Read halsey text version

CENTRE FOR RESEARCH INTO ELECTIONS AND SOCIAL TRENDS

CREST

Working Paper Number 70 June 1999

Twentieth Century Trend in Social Mobility in Britain

By Anthony Heath and Clive Payne

The Centre for Research into Elections and Social Trends is an ESRC Research Centre based jointly at the National Centre for Social Research (formerly SCPR) and the Department of Sociology, University of Oxford http://www.crest.ox.ac.uk

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Abstract

This paper examines trends in social mobility for men and women born in the first sixty years of the twentieth century. It finds that intergenerational mobility rose during the course of the century for both men and women. Much of this reflects the growing size of the salariat and the declining size of the working class, but there is some evidence to suggest that there was greater fluidity of intergenerational movement between social classes amongst men born after 1940 compared with earlier cohorts.

1

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

This paper is about the intergenerational mobility of men and women. That is to say, we investigate how far, and in what way, the social positions of adults have changed from those of their parents. Studies of this kind enable us to chart the trends in the 'openness' of society. For example, do children largely follow in their father's footsteps, or do people move freely up and down the social ladder? An open society in which people's positions do not depend upon ascribed social characteristics such as their social origin is often thought to be both more efficient - better use being made of the available talent in society - and more in accordance with social justice. While the arguments about the relationship between social mobility, economic efficiency and social justice are by no means uncontroversial (see for example Marshall et al 1997), there is nonetheless considerable interest in documenting trends in the openness of society.

Following recent practice in sociology (eg Goldthorpe 1980), we will focus on the social class positions of men and women in Britain, and we shall compare these positions with those of their fathers. While it would be wrong to equate social class with the broader concept of social position, social class is nonetheless a powerful index of people's social advantages and disadvantages, their likely income and material conditions, their security and prospects, their health and their lifestyles (see Reid 1998). It is probably the most useful single index of social position. Our analysis will give us a picture of the overall mobility experience of the population as a whole. There is a separate tradition of research which has focused more narrowly on recruitment to elite positions, such as Parliament and to higher levels of the civil service, business, church and military. However, it is beyond the scope of this paper to explore patterns of elite recruitment. (Heath 1981 gives a summary of previous research in this field.)

Again, following conventional practice, we shall compare people's social positions with those of their fathers. Sociologists have in the past regarded the family as the unit of class stratification and have taken the father's position to be the best guide to the social class of the household as a whole. With the growing number of female-headed households in Britain and of dual-career families in which husbands and wives have full-time occupations, this is becoming a more 2

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

controversial assumption, and at some stage in the future sociologists will need to take more account of mothers' social class positions. However, in a paper concerned with historical trends, it is sensible to follow the conventional procedure. (Sorensen 1994 gives a comprehensive review of this debate.)

The first studies of social mobility appeared at the beginning of the century. They were small scale, focusing on the recruitment to particular occupations (Chapman and Marquis 1912, Chapman and Abbott 1913, Ginsberg 1929). They showed considerable openness, but one could not generalize their findings to the society as a whole. The first major nationally-representative study was that of Glass and his associates - the 1949 mobility survey. This provided a picture of mobility trends over the first half of the century (Glass 1954, Kelsall and Mitchell 1959). It portrayed a society with considerable inequality in mobility chances, and with especially high rates of social closure at the apex of the class structure. "... the general picture so far is of a rather stable social structure, and one in which social status has tended to operate within, so to speak, a closed circuit. Social origins have conditioned educational level, and both have conditioned achieved social status. Marriage has also to a considerable extent taken place within the same closed circuit" (Glass 1954, p21). The data also suggested that there had been some net downward mobility (although there were some problems here since the 1949 study compared respondents' current occupations with their fathers' last main occupation).

1

Glass and his associates expected mobility to increase in the second half of the century, due to increasing equality of opportunity. They were writing shortly after the 1944 Education Act, which had provided for free secondary education for all, abolishing the fees that some students had previously had to pay in pre-war grammar schools. It was widely hoped at the time that greater equality of opportunity for educational success and hence for occupational advancement would follow.

The situation during the third quarter of the century has been described by the work of Goldthorpe and his associates, based on a 1972 survey of men's mobility in England and Wales (Goldthorpe 1980, Goldthorpe and Payne 1986, Heath 1981). (There were also parallel studies 3

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

in Scotland and Ireland; see Payne, Ford and Robertson 1976, Hout 1989.) The 1972 study told a very different story from the 1949 survey. By 1972 there was increasing 'room at the top' and a considerable surplus of upward over downward mobility. The driving force behind this change was the expansion of professional and managerial occupations and decline of manual occupations. Hence mobility rates, especially upward mobility, increased substantially in the post-war period.

However, Goldthorpe also suggested that fluidity, or relative mobility rates, had not changed very much. That is to say, the observed changes in the overall rates of class mobility were primarily due to changes in the distributions of fathers and sons in the class structure, not to equalization of the terms on which people from different origins competed. In this sense, although chances of upward mobility had increased for everyone, the relative competitive chances of people from different social origins had not greatly changed. In this sense Glass's hopes for increased equality of opportunity and openness in society had not been fulfilled. (For a more detailed examination of Glass's thesis, using the 1972 data but focussing on education, see Halsey et al 1980, Kerckhoff and Trott 1993.) This has been one of the most controversial conclusions in mobility research. Many theorists had expected that the logic of industrialism would lead to greater openness or fluidity in society. And some empirical research (esp Treiman, Luijkx and Ganzeboom 1989) had claimed to find that social fluidity had indeed increased over time. We shall explore this issue of trends in fluidity, or relative mobility rates, later in this chapter, using recent data which enables us to bring the trends almost up to the end of the century, in 1997. First however we shall focus on the trends in absolute mobility.

Data and methods

To explore British experience of social mobility during the twentieth century we use the cumulated files for the 1964-97 British Election Surveys (BES). 4

2

These are nationally-

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

representative sample surveys of the British electorate, excluding Northern Ireland, conducted after each of the ten elections from 1964 to 1997. (Breen and Whelan 1999 give an up-to-date account of trends in mobility in Northern Ireland and in the Republic of Ireland.) The BES is the longest-running academic survey series in Britain.

3

Unlike the 1972 study used by

Goldthorpe they cover both men and women. They have been conducted and coded in closely comparable ways and hence it is reasonable to pool them. The British Election Surveys do not of course reach as far back as the 1949 survey, but unfortunately the coding procedures of the 1949 survey have not been preserved and it has proved impossible to establish comparability between the 1949 survey and subsequent research (Ridge 1974, Macdonald and Ridge 1972, Hope 1981). It is not therefore possible to add the 1949 survey to the cumulated BES files.

In order to chart trends over the whole century we use cohort analysis, that is we use the older respondents in our surveys to tell us what mobility was like earlier in the century. Since, in all the Election surveys, retired people reported their last main occupation, we actually have a few respondents from the 1964 and 1966 surveys who entered the labour market at the beginning of the century. We use the same pattern of ten-year birth cohorts as did Glass and thus distinguish the following birth cohorts:

1) 2)

people born before 1900 and entering the labour market early in the century; born 1900-09, entering the labour market around the time of the first world war with their careers spanning the depression; born 1910-19, entering the labour market during the depression and having their careers interrupted by the second world war; born 1920-29, entering the labour market around the second world war but their careers benefiting from the 'long boom' of the post-war period; born 1930-39, entering the labour market after the second war and again benefiting from the 'long boom'; born 1940-49, benefiting from the free secondary education of the 1944 Education Act and entering the labour market around the 1960s; 5

3)

4)

5)

6)

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

7)

born 1950-59, our youngest cohort whose careers cover the last third of the century.

There are some well-known problems with cohort analysis. First of all, there is the problem of differential mortality and migration. For example, survivors from the 1920-29 birth cohort and interviewed in, say, 1997 may not be representative of the original cohort born before the war. Some of the original birth cohort may have emigrated, and others may have died. To deal with this problem we have compared our results drawn from later surveys with those for the same birth cohort but drawn from earlier surveys.

4

The second main problem is that mobility is a life-time process: young people tend to enter the labour market in lower positions and subsequently get promoted. So there is the risk that, when we compare birth cohorts, we are comparing people at different stages of their careers and intergenerational processes will thus become confused with intra-generational ones. To deal with this problem we consider only people who have reached 'occupational maturity' at the time of the survey. Following Goldthorpe (1980) we assume that there is relatively little career mobility after the age of 35 and that most people over this age will have reached 'occupational maturity'.

5

Glass in his work on the 1949 survey made a more conservative cut-off at the age of 50, but our checks suggest that the trends are much the same whether we use the younger or the older cutoff. By using the younger cut-off we can of course increase our sample size and explore mobility experiences of more recent birth cohorts.

We cannot claim that our solutions to these problems are perfect, and so all conclusions, especially those about the oldest and youngest birth cohorts must be treated with caution. On the other hand, the problems of alternative methods are probably even greater, and the fact that we can compare results from several different surveys means that our results are likely to be more robust than those of any previous investigators.

6

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

The class schema

To measure the class positions of respondents and their fathers we use a modified version of the class schema devised by Goldthorpe and his associates. Goldthorpe distinguished seven classes, but we have combined two of these - the foremen/technician and skilled manual classes (Goldthorpe's classes V and VI). We do this because there are relatively few women in either of these classes and the numbers become too small for effective analysis. We thus use the following class schema (retaining Goldthorpe's numbering):

6

I II

Higher salariat (professionals, managers and administrators in large enterprises) Lower salariat (semi-professionals, managers and administrators in small enterprises) Routine white-collar workers

III

IV

Petty bourgeoisie (farmers, small employers and own account workers)

V/VI Higher working class (manual foremen, technicians and skilled manual workers) VII Lower working class (semi and unskilled manual workers including agricultural workers). first the salariat

Broadly speaking this class schema distinguishes three main groupings:

(subdivided into higher and lower levels of classes I and II) largely consisting of salaried employees with relatively secure employment, some promotion prospects and various staff benefits such as pension and sick pay schemes; second the petty bourgeoisie (class IV), consisting of independents who are directly exposed to market forces and are not cushioned by the presence of the bureaucratic employers of the salariat; and third the rank-and-file manual workers of the working class (again subdivided into higher and lower levels of classes VI and VII) with relatively higher risks of unemployment, poorer promotion prospects and fewer fringe benefits. Goldthorpe argues that the routine white-collar workers (class III) can be thought of as 7

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

marginal to the salariat, sharing some of its characteristics but with lower pay and privileges, while the foremen/technician class (class V) can be thought of as marginal to the working class.

An important point to note about this class schema is that it is not a completely hierarchical one. Unlike the scale used by Glass (the Hall-Jones scale), it is not intended to be a hierarchical measure of social status where occupations can be ranked on a single dimension of 'social standing'. In particular, in Goldthorpe's schema, Classes III, IV, V and VI cannot be ranked in any straightforward way relative to each other. Classes V and VI, for example, may have higher take-home pay than classes III and IV, but their promotion prospects will be inferior to those of class III and their opportunities for acquiring wealth and assets inferior to those of class IV. These classes have different employment relations from each other, and their members have distinct mobility experiences, but we should not attempt to rank them relative to each other.

This means that, when we want to talk about upward or downward mobility in a vertical sense we ought to exclude movement between classes III, IV and V/VI. The movements between these classes can better be thought of as horizontal rather than vertical movements.

8

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Table 1 Class profiles of men aged 35 and over at the time of the survey (column percentages)

Social Class I II III IV V/VI VII N pre 1900 7.0 11.0 7.9 12.2 31.9 30.0 (417) 1900 -09 10.1 9.2 8.3 10.3 30.3 31.7 (941) 1910 -19 11.6 11.2 9.1 8.7 30.8 28.7 (1393) 1920 -29 16.1 13.1 6.9 8.9 31.9 23.1 (1840) 1930 -39 17.1 15.0 5.8 13.3 25.6 23.2 (1617) 1940 -40 21.6 16.3 4.3 15.0 22.6 20.2 (1095) 1950 -59 23.2 19.1 4.5 15.3 22.0 15.8 (565)

Sample: men aged 35 and over at the time of the survey and on the GB electoral registers. Retired or economically inactive men are assigned to classes on the basis of their last occupation. Economically inactive men (other than retired) are excluded. Table 1 charts the changing distribution of our male respondents in each birth cohort.

7

The

overall picture is a familiar one from studies of the labour market : there is a gradual expansion of the salariat, rising from 18% of the oldest, pre-1900, birth cohort to nearly 41% of the younger, 1950-59 cohort. The expansion was relatively gradual over the three oldest birth cohorts but then accelerated among the cohorts that benefitted from the long boom of the postwar period. If there has been an expansion of the salariat, then there must be compensating contractions elsewhere, and as we can see it is the working class (classes V-VII) that has taken the brunt: in our oldest birth cohort 62% of the respondents held working-class positions, but by the time of the youngest cohort only 37% did so.

One question that can be raised about this picture is whether the classes have really remained comparable over time (Crompton 1981, Goldthorpe 1981). To some extent, as the classes have changed in size, so they have also changed in character. Certainly, the salariat of the youngest birth cohort contains occupations such as computer programmer that simply did not exist among 9

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

the earlier birth cohorts. Other occupations, such as the public-sector professionals, will have seen their privileges decline relative to other salaried employees. However, it is likely that much of the change will have taken place within the broad classes that we have distinguished: while the situation of particular occupations will have changed for the better or worse, the salariat as a whole is still broadly characterised by lower rates of unemployment and more favourable terms and conditions than the working class. Table 2 Class profiles of women aged 35 and over at the time of the survey (column percentages)

Social Class I II III IV V/VI VII N pre 1900 1.6 11.4 23.9 11.4 14.7 37.0 (184) 1900 -09 2.5 13.8 27.6 5.3 10.4 40.3 (513) 1910 2.3 13.5 30.9 5.1 10.8 37.4 (868) 1920 3.7 12.2 34.5 3.0 9.3 37.3 (1311) 1930 4.3 16.2 37.0 4.5 7.4 30.6 (1121) 1940 6.7 19.3 36.3 6.6 6.3 24.7 (911) 1950 -59 9.4 27.2 34.7 5.8 4.1 18.9 (428)

Sample : women aged 35 and over at the time of the survey, and on the GB electoral registers. Women who have never had a job or who describe themselves as 'looking after the home' are excluded. Women who describe themselves as 'retired' are assigned to classes on the basis of their last occupation. Table 2 shows the changing class distribution of women. Again the overall picture is a familiar one: women are concentrated in the lower white-collar work of class III and are underrepresented (relative to men) in the higher salariat and in the higher levels of the working class. However, although the overall distribution is very different, the trends are much the same as for men: women's employment in the salariat has been expanding while that in the working class has been contracting. The one notable divergence between the two tables is that women's employment in Class III has also been expanding while men's has been contracting. This indicates that there has been a major 'feminization' of Class III and the argument that classes do not retain the same character over time may well apply to Class III more than it does to any 10

8

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

other class. Table 3 Class profiles of respondents' fathers

Social Class I II III IV V/VI VII N pre 1900 4.1 8.2 3.9 17.8 34.9 31.1 (562) 1900 4.4 6.7 4.3 16.1 36.1 32.3 (1357) 1910 5.2 6.8 4.8 16.3 35.1 31.8 (2074) 1920 5.2 7.5 5.5 14.9 35.2 31.7 (2882) 1930 6.3 6.4 5.8 14.5 37.8 29.1 (2502) 1940 9.1 8.6 6.2 13.0 36.8 26.4 (1821) 1950 -59 10.8 10.5 5.8 15.6 32.6 24.7 (902)

Sample: both men and women respondents, men as in the note to Table 1 and women as in the note to table 2. Table 3 then shows the changing distribution of the respondents' fathers. Although the fathers and respondents are by definition a generation apart, we cannot neatly allocate fathers to birth cohorts, since of course father's age at the time of the child's birth will vary both between respondents and over time. However, the broad outlines of table 3 are quite similar to that of the earlier cohorts in table 1. We see the same predominance of fathers in the working class that we saw among our older birth cohorts of respondents, and we see a gradual expansion of the salariat over time. But we do not see, and should not expect to see, in the fathers' generation the more rapid expansion associated with the 'long boom' of the post-war period.

9

Men's outflow mobility

In the study of social mobility it is helpful to distinguish between outflow and inflow mobility. In the case of outflow mobility, we examine the class destinations of respondents from different social origins, and we calculate the row percentages. Table 4 gives an example of an outflow table. Analysis of outflow mobility is useful for exploring the question how similar are the 11

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

chances of people from different origins. For example, how equal are the chances of people from salariat and working-class origins of gaining access to the privileged ranks of the salariat?

In contrast, the analysis of inflow mobility is concerned with the social composition of the different classes, and is based on column percentages. Table 10 gives an example of an inflow table. Inflow analysis thus tells us where the current occupants of a particular class came from. Is a class homogeneous in its composition, or does it include people from a wide variety of social origins? These questions about the social composition of the classes may have

implications for the extent of 'class formation' and the potential for collective class action (Goldthorpe 1980, De Graaf, Nieuwbeerta and Heath 1994).

We begin with outflow mobility and then turn in the next section to inflow mobility. Before looking at the trends over time, however, it is useful to look at the picture for the cumulated sample as a whole. The overall table for men is given in table 4. Table 4 Men's outflow mobility (row percentages)

Father's Class I I II III IV V/VI VII All 46 30 27 15 12 9 16 II 23 29 21 13 12 9 14 III 5 10 12 7 7 5 7 Son's Class IV 11 10 9 29 8 7 11 V/VI 8 12 19 20 38 32 28 VII 6 9 13 17 23 38 24 N (447) (535) (381) (1103) (2626) (2204) (7296)

Sample: see note to Table 1. The main diagonal running from the top left to bottom right of table 4 shows the proportions from each origin class who were intergenerationally stable, that is who followed in their father's 12

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

footsteps. Respondents who lie in the cells above and below the diagonal can be thought of as intergenerationally mobile. Looking down the diagonal, we see the highest chance of following in father's footsteps is at the apex of the class structure, where 46% of sons from Class I origins themselves had moved into class I positions at the time they were interviewed. The percentage then falls sharply as we move down the diagonal, reaching a low of 12% among men from class III origins. The percentage then rises again to 29% in the petty bourgeoisie, 38% in the upper working class and 38% again in the lower working class. The general character of this picture is the same as that which Glass had found fifty years ago: "the highest rigidity is found in the professional and high administrative cadres, and the least in the ... routine-nonmanual category. The latter category is, in fact, a kind of valley, the rigidity increasing on each side" (Glass, 1954, p19).

10

Looking down the first column, we see that men from class I backgrounds had far superior chances of reaching class I themselves than did men from lower working class backgrounds: 9% of men from lower working-class origins were in class 1 compared with 46% of those from higher salariat origins. Similarly, looking down the last column, men from class I backgrounds had better chances of avoiding demotion to the lower working class than did those in any other class.

Short-range mobility is also more common than long-range movement. Thus it was more common for men from class I origins to move the short step down to class II than it was for them to move all the say down to the working class. Similarly, men from class VII origins were much more likely to move the short distance to the upper working class than they were to achieve long-range mobility to the salariat. Again, this parallels Glass's findings from the 1949 study.

On the other hand, unlike the findings from 1949, there has clearly been a surplus of upward over downward mobility. The percentages below the diagonal are generally rather larger than those above the diagonal. 13

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

How has this overall picture changed over time? In table 5 we show, for each of our birth cohorts, the proportions of intergenerally stable men from each class origin, ie the proportions lying on the main diagonal. For example, the cell in the top right corner, for the 1950-59 birth cohort, shows that 68% of men in this cohort from class I origins themselves were in class I positions at the time they were interviewed. The next cell down shows that 40% of men in this cohort from class II backgrounds were themselves in class II when they were interviewed, and so on. In the bottom two rows of the table we combine the upper and lower salariat and the upper and lower working class respectively. This enables us to see how many men from the salariat or from the working class as a whole followed in their father's footsteps.

Table 5 Trends in intergenerational stability (percentage remaining in the same class as their father): men

pre 1900 Class I Class II Class III Class IV Class V/VI Class VII (14) 19 (31) (15) 34 (74) 49 (144) 53 (119) 1900 -09 36 (36) 22 (55) 21 (38) 32 (136) 38 (319) 47 (303) 1910 -19 39 (72) 33 (92) 21 (58) 25 (208) 41 (445) 46 (417) 1920 -29 48 (87) 28 (124) 15 (86) 28 (248) 40 (622) 34 (536) 1930 -39 47 (100) 38 (88) 6 (85) 31 (239) 34 (536) 35 (437) 1940 -49 44 (90) 22 (93) 7 (65) 28 (124) 31 (385) 28 (263) 1950 -59 68 (48) 40 (51) 1 (34) 31 (75) 35 (175) 33 (129)

Salariat (1 + ll) Working class (V/VI + VII)

51 76

49 75

61 73

69 67

67 60

62 53

73 50

Sample: see note to Table 1. Figures in brackets give the Ns for the origin class.

14

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Some of the percentages in table 5 are based on rather small numbers of respondents, and so there will be considerable sampling variation. Nevertheless, there are some clear trends. First, the proportion of sons from class I and II backgrounds who have been able to achieve the same positions as their fathers has tended to increase over time. In the oldest birth cohort 51% of sons from salariat backgrounds secured positions in classes I or II themselves, while by the youngest birth cohort this figure had risen to 73%. Secondly, there has been a comparable decline in stability in the working class. In the oldest birth cohort 76% of men from workingclass origins were intergenerationally stable, but this figure had fallen to 50% by the time of the youngest cohort. Thirdly, and contrastingly, intergenerational stability in the petty bourgeoisie has been pretty well unchanged over the whole period, oscillating around 35%. These trends can be checked against the data of the 1949 and 1972 studies; the figures for the corresponding cohorts in the different studies are comfortingly close, except for Glass's youngest cohort who had not at that time reached occupational maturity (Glass 1954 p 186).

From these tables we can calculate some summary indices of mobility. Thus the first row of table 6 shows the overall proportion in each birth cohort as whole who remained in the same class as their fathers. The figure falls fairly steadily from 43% in the oldest cohort to 28% in the penultimate cohort. There is then a rise in the final, youngest cohort. The decline in the

percentage who were inter-generationally stable should not surprise us. While, as Table 5 showed, the increasing mobility out of the working class was matched by increasing stability in the salariat, the working class was for much of our period considerably larger than the salariat. Hence the overall figure of stability tended to decline, reflecting the numerical predominance of men mobile out of the working class. In the youngest cohort, in contrast, the salariat has actually grown somewhat larger than the working class; hence the increase in overall stability.

15

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Table 6 Summary indexes of mobility: men pre 1900 Percentage stable upwardly mobile downwardly mobile horizontal movements N 43 27 20 10 (397) 1900 1910 1920 1930 1940 1950 -59 35 42 13 10 (533)

39 29 21 11 (890)

38 30 20 12 (1303)

35 39 17 9 (1778)

34 38 18 10 (1572)

28 42 19 11 (1074)

Sample: see note to Table 1. Table 6 also shows the trends in upward and downward mobility. Recall that we must treat movements between classes III, IV and V/VI as horizontal rather than vertical. Upward

mobility therefore excludes these horizontal movements but includes all movements out of class VII, all movements into class I and all movements from below into class II. As we can see, there has been a substantial net surplus of upward over downward mobility throughout our period, the surplus in fact tending to rise and reaching a maximum in our youngest birth cohort of 29 points.

This pattern is largely driven by the changes in the class structure which we described earlier. As we argued in the introduction, there has been increasing 'room at the top' and hence an increase in upward mobility is arithmetically inevitable.

16

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Women's outflow mobility

The picture for women is rather different from that for the men. Comparing table 7 with table 4, we can see that fewer women than men followed in their father's footsteps into the higher salariat, into the petty bourgeoisie or into skilled manual work. Instead there is a lot of movement from all classes alike into the lower white-collar work of class III These patterns largely reflect the differing class distributions of men and women: since there are so many more women than men in class III, it is hardly surprising that there is more mobility into it. However, after taking account of the differences in their class distributions, there are still some significant differences in women's mobility patterns from those of men. In particular, women show a weaker tendency to inherit positions in the petty bourgeoisie; they also less likely to follow in father's footsteps into either the upper or the lower working classes.

11

Table 7 Women's outflow mobility (row percentages)

Father's Class

Daughter's Class I II 32 32 23 17 13 10 16 III 35 41 42 34 35 29 34 IV 9 5 5 10 4 3 5 V/VI 3 4 4 7 10 11 9 VII 7 13 17 28 35 46 33 N (314) (373) (271) (722) (1712) (1413) (4804)

I II III IV V/VI VII All

14 6 9 5 3 2 4

Sample: see note to Table 2. Table 8 shows the trends over time in women's intergenerational stability. Although, as we have just seen, the absolute levels are very different from men's, the trends are more similar. Thus there has been a trend towards greater stability for women from salariat origins, the percentage 17

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

stable rising from 33% in the oldest cohort to 50% in the youngest, albeit somewhat less powerful trend than that for men. There is also a trend towards reduced stability for women from working-class origins, the percentage stable falling from 68% to 30%. This is a rather more dramatic trend than for men. Table 8 Trends in intergenerational stability (percentage remaining in the same class as their father): women

Father's class Class I Class II Class III Class IV Class V/VI Class VII

pre 1900 (9) (15) (7) 19 (26) 17 (52) 46 (56)

1900 8 (24) 27 (37) 50 (20) 12 (82) 16 (171) 58 (136)

1910 10 (36) 32 (49) 42 (43) 14 (130) 11 (283) 47 (242)

1920 15 (61) 33 (93) 35 (73) 6 (183) 13 (392) 53 (377)

1930 15 (58) 36 (73) 49 (61) 10 (124) 7 (409) 41 (291)

1940 15 (76) 28 (63) 40 (48) 7 (112) 8 (285) 37 (217)

1950 -59 14 (50) 34 (43) 24 (19) 10 (66) 4 (119) 30 (94)

Classes I, 11 Classes V, VI and VII

33 68

40 64

37 56

40 57

42 44

43 41

50 30

Sample: see note to Table 2.

18

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Table 9 Summary indexes of mobility: women

pre 1900 Percentage stable upwardly mobile downwardly mobile horizontal movements N 31 24 30 15 (219) 1900 -09 29 22 32 17 (602) 1910 -19 26 27 30 18 (1000) 1920 -29 28 23 28 21 (1403) 1930 -39 22 29 27 22 (1164) 1940 -49 20 32 26 22 (886) 1950 -59 17 36 27 20 (298)

Sample: see note to Table 2.

Turning to the summary indexes for women's outflow mobility, we see that the trends are roughly similar to the men's, although the levels are different. Thus there is in all birth cohorts more downward mobility and less upward mobility for women than for men; but over time there has been a marked increase in upward mobility, just as there has been for men, and there has been a modest decline in downwards mobility too. Thus for women upwards mobility has increased by twelve percentage points compared with a fifteen point rise for men over this period. Women's downward mobility has declined by three percentage points, compared with a decline of seven points for men.

Horizontal movements, between classes III, IV and V/VI, have however increased somewhat more for women than for men, and so the net result is a slightly larger decline in immobility. In the case of women, immobility has fallen by fourteen points from 31% to 17%, whereas for men the decline was eight points from 43% in the oldest birth cohort to 35% in the youngest. Again, it is very important to emphasize that these changes have been very largely driven by the changing class distributions over time of the women and of their fathers. There has been less 19

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

room at the top for women, and so their gains in upward mobility have been less dramatic than men's. But there has been much more room in the middle for women as routine nonmanual work has become feminized.

Inflow mobility

We now turn to the analysis of inflow mobility. As we explained earlier, inflow mobility tells us about the composition of social classes and tells us how diverse a class's members are in their social origins. It thus addresses questions about the extent of class formation, the assumption being that a class which is homogeneous in its composition will exhibit a more distinct subculture of its own.

In the analysis of inflow mobility we follow a rather different strategy from that of outflow mobility. If we are interested in how homogenous a class is, then the issue of occupational maturity is no longer so relevant. The fact that a class may have a lot of young people in it at an early stage of their careers still tells us something about the composition of that class.

Nor is birth cohort analysis quite so relevant to the analysis of trends over time in class composition: members of several different birth cohorts may all be members of a class at the same moment in historical time. We therefore look at each survey separately rather than conducting a pooled analysis of birth cohorts.

Finally, there is no substantive reason to distinguish men from women, unless we believe that men and women form different class subcultures. (In fact the separate inflow tables for men and women are very similar.) We therefore pool both men and women. However, it could also be argued that, from the point of view of class formation, it may be better to treat the family as the unit of stratification and to assign husbands and wives to the same class position. How this should be done is a contentious matter (see Sorensen 1993), but it is unlikely that alternative methods will produce greatly differing pictures of the trends over time. 20

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

We begin, as with the analysis of outflow mobility, by looking at the pattern for the sample as a whole in table 10. We then turn in Table 11 to the trends over time.

Table 10 Inflow mobility (column percentages)

Father's class I I II III IV V/VI VII N 20.7 15.7 8.5 13.4 26.7 14.9 (1885) II 13.4 16.8 7.8 14.0 29.9 18.1 (2589) III 8.6 11.0 7.4 13.5 35.1 24.6 (3388) Respondent's Class IV 7.7 6.5 4.3 36.2 25.8 19.5 (1367) V/VI 3.0 4.4 3.4 9.9 46.3 33.1 (3673) VII 2.3 3.6 3.2 11.5 36.2 43.1 (4589) All 7.7 8.7 5.4 14.1 35.3 28.8 (17490)

Sample: economically active or retired men and women on the GB electoral registers. Retired or economically inactive respondents are assigned to classes on the basis of their last occupation. Those looking after the home are excluded. Respondents aged 21 and over in 1964, 18 and over in later surveys. Table 10 shows how many members of each class are what might be termed 'second generation', that is how many originated in the same class that they currently occupy. This shows a remarkable contrast with the outflow tables: for example, very few members of the salariat are 'second generation'. Most are newcomers who have been upwardly mobile, with particularly large number of newcomers from the working class. The routine white-collar class (class III) then shows a particularly diverse composition, the social origins of its members closely matching the origins of the sample as a whole (as shown in the final column of the table).

In contrast, the petty bourgeoisie (class IV) shows a relatively high proportion of secondgeneration members, as do the upper and lower working classes. Indeed, if we combine classes V, VI and VII together we find that around 80% are second generation members. This contrasts with around 30% for the salariat. The key conclusion, then, is that the salariat is much more 21

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

diverse in its origins than the working class. This is a very different story from the one Glass told, when he claimed that "... the general picture so far is of a rather stable social structure, and one in which social status has tended to operate within, so to speak, a closed circuit."

Is this because British society has changed since the 1949 survey? We attempt to answer this question in table 11 which charts inflow mobility from a selection of the BES surveys. Table 11 Percentage of second-generation members of each class: men and women

1949 Class I Class II Class III Class IV Class V/VI Class VII 1964 13 (76) 16 (135) 7 (181) 47 (71) 43 (284) 51 (358) 1970 21 (84) 18 (164) 4 (207) 44 (91) 41 (325) 56 (391) 1979 22 (164) 12 (185) 9 (227) 36 (110) 47 (344) 41 (372) 1987 22 (320) 16 (441) 7 (573) 29 (210) 49 (475) 39 (625)

1997

24 (307) 18 (335) 3 (359) 30 (181) 44 (315) 44 (369)

Salariat Working class

51 (664) 77 (2807)

29 79

27 82

30 79

34 78

37 75

Sample: as in note to table 10. Figures in brackets give the base Ns.

Table 11 shows remarkable consistency in the inflow figures from 1964 to 1997.

The

percentage of second-generation members in the salariat has remained around 30%, while that in the working class has remained around 80%. The one notable change is the decline in the proportion of second-generation members in the petty bourgeoisie. 22

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

As we argued earlier, the classification of social status used by Glass and Hall is not comparable with that used in the later surveys. However, we can make a rough approximation, and this suggests that, in 1949, the salariat had a much higher proportion of second-generation members (although still not as high as the working class). This must be treated with caution, but it makes reasonably good sense: in 1949 the salariat had not yet started to expand. It was the continued expansion in the later post-war period that required the salariat to recruit so many people from below, but before the expansion began it is likely that social status did indeed tend to operate within a more closed social circuit.

12

Relative mobility rates

The trends which we have observed so far have been largely influenced by the changes in the class distributions of the respondents and their fathers. In particular the increasing room at the top has meant that there could simultaneously be improved chances of people from privileged backgrounds remaining in the salariat and also improved chances of people from less privileged backgrounds gaining access to the salariat.

However, it is also interesting to ask whether the changes in the distributions is the whole story or whether there has been an increase in interchange between the classes over and above that which would have been expected from the changing distributions. An earlier generation of sociologists divided up total mobility into 'forced' or 'structural' mobility and 'exchange' or 'pure' mobility. The structural mobility was that amount which was required by the changing thus if there was increasing room at the top, some upward mobility was

distributions;

inevitable. Exchange mobility covered all the rest of the movement up and down; by definition the upward component of exchange mobility would be balanced by an equal downward movement.

23

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

It is now customary to think of this exchange mobility as the underlying fluidity in the mobility process, and it is measured by odds ratios. Odds ratios are not directly affected by the changes in the class distributions of the respondents and their fathers, and thus are a convenient way of measuring fluidity. They can be thought of as measuring the relative risks (or odds) of people from two different origins reaching a given destination and of avoiding another destination.

In a mobility table, there are many possible odds ratios that can be calculated. For simplicity we shall focus on what are termed the symmetrical odds ratios. Thus in table 4, 46% of men from class I origins stayed in class I while 6% were downwardly mobile into class VII; this can be expressed as odds of 7.7:1 (since 46/6 = 7.7). However, only 9% of men from Class VII origins reached class I themselves while 38% stayed in class VII, giving odds of 1:4.2 (since 38/9 = 4.2). The ratio of these two odds is 32.3:1.

An odds ratio of 1 would indicate that the two classes concerned had equal competitive chances of reaching one destination and avoiding the other. The odds ratios would all be 1 in a society where social origins were unrelated to class destinations. The larger the odds ratio, the more unequal the competition, and the ratio of 33:1 suggests that the competition is extremely unequal.

24

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Table 12 Symmetrical odds ratios: men

pre 1900 1900 10.0 1910 19.0 1920 14.0 1930 10.3 1940 5.6 1950-59 7.7

salariat: working class salariat: petty bourgeoisie petty bourgeoisie: working class

16.0

5.5

7.0

6.7

9.4

6.8

4.1

7.3

9.5

11.0

9.6

8.2

5.8

3.2

4.9

Sample: as in note to Table 1.

Table 12 illustrates the trends by focussing on three symmetrical odds ratio: that between men from salariat (classes I and II combined) and working-class origins (V, VI and VII) to reach the salariat and avoid working-class destinations; that between men from salariat and petty bourgeois origins to reach the salariat and avoid the petty bourgeoisie themselves; and that between people from petty bourgeois and working-class origins to reach the petty bourgeoisie and avoid the working class.

As we can see, the most unequal competition is that between men from salariat and workingclass origins where the odds ratios have been around 10:1. The competition between men from salariat and petty-bourgeois origins is rather more equal in general, with odds ratios around 6:1, while that between men from petty bourgeois and working-class origins is of broadly similar magnitude.

However, there does seem to be some suggestion that the odds ratios have declined somewhat over the period. In particular, the working class seems to have improved its competitive situation vis-a-vis the other two classes. Thus the salariat:working class odds ratios falls below 10:1 in the youngest two cohorts, while the petty bourgeoisie:working class odds ratio falls 25

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

below 5:1. In both cases, the two youngest cohorts exhibit the smallest odds ratios in the series. Table 13 Symmetrical odds ratios: women

pre 1900 1900 17.2 2.1 1910 15.2 1.0 1920 13.4 1.5 1930 7.3 1.7 1940 10.6 1.3 1950-59 5.8 0.6

salariat: working salariat: routine nonmanual routine nonmanual: working

-

-

6.5

4.5

1.9

3.3

2.7

3.2

Sample: as in note to table 2. The figures for the oldest cohort have very large confidence intervals and are therefore not shown. In the case of women we focus not on the petty bourgeoisie, of which few women are members, but on the routine non-manual class. As we can see from table 13, the salariat:working class odds ratio is broadly comparable to that for the men, although the downwards trend is less apparent. The odds ratio for the salariat: routine nonmanual contrast, however, becomes close to unity, while the working-class:routine nonmanual ratio is also quite low.

Odds ratios, like other statistical measures based on sample surveys, are subject to sampling error, and the odds ratios reviewed in tables 12 and 13 are only a small selection of all the possible ratios that we could calculate. Fortunately, we can conduct a general statistical test of whether the complete set of odds ratios has remained constant across birth cohorts by using loglinear models. The results are reported in table 14.

26

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Table 14 Loglinear models of trends in social fluidity

Men Women

Chi2 F,R,C +FC,RC +FR Unidiff 2050.1 1642.4 217.8 192.0

df 235 175 150 144

p .000 .000 .000 .0002

Chi2 1078.6 742.6 175.9 167.2

df 235 175 150 144

p .000 .000 .055 .09

Sample: as in notes to tables 1 and 2. For technical reasons concerned with model fitting, 0.01 has been added to all empty cells. We fit a sequence of models. (For a more detailed discussion of fitting trends to multi-way tables see Payne, Payne and Heath 1994.) The first model postulates that the three variables of father's class (F), respondent's class (R) and cohort (C) are independent of each other. This model gives a very poor fit to the data. The second model postulates that the class distributions of both respondents and their fathers have changed across birth cohorts, but that father's class and respondent's class are independent of each other. This model gives a somewhat better fit but is still a long way short of accounting for the observed patterns.

The third model then postulates that there is, in addition, an association between father's and respondent's class but that this association (as measured by the full set of odds ratios) has remained constant across birth cohorts. This is the key model for testing whether fluidity has remained constant over time and we can describe it as a 'constant fluidity model'. The model makes a dramatic improvement in fit compared with the second model, but in the case of men it still falls somewhat short of an acceptable fit to the data.

The story told by Table 14, then, is that we should reject the hypothesis that, in the case of men, the odds ratios have remained constant across the birth cohorts. The discrepancy between the predictions of the model and the observed mobility patterns is greater than could be expected by 27

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

chance. In the case of women, however, the hypothesis cannot be confidently rejected since the discrepancies between the model and the observed data could have occurred by chance (at the conventional 0.05 level). In other words, there seems to have been some change in men's fluidity, but possibly not in the case of women.

Our third model provides a global test of the hypothesis that social fluidity has been constant across birth cohorts; it does not tell us what form the changes have taken. However, we can gain some more insight into the pattern of the changes in fluidity if we fit the 'uniform difference' model. This is a log-multiplicative model that was developed independently by Xie (1992), van der Heijden and Jansen (1992) and by Erikson and Goldthorpe (1992). The model starts from the assumption that there is a general pattern to the association between father's class and respondent's class (as defined by the odds ratios) which persists over time even though the overall strength of the association may vary from survey to survey. The model imposes the constraint that all the odds ratios either uniformly increase or uniformly decrease as we move between surveys, and it estimates a 'uniform difference' parameter for each survey. (Essentially, these parameters are multipliers which operate on the full set of odds ratios.) These parameters can be thought of as measuring the extent to which the association between father's class and respondent's class strengthened or weakened in each survey.

Table 14 shows that, in the case of men, the uniform difference model gives a significant improvement in fit over the 'constant social fluidity' model. Figures 1 and 2 plot the uniform difference parameters (together with the 95% confidence intervals) for men and women respectively over our seven birth cohorts. We set the first cohort to one, and the subsequent parameters show the changes in fluidity relative to the first cohort. In the case of men we can see a fairly clear downward trend over time, albeit one that is reversed among the final birth cohort. All the estimates are less than one, indicating that the association between father's and son's class is weaker in the younger birth cohorts; that is, the odds ratios tend to become smaller (since they are being multiplied by a parameter that is less than one). A weakened association is of course another way of saying that fluidity has increased. 28

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

29

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

In the case of women there are also hints of a downward trend although the changes are smaller and, as we noted earlier, we cannot strictly reject the hypothesis that women's fluidity has remained constant over time.

The result for men is at some variance with the conclusions reached by Goldthorpe (1980) and Goldthorpe and Payne (1986) who had found that fluidity had remained constant across birth cohorts in Britain. Goldthorpe's work was based on a 1972 survey of men (later updated by the 1983 BES) and did not cover either our two earliest cohorts or the most recent one.

13

The

discrepancy between his results and ours, therefore, may simply reflect the different time periods covered. It is possible that, over the longer time period available to us, there has been a real, albeit small, increase in the openness of British society that was not visible over the shorter time-period covered by Goldthorpe's work.

However, we should also recognise that there might be methodological explanations for the apparent trend. In particular, the data on the earliest birth cohorts come from reports given by the oldest respondents and they may be subject to some systematic errors. While cohort analysis is the only practical way to investigate trends in mobility over the twentieth century, the limitations of cohort analysis must be borne in mind.

Acknowledgements

The British Election Surveys have been made available by the Data Archive at the University of Essex. We are grateful to them and to the original investigators, who include David Butler and Donald Stokes, Ivor Crewe, Bo Sarlvik and David Robertson, and Anthony Heath, Roger Jowell, John Curtice and Pippa Norris (1983-97). The BES has been funded by a number of bodies, and we would especially like to thank the ESRC and the Gatsby Charitable Trusts.

30

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Bibliography Breen, Richard and Christopher T Whelan (1999) Social mobility in Ireland: a comparative analysis. In A F Heath, R Breen and C Whelan (eds) Ireland North and South: Perspectives from Social Science. Proceedings of the British Academy. Oxford: Clarendon Press. Chapman, S J and Abbott (1913) The tendency of children to enter their father's trades, Journal of the Royal Statistical Society 76: 599-604. Chapman, S J and Marquis (1912) The recruiting of the employing classes from the ranks of the wage earners in the cotton industry, Journal of the Royal Statistical Society 75: 293-306. Crompton, R (1980) class mobility in modern Britain. Sociology 14: 117-9. De Graaf, N.D., Nieuwbeerta, P. and Heath, A.F. (1994) 'Class mobility and political preferences: individual and contextual effects.' American Journal of Sociology 100: 997-1027. Erikson, R and Goldthorpe, J H (1992) The Constant Flux: A study of Class Mobility in Industrial Societies. Oxford: Clarendon Press. Ganzeboom, H B G, Luijkx, R and Treiman D J (1989) Intergenerational class mobility in comparative perspective, Research in Social Stratification and Mobility 8: 3-79. Ginsberg, H (1929) Interchange between social classes. Economic Journal: 554-565. Glass D V (1954) Introduction. In D V Glass (ed) Social Mobility in Britain. London: Routledge & Kegan Paul. Glass D V and Hall J R (1954) Social mobility in Great Britain: a study of inter-generation changes in status. In D V Glass (ed) Social Mobility in Britain. London: Routledge & Kegan Paul. Goldthorpe J H with C Llewellyn and C Payne (1980) Social Mobility and Class Structure in Modern Britain. Oxford: Clarendon Press. Goldthorpe J H (1980) Reply to Crompton, Sociology 14: 121-3. Goldthorpe, J H and C Payne (1986) Trends in intergenerational class mobility in England and Wales 1972-1983. Sociology 20: 1-24 Halsey, Heath and Ridge (1980) Origins and Destinations. Oxford: Clarendon Press.

31

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

Heath, A F (1981) Social Mobility. Glasgow: Fontana. van der Heijden, P G M and Jansen, W (1992) A class of models for the simultaneous analysis of square contingency tables. Pp 125-30 in L Fahrmeir et al (eds) Advances in GLIM and Statistical Modelling. Berlin: Springer. Hope, K (1981) Trends in the openness of British society in the present century, Research in Social Stratification and Mobility 1: 127-70. Hout, M (1989 Following in Father's Footsteps: Social Mobility in Ireland. Cambridge, Mass.: Harvard University Press. Kelsall, R K and Mitchell, S (1959) Married women and employment in England and Wales, Population Studies 13:19-33. Kerckhoff, A C and Trott, J M (1993) Educational attainment in a changing educational system: the case of England and Wales, in Y Shavit and H-P Blossfeld (eds) Persistent Inequality: Changing Educational Attainment in Thirteen Countries. Boulder, Colorado: Westview Press. Macdonald, K I and Ridge, J M (1972) Social mobility. In A H Halsey (ed) Trends in British Society since 1900. Basingstoke: Macmillan. Marshall, G, Swift, A and Roberts, S (1997) Against the Odds? Social Class and Social Justice in Industrial Societies. Oxford: Clarendon Press. Payne, C, Payne, J and Heath, A F (1994) Modelling trends in multi-way tables. Pp 41-74 in Dale, A and Davies, R B (eds) Analyzing Social and Political Change: A Casebook of Methods. London: Sage. Payne, G, Ford, G and Robertson, C (1976) Changes in occupational mobility in Scotland, Scottish Journal of Sociology 1: 57-79. Reid, Ivan (1998) Class in Britain. Cambridge: Polity Press. Ridge, J M (ed) (1974) Mobility in Britain Reconsidered. Oxford: Clarendon Press. Sorensen, Annemette (1994) Women, family and class, Annual Review of Sociology 20:27-47. Xie, Y (1992) The log-multiplicative layer effect model for analysing mobility tables, American Sociological Review 57: 380-95.

32

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

1. In the 1949 survey respondents were asked about their current occupation whereas they were asked to report on the last main occupation of their fathers. For younger respondents, then, the occupational data refer to an earlier stage of their occupational careers from that of their fathers. Given the general tendency of men (although less so of women) to move upwards in the course of their careers, this procedure is likely to underestimate the eventual amount of upward mobility that will take place among the respondents. 2. Most of the surveys used the electoral register as the sampling frame and then drew clustered random samples from this frame. In some cases alternative frames have been used, such as the post office address files in the case of the 1997 survey, but with the use of appropriate filters it is possible to establish electorate samples throughout. The electorate will not be identical to the adult population: there will be some residents who fail to register and others who are not eligible. The 1992 and 1997 surveys over-sampled in Scotland in order to permit more detailed investigation of Scottish voting behaviour, and these two surveys have been weighted so as to make them representative of Great Britain as a whole. The total (weighted) sample size is 25,573. After age and birth cohort selections this falls to 17,595 (8311 men and 9283 women). 3. The great strength of the election surveys, for our present purpose, is that they have all carried out a detailed coding of both respondents' and fathers' occupations, using the official government classifications of occupations. These classifications have been revised for each new Census, and this will have introduced some element of incomparability over the series. However, the checks that we have been able to cary out suggest that, at least when occupations are aggregated into social classes, the errors are of an acceptable level. 4. For example, in the case of men in cohorts 3,4 and 5 it is possible to check our results against those obtained by the 1972 survey. 5. Goldthorpe used the notion of 'occupational maturity' in his analysis of men's mobility. The concept may have less relevance for women's mobility, since many married women leave the labour market, or work part-time, when their children are young and then return to paid employment, but at much the same level as before, in their mid-thirties. In the absence of any better alterntive we use the same cut-off for the women as for men. 6. We do however depart from Goldthorpe's nomenclature. First, we use the term 'salariat' rather than his term 'service class' to refer to classes I and II. Second, we broaden the term 'working class' include class V along with classes VI and VII. Goldthorpe terms class V one of the intermediate classes, but our own preference is to group it with the other manual classes (with whom we believe it has more in common). 7. Men who have never had a job, or who gave inadequate occupational data, are excluded. There were 148 men in total who have therefore been excluded.

33

Twentieth Century Trends in Social Mobility in Britain by Anthony Heath and Clive Payne

8. The earlier BES surveys did not ask for the previous occupation of married women who described themselves as housewives, and accordingly we have excluded such women throughout. This makes good substantive sense too as any previous occupation may have come from the beginning of the woman's career. However, in the case of surveys which ascertained the previous occupations of women who described themselves as 'retired', we have treated such women in the same way as men and have included their previous occupation. These occupations will have been their last main occupation before retirement. 9. Throughout the surveys respondents were asked to give the details of the occupation that their father had had when the respondent was growing up. This is preferable to the practice followed by Glass in the 1949 survey of asking about father's last main occupation. We combine both men and women since there is no statistically significant difference in the class distributions of the fathers of the men and women in our selected sample. 10. In this quotation Glass is summarizing his findings from an analysis which used the index of association, rather than the simple percentages on the diagonal which we have used. While there are some well-known problems with the index of association (Billewicz *** ), Glass's remarks as quoted actually apply quite well to the simple percentages reported by Glass and Hall. 11. These conclusions are based on a loglinear analysis which controls for the marginal distributions. The model which postulates that the association between father's and respondent's class is the same for women as it is for men is on the borderline of statistical significance: chi square=38.0 for 25 degrees of freedom, p = .046. However, inspecting the residuals for this model we see that there are highly significant residuals for the diagonal cells representing class inheritance of petty bourgeois positions (-3.21 for women), of upper working-class positions (2.49) and of lower working-class positions (-3.24). 12. The figures for 1949 in table 11 include both men and women and are based on figures presented in Glass and Hall (1954) and Kelsall and Mitchell (1959). 13. The 1972 study covered men aged 20-64 and it is thus possible to track cohorts born from 1908 to 1952. However, the younger men will not have reached occupational maturity by 1972 and therefore Goldthorpe's cohorts effectively overlap only with our middle three cohorts.

34

Information

halsey

35 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

174080


You might also be interested in

BETA
Microsoft Word - Platt paper submission.doc
FOC journal
2010-2012 UTSA Undergraduate Catalog
2-Cover 2