Education and Social Mobility: The OED Triangle
Education and Social Mobility
91
We noted in the Introduction that as well as a broad political consensus existing on the need to increase social mobility, it is also generally supposed that in achieving this goal it is educational policy that is of crucial importance. This supposition is built into all the ministerial speeches and governmental policy statements on mobility that we referred to, into the series of annual reports and policy recommendations that have been made by the Social Mobility Commission, and also into the programmes of various 'third sector' organisations concerned with social mobility.1 In Chapter 2, in focusing on absolute class mobility, we examined the argument that it is primarily through educational expansion and reform that absolute mobility rates can be brought back closer to the pattern prevailing in the golden age when upward movement predominated over downward: that is, by building up a highly qualified labour force that will then pull in 'top-end' jobs to Britain from all parts of the global economy. We gave reasons for regarding this argument as unrealistic. In the present chapter our focus moves to the role of education in regard to relative rates of class mobility: that is, to the question of how far through education a significant reduction can be achieved in the inequalities - in some instances, as we have seen, the very marked inequalities - that exist in these rates, so that a new, more 'open' mobility regime can be brought into being, and one that can make some claim to meritocratic legitimation.
On the face of it, the idea that education plays a major role in social mobility might appear quite obvious. And, as will subsequently emerge, there is indeed ample evidence that in present-day Britain educational attainment is a major, even if not always an overriding,
1 For example, attached to the logo of the Sutton Trust is the phrase 'Improving social mobility through education' and the Social Mobility Foundation describes its main aim as being that of providing support in their educational careers for young people from low-mcome backgrounds.
___.......j, weakening ^ strengthening
association association
Figure 5.1 The OED triangle: requirements for education to create greater social fluidity
factor in determining who is mobile or immobile - that is, which individuals. But what has then to be further recognised - though it is in fact widely overlooked - is that it in no way follows automatically from this that education will be of similar importance in determining the amount of mobility within society at large. It is essential here to make the distinction between the individual and the societal levels of analysis. This is because education can serve to promote mobility at the societal level, through creating a greater equality of relative mobility chances or, that is, greater social fluidity, only if a number of conditions are met.
In treating the often rather complex issues that arise in this regard, sociologists refer to what is known as 'the OED triangle': that is, the triangle of associations that exists between individuals' social origins (0), their educational attainment (E) and their eventual social destinations (D). If education is to play the role in increasing mobility that is generally envisaged for it in current political discourse, what is necessary is that the associations within the OED triangle should change over time on a particular pattern. This pattern is shown in Eigure 5.1.
As is indicated, the OE association has to weaken. As a result of policies of educational expansion and reform aimed at widening equality of opportunity, differences in individuals' levels of educational attainment related to their social origins should diminish. At the same time, the ED association has to strengthen. Educational attainment has, through the decisions made by employers, to be the key determinant of individuals' social destinations. And, finally, the 'direct' OD
92
Social Mobility and Education in Britain
Education and Social Mobility
93
association - that which is not mediated through education - should like the OE association, also weaken (or at all events not strengthen so as to offset the changes in the OE and ED associations). Given these conditions, the overall association between origins and destinations will then itself weaken and social fluidity will increase. Under models such as those discussed in the two previous chapters, all the odds ratios underlying intergenerational mobility tables that express inequalities in relative mobility chances will tend to move closer to 1 - the value implying perfect mobility.
It might be added that among theorists of postindustrial society writing in the later twentieth century from a broadly 'liberal' position, the view was widely held that changes within the OED triangle on the lines shown in Figure 5,1 were readily predictable if not indeed already under way. On the one hand, the demand for an increasingly better educated and trained labour force, stemming from technological and economic advance, would, in conjunction with democratic pressures for greater equality of opportunity, ensure that the appropriate progressive development of educational systems took place. And, on the other hand, employing organisations would, as a requirement of their operating efficiency, have to base their recruitment procedures increasingly on formal educational qualifications rather than on any other criteria. The postindustrial society, it was claimed, was 'in its logic' an education-based meritocracy.2
However, one may ask how far, at least in the British case, this liberal scenario has been borne out, and in turn how far support is given to the politically prevalent idea that education is the key driver of mobility at the societal level. In Figure 5.2 another version of the OED triangle is shown that reflects the findings of a substantial body of sociological research in Britain covering the period since the Second World War and in which origins and destinations have been treated primarily in terms of social class.
The following points emerge. First, so far as the OE association is concerned, the most comprehensive studies have in fact revealed a tendency for inequalities in educational attainment related to class origins to narrow, even if only slightly and mainly at lower educational levels. In this respect, therefore, the liberal view can claim at least some
2 The phrase comes from the work of the best-known theorist of postindustrial society, the American sociologist Daniel Bell (1972, 1973).
* D
__________> weakening -> no change
association
Figure 5.2 The OED triangle: typical results from sociological research
degree of confirmation: Figure 5.2 replicates Figure 5.1. But, second, a very different situation arises with the ED association. It has been quite regularly found that this association, as indicated in Figure 5.2, also weakens rather than strengthening as in Figure 5.1. And, third, studies of the direct OD association have detected no consistent weakening or strengthening.3 The upshot, then, is that from the research findings represented in Figure 5.2 no firm conclusion can be drawn as regards the overall OD association. What is implied is that any equalisation that may have been achieved in educational attainment in relation to class origins has been in some degree or other offset, so far as its
3 On the OE association, the most important papers are Jonsson and Mills (1993), Jonsson, Mills and Muller (1996) and Breen et al. (2009, 2010); on the ED association, Goldthorpe and Mills (2004), Breen and Luijkx (2004b), Jackson, Goldthorpe and Mills (2005) and Goldthorpe and Jackson (2008); and on the direct OD association, Goldthorpe and Mills (2004) and Vandecasteele (2016). A good deal of this research has a comparative, cross-national dimension, and the results reported for Britain for the most part follow the general pattern. But one apparent exception may be noted. Evidence that educational expansion may serve to increase mobility has mainly related to a three-way 'interaction effect' within the OED triangle. It has been found that the OD association is strongest among those with the lowest levels of education and weakest among those with the highest levels. Thus, as the number of individuals attaining higher levels of education increases, so too does social fluidity, simply through a 'compositional' effect (see Breen, 2010). However, our analyses of the birth cohort data reveal that while this interaction effect is present over the early years of a working life, at least in the 1970 cohort, it has disappeared by age 38 (Bukodi and Goldthorpe, 2016). A possible explanation for this British 'exceptionalism' is suggested in Chapter 9, n. 11 below.
94
Social Mobility and Education in Britt
am
potential impact on social fluidity is concerned, by a decline in the class returns that education brings.
The explanation for the weakening ED association that has been most often put forward is that of overqualification or 'credentials inflation', and we have already noted in Chapter 2 that there is evidence of such a situation now developing within the British labour force. This was an outcome simply not envisaged by the theorists of postindustrial society, and it is one that has been almost entirely disregarded in attempts to present policies of educational expansion and reform as being of leading importance in raising levels of social mobility.4 It is true that difficulties arise over what the concept of overqualification actually entails and that, in consequence, estimates made of its extent and trend are open to some dispute. But with the growth of the managerial and professional salariat slowing down (cf. Figure 2.1) while the number of individuals with higher-level educational qualifications steadily rises, some change in the relationship between qualifications and class positions would seem scarcely avoidable.
We will return to this matter later in the hope of providing a degree of clarification. For the moment, however, the relevant point remains that, as they stand, the findings represented in Figure 5.2 are indeterminate as regards the question of whether or not in Britain, over recent decades, changes in the associations within the OED triangle have been such as to create greater mobility. For this reason - among others that will become apparent - we have been led to ask whether it might be possible to gain some better understanding of the situation through a reconsideration of the way in which educational attainment, in its relation to social mobility, is conceptualised and measured.
In research of the kind that underlies Figure 5.2 education has for the most part been treated in what might be described as absolute terms, and measured according to the number of years of education that individuals have completed or the highest level of educational qualification that they have obtained. This approach would appear appropriate insofar as education is being viewed as a consumption good and therefore as an 'absolute' good in the sense that its value to one individual is unlikely to be affected by the extent of its
4 We can find no reference to the possibility of overqualification in any of the series of annual reports of the Social Mobility Commission.
Education and Social Mobility
95
consumption by others. If, for example, by taking a university degree, someone discovers the joy of knowledge and expands their intellectual and cultural horizons, this outcome need be little affected by the number of others who have the same experience. But in analyses of associations within the OED triangle education is in effect being treated not primarily as a consumption good but rather as an investment good - that is, in relation to individuals' economic futures in the labour market. And this being so, it could then be thought preferable to treat education not as an absolute but rather as a 'positional' good in the sense that the value of an individual's particular level of educational attainment will be directly dependent on the levels of attainment of others. If someone has a university degree, its value in the labour market will be greater if, say, only 10 per cent of everyone in the labour force have degrees than if 40 per cent have degrees.5 And it would in turn appear appropriate, in analyses of the kind in question, to measure education in positional or, in other words, relative rather than absolute terms: that is, to recognise that what matters in this case is not simply how much education individuals have but how much relative to others and, in particular, relative to those others with whom they will be in most direct labour market competition.
To pursue this issue, we have drawn on the same dataset from the British cohort studies we have used in our analyses of absolute and relative rates of class mobility, and we have examined the OE and the ED sides of the OED triangle with educational qualifications being measured in both absolute and relative terms, so as to be able to see what differences may arise. For the purposes of this comparison, we focus on men in order to avoid difficulties in the interpretation of results that women's more complex employment histories would be likely to cause. But we can see no reason why the conclusions we reach should not in principle apply in the case of women also.
Our absolute measure of individuals' educational qualifications is provided by the eight ordered categories that are shown in Table 5.1, ranging from 'no qualifications' to 'postgraduate qualifications'. However, the rather small size of the 1946 cohort means that in order to maintain adequate numbers in our analyses, we have to collapse the
5 The pioneering and still highly influential work on the distinction between absolute and positional goods is Hirsch (1977). On education as a positional good, see Wolf (2002), an important but, in the prevailing political climate, unduly neglected book.
96 Social Mobility and Education in Britait
Tabic 5.1 Distribution (%) of male cohort members by highest level of educational qualification (absolute measure)(a)
		Cohort	
Level of qualification	1946	1958	1970
1. No qualifications	34	21	15
2. Sub-secondary (below 0 level or GCSE, NVQ1)	5	10	7
Total	39	31	22
3. Lower secondary - low performance (1-4 O level	19	21	21
or GCSE passes, NVQ2)			
4. Lower secondary - high performance (5+ O level or	17	18	21
GCSE passes or 1 A level pass, NVQ 3)			
Total	36	39	42
5. Higher secondary (2+ A level passes)	2	4	3
6. Lower tertiary (tertiary sub-degree qualification,	1-1	12	15
NVQ 4)			
Total	16	16	17
7. Higher tertiary (degree, NVQ 5, 6)	8	11	15
8. Postgraduate	I	2	4
Total	9	13	19
Total	100	100	100
N	2394	7219	5979
Note (a) All vocational qualifications are either National Vocational Qualifications (NVQs) or are given as their NVQ equivalents Source: Bukodi and Goldthorpe (2016)
eight categories to four, as is indicated. It can be seen from the distributions given that, as would be expected, there is a tendency for the overall level of qualification to increase across the cohorts. In particular, the proportion of men with no, or only sub-secondary, qualifications falls while the proportion with higher tertiary qualifications rises.
Our relative measure is then derived, as is shown in Table 5.2, from further fourfold collapses of the eight categories of Table 5.1 but collapses that change from cohort to cohort according to the proportions of cohort members holding the qualifications that the categories comprise, and with the aim of producing relative categories of as
Education and Social Mobility
Table 5.2 Collapses of eight categories of educational qualifications to produce relative levels
Cohort
Level of qualification	1946	%	1958	%	1970	%
Lowest level	1	54	1,2	31	1,2	22
Next to lowest level	2,3	25	3	20	3	21
Next to highest level	4	17	4,5	22	4,5,6	38
Highest level	5,6,7,8	23	6,7,8	25	7,8	19
Total		100		100		100
N		2394		7219		5979
Source: Bukodi and Goldthorpe (2016)
similar size as possible given the 'lumpiness' of the distributions shown in Table 5.1. In fact, the proportions of men at each relative level range from 17 to 38 per cent, but in seven cases out of the twelve they fall between 20 and 30 per cent.
The changes across the cohorts amount to the following. With the 1946 cohort only men with no qualifications are placed at the lowest relative level but with the 1958 and 1970 cohorts those with no more than sub-secondary qualifications are also placed at this level. With the 1946 cohort the next to lowest level comprises both those with sub-secondary and those with low performance secondary qualifications, while with the 1958 and 1970 cohorts only the latter are at this level. Men with high performance secondary qualifications are at the next to highest level in all three cohorts but also placed at this level are men in the 1958 cohort whose highest qualifications are A levels and men in the 1970 cohort with A levels and tertiary sub-degree qualifications. In turn, the highest relative level comprises with the 1946 cohort all men with A level or higher qualifications, while with the 1958 cohort only men with tertiary qualifications are included, and with the 1970 cohort only men with degree-level qualifications. In short, one
Social Mobility and Education in Britain
could say that as the proportion of cohort members with degrees increases, other qualifications tend to be pushed down the relative levels.6
In our analyses of the OE and ED associations, using our four-level absolute and relative measures of qualifications, we treat class origins and class destinations on the basis of the fivefold 'hierarchical' collapse of NS-SEC that is indicated in Table 1.1. We thus have 5x4 and 4x5 tables that, respectively, cross-classify origins by qualifications and qualifications by destinations for each cohort. We then apply to these tables versions of the same three statistical models we previously applied in investigating possible changes in relative mobility rates: first, the independence model proposing no association and under which all odds ratios underlying the tables are equal to 1; second, the constant association (CA) model proposing an association but one that is unchanging across the cohorts, with all corresponding odds ratios taking the same value; and, third, the uniform difference (UNID1FF) model proposing an association that, from cohort to cohort, becomes uniformly weaker or stronger, with all odds ratios moving close to or further from 1 by some common factor.
Considering the OE association with our absolute qualifications measure, we find, not surprisingly, that the independence model shows a significant and substantial lack of fit to the data: over 11 per cent of all men arc misclassified. That is to say, an association clearly does exist between men's class origins and the level of their educational attainment according to the four categories of Table 5.1. The CA model does then significantly improve on the independence model, misclassifying less than 3 per cent of all cases. But this model still does not give an entirely acceptable fit to the data and it is in turn significantly improved upon by the UN1DIFF model. This misclassifies less than 2 per cent of all cases and the p parameters that are returned - that is, the factors by which the odds ratios expressing the OE association have to be multiplied - are as plotted in the left-hand panel of Figure 5.3. A weakening association is apparent: the point estimates fall across the cohorts and quite sharply so between
* For the 1958 and 1970 cohorts, men's level of qualification was determined at age 37 but for the 1946 cohort, because of data limitations, at age 26. For full details of the construction of the absolute and relative educational measures and of the analyses discussed in the following paragraphs, including results of robustness checks on the results obtained, see Bukodi and Goldthorpe (2016).
and Social Mobility
Level of qualification, absolute
99
Level of qualification, relative
1958
1970 1946 Birth cohort
1958
1970
Source: Bukodi and Goldthorpe (2016)
Figure 5.3 Association between class of origin and level of qualification by cohort: UNfDIFF parameters with 95% confidence intervals
the 1958 and 1970 cohorts, in which case no overlap of the confidence intervals occurs. Our results here do therefore serve to confirm those of previous research relating to the OE association, as represented in Figure 5.2.
However, when we repeat the same sequence of modelling but now using our relative measure of qualifications, we have a different outcome. As before, the independence model fits badly, with again over 11 per cent of all men being misclassified; and, also as before, the CA model makes a large improvement, misclassifying only a little over 2 per cent. But what we further find is that now the UNIDIFF model does not significantly improve on the CA model. The point estimates of the B parameters, as plotted in the right-hand panel of Figure 5.3, do fall slightly across the cohorts but there are wide overlaps of the confidence intervals around them, so that it is the model of constant association that can most safely be accepted. In other words, the weakening association between men's class origins and their qualifications that shows up when education is measured in absolute terms is not reproduced when qualifications are measured in
100
Social Mobility and Education in Britt
■am
UNIDIFF parameter (ß) 1.2
1.1
1.0 H
0.9 0.8 0.7 0.6 0.5 0.4
Level of qualification, absolute
Level of qualification, relative
1946      1958      1970 1946 Birth cohort
1958
1970
Source: Bukodi and Goldthorpe (2016)
Figure 5.4 Association between level of qualification and class of destination by cohort: UNIDIFF parameters with 95% confidence intervals
relative terms - that is, when education is envisaged as an investment good in regard to individuals' economic futures.
Turning then to the ED association, we follow the same analytical approach as with the OE association except that in our modelling we now control for the OE association on the basis of the results reported above. When we treat the ED association on the basis of our absolute measure of qualifications we find that while the CA model greatly improves on the independence mode! and does in fact give a statistically acceptable fit to the data, we still obtain a significantly better fit by moving to the UNIDIFF model. And from the $> parameters that are returned, as displayed in the left-hand panel of Figure 5.4, it can be seen that it is a weakening association that is shown up, and especially between the 1946 and 1958 cohorts. Thus, just as with the OE association, our results are here in line with those of previous research in which education has been measured in absolute terms: the class returns to educational attainment would appear to be falling.
Again, though, when we work with our relative measure of qualifications a different situation is indicated. The CA model once more
Education and Social Mobility
101
gives a big improvement on the independence model but in this case the further improvement made by the UNIDIFF model on the CA model is only marginally significant, and further, as is shown in the right-hand panel of Figure 5.4, the p parameters returned do not indicate any change of a directional kind. There is a possible weakening of the £D association between the 1946 and 1958 cohorts but then a strengthening between the 1958 and 1970 cohorts. If therefore, on the basis of our relative measure, any change at all occurs in the class returns to education, it would be best seen as taking the form simply of rather minor and short-term fluctuation, in response, it could be supposed, to changing labour market conditions. As regards the suggested dip in the ED association for the 1958 cohort, we do in fact know that men in this cohort, and especially those who had been longest in education, entered the labour market at a time of severe recession and high unemployment - the early 1980s - and that this had a lasting negative influence on their occupational careers.7
Finally, we may add that in all the analyses undertaken, whether qualifications are treated in absolute or relative terms, no evidence emerges of any change in the direct OD association - that which is not mediated via education. This would appear essentially stable.
From our comparison of changes in the associations within the OED triangle using our two different measures of educational attainment, we can then draw the following conclusion. As well as the relative measure appearing the more appropriate in this context, it leads to results that are open to a far more straightforward interpretation than those obtained using the absolute measure. Rather than having to suppose that some equalisation in educational attainment in relation to class origins is offset, to a greater or lesser extent, by a decline in the class returns to education, and with no change occurring in the direct origins-destination association, we can simply envisage a situation in which all the associations in the OED triangle are unchanged, or, at most, change in only limited and directionless ways. And this is of course a conclusion that is entirely consistent with what we have shown in Chapters 3 and 4 concerning the absence of significant change in the case of the men in our cohorts in either the level or the pattern of relative rates of class mobility. In other words, what is again
For further details, see Bukodi and Goldthorpe (201 la).
102
Social Mobility and Education in Brita
m
education and Social Mobility
103
pointed to is the essential stability of the endogenous mobility regime -or, that is, the resistance to change that it offers.
If this interpretation of the empirical evidence is accepted, we are thus brought back to the question, already introduced in the conclusion to Chapter 3, of how it could come about that over a period in which educational expansion and reform have been more or less continuous, a reduction in class-linked inequalities in educational attainment leading to increased social mobility has not been achieved. To pick out only the major developments over the years relevant to our analyses, the Butler Act of 1944 introduced free secondary education for all and raised the school-leaving age from 14 to 15; during the 1960s the selective, tripartite system of secondary education was progressively changed to a comprehensive system in order to prevent 'social segregation' and increase equality of opportunity; following the Robbins Report of 1963 a major expansion of university education occurred with the aim of making this available for all of those 'qualified by ability and attainment'; in 1972 the school-leaving age was raised again, to 16; and from the 1980s a further phase of university expansion began. These developments were associated with a substantial advance in levels of education and qualification within the population at large, as is reflected in Table 5.1. But why did they contribute so little to realising the liberal scenario of steadily rising social fluidity resulting from the creation of an education-based meritocracy?8
The explanation we would give starts out from the proposition that employers, on the one hand, and in turn parents and their children, on the other, do themselves in many contexts view education in relative rather than in absolute terms. And, from this, the persisting strength of the associations within the OED triangle that
8 It may be noted that in the case of two of the measures referred to in the text, specific evaluations of their impact on social mobility have been made - with negative results. Boliver and Swift (2011) have exploited the fact that children in the 1958 birth cohort passed, in almost equal numbers, through tripartite and comprehensive secondary school systems and, on the basis of a careful comparison, find that the mobility chances of the two sets of children, whether considered in terms of class or income, differed very little. Sturgis and Buscha (2015) have examined the effect of raising the school leaving age in 1972, using the data of the ONS Longitudinal Study, and conclude that while an increase followed in levels of educational attainment among the population at large, no change was discernible in rates of intergenerational mobility in terms of class or social status. Earlier, Halsey (1977) found no evidence that the 1944 Act had served to reduce educational inequalities or thus to promote social mobility.
we have demonstrated when using a relative measure of educational qualifications can be seen to follow.
As regards employers, what may be supposed is that as well as treating educational qualifications as certifying forms of knowledge and skill that are required for particular occupations, they also treat them as indicators, or 'signals', of certain more general attributes of potential employees - such as their self-control, perseverance and capacity to learn; that is, as indicators of individuals' trainabiliry and productivity once in employment. Qualifications understood in this way then serve employers as a means of ranking potential employees in what may be thought of as a 'labour queue', which they will seek to match up to the jobs they have to fill - their 'job queues' - starting with the best qualified of those available and working downwards. In other words, it is the relative aspect of qualifications that counts. If there is an increase in potential employees with higher-level qualifications, so that their number extends further down the labour queue, then, assuming no accommodating change in the state of demand, more of them will be taken on at lower levels in employers' job queues than was previously the case - and with the effect of 'bumping down' in turn all those with inferior qualifications.9 On an absolute view of qualifications, a weakening of the ED association within the OED triangle will thus show up: class returns to educational attainment will appear to fall and overqualification to be present. But on a relative view - that is, with due account being taken of the changing distribution of qualifications within the labour force - no weakening of the association need occur. Apart from any effects of changes in demand, the same relative level within the distribution can still attract the same class returns.10
9 There is mounting evidence that in Britain today - following on from the major expansion of the tertiary educational sector in the 1990s - employers are significantly differentiating among graduates, at least in terms of the pay they receive, according not only to their class of degree and field of study but also to the university they attended (Green and Zhou, 2010; Britton et al., 2016). The increasing variance in earnings returns to a degree that has resulted means that the average return becomes less and less informative.
10 We draw in the foregoing paragraph on work in labour economics guided by what are known as signalling theory and job competition theory, which seem to us far more illuminating in regard to sociologists' empirical findings on associations within the OED triangle than the prevailing orthodoxy of human capital theory. These theories, it may be added, do allow for the existence of overqualification in the sense that certain levels of qualification may become 'devalued' and also in the sense that qualifications may be necessary to get a
104
Social Mobility and Education in Britain
Insofar, then, as employers do view qualifications relatively, and act accordingly, parents and children, at least to the extent that they see education as an investment good, will be under evident constraint to follow suit. Thus, where attempts arc made through public policy to widen and equalise educational opportunities, parents in more advantaged class positions will not be unapprcciative of the implications for their own children's life-chances. They may be expected to respond by drawing on their superior resources in order to take 'defensive measures': that is, measures designed to ensure that, in the context of generally rising levels of educational attainment and qualification, their own children retain their competitive edge. And there is indeed no shortage of evidence of such a response in the British case. While resort to the private schools sector has always been favoured by the wealthiest, it is clear that over recent decades a steadily growing number of parents with adequate means have sought to protect their children's relative positions within the state sector in a number of different ways: by paying for high-quality preschool child care, purchasing houses in areas with high-performing state schools, engaging home tutors, and providing their children with a wide range of educationally relevant extra-school activities and experiences. In other words, possibilities for what has been aptly called 'the commodification of opportunity' have been increasingly taken up.'1
We are now in a position to say more about the political consensus that prevails on the need to increase social mobility and on educational
job that are not necessary to do the job. For further discussion, see Goldthorpe (2014).
The idea of 'the commodification of opportunity' comes from the American sociologist, David Grusky. Several studies of parental strategies of the kind in question have been sponsored by the Sutton Trust. Francis and Hutchings (2013) give a good idea of their general spread, and Kirby (2016) focuses on the growth of private tutoring or 'shadow education'. Both smdies very probably underestimate the association with parental resources in relying on inadequate measures, such as the Market Research Society social categories or the 'free school meals' proxy for parental income (see further Chapter 6, n. 5). Jerrim (2017a) is a better-grounded investigation and indicates that private tuition plays an important role in providing a 'safety net' for children from better-off families who are in apparent danger of educational underperformance. A further study, Cullinane et al. (2017), reveals how many high-performing state comprehensives are in effect socially selective, partly as a result of the social composition of their catchment areas but also as a result of their admissions procedures which, it appears, more advantaged parents can often successfully 'game'.
Education and Social Mobility
105
policy as being the prime means to this end. As we observed in the Introduction and again at the end of the previous chapter, the concern to promote mobility can be understood as a response to rising economic and social inequality of condition - a response that is seen as politically more manageable and, at least from some points of view, more desirable than attempting to reduce inequality itself. And insofar as a contradiction might be seen to arise between widening opportunity while inequality also widens, it is further supposed that education is the way through which any such contradiction can be overcome. It would be difficult to find a better encapsulation of the thinking here involved than in two famous quotes from the New Labour era: Peter Mandelson's acknowledgement that he was 'intensely relaxed about people getting filthy rich' and Tony Blair's declaration that in seeking 'opportunity for all' his priorities were 'education, education and education'.'2
However, the results of the research that we have reviewed would strongly suggest that to look to education to break the link between inequality of condition and inequality of opportunity is to ask of it far more than it can alone deliver. What has to be realised is that if educational policy is designed to equalise educational opportunity with the further aim of thus equalising relative rates of mobility, then the zero-sum game that we have shown to he involved in the latter case is, as it were, simply brought forward. To the extent that some would benefit from the policy - that is, end up in a better relative position in the qualifications hierarchy - others would correspondingly have to
12 Mandelson was speaking in 1998 to a meeting of US industrialists. He modified his views after the 2008 financial crisis, conceding that more concern had to be shown over rising economic and social inequality resulting from globalisation. Blair apparently first used his priorities phrase in his leader's speech at the Labour Party conference in 1996, but reused it several times later. John Major took the opportunity to respond by saying that these were his priorities too 'but not necessarily in that order'. Lord Adonis (2012) takes Blair's words as the title of his book describing, and proclaiming the success of, the policy of converting state comprehensives into independently managed 'academies', of which he was the chief architect while in the Cabinet Office and then Minister of State for Education from 2005 to 2008. While it is still too eatly to assess what effects, if any, this policy has had on social mobility, the most detailed evaluation of it to date, by the Education Policy Institute and the LSF. (Andrews and Pcrera, 2017), concludes that its results have been very variable and that 'academies are not a panacea for school improvement'. It is of interest that the Executive Chairman of the EPI is David Laws, who was from 2012 to 2015 Minister for Schools in the coalition government.
106
Social Mobility and Education in BritaitM Education and Social Mobility
107
lose out. This being the case, and recalling the psychology of l0ss aversion referred to in Chapter 3, it is only to be expected that a reaction will come from parents in more advantaged positions, and that where policy appears likely to create a situation in which their children's interests are under threat, they will draw on their superior resources so far as is necessary in order to take countervailing action.
We would not wish our position here to be misunderstood. We recognise - indeed we have emphasised - that during the period to which our research relates educational expansion and reform did substantially raise the overall educational level of the British population. Opportunity certainly has been widened in the sense that more men and women than ever before have been able to fulfil more ambitious educational aspirations, with both intrinsic benefits for themselves and wider economic and social benefits. And we would further recognise that many educational policy initiatives proposed with the aim of increasing mobility could in fact have positive consequences from a purely educational standpoint: for example, the extension and upgrading of preschool programmes for children from disadvantaged backgrounds, higher-quality vocational education, more transparent admissions procedures to elite universities and wider provision for lifelong learning. But what we would question is whether through educational measures alone a greater equality of opportunity is likely to be created in the sense of a significant reduction in inequality in relative mobility chances. The historical record of the last half-century or more, as we can reconstruct it through our research, makes it difficult to avoid the conclusion that educational policy, directed towards expansion and, for the most part, egalitarian reform, has had very little effect in weakening the association that exists between individuals' class origins and their class destinations.
We cannot of course rule out the possibility that educational policy of some new and more effective design could have a greater societal impact in the future than has been the case in the past. However, an alternative possibility has also to be recognised, and one that seems to us sociologically more plausible: that is, that in any society with a capitalist market economy, a nuclear family system - even if one less stable and thus more complex than previously - and also a liberal democratic polity, a limit exists to the extent to which relative mobility chances can be made more equal by means of educational policy on which some broad degree of consensus might be possible, and that, as
this limit is approached, any further advances will increasingly require torms of intervention that will be of a politically far more controversial
and contested kind.13
We will return to and develop this line of argument in Chapter 10, in the context of our analyses of social mobility in Britain in a cross-national comparative perspective, and we will take up some of the wider implications that arise in our concluding chapter. However, in the chapters that directly follow our concern is to draw further on the experience of the men and women in our birth cohorts in order to provide a series of more detailed analyses of the associations that exist within the OED triangle.
13 It should be added that even if educational policy loses its capacity to further reduce inequalities in relative mobility chances, it could of course still serve to widen them. Such 'one-way' effects are not uncommon in public policy: to use the old analogy, one may be able to pull on a string without being able to push on it. Certain developments in the recent past that might be thought likely to increase inequalities - though no decisive evidence has as yet been produced - are the reduction in funding for Sure Start, the ending of Educational Maintenance Allowances, and the shift in higher education from free tuition and maintenance grants to tuition fees and student loans.