently reported were, over earlier periods, those of trade union official, barrister and academic but, post-1997, those of political advisor, organiser or researcher come to the fore.

A significant contrast thus arises. In the case of the composition of the elite in terms of class origins change was most evident within the Conservative leadership in the form of decreasing homogeneity. But in the case of composition in terms of class position before entry into parliament, it is on the Labour side that change is most marked in the form of increasing homogeneity.

As regards the representativeness of the political elite in terms of the class positions of its members’ previous employment, Fig. 4 shows comparisons with estimated population distributions, again on a birth cohort basis.Footnote 6 As the IDs indicate, the lack of representativeness, although showing some tendency to decline—which is chiefly the result of the expansion of Class 1 positions in the population at large—remains substantial. While this lack is more marked on the Conservative than on the Labour side, it could be thought to be of greater consequence in the latter case. While around 60% of the population are in, or retired from, employment in less advantaged, non-managerial, non-professional class positions, and thus a major potential source of Labour’s electoral support, only 6% of the party leadership born after 1950 came into parliament from such positions.

Fig. 4
figure 4

Comparing of distributions of social class (%) between population at large(a) and political elite, in three birth cohorts. Note(a)Source of information for cohorts 1860–1919 and 1920–1949: Oxford Mobility Study, 1972, covering men aged 20–64. Source of information for cohort 1950: Bukodi et al. (2015), averaging over distributions in two British Birth Cohort studies (NCDS and BCS70) and in a quasi birth cohort of individuals born in 1980–1984

Full size image

In sum, over the period we consider, the social homogeneity of the political elite in terms of the class backgrounds of its members diminished somewhat and so in turn did its lack of descriptive representativeness, although this remains marked and more so on the Conservative than on the Labour side. However, in certain respects the leaderships of the two parties appear to be moving in different directions. In their class origins, members of the Conservative leadership are becoming less homogenous and rather more representative, while in their previous class positions members of the Labour leadership are becoming more homogenous and less representative.

Party convergence in the social composition of the political elite?

The changes described in the previous section in the social class backgrounds of the Conservative and Labour components of the political elite in some ways point to a growing similarity between the two but in other ways less obviously so. As noted, the possibility of convergence was earlier envisaged by Johnson (1973) but was called into question by Burch and Moran (1985). More recently, it may be noted, somewhat similar disagreements have arisen over the emergence of a so-called ‘political class’. That is, a collectivity, including party leaderships but also MPs, and a range of other occupants of the ‘Westminster village’, who are career politicians, similar in their class backgrounds and education, differing little in their political ideologies, and more concerned with their own advancement than with furthering the interests of the groups in society they are supposed to represent (for a polemic, see Oborne 2008, and for a more balanced discussion, Allen 2018).

In addressing this issue, we begin by comparing in Fig. 5 class origins distributions for the Conservative and Labour components of the political elite, while also making certain distinctions within classes. To preserve adequate numbers, we now collapse the periods we distinguish to three.

Fig. 5
figure 5

Distribution (%) of political elite by parental class and subdivisions and indices of dissimilarity between parties, by period

Full size image

If we consider the distributions on the basis of the fourfold collapse of NS-SEC, some evidence of convergence does appear. The IDs fall across the three periods, even though the difference between the distributions remains quite wide. The convergence is largely the result of the decline within the Conservative leadership of individuals of Class 1 origins between the second and third periods. This outweighs the differentiating factors of the increase in those of Class 4 origins, which decrease on the Labour side, and the stable proportion among Labour of those from Class 6 and 7 origins. However, if we examine subdivisions within classes, giving an eightfold categorisation, little evidence of party convergence is present. This results chiefly from the fact that among elite members who are of Class 1 or 2 origins, those on the Conservative side are always more likely to come from the families of employers or of managers than from the families of professionals, while on the Labour side the reverse is the case. As is emerging in other respects, this occupational division within what could be broadly described as the salariat proves to be consequential.

In Fig. 6, we go on to consider distributions of the class of elite members’ own employment prior to their entry into parliament. In this figure, where we work again with the threefold categorisation of Classes 1, 2 and 3–7, the IDs returned clearly point to convergence. This is the result primarily of the Labour leadership, as earlier shown, moving steadily closer to the Conservative leadership in becoming increasingly made up of individuals who had previously been in Class 1 or 2, managerial or professional, positions. However, when in Fig. 7, we introduce subdivisions within such employment, it is again the case that, at a more detailed level, the evidence for convergence disappears. What can be seen is that differences between the Conservative and Labour components of the elite persist in two main respects.

Fig. 6
figure 6

Distribution (%) of political elite by social class before entry into parliament and indices of dissimilarity between parties, by period

Full size image
Fig. 7
figure 7

Distribution (%) of political elite by social class and subdivisions before entry into parliament and indices of dissimilarity between parties, by period. Note(a)Also includes large employers

Full size image

First, among those in the elite who previously held managerial positions, members of the Conservative leadership prove to have been employed, and to an increasing extent, in private sector rather than in public, or third sector, non-profit oriented organisations. In contrast, members of the Labour leadership have always been predominantly employed in organisations of the latter kind—for example, in public services, national and local government and in the running of trade unions, professional associations and charities. Second, among those previously employed as professionals, a persisting party difference shows up in terms of a distinction between ‘technical’ and ‘sociocultural’ professions (Häusermann, and Kriesi 2015; Oesch and Rennwald 2018). The former includes lawyers, health professionals, engineers, accountants and financial and management consultants; the latter, lecturers and teachers, social workers, journalists and all professionals in the arts. It can be seen that within the Conservative leadership, those individuals who were previously in professional positions have always been more likely to fall on the technical rather than the sociocultural side of the divide, while within the Labour leadership there is no great difference.

If, then, some convergence is apparent among the individuals making up the Conservative and Labour leaderships in their broadly defined class backgrounds, it can still be doubted if this translates into any great increase in similarity in their previous experience of, and socialisation into, family and working life. Rather, the two leaderships can be regarded as still to a large extent set apart in the socio-economic environments of the families in which they grew up and in the occupational cultures in which they were involved before entering parliament.

In earlier studies, the possibility of party convergence within the political elite was seen as being furthered and expressed through an increasing similarity in the educational careers of elite members as educational qualifications became of growing importance for access to the elite. We next examine how far this is the case.

In Fig. 8 we show relevant distributions by type of secondary schooling. The IDs reported do indeed indicate an increasing similarity, especially as between the second and third periods. This largely results from a decline in the proportion of those in the Conservative leadership going to the Clarendon schools, including Eton, though not to other private schools, and an increase in the proportion going to state schools.

Fig. 8
figure 8

Distribution (%) of political elite by type of secondary school and indices of dissimilarity between parties, by period

Full size image

However, when we turn to tertiary education, the situation, as shown in Fig. 9, is more complex. Convergence certainly occurs in that by the 1997–2021 period the proportion in the Labour leadership who had not attended university had fallen to the same low level as in the Conservative leadership. But as regards university or type of university attended, the IDs show that the difference between the two leaderships actually widened between the first and second periods distinguished and still remains higher in the third period than in the first. This pattern, it can be seen, results primarily from an increase on the Labour side in those going to universities outside of the Oxford–Cambridge–London ‘golden triangle’ that comes some time before a similar, though weaker, shift on the Conservative side. Nonetheless, for the most recent, 1997–2021, period it is the case that twice as many individuals in the Labour leadership went to ‘non-elite’ universities, outside of the golden triangle and of the Russell Group,Footnote 7 than did individuals in the Conservative leadership.

Fig. 9
figure 9

Distribution (%) of political elite by university and indices of dissimilarity between parties, by period

Full size image

Finally, it is also of interest to see if among elite members who went to university there is any party convergence in the subjects they studied. In Table 4, we show distributions, by period and party, across three broad subject areas: humanities, social sciences and law, and natural science, technology, engineering and mathematics (STEM subjects). The IDs show that some convergence has occurred. There is in general a decrease in those studying humanities and an increase in those studying social sciences and law, but this change is more marked on the Conservative side, where the humanities were previously more popular, thus bringing the two sides closer together. As regards the social sciences, a good deal has been written journalistically about the numbers of the political elite who have taken the Oxford PPE degree (see, e.g. Beckett 2017). It is indeed the single university degree course that has been most often followed among the elite as we would define it, but still by no more than 14% of all in our target population. As Table 4 shows, there is a tendency for it to have been more frequently taken by those on the Labour than on the Conservative side, although this difference is declining and thus the popularity of the degree overall.Footnote 8

Table 4 Distribution (%) of political elite by degree subject and indices of dissimilarity between parties, by period
Full size table

Convergence in the case of educational careers is then, as in the case of class backgrounds, more apparent on a broad view than when greater detail is taken into account. We would thus conclude that even for the most recent period that we cover, the two sides of the political elite still show far from negligible differences across a range of social attributes. The much discussed rise of the professional or ‘career’ politician (King 1981; Riddell 1993; Jun 2003) does find some reflection in the composition of cabinets and shadow cabinets. But the idea of a socially undifferentiated political elite, forming part of an internally cohesive political class, can be given little support. Those on the Conservative side of the elite are clearly the more likely to have come from the families of employers or managers, to have been privately educated, to have gone on to Oxford, Cambridge or other elite universities, and then to have previously worked in management in the private sector or, as professionals, in technical rather than sociocultural fields. Those on the Labour side are clearly the more likely to have come from either professional or working-class families, to have been state educated, to have attended non-elite universities, and to have previously worked in management in the public or ‘third sector’ or, as professionals, in sociocultural as much as in technical fields.

Questions of representation

As observed in the Introduction, earlier students of the social composition of cabinets were primarily concerned with the extent to which disparities between their class composition and that of the electorate at large compromised the idea of representative democracy. In contrast, in more recent research, the concern has been with how the representativeness of the political elite, now usually understood more widely as MPs, is viewed by members of the electorate and with the possible consequences for their political attitudes and behaviour. Evidence has indeed accumulated that representation does matter to voters and, further, that questions of representation create more serious problems for Labour than for the Conservatives. We briefly review this evidence and then consider the relevance of our own research—focusing on party leaderships rather than MPs—to the issues that arise.

In a pioneering paper, Cowley (2013, and see also Campbell and Cowley 2014) notes that in academic discussion of ‘the politics of presence’ (Phillips 1995), much has been claimed about which groups lack appropriate representation, in the sense of presence, within the political elite, but that little attention has been given to the views of the public. On the basis of national survey research, Cowley finds that voters do care about who politically represents them, do wish to see greater social diversity in parliament and, in particular, want greater representation by individuals with lived experience similar to their own and as grounded, most importantly, in locality and in social class. Such ‘self-identification’ effects lead to working-class men and women being especially concerned to have more working-class representation in parliament.Footnote 9

The significance of this latter finding is then underlined by the work of Heath (2016) and of Evans and Tilley (2017, Chap. 8) on the sources of the marked increase in working-class abstention in general elections that has become apparent from the end of the last century. Although following somewhat different analytical strategies, these authors reach essentially the same conclusion. The main source of the growing tendency for former working-class Labour voters to stay at home on election-day is their ‘social alienation’ from the political process, independently of policy issues. And crucial to this alienation is the sense that with the falling numbers of Labour MPs who can claim to have adult experience of working-class life—from almost 40% in 1964 to under 10% by 2010 (Evans and Tilley 2017, Fig. 6.8)—many working-class people no longer believe that there is a party that represents them. Thus, ‘large swathes of the working class have simply stopped voting’ (Heath 2016, p. 1070); or, as Evans and Tilley (2017, p. 186) put it, ‘The new party of the working class is no party at all’.

How, then, are our own findings to be related to the foregoing? It is difficult to see why the views of the public on the representativeness of party leaderships should differ significantly from their views on the representativeness of a party’s MPs. Party leaders are more often in the public eye, become better known than most other MPs, and are therefore likely to contribute more to their parties’ images.

Accepting this, the first impression to be gained from the findings we have reported might be that a somewhat paradoxical situation exists. Over the period we cover, the Labour component of the political elite is both less homogenous as regards its members’ class backgrounds and at the same time more descriptively representative of the electorate than is the Conservative component. Yet Labour would appear to be the more seriously threatened by problems of representation, in the form of working-class alienation and non-voting.

To see matters more clearly, it is necessary to go beyond questions of descriptive representation to consider those of substantive representation (Allen 2018, pp. 7–8). The former relate to representation within the elite in the sense of the presence there of individuals with particular characteristics; the latter relate to representation in the sense of there being within the elite individuals with a commitment to the particular interests of different social groupings. And in this connection, the social composition of party leaderships, responsible for the formation and implementation of policy, could be regarded as being of prime importance.

It is on the Conservative side, we have shown, that change in the class origins of the leadership, leading to reduced homogeneity, has mainly occurred. The previous dominance of those of Class 1 origins has weakened and more members of the leadership have come from Class 2 and, especially, from Class 4 origins. However, so far as the effective representation of class interests is concerned, little difference would seem to be implied. The main shift is that the numbers of those in the leadership coming from the families of large employers in Class 1 have fallen relative to those coming from the families of small employers and other ‘independents’ in Class 4. That is to say, there has been a growing element of individuals drawn from a social class noted for its strong pro-business and low-tax and small-state attitudes—and also, for its more general support for ‘radical right’ positions (Evans and Mellon 2016). Thus, so far as class origins are concerned, there is little reason to suppose any decline in the presence in the leadership of individuals so socialised as to be sympathetic to business and related interests. Moreover, we have also shown that in their employment before entering parliament, members of the Conservative leadership, while always in the large majority of cases having held Class 1 positions have, to an increasing extent, been engaged in management in profit-oriented, private sector organisations. In other words, whatever may be the consequences of change in the class composition of the leadership so far as descriptive representation is concerned, it could still be taken to provide quite durable substantive representation of the interests of individuals in economically more advantaged social groupings, from among whom the Conservatives would expect to receive their strongest electoral support.

However, change in the composition of the Labour leadership has taken a somewhat different path. After the immediate post-war period of Labour government, shifts in the class origins of its members have been relatively slight. Change has rather been concentrated in the class positions that they have held before their entry into parliament, and change has in this case led to increasing homogeneity, especially from the 1997–2010 period of New Labour rule. The proportion of those who previously held Class 1 positions has become almost as high as on the Conservative side—although with significant differences according to sector and occupation—while the proportion of those previously employed in working-class positions, or indeed in any other than managerial or professional positions, has fallen to a more or less negligible level. In other words, the decline in the presence of those with direct experience of working-class life is even more marked within the party leadership than within the body of Labour MPs.

Labour may then lay claim to greater descriptive representativeness at elite level than the Conservatives, but the position of the Labour leadership as regards the substantive representation of the class interests of those from whom the party has historically looked for a major part of its electoral support is far less clear than on the Conservative side. Survey evidence indeed shows (Evans and Tilley 2017, pp. 130–132) that not only within the working class itself but among the population at large, there has been a marked decline in the numbers of those who see the Labour Party as one that ‘looks after’ the interests of working-class people.

Insofar as significant working-class representation might still be claimed within the Labour leadership, it could then only be, so to speak, at one generation removed—that is, through those individuals upwardly mobile from Class 6 and 7 origins. There is evidence for Britain (Vivyan et al. 2020) that voters in general tend to prefer politicians who are of relatively humble origins and that ‘affinity biases’ also exist so that among working-class voters there is an especially strong preference for politicians of working-class origins. The suggestion is made that descriptive representation at elite level in terms of class origins may therefore ‘help to mitigate any popular dissatisfaction’ arising from the decline in the number of politicians with their own experience of working-class life (Vivyan et al. 2020, p. 11). However, little evidence has been produced of such an effect actually occurring, and among working-class voters it is clearly the lack of representation by those who have themselves actually lived in working-class localities and been employed in working-class jobs that appears to be the main source of dissatisfaction. There is, moreover, research to suggest (O’Grady 2019) that, over the years of New Labour government, Labour MPs with such characteristics were more likely to support policies that were in working-class interests—and even in opposition to the party leadership—than were other MPs, whatever their class origins, who had previously been only in managerial or professional employment.Footnote 10

Finally, it is relevant to note that Labour’s problem regarding working-class representation is not simply one of academic interest but is a live issue within the Party itself. Under Tony Blair, with recognition being given to the numerical decline of the working class, the problem was played down, and attempts were made to increase the attractiveness of New Labour to the electoral ‘centre ground’ by favouring parliamentary candidates with managerial and professional backgrounds. It is possible that this contributed to the Party’s electoral successes. But with Labour in opposition after 2010, a reversal occurred and increasing the number of working-class candidates became a specific objective. Ed Miliband stated that ‘we need a politics where politicians look like the constituents they represent’ (speech, July, 2012). John Trickett, the shadow Cabinet Office Minister, was charged with carrying out a search for suitable working-class candidates, and the National Executive Committee (NEC) changed its rules so that in making their decisions selection panels were required to take social class into account as well as gender and ethnicity (Cowley 2013, p. 129).

It does not, however, appear that these initiatives had any great measure of success, and there is certainly no indication of their effects filtering through to the composition of the party leadership. A renewal of efforts to increase working-class representation might have been expected as a result of the loss of ‘Red Wall’ seats in the 2019 General Election. But it would seem that, under Keir Starmer, there has, rather, been a reversion to the Blairite view that longer-term electoral advantage lies in downplaying the image of Labour as a working-class party—even if with some degree of at least verbal equivocation. Michael Crick’s ongoing research into the selection of Labour candidates for the next General Election (@tomorrowsmps) is of interest. His findings suggest (interview Conservative Home, March 15, 2023) that, given concerns within the NEC over party discipline in the context of a hung or tight parliament, its influence is directed towards ‘purging’ not only possible candidates thought to be unduly left-wing but also those with a strong trade union background in favour of ‘the middle-class and dull’. It would therefore seem unlikely that any increase in working-class representation within the leadership of the Labour Party will be evident in the foreseeable future. The electoral consequences are less predictable.

Conclusions

In this paper, we have sought to extend the tradition of twentieth-century studies of the social composition of the British political elite in relation to problems of representation, but with two developments. First, we have included members of shadow cabinets as well as cabinets in our definition of the elite, so that in effect it covers the leaderships of the two main political parties, and we have widened the analysis of social composition to include important sectoral and occupational divisions within classes that allow the identification of new sources of elite differentiation. Second, we have focused on the relation of our findings to current discussion of the implications of representation for political attitudes and behaviour among the electorate at large.

We started out with the aim of obtaining information on the social class backgrounds and also on the educational careers of all members of cabinets and shadow cabinets from 1945 down to 2021. We achieved overall an 86% coverage, with little indication of any selection bias. On this basis, we are able to show that the political elite as a whole has become somewhat less homogeneous in class origins over the period in question and at the same time somewhat more representative of the electorate, although large discrepancies still exist. However, the pattern of change differs notably as between the Conservative and Labour components of the elite. On the Conservative side, the proportion of elite members coming from Class 1 origins, and especially from the families of large employers, has declined while the proportion coming from the families of small employers and other ‘independents’ has increased. In contrast, on the Labour side, change is most marked in the class positions held by members of the elite before their entry into parliament: that is, in the direction of greater homogeneity through the rising proportion having held Class 1, higher managerial and professional positions. Thus, although over the period covered the Labour leadership is always more heterogeneous than the Conservative in this respect, and thus somewhat more representative of the electorate, the difference is diminishing.

Both of the trends in question suggest that the leaderships of the two parties are in general converging in their class composition. But it is important to go into greater detail. If differences within classes, by sector and occupation, are taken into account, convergence is no longer apparent. Members of the Conservative leadership who are of Class 1 or 2 origins are always the more likely to have come from the families of employers or managers, while their counterparts in the Labour leadership are always the more likely to have come from professional families. Further, in their previous employment, members of the Conservative elite who were managers were far more likely to have worked in private sector organisations than in public or third-sector ones, while the reverse is the case for Labour. And among elite members who were previously professionals the Conservatives have been in technical rather than sociocultural professions while there is little difference in this regard on the Labour side.

Increasingly similar educational careers have often been seen as a source of growing convergence in the backgrounds of the Conservative and Labour leaderships, and we do show some evidence of such a trend as regards schooling, chiefly resulting from a decline within the Conservative leadership in the numbers having attended elite Clarendon schools. But the greater frequency of private schooling among Conservatives is otherwise little changed, and there is likewise little change in differences in tertiary education. Although the political elite has become predominantly a graduate elite, attendance at Oxford, Cambridge or at other elite universities is still clearly more common in the case of its Conservative than of its Labour members.

Turning to questions of representation, recent research has established that voters would like greater representation by individuals with similar lived experience to their own, especially in the contexts of locality and class. And what has also emerged is that problems in this regard prove to be more serious for Labour than for the Conservatives: that is, in the form of apparent working-class alienation from the political process, as expressed most directly in a rising rate of abstention at General Elections.

To understand this situation, we have argued that it is necessary to think in terms of substantive as well as descriptive representation. In this regard, the social composition of party leaderships, given their public salience and their dominant influence over policy, could be seen to have a particular importance. In the case of the Conservative leadership, the changes shown in the class origins and previous employment of its members are not such as to suggest any weakening in its capacity or readiness to represent business and managerial interests and those of individuals in more economically advantaged class positions. Indeed, the increasing proportion of cabinet members from a small business owning background suggests that there is a social basis for a more right-wing position on both economic and social issues. But in the case of the Labour leadership, the changes are far more problematic. While we have shown that there has been no decline in the proportion of its members coming from working-class origins, it is not clear if this can compensate, so far as working-class voters are concerned, for the almost complete disappearance of those with adult experience of working-class life. That this is an issue of concern within the party itself is indicated by the shifting positions that have been taken up over the extent to which, from the standpoint of electoral advantage, Labour should seek to bring more working-class men and women into Parliament and to present itself as a party with a prime responsibility for representing the working-class. How this issue will be resolved remains, for now, highly uncertain.