i8 Social Mobility* RICHARD BREGN AND DAVID ROTTMAN II! INTRODUCTION Neither the position of families nor individuals within the class structure nor that structure itself, remain constant over time. Individuals and families change their class position and the class structure itself evolves, as some occupations decline and others become more numerous. Doth these sorts of change have been intensively studied by sociologists and other social scientists. Examining the development of a ctass structure over time involves adopting a historical perspective, as in the work of Przeworski ct al. (1980) or Wright and Martin (1987). The extent and the way in which families move through the class structure—between positions in it, in other words—is the subject matter of the study of social mobility. Social mobility has long been a central topic of sociological inquiry, and has been particularly actively pursued over the past 25 years. In this chapter our aim is to explain what the study of social mobility is, to give a brief explanation of the methods used in social mobility analysis, and to summarize the main results of recent research. Before we begin, * Fine published in 1995; from Ci,iu SnJliJitminn: A CurnfiiMlire {'tnperiirc. I I I 1 J!.} WHO GETS WHAT? however, we need first to set the scene by saying something about the temporal dimension of soda! class. Ill SOCIAL MOBILITY When we examine social mobility—-that is, how and why people or families change position in the class structure—we are usually interested in two things, first, in the nature of mobility: bow much change in class position is there; how far do people move from their original class; is there more upward or downward mobility in society; haw does mobility affect people's behaviour and attitudes (for example, when people are mobile out of ciass X and into class Y, to what extent do they continue to behave (for example to vote) in ways that are typical ofclass X—from which they have come—and to what extent do they take on the behaviour typical of the class to which they move—class Y?)? Second, we are interested in the consequences of mobility for the class structure. For instance, if we take one class virtually all of whose members have always been in that class and whose families were before them, and contrast it with another class which is chiefly made up of families which have been mobile into that class from outside it, how ate the two likely to differ? In particular, will the members of the more 'closed' of the two classes be more likely to view themselves as constituting a class for themselves, in Marx's terms, than will the members of the more open class? These two perspectives on mobility are, as we might have anticipated, not entirely distinct. Classes are, after all, made up of families and the individuals within them: they both comprise a class and are influenced by being members of it, in much the same way that individual actions:' are shaped by the existence of constraints but also help, to a greater or lesser extent, to change or maintain these constraints. INTER-GBNERATIONAL MOBILITY The most commonly studied form of class mobility is termed inter-generational mobility. j1 This takes the form of a comparison of a person's current social class with the class that his or her family occupied at the time the petson was growing up. The vast majority of studies of inter-generational mobility have analyzed data for men: typically, then, the comparison is between the class position occupied by a man's family at the time lie was growing up (say, at age 16) and. v the class position he currently occupies. When these studies are carried out on populations or ■::<: large samples the relationship between the two is shown as a two-dimensional cross-tabulation ■ i (see Table 1). This table shows us the number of men who fall into each combination of cur- ,;: rent class and the class they were part of when they were growing up. Such tables are some- ,:■ times called origin by destination tables, since the name 'class origin' is usually given to the class they occupied when they were growing up, ; and 'class destination' is the name given to their .' current class. Hence the process of mobility 15 ; conceived of rather like a journey or a flow from ■ : an origin to a destination. The labelling of the : two margins of the table in this way overcomes a problem of interpretation that had caused some problems in mobility analysis', this is that, despite . t" —- Sacinl Mobility 315 Iff WBLE 5 THREE-CLASS MOBILITY TABLET MEN IN ENGLAND ÄND WALES 1372 Ills Current (tlcstiitiuisrit) class r 2 3 Total ORIGIN CLASS 1 731 322 IB9 1242 2 857 1 140 1109 3106 3 787 1386 2915 5086 1 ~ Total 2375 2848 4213 9436 : iiote: Classes are: 1 = Service;! = Intermediate;3 = Working, sounds: Calculated from Goldlhorpe et al, (19S0/87),Table 2.2. the term 'inter-generational' mobility, the distribution of men across the origin classes did not represent the class structure as it was at any particular point in time or for any particular generation. Mobility data is gathered from a survey of men (orpeople) in the currentpopulation (sometimes only the current working population) and thus, if it is representative of that population, it cannot, in its distribution over the origin classes, be representative of any other population (except by chance)—such as a particular generation or age group, or even of the population of fathers or families of men currently in the workforce (Duncan 1966). Therefore, mobility tables do not show us how one generation's class distribution evolves into the next generation's; rather, they show us how the classes men start out in (their origins) relate to the class they are in at the time of the survey (class destinations'). Table 1 refers to a sample of 9436 men in England and Wales interviewed in 1972 for the Oxford Mobility Study (Goldthorpc 1980/87). It is immediately evident, of course, that what the table looks like will depend very much on how many classes we identify: here we have used a three-class categorization, and, as is conventionally the case, the same three classes are identified for origins as for destinations. These three classes are termed the service, intermediate and working class, respectively * * *.;Tlie working class comprises men in largely manual occupations, whether these arc considered skilled or not. The service chss is made up largely of professionals, managers, administrators, supervisors of white-collar workers and owners of capital. In the middle, the intermediate class comprises other white-collar workers—such as clerical workers, salespersons, employees in services—small proprietors, such as farmers and smallhoiders, the self-employed (who do not employ others), I lb label them class dcsticiaiion] may also lie misleading, 21 Sotetisan (i'JHfi) hai argutrii. The ntctl in 3 mnbility sample typically rauste i:l aye from IS or 21 ID retirement age unison?, .1111t their current daw cannot be considered a destination (at least in the sense affinal destination) for many of the yOL3 lt(jcr men. 1 We could, of Culme, have used a filler class classification—say into resell classes, as Goldthorpc dues for mult nfllis analyses (see also Mariliall 1990: Chapter 2). table 2 PERCENTAGE OUTFLOW MOBILITY TABLE.* MEN IN ENGLAND AND WALES 1972 Destination class Tutal ORIGIN CLASS 59 28 15 26 37 27 15 36 57 100 10! 99 hote: Classes as Table I. Percentages are by row—row totals may not add to 1C10 because of roundiiin souhce: As Table I. and lower-grade technicians and supervisors of manual workers.1 This table tells us that there are 731 men who were born into class 1 and, at the time or the survey, were in class 1; there were 322 men also born into class 1 who had moved to class 2; and so forth. Since it is difficult to interpret what these numbers mean when they are presented this way, they are usually given as percentages. If we calculate the percentages along the rows, we get the percentage of all men of a given origin class in each destination class. This is termed an 'outflow' table. Such a table for the England and Wales sample looks like Table 2. This table tells us that, for example, ofallmen originating in the working class, 15 per cent moved into the service class; 27 per cent moved into the intermediate class; and the remaining 57 per cent stayed in the working class (they were 'immobile'). Another way of interpreting this is to say that the probability of a man, who was born into the working class moving into, say, the intermediate class, was 0.27. 3 The three-claw cnrennraaiinn tiled here coircsponds to That used by Goldthnrpe er til. {ty»l)/o7) arid not in that fuunr} ill Erikion and Goldthorrie {l°92h:3n-n). In term* of the The other way of percentaging the table ts to do it by columns: this gives us the percentage of men in a given destination class who come from the various origin classes. This is termed an 'inflow' table. An inflow table looks like Table 3. We interpret this table by, for example, noting that, of the current occupants of the intermediate class, t i per cent came from service-class on- ■ gins; 40 per cent came from intermediate-class 1 origins; and 49 per cent came from working- ; class origins. These two different ways of percentaging a ■'■ table yield different insights into mobility. The inflow table tells us about the current composi- I tvon of the classes, in terms of where members / of the class came from. So, they tell us how het-erogenous each class is in its composition. In 4; these data an interesting contrast is provided by the service class and the working class. The ser-.-fj vice ctass is very heterogenous in its compost-tion, being marie up almost equally of men from ;| all three origin classes, while the working class j is much mare homogenous; over two-thirds of original Goltftliurpe classes (as shown ill TlWr 3.1) the three ■ classes usctl here arc made isp ai follows: Service class; I and It!;; Intermedial!: dais: III. IV anil V: Working class: VI am! VII. mole 3 PERCENTAGE INFLOW MOBILITY TABLE: MEM IN ENGLAND AND WALES 1A7J t Destination clnas 3 3 ORIGIN CLASS 1 31 11 5 2 36 40 26 3 33 49 69 Total 100 100 100 hote: Percenlagcs art by column-souRcev As Table 1. column totals may not add to 100 because of rounding. men in this class were also born into this class. It is relevant to bear in mind that the service class expanded over this period—hence one would expect that it would be heterogenous in its composition not least because there are not enough men of service-class origins (1242 in Table 1) to fill the number of service-class destination positions (2375). Conversely, the working class destination is much smaller than the working-class origin, and the reverse argument applies. As the number of positions in this class contracted, we should expect that the remaining positions would have been filled by those with origins in that class, rather than by outsiders moving in. We should expect, furthermore, that dsfler- < ences of this kind in the composition of classes would have consequences for the formation of 'class consciousness'. The members of a class ■■ which is relatively homogenous with respect to the class origins of its members are, ail other tilings being equal, probably rather more likely to be aware of themselves as constituting a dis- tinctive class than are the members of a class who are diverse in respect of their class origins.* 'In general, the greater the degree of "closure" of mobility chances—both intergenerarionally and within the career of the individual—the more this facilitates the formation of identifiable classes' (Guldens 1973:107). However, shared class origins are only one factor which may contribute to an awareness of class (see * * * Giddens 1973). The outflow table, on the other hand, tells us the chances of ending up in a particular destination class, given that a man started in a certain origin class. We can then make a comparison of these chances as between different origins. For example, the chances of a man born into the working class getting into the service class are 0.15 or 15 per cent; while the chances of a man of service-class origins staying in that class 3re 59 per cent. So, men from the service class are much more likely to be found there than are men born into the working class. Hence, the outflow table \ Recall tijat ease of muhiliiy is one of ihc laclnrs lhal Weber uicr to dittin^tisli the existence of social clastct ottt of t-rouris □fccoriamic classes. JtS WHO GETS WHAT? Social Mobilily Jiff provides us with a ready means of examining class differences in mobility chances, or, to put it in slightly different terras, the strength of the relationship between where you start out in the class structure {your class origin) and where you go to (your class destination). However, when we make these comparisons of mobility chances as between different origin classes, we do not usually do so in terms of probabilities (or percentages flowing into a particular destination class from a given origin class); rather, we calculate chances in terras of odds. This idea will be familiar to any readers with an interest in gambling. Instead of looking at the probability that a man of intermediate class origins ends up in the service class, we look, instead, at the adds that such a man ends up in the service class miller than another class. So the probability of being in the service-class destination is .28, while the odds of being in that destination class rather than in, say, the intermediate class are 0.75. This figure is equal to the number (or percentage) who endup in the service class divided by the number (or percentage} who end up in the intermediate class.5 So, when we make the comparison across different origin classes, we do this in respect of the adds of being in one destination class, rather than another. If we then want to compare the mobility chances of men from service-class origins with those of men from intermediate-class origins, say, we do this in terms of the odds of their arriving at one destination class rather than another. For example, we can compare the odds of entering the service-class destination rather than the intermediate-class destination, as between men of service-class and intermediate-class on--gins. In these data, then, the odds for men of service-class origins are: 731 {= number of men in service-class destination from service-class origins) divided by 322 (~ number of men in: intermediate-class destination from service-class origins) = 2.27; while for men of intermediate-class origins they are 857 (= number of men in1 the service-class destination from intermediate-class origins) divided by 1140 (= number of men in intermediate-class destination from intermediate-class origin) = 0.75. We compare: these two odds by simply taking their ratio: this; yields a measure called the 'odds ratio' which, in this case, is equal to 3.03. The odds ratio is the conventional measure of inequality in access to particular class destk nations from different class origins. Odds ratios are usually set up so that they measure the odds of getting into a 'higher' or more desirable destination class, relative to getting into a lower, or less desirable, class. Odds ratios can be readily interpreted as a measure of how the odds of getting into a more desirable class relative to getting into a less desirable one differ as between different origin classes. Equality of access to a more desirable, rather than a less desirable, destination class, as between different origin classes, would give rise to an odds ratio of one (since both origins would have the same odds). If the odds ratio is more than one, this reflects greater advantages to the origin class whose odds form the numerator of the ratio, while an odds ratio ; less than one indicates that the advantages ; 5 5n 20/37 from the ugiHow table (Table 2) ii (allowing for rounding error) eiiual u> 857/1 HO from the nrtßirwl table (Table S), tjbie 4 ALL POSSIBLE ODDS PATIO IN THE THREE-CLASS ENGLAND AMD WALES MOBILITY TABLE Destination clnss Origin clnss Odds ratio lv2 I v 2 3.03 lv2 1 v3 3.9H lv2 2v3 1.32 lv3 1 v2 5.03 1 v3 1 v3 14.33 lv3 2v3 2.85 2v3 1 v2 1.65 2v3 1 v3 3.54 2v3 2v3 2.15 accrue to the destination class whose odds form the denominator. It might seem that there is likely to be a plethora of possible odds ratios that we could calculate for any table. In Table 4 we show all the possible odds ratios in the three-class table for England and Wales.'' So, for example, the odds of being in the intermediate-class destination rather than the working-class destination are 2.15 times greater for men of intermediate class origins than for men of working-class origins. However, it turns out that not all of these odds ratios arc independent of one another. In a mobility table using M classes (M = 3 in our example) there are (M-l)- independent odds ratios. Sd, in our three-class table, there are four independent odds ratios. If we know these, then we can calculate all die rest. So, for example if we take any pair of destination classes (say classes 1 and 3) we can calculate the odds ratio as between origin classes 2 and 3 (which is 2.85} from a knowledge of the odds ratios involving origin classes 1 v 2 (5.03) and 1 v 3 (14.03). In this case we divide the latter by the former to yield 2.S5.7 ill STRUCTURAL AND III EXCHANGE MOBILITY We have spent some time discussing odds ratios because they turn out to play a central role in mobility table analysis.We will now explain why. Many mobility analysts have argued that the mobility we see in a mobility table can be explained as the result of two processes. These arc sometimes called structural and exchange mobility. The idea behind structural mobility is quite simple. If we take a given society, the amount of inter-generational class mobility that we observe will depend, to a very great extent, upon the degree of change in the class or occupational structure of that society. So, a society which was developing rapidly should show a lot of mobility, not least because many occupations would be declining in importance and thus men whose father held one such occupation would have very great difficulty in pursuing the same occupation. They would, in a sense, be forced to be mobile out of that class or occupation by virtue of the fact that the occupational positions were not there for them to fill. Sometimes this kind of mobility is called 'forced mobility'. The difference between the origin ciass and destination class distributions in a mobility table is sometimes taken as a measure of the extent of this. So, in the three-class table (Table t), we see 6 We could alio invert all these njilj ratio* tu yield .r much larger total, lint this wotiiri be of no interen. 7 We tec why tills futluwi ifwe write the oiliii ratio for any two origin chJici til tile form ofa ratio, tliu] 1/2 and 1/3. Ir therefore follow* that ilis ratio 2/3 n simply 1 /3 divttieil by 111. J20 WHO GETS WHAT? Social Mobility 321 many more origin than destination positions in the working class, and rather more destination than origin positions in the service class. So, the suggestion is that men must have had to move out of the working class because it is contracting, and, equally, men must have been 'drawn into' the service class as it expanded. It was usually argued that this process operated indcpendendy of other processes of social mobility. In particular, it operated independently of processes of exchange mobility, which was concerned with how different class origins influenced mobility, and the inequalities in mobility chances that derive from different class origins. The reason that different origins confer different chances of mobility is because they provide people with different resources for mobility. So, people born into mare advantaged classes generally acquire higher levels of formal qualifications, and, in addition, may have other resources (such as kinship links or friendship networks) which they can use to help them acquire a more desirable class position, in his analysis of the English and Welsh mobility data Goldthorpe (1980/87:99) developed a mobility model in which he argued that patterns of social fluidity (in other words, patterns of inequality of access to particular class destinations as between men of different class origins) were shaped by three factors. These arc, he argues, the relative desirability of different classes as destinations; the barriers to entry to these classes; and the resources attached to different class origins which allow these barriers to be overcome and the more desirable destina- tions to be entered (and the less desirable ones :■ to be avoided).6 So, people seek to gain entry to more highly desired destination classes; to do this they must overcome a variety of barriers to entry (such as the requirement to possess certain educational or other credentials; or the acquaintance ofparticular individuals), using the resources that they have acquired as a result of their origin-class position. From this it follows that if resources were 1 more or less equally distributed (so that the-; resources one had for mobility did not depend upon one's origin class) there should be a good ■:: deal of inter-gcnerationaf class mobility in society. Hence, an egalitarian society (in the sense of -one in which there was equality of condition as between people of different class origins) should be a society displaying high rates of mobility. In particular, of course, the chances of people born into a given class stayingin that class wouldbe no better than the chances of people born outside .-that class entering it. Therefore more equal societies should display more social mobility. However, we have already seen that the : amount of mobility in a society also depends upon the amount and speed of occupational or. class change. Hence, a society which was very : unequal could, it appeared, display a high rate : of mobility provided that the pace of structural'; change were fast enough. The problem is to disentangle these two effects: how much mobility is due to structural change, and how much reflects the degree of equality—or, as it is sometimes called, openness—in society? The posing R * * * Rcsoviices arc used to overcome harness (cniistratnlsj in Rider to try to secure the nioit desired destination {[lie most prcfctreil alternative}. of this question then led to a number of attempts, by sociologists, to partition the total amount of mobility in a given observed mobility table into a component due to structural mobility (which was, in some fashion, linked to changes in the marginal distributions of the mobility table— that is, to the difference between the origin and destination distributions) and some component due to exchange mobility.1 None of these attempts, in the 1970s and early 1980s, were particularly successful. In the 1970s, however, a number of sociologists began to point out that odds ratios might be useful in this context, since they certainly ■measured inequalities in access to different class destinations arising from different class origins. Furthermore, odds ratios arc independent of the marginal distributions of the mobility table. This means that, if two societies have the same level and pattern of class inequality in relative mobility chances, the fact that one of them has experienced rapid changes in the class structure (and, perhaps as a result, the origin and destination distributions are more unalike in one country's table than in another's) will not affect the fact that the pattern and magnitude of their odds ratios will be the same, Drawing on this, Goldthorpe and his co-authors (Goldthorpe 1980/87; Erikson, Goldthorpe and Portocarero 1979, 1982) abandoned the structure/exchange distinction and replaced it with an emphasis on absolute mobility (the actual mobility observed in the mobility table) and social fluidity (measured in terms of odds ratios), somettmcscalled'relativemobility'.1" There is no attempt, in this approach, to partition absolute mobility into some part due to structural change and some to exchange mobility." Nevertheless it provides a framework in which it is passible to identify societies which display very high rates of absolute mobility, together with low social fluidity or low 'societal openness' as reflected in large odds ratios. A very good example of this is provided by social mobility data for Sao Paolo, Brazil, collected by Hutchinson (1958) and later presented and used by Sobel, Hout and Duncan (1985:366). This shows massively high rates of absolute mobility, arising from very rapid and large changes in the class structure together with very large odds ratios reflecting a high level of inequality in access to more desired class destinations as between men of different class origins. 5 Some csasiiples include llaiclririit 197-1; Hope S'JHS, 1982: ami McClessdols 1977, l%(l. For a crilstjsse of this apptoach ice Solid 198i. to U watalsn recognized thai udds ratios, which measure social fluidity, ate directly captured in the parameters of leg-linear models which, since the 1970s, have been used to model mobility tables (scc fietsberrr 1977; GuosJmat; 197°). The parameters [hat ate estimated in log-linear mtitlellimj fall into [Wo kinds; main effeet parameters, and interaction, or association, parameters. These latter depend upon the nature am] extent of the statistical relationship between origins and destinations. Odds ratios arc function! of these parameters, arid oot of the main effects. So. furesample, if all odds nitios are one, the attociatirjll parameters of the log-Linear mndel will aUbcicm (in rhclottfotlnofthe model). 11 Theitructnre/cjtchanpcdislijieiinnwailhciiTCvivedbySobcl, 1 lout and Duncan (19HS), They present an riettatu reformulation of the concepts, arguing that exchange mobility refers to equal reciprocal flows between pairs of classes (for example (he Elotv frnltl origin class A tn destination class 11 arid front arirtin 11 tn destination A); and that structural mobile ity is captured, sis origin-specific parameters that make such flowi unequal (so that flic flow from A lo □ may, for example, exceed that from D lo A). The difficulty with this model is llut it implies thai tiie uhtcrved inubilily lablc should display tile properly af'ipiasi-syimiietry' {see Dishiip rj til. 1975 for a definition of (his technical Icrtu). While a number sif mobility tables do indeed display this property, many do not. In such cases, a thild type of mobility—a residual category-hat In he invoked its older to account for the observed flows. WHO GETS WHAT? ABSOLUTE MOBILITY The study of absolute mobility places the focus on changes in the class structure over time (such as the contraction or expansion of classes). We have already noted the major trends in this respect during this century in most of the industrialized countries of the world: the decline in farming and farm-related jobs, and in unskilled work, with increases in skilled and white-collar jobs. Tilt timing of tins transition has, however, varied. We can gain same indications of this by comparing the origin and destination distributions for a number of tables from different countries.'- Table 5 shows this comparison for four of the European countries taken from the CASM1N data set—Sweden, England and Wales, the Republic of Ireland, and Poland. Here we have moved to a five-class classification, in order to bring out some of the salient differences between these societies. These five classes (with the Erikson and Goldthorpe 11-class schema classes in parentheses) are: 1. White-collar workers (I, II, Ilia and Illb). 2. Petty bourgeoisie (tVa and lVb). 3. Farmers and farm workers (IVc and Vllb). 4. Skilled workers (V-t-VI). 5. Non-skilled workers (Vila). (See Erikson and Goldthorpe 1992b: 38-9.) We have chosen these four countries because they represent four different sorts of society— England and Wales having been industrialized for a long period, Sweden being a society which has experienced a long period of social demo- cratic government and which has, accordingly, developed possibly the world's most consprehen-sive system of social welfare (broadly defined). Tlie Republic oflrcland is a late-industrializing S; nation that retains a substantial dependence on ;v agriculture, and Poland was, at the time these .X data were collected, a state socialist country. Many of these differences arc reflected in the .; comparison of the origin and destination distri- « buttons of their respective mobility tables. I:or example, it is noticeable that in England and i Wales, origin classes 1 (white-collar workers) j and 4 (skilled manual workers) are much larger V (in percentage terms) than in the other three countries, reflecting Britain's earSier indtlstri-alization. In Sweden the destination distribu- ,i tion of class 1 is of comparable size to England j and Wales, but in Ireland and Poland it remains ■* much smaller. The relative lack of men in class s 2 (petty bourgeoisie) in the Polish origin and o destination distributions is hardly surprising in a state socialist country. In the other countries the distributions of this class are quite similar, ft :; is in class 3 (farmers and farm workers) that we find major variation, particularly inr.hede5tifi.i- : tion distributions where the contrast is between \: the two countries which retain substantial dependence on ngricultu re (Ireland and Poland) ; and whose industrialization has been very late and the other two. However, the decline in the importance of farming in Sweden has been ■ both recent and very rapid indeed as we can see -by comparing the origin and distributions for.. class 3 here. Conversely, agriculture declined in significance in England and Wales long before 12 h is important m teileialc that tliij miiy apptiuimalca change! in llie elan UMUtnilniial ilniiturei over time. Til examine inch iltanpe fotmally tve sllmili! cumi»arr taiuple iurveys nfthe labmiriWe at twit JiunUs ill tittle. i 2 3 4 5 Sweden Origin 14 11 26 24 25 '"' (N = 2103) Destination 32 8 5 30 24 £ »W Origin 21 iO 5 39 26 (N = 9434) Destination 34 8 2 41 15 Republic of Origin 11 10 39 14 27 Sudani! Destination 23 a 21 20 27 (N = 5 992) Poland Origin 10 3 53 18 16 (N = 32109) Destination 20 2 25 31 22 Social Mobility 323 TVUUr, 5 PERCENTAGE ORIGIN- AND DESTINAT.ON-Ct.ASS DISTRIBUTIONS." SWEDEN, ENGLAND AND WALES, REPUBLIC OF IRELAND, POLAND Classes ...source; CASMIN data set. the period covered by these data: here class 3 makes up only a very small part of both the origin and destination distributions. Finally, class 5 {unskilled workers) shows a good deal of cross-national variation, particularly in a comparison of Poland with the other three countries. Here we see that it is relatively under-represented in the origins, but is the only country in which this class is larger (in relative terms) in the destination distribution. Again, it seems likely that the unusual position ofPoland is associated with its post-war experience of state socialism. Table 6 shows the percentage composition of the destination classes in terms of origin class: in other words, the table shows what percentage of men in a given class come from each of the origin classes. We limit ourselves to highlighting two points. First, the class which has shown the greatest growth in these four countries— the white-collar class 1—also shows substantial heterogeneity of composition. This is particularly striking in Ireland. Nevertheless, with the exception of Poland, the origin class which is most over-represented among the incumbents of this destination class is class 1 itself. This is part of a more general feature of these four tables, namely that for all except the relatively small classes, it is the corresponding origin class that supplies the largest share of members of a given destination class. The exception to this is Poland, where this is true only of the farming class. The second trend is the remarkable degree of self-recruitment ami class closure in class 3 (farmers and farm workers). This is particularly pronounced in Poland and Ireland, where this class remains very large. The reasons for this high degree of self-recruitment are easy to find: by and large farms are inherited, either legally or ik facto and the same is true of jobs as farm workers. Except in England and Wales there is a good deal of mobility from class 3 into all the other classes, reflecting the 'forced' outward mobility of those born into a declining class. WHO GEtS WHAT? TABLE 6 PERCENTAGE ,™ TABLES PHOM SWEOE», ENGWHP AND WALES, ^ °» IRELAND, Destination classes 8 32 21 32 {in oilie* wmich, in a iclatiunsiiiji between origin dsn uitl mbjettivc anejimciltl of tlic desirability of fjificrcnt dcitinatjam). jty. Recall that in studying social fluidity we ate using odds ratios to measutc the differences between people of different origin classes in their chances of access to more rather, than less desirable destination classes. This is, therefore, a useful measure of the degree of societal openness, since if there were no differences in this respect ■ between men Df different class origins, all odds ratios would be equal to one. Such differences as exist are usually attributed to inequalities in the possession Df mobility resources as between different class origins." When we come to try to judge whether or not a society displays much or little 'open-"ness' of this kind, we can adopt one (or both) of two yardsticks. First, we could compare the observed set of odds ratios with the yardstick of total equality where all odds ratios would equal □nc. The latter is sometimes called a situation of perfect mobility, and it arises when there is no relationship between class origins and class destinations—that is to say, between the class a person starts out in and the one he or she is currently in. However, since all societies are some considerate distance from displaying perfect mobility, a possibly more useful perspective is provided by international comparisons which ask: How open is one society compared with another society? One of the most famous hypotheses in sociology addresses exactly this question. The so-called Featherman-Joncs-Hauser (FJH) : hypothesis argues that a basic similarity will be found in social fluidity in all industrial societies 'with a market economy and a nuclcarfamily system' (Featherman, Jones and Hauser 1975:340), This innocuous-seeming formulation has, if it is true, some very important ramifications. Many societies have expended a good deal of effort and resources on policies designed to increase societal openness by, for example, providing free education, medical care, and, more generally, the panoply of the welfare state. The FJH hypothesis suggests that whether a state pursues such policies or not has no consequences for the level of social fluidity that it will display. The bulk of the many papers that have used comparative data to test the FJH hypothesis have arrived at much the same conclusions. These arc that, first, tile greatest differences between societies in mobility arc in the area of absolute mobility. This is not stirprising, given the different rates of structural change in societies, as we noted earlier. Second, there are very great similarities in the degree of openness in different societies. There are statistically significant differences in fluidity between them, but these tend to be relatively small for the most part. This finding has largely been born out by the results of the most painstaking and detailed comparative mobility project yet undertaken, the CASMIN project. In discussing the results of this research, Erikson and Goldthorpe find it necessary to modify the FJH hypothesis somewhat. Their conclusion is that: A basic similarity will be found in patterns of social fluidity ... across all nations with market who Girrs what? economies and nuclear family systems where no sustained attempts have been made to use the power of the modern state apparatus in order to modify the processes or the outcomes of the processes through which class inequalities are intergenerationally reproduced. (Erik-son and Goldthorpe 1987b: 162} What is notable about this modification is that, while retaining the emphasis on the high degree Df commonality that apparently exists across industrialized nations in their pattern of social fluidity, it allows for the possible impact of state intervention. Erikson and Goldthorpe (1992b: 178) argue that it is the attaining of greater equality of condition that best promotes high rates of social fluidity; that is, ifincquali-tics in the conditions of life enjoyed by people are small, fluidity will be high. Thus, for example, policies of taxation and redistribution that seek to reduce the level of inequality in the distribution of income and in living standards, are likely, all other things being equal, to promote greater social fluidity. SOCIAL FLUIDITY IN EUROPE Differences in social fluidity will be most evident in comparisons involving the extremes. In Etikson and GoSdthorpe's analysis of the CASM1N data the extremes of societal openness in Europe are represented by Sweden (most open) and Poland and the Republic of Ireland (least open), with England and Wales falling in the middle. In Table 7 we show the outflow tables for our four countries—Sweden, England :-and Wales, Republic of Ireland and Poland—\:i with a view to comparing theirsocial fluidity. : Recall that an outflow table tells us the per- :.; centage of men from each class origin who ; entered each destination class. So, in Table' 7 we see that, in Sweden, of men born into .'-class 1, 64 per cent had class 1 as their destisia- :~ tion class. The striking feature of this table is..::! that, far both white-collar (class 1) and skilled worker (class 4) destinations in all countries, ;íí the highest probability of being found in that..-class is enjoyed by men who were born into ,. it. In all countries the strength of the link ■ between origins and destinations in the white-collar class (as measured in this way) exceeds •'• that observed in the farming class, and, in all :.J; countries except Ireland, so does the strength ■: of this link among the skilled working class. These figures suggest that in both these classes very effective mechanisms exist through which.,;1 class position can be transmitted from father to son, despite considerable heterogeneity in the composition of these classes as destinations ,-' (particularly in the case of class 1, as revealed by the inflow table). To undertake a proper comparison of social . fluidity between these four countries would require that we model the pattern of odds-ratios in each table using log-linear models, js!. Erikson and Goldthorpe (1987a and b, 1992b) do in their analyses. Here, however, we will;'! compute some illustrative odds ratios.14 So, for.-example, take the extreme odds ratios—that .. [ 4 M we noted earlier, uiljs raticu Ľan he computed fcum Either a lattt: of ireijiieircy cinirlti (TaHc 1 Cur example) or from taille* [if inflow or unttldw iietcetita^its. ■Sofia/ Moiúliíy 327 ™E 7 POUrT" °UmOW TABL£S — —' — «■» — OF ,HElAMD, (a) sweden (b) england and wales (c) republic Of IRELAND (o) poland 6 as a late industrializing, semi-peripheral u re in which free post-primary education and very— many other welfare state programmes were introduced only around the time that Uiese mobility data were collected or afterwards. -As a result, inequalities of condition between" families were particularly marked (Breen tt at. 1990) and thus, following Erikson and Gold-thorpe's (1992b) argument cited earlier, a finding of low levels of social 6u idity is perhaps not ■ surprising. What of Poland? Despite its post-war history Poland displays less social fluidity than, either Sweden or England and Wales—but, once again, this accords with what we know of the impact of state socialism on social mobility. Broadly speaking, levels ofsocial fluidity in state socialist societies are similar to those found in capitalist societies, albeit with some differences as to which classes are advantaged and which disadvantaged. Furthermore the persistence, in both Ireland and Poland, of a large agricultural .■. sector in which inheritance is of overwhelming importance in acquiring a position as a farmer or firm worker, acts to reduce the overall level of social fluidity in these countries. Finally, in our discussion of social fluid- - ity we might ask; Where does America fit into this picture? A view of America as the 'land of :: opportunity' has existed for several centuries, ■• ■ In this view, America is seen as lacking the kind • of rigid class structure felt to be characteristic of ■■; European societies and as presenting opportunities for personal advance that older countries could not. So, for example, the idea that class 'relations in America were distinctively different was found in Sombart's (1907/76) thesis of "'American exceptionalisns'. While some studies can be seen as supporting this position (Miller 1960, and, particularly, Blau and Duncan 1967), . recent research by Erikson and Goldthorpe ... (1985, 1992b:321) has led to the opposite conclusion; namely that 'no very convincing case for American exccptionalism ... can be made out', and No matter how distinctive tire United States . . . may be in (its) economic and social histories ... or in the ideas, beliefs and values concerning mobility that are prevalent ... it could not, on our evidence, be said that (it) difli'r(s) more widely from European nations in ... mobility than do the European nations among themselves. (Erikson and Goidthorpe 1992b:337, parentheses added) 15 The technicalities nfthis measure need not concern us; interested reader* thould consuls Breen 1994 fordctatls. 16 Though ErifclDIl and Goldthorpe (1992b: 177-9) ire at pains (n argue that alrllouj-U stale intervention may influence fluidity in particular instances (as in .Sweden) this docs not allow us to conclude that rule by a particular sort nf political party {e.g. Social Democrats) will necessarily give rise to a ch'niiictivepaitcrn offluitliiy in all countries where that kiniS of political patty is in pnwer. tn other words, there tioes not Mi«, Tut example, a generic 'social democratic' pattern of social fluidity. 17 tn this chapter we have concentrated on intcr-gencrationa! mobility: however, in recent yean a great deal of interest lias amen in the study orililra-tjcneraiiorial mobility. While at Its simplest this involves 1 tabular comparison of the class an individual occupied on entry to the labour force with the clajr he or she oceupiei ar some rater point in time (and is Social Mobility 329 III CONCLUSIONS The study ofsocial mobility is, we believe, a powerful research programme that teifs us a great deal about the nature of modern societies and the position of classes within them." However, it is hardly to be wondered at that, having been an area of active research for many years, it has generated many critiques. But what is less obvious perhaps is that the great majority of the criticisms levelled at mobility studies are not criticisms of mobility research per re but, rather, of the framework within which it is pursued. That is to say, most critiques of mobility research concern either the issue of the adequacy of the class classification used or the question of the appropriate unit of class composition. In this sense they are criticisms of mobility research at passmU. * * * Aside from these very important questions there are two specific criticisms of mobility research itself which we should mention. The first of these concerns the neglect of women in much social mobility analysis, it has been argued (by Hayes and Miller 1993 for example) that the concentration on men distorts the picture we have of mobility in modern societies. While this would be a very damaging criticism if it were true, there is much evidence to suggest that the inclusion of women in mobility studies does not change conclusions about social fluidity based on men-only studies (for example Marshall thus analogous to theorigin—dcsrinalion inter-itencrarional mobility table), mote sophisticated approaches arr also Died which seek tn analyze rhe sequence and timing of transitions between jobs or claries that people experience iluiitiij their life course (far example AJlmendinrtcr 19iV>). jjo WHO CHTS WHAT? Sofia/ Mobility 331 ct ill. 1988). While it is vety obvious that there is a marked difference in the occupational and class distributions of men and women, patterns of social fluidity among women appear to be very similar to those found among men. in other words, differences in mobility chances between women of different class origins are much the same as those found between men of different class origins. The second criticism is one which has been made by Poulantzas (1975) who argues that mobility research is fundamentally mistaken in placing its emphasis on the movement of individuals between class positions, when what should be focused on is the structure of, and the functions carried out by, these positions. While such an argument accords well with the views of 'structural Marxists' it is difficult to credit it with any force. In its concern with absolute rates, mobility research docs indeed examine the structure of class positions in society. All researchers in the area accept, for example, that the single largest 'cause' of mobility flows during the last 100 years has been the contraction in the number of positions available in agriculture. On the other hand, in any society with a division of labour linked to unequal rewards the question of how people and families are distributed over these positions will be of central significance for an understanding of the way in which life chances are allocated. Furthermore, the study of how much openness of access exists to these various positions, as measured by social fluidity, tells us something very basic about the nature of the modern nation-state. To argue, as Poulantzas seems to, that a concern for such matters is misplaced is, at best, a statement of somewhat eccentric preferences. Ill REFERENCES Blau, P. and O.D. Duncan, 1967, The American OrntiinliiHJii/ Structure, New York: John Wiley, Breen, R. 1987, 'Sources of Cross-National Variation in Mobility Regimes-. English, French and .-Swedish dataReanalyscd'. Siirioftyy, 22,1:75-90., Breen, R., D.P, Hannan, D.B. Rottman, and C.T. Whelan, 1990, Understanding Conleniporary ;. Ireland: State, Class and Development in the Republic ■ of Ireland, London: Macmillan. Duncan, O.D. 1966,'Methodological Issues in the Analysis of Social Mobility', in N.J. Smelscr andS.M. Lipset (cds) Social Structure and Mobility ■■ in Economic Development, New York: Aldiue. Erikson, R. andJ.H. Goldthcrpe, 19S5, 'Are American Rates of Social Mobility Exceptionally High? New Evidence on an Old Issue', European Sociological Review, 1, 1: i-22. Erikson, R. andJ.H. GoUithorpe, 1987a, 'Commonality and Variation in Social Fluidity in Industrial Nations, Part I!.'A Mode! for Evaluating the "FJM Hypothesis" ', Hiim/miii Sociological : Review, 3, 1: 54-77. Erikson, R. and j.H. Goldthorpe, 1987b, 'Commonality and Variation in Social Fluidity in Industrial Nations, Part I: The Model of Core Social Fluidity Applied', Eumpetiti Sociologies Rci'ietv, 3, 2; 145-66. Erikson, R. and J.H. Goldthorpe, 1992a, 'Indi-vidual or Family? Results from two Approaches to Class Assignment', Acta Sothhgita, 35:95-106. - ' " Erikson R. andJ.H. Goldthorpe, 1992b, The i . ". Cmumt Flux; A Smdy of Class Mobility in Indm-trh'i Sonnies, Oxford: Clarendon Press. * ' , Erikson, R.,J.H. Goldthorpe, and L. Porto- Nations'. England, France and Sweden', British Journal of Sociology, 33, 1; 1-34. Erikson, R. and S. Pominen, 1984, "Social - Mobility in Finland and Sweden: A Comparison i of Men and Women', in R. Alapuro (ed.) SimiH , ^ Stoles in Coinpomive Perspective, Oslo: Norwegian V' University Press. ', Featherman, D.L., F.L.Jones, and R.M. Hauser, v' 1975, 'Assumptions of Social Mobility Research in the US: The Case of Occupational Status', Social Science Research, 4: 329-60. Giddens, A. 1973, The Class Striaton of the Advanced Societies, London: Hutchinson. Goldthorpe, J.H. with C. Llewellyn and C. ' ' Payne, 19B0 (second edition 1987), Social Mobility mid Clus: Structure in Modem Britain, Oxford: I Clarendon Press. ■ ' Hayes, B.C. and R.L. Miller, 1993, 'The t. Silencetl Voice: Female Social Mobility Patterns '- with Particular Reference to the British Isles', \ i_ BrilisliJournalofSocwbgy,-\'),A:653-72. Hazelrigg, L.E. 1974, 'PartitioningStructural Effects and Endogenous Mobility Processes in the Measurement of Vertical Occupational Status Change', Acta Sociologica, 17,2: 115-39. Hope, K. 1981, 'Vertical mobility in Britain: A Structured Analysis', Sno'iiliii;)', 15, 1: 19-55. Hope, 1C 1982, 'Vertical and Non-vertical Class Mobility in Three Countries', American Sociology caiRei'kiv, 47, 1: 99-113. Hutchinson, B. 1958, 'Structural and Exchange Mobility in the Assimilation of Immigrants to Br.17.il', PiipiiditKni Studies, 12:111-20. Marshall, G. 1991), In Praise of Sociology, London: Unwin Hytnan. Marshall, G., 1-1. Newby, D. Rose and C. Vogler, 19HR, Social Class in Modern Britain, London: Unwin Flyman. McClendoii, M.J. 1977, 'Structural and Exchange Components of Vertical Mobility', WmcriVitn Sociological Rcmm', 42, 1: 56-74. McClendDn. M.j. 1980. 'Structural and Exchange Components of Occupational Mobility: A Cross-National Analysis', The Socbhgiail Quarterly, 21: 493-509. Miller, S.M. 1960, 'Comparative Social Mobility', Cmtenl Sociology, 9: 1-89, Poubntzas, N. 1975, Classes iu Contemporary Capitalism, London: New Left Books. Przeworski, A., R.R. Barnett, andE. Underbill, 1980, "The Evolution of the Class Structure of France, 1901-1968', Economic Development and Cultural Change, 28, July: 725-52, Sobel, M.E. 19SJ3, 'Structural Mobility, Circulation Mobility and the Analysis of Occupational Mobility: A Conceptual Mismatch', AjncticMi SociiifigW Review, 48, 6: 721-7. Sobel, M.E., M. Hout, andO.D. Duncan, 1985, 'Exchange, Structure and Symmetry in Occupational Mobility', American Journal of Sociology, 91, 2: 359-73. Sombart, W. 1907/76, M-Viy is there no Socialism in the United States?, London: Macmillan. Sorcnson, A.B. 1986, 'Theory and Methodology in Social Stratification', in U Himnielstrand uul.l Sociology; From Cmis to Science, London: Snj Wright, E.O. and B. Martin, 1987, 'The Tran>-fotmation of the American Class Structure', American Journal of Sociology 93> 1: 1_29, 19 Finding Work: Some Basic Results* MARK GRANOVETTIiK How people find jobs is a prosaic problem—but exactly for this reason, it relates closely to important issues in sociology and economics. Under the rubric "labor mobility" in economics, and "social mobility" in sociology, how people move between jobs and between occupations has received much study; but surprisingly little detailed attention has been given to the question of how individuals become aware of the opportunities they take. Most studies are either highly aggregated or highly individualized. At the macro level, excellent monographs detail the statistics of men flowing between categories + * *; at the micro level, Dther studies offer plausible psychological and economic motives lor particular individuals to want to change jobs * * *. Important as these concerns are, they are not those of the present study. Rather, I have chosen to concentrate on the issue of how the information that facilitates mobility is secured ant) disseminated. This question lies somewhere between the micro- and macro-level concerns described above, and is a potentially crucial link in their integration; it is an important part of the study of the humiliate causes of mobility, and, as in other social science problems, failure * Pitsi jmWiihtfil iii 1'J7-I; from Geaii^ ajn(>: A Stu/iy rfCantttit ami Girrc/j.