John H. Goldthorpe

1. Theories of Change in Class Stratification

In this chapter my concern is with theories of long-term change in the class stratification of advanced Western societies. Such theories fall into two main types on which I shall concentrate: (i) those that form part of more general liberal theories of the development of industrialism or "postindustrialism"; and (ii) those that form part of more general Marxist theories of the development of capitalism or " late-capitalism." In this first section of the chapter, I review these theories in sufficient detail to bring out the wide differences that exist between them. In the following section, I examine how well they fare in the light of recent research. I focus my attention here—without apology—on research that has produced quantitative results, since, as will be seen, most of the issues that arise in comparing the two types of theory are ones of an inescapably quantitative kind.^[1]

The main conclusion that I reach is that neither liberal nor Marxist theories stand up at all well to empirical testing, and that the claims of both must now in fact be seen as in major respects invalid. Consequently, in the final section of the chapter, I am led to ask whether anything of a general kind can be said about why these theories should have proved to be so unsuccessful and, if so, what lessons may be learned from their failure.

The trends of change proposed by the theories that I wish to examine can be usefully treated as occurring (i) in the division of labor or, more precisely perhaps, in the structure of employment; (ii) in the class structure; and (iii) in rates and patterns of social mobility within the class structure. Some amount of variation is of course to be found among both liberal and Marxist theories, and I cannot pretend here to do full justice to this. Rather, I shall concentrate on those features of liberal theories on the one hand of Marxist theories on the other that would appear to be central to the genre. On the liberal side, I draw most heavily on Kerr et al. ([1960]) 1973) and also Kerr (1983), Bell (1967, 1973), and Blau and Duncan (1967); on the Marxist side, I draw on Braverman (1974), Carchedi (1977), Wright (1978, 1985), and Wright and Singelmann (1982). These sources will not be further cited except where direct quotations are made from them.

1.1. Changes in the Structure of Employment

It is a central claim of liberal theories that in the course of the development of industrial and postindustrial societies the structure of employment is, in net terms, progressively "upgraded." The proportion of the total labor force engaged in work that requires minimal skills and expertise—and in turn minimal training and education—tends steadily to decline, while a corresponding expansion occurs in the types of employment that call for various new skills and often for expertise grounded in theoretical knowledge, and that may moreover entail the exercise of authority and responsibility. Two major sources of this upgrading are identified. First, the continuing advance of technology in all forms of production tends to eliminate merely "laboring" or highly routinized jobs, while creating others that must be filled by technically and scientifically qualified personnel. Second, the tendency, analyzed by economists such as Clark (1957), Kuznets (1966), and Fuchs (1968), for labor within a growing economy to shift intersectorally—initially from agriculture and extractive industries to manufacturing but them from manufacturing to services—also generates an increased demand for high-level skills, both technical and "social." Furthermore, the organization of services, public and private, tends to multiply administrative and managerial as well as professional positions.

In direct contrast—indeed, in more or less explicit opposition—to this thesis of the progressive upgrading of employment, Marxist authors claim that the long-term trend of change in Western societies is in fact for employment to be systematically degraded. Within the context of the prevailing methods of industrial organization, as informed by modern "scientific management," the consequence of technological advance is that one type of work after another, in all sectors of the economy alike, is effectively "deskilled," and this tendency far outweighs any increase that may occur in professional or managerial positions. Thus, Marxists argue that a large part of the growth of nonmanual or white-collar employment that has so impressed liberal authors is made up of jobs that are already degraded and that, in terms of skill levels and autonomy as well as pay and conditions, are in no way superior to the manual jobs they have replaced.

This systematic degrading (or, more emotively, "degradation") of labor results ultimately, in the Marxist view, from the fact that capitalist production entails the exploitation of workers by their employers. This means that, under capitalism, the labor process must be organized in such a way as to meet not only the requirements of productive efficiency but also the requirements of the social control of labor. And in this latter respect, the possibility of using new technology and managerial methods in order to remove skill—and thus autonomy and discretion—from the mass of employees and to concentrate them in the hands of management is one that employers will seek always to realize. In sum, while liberals view change in the structure of employment as essentially an adaptive response to technological and economic progress, Marxists would rather see it as reflecting the outcome of an abiding class struggle.

1.2. Change in Class Structure

As regards change in the class structures of Western societies, liberal theorists stress tow main developments, both of which they would regard as following directly from the net upgrading of employment. They point, on the one hand, to the contraction of the working class, in the sense of the body of manual wage-earners and, on the other, to the emergence of what has most often been labeled the "new middle class" of nonmanual, salaried employees—as distinct from the "old middle class" of proprietors and "independents," whose social significance diminishes as large-scale enterprises increasingly dominate the economy. The class structure of industrial societies is in fact seen as coming increasingly to reflect the social hierarchy of the modern corporation, although at the same time across all classes a reduction occurs in the more evident inequalities—for example, in incomes and in standards and styles of living. Indeed, for some authors (for example, Duncan 1968), the very concept of a class structure is inappropriate to the gradational complexity found under advanced industrialism and is better replaced by that of a continuum of "socioeconomic" status.

Again, the position taken up by Marxist theorists is a quite contrary one, deriving as id does from their emphasis on the degrading rather than the upgrading of the job structure. Instead of a decline in the size of the working class, what they see in train is a process of widening "proletarianization" that will maintain the working class in a position of at least numerical, if not sociopolitical, preponderance.^[2] In classical Marxism, the idea of proletarianization was applied specifically to the "driving down" of elements of the petty bourgeoisie into the ranks of wage labor; but in more recent analyses its meaning has been extended to that it has in effect become synonymous with those changes in the content of work and in working conditions that are claimed by the "degradation of labor" thesis—and especially as this thesis is applied to work of a nonmanual kind. Marxist authors would then reject the view that the long-run tendency is for the class structures of Western, late-capitalist societies to display increasing gradational complexity. While the growth of the professional, administrative, and managerial salariat may be acknowledged as having created some problems of both theory and practice (the problems of "middle layers," "dual functions," "contradictory class locations," etc.), such problems tend to be regarded as transient ones that are in fact finding their solution as the simplifying logic of proletarianization continues to work itself out. As Braverman puts it:

The problem of the so-called employee or white-collar worker which so bothered earlier generations of Marxists, and which was hailed by anti-Marxist as proof of the falsity of the "proletarianization" thesis, has thus been unambiguously clarified by the polarization of office employment and the growth at one pole of an immense mass of wage-workers . The apparent trend to a large nonproletarian "middle class" has resolved itself into the creation of a large proletariat in a new form. (1974, 355, emphasis in original)

1.3. Changes in Social Mobility

For liberal theorists, industrial societies are highly mobile societies. The general assumption is that economic development and social mobility are positively associated: the more economically developed a society, the higher the rates of social mobility that it will display. Explanations of how this association comes about take somewhat different forms (Goldthorpe 1985a), but two processes are clearly given major emphasis (Treiman 1970). First, it is observed that the transformation of the structure of employment, which, as already described, is seen as following from technological and economic advances, mush in itself generate high levels of occupational and class mobility. Moreover, because a net upgrading of employment takes place, the pattern of mobility will in turn show a net upward bias. In particular, the expansion of the professional, administrative, and managerial occupations of the new middle class is too great to be met entirely by self-recruitment and mobility from the old middle class, and in fact greatly enlarges the opportunities for the social ascent of individuals of working-class origins.

Second, it is argued that, quite apart from the effects of such structural change, mobility is also increased as a result of changes in the criteria and processes of social selection. The functioning of a modern industrial society entails a progressive shift away from "ascription" and toward "achievement" as the leading principle of selection, and this principle becomes embodied in the procedures of educational institutions and employing organizations. In turn, the operation of such "meritocratic" selection promotes greater openness and equality of opportunity in the sense that individuals' levels of educational and occupational attainment become less closely correlated with the attributes of their families or communities of origin. Because of the prevailing high levels of mobility, the potential for class formation and class conflict in modern societies is seen as steadily declining. The more readily that lines of class division are crossed, the less likely class is to provide the basis on which collective identifies and collective action will develop. Mobility performs important legitimizing and stabilizing functions. To quote Blau and Duncan (1967, 440), "Inasmuch as high chances of mobility make men less dissatisfied with the system of social differentiation in their society and less inclined to organise in opposition to it, they help to perpetuate this stratification system, and they simultaneously stabilise the political institutions that support it."

Marxist theorists do not, in the case of mobility trends, provide a matching set of counterarguments to those that are advanced by liberal authors. Rather, major disagreement is apparent over the importance of mobility per se. Thus, where Marxists do not ignore the question of mobility altogether, they either attempt to dismiss it as one prompted more by ideological than by social-scientific concerns (see, e.g., Poulantzas 1974, 37) or they treat it as being one of no great consequence on the grounds that mobility between different class locations in fact occurs to only a very limited extent. Thus, for example, Carchedi (1977, 173) asserts that "generally speaking, a member of the working class remains a member of the working class" and that "the same applies to both the middle classes and to the capitalist class." The only way of any real significance in which individuals may change their class location is collective and "through changes in the inner nature" of the positions they occupy. That is to say, individuals move not so much between classes as with classes. And the major instance of mobility in this sense evident within contemporary capitalism is one involving clearly downward movement, namely, "the case of the proletarianization of the new middle class" through the "devaluation" of its labor power. For Marxist authors, then, there is nothing in the dynamics of the class stratification of modern societies that tends in itself to undermine the possibilities for class-based action—even though this action may be inhibited by other "ideological" or "conjunctural" factors. On the contrary, as the degradation of work and the proletarianization of workers continue inexorably, class differences widen and class interests come into still sharper opposition.

2. Review of the Evidence

In coming now to assess the claims made by liberal and Marxist theories in the light of research findings, I proceed under the same headings as in the previous section. Apart from being convenient, to move through this same sequence of topics—from the evolution of the structure of employment via changes in class structure to trends in class mobility—should help bring out a certain coherence in the empirical results on which I draw.

2.1. Changes in the Structure of Employment

The Marxist thesis of the degradation of labor was developed as an explicit critique of liberal claims of a progressive upgrading of employment. It would, however, seem fair to say that in this respect it is Marxists rather than liberals who have been forced onto the defensive whenever the issue between them has been treated on an empirical basis. The major difficulty for the Marxist position has come form trends displayed in the official employment statistics of the more advanced industrial nations. These show—with considerable regularity—that the greatest increases in nonmanual employment over recent decades have occurred not in relatively low-level clerical, sales, and personal service grades but rather in professional, administrative, and managerial occupations. Furthermore, the major decreases in manual employment have been in the less-skilled rather than the more-skilled categories. Such statistics create in themselves a strong prima facie case against the degrading thesis and have thus forced Marxists into attempting counterarguments. These have taken two main lines.

First, some Marxists point out (see especially Wright and Singelmann 1982) that the discrepancy between the claims of the degrading thesis and the trends apparent in employment statistics could arise through changes taking place in occupational distributions at the societal level that are independent of changes in the organization of production within particular enterprises. Such changes could result simply from shifts in the division of total employment among different industries and sectors that themselves possess different occupational structures. In other words, it is possible that the increase in professionals, administrators, and managers is the consequence largely—or entirely—of the growth of the services sector, within which such occupations have always been more prominent than within, say, the heavy industries, which are now typically in decline. But, it may then be maintained, the effect of degrading are in this way only masked by countervailing tendencies that cannot continue indefinitely; and, as the "service economy" reaches its full development and shift effects on the structure of employment fall off, the reality of degrading will be made increasingly apparent.

However, while this line of argument is analytically sound, it does not of course follow that it is empirically valid, and in fact evidence has of late mounted strongly against it. Although Wright and Singelmann (1982) were able to produce some supportive results for the United States in the 1960s (Singelmann and Browning 1980), later analyses for the United States in the 1970s reported by Singelmann and Tienda (1985) and analyses for a number of European nations (Gershuny 1983) have all yielded results that run in a clearly contrary direction. What is here shown is that the major trends evident in national employment statistics persist even when all interindustry shifts are allowed for. That is to say, net upgrading tendencies must be seen as resulting not only from such shifts but further from technological, organizational, and other changes determining the actual occupational mix at the level of production units.

The second way in which exponents of the degrading thesis have sought to overcome the problem posed by the data on occupational distributions is to claim that the thesis relates not simply to shifts of workers among occupations but further to changes occurring within occupations themselves—that is, to changes in their specific technical and social content. This arguments is difficult to either confirm or refute in any direct way. Detailed investigation of work-tasks and roles of the kind that is called for is scarcely feasible on an economywide basis and has in fact been limited to case studies—the results of which, not surprisingly perhaps, have proved quite inconclusive.^[3] However, one unique data set has afforded the opportunity for an indirect test of the degrading thesis, understood in the way in question and in the context of a national society. In each of the three "rounds" of the Swedish Level of Living Survey, carried out in 1968, 1974, and 1981, a sample of the Swedish population was asked a range of questions on their work and working conditions. These results have recently been analyzed in the light of expectations derived from the degrading thesis (Åberg 1984a, 1984b). Little or no support emerges for the idea that the Swedish labor force experienced degrading over the period covered. For example, levels of work-related training and education generally increased, fewer employees reported a boring routine or a serious lack of autonomy in their jobs, and more found their work mentally (but not physically) demanding.

Efforts to reconcile the degrading thesis with the dominant trends apparent in employment statistics are thus scarcely convincing, and it is notable that some former adherents of the thesis now appear ready to abandon it (see Singelmann and Tienda 1985). It is not of course in question that at any one time in a technologically and economically dynamic society some kinds of work will be in the process of deskilling or degrading in some sense (even if not necessarily for reasons that Marxists would advance). But neither would there seem grounds for questioning the thesis that as old skills, autonomies, and responsibilities disappear, they are replaced—and indeed more than replaced—by new ones, as liberal theorists would maintain.

Does it then follow that in rejecting the degrading thesis, one is at the same time required by the empirical evidence to accept the liberal argument that labor forces are being progressively upgraded? Much depends on what exactly this argument is taken to imply. The declined of manual wageworkers and the expansion of the professional, administrative, and managerial salariat could plausibly be regarded as generic processes of advanced industrial societies. But other developments have of late become apparent in such societies that do not accord well with the more optimistic liberal scenarios that developed in the postwar years of the "long boom," which saw upgrading in effect at all levels of the occupational structure and as leading toward the eventual "professionalization of everyone." Most obviously, one may point to the return since the mid-1970s of large-scale and long-term unemployment to most (though not all) Western societies. This means that even though the balance of change may be in favor of upgrading among jobs that exist, job loss is for many communities, families, and individuals now the central economic reality.^[4] The fact that unemployment remains, as in the past, heavily concentrated among those who were previously wageworkers brings out the persisting differentiation in economic life-chances between this group and the those employees who enjoy the much greater security typically afforded by the "conditions of service" of bureaucratic personnel (Goldthorpe 1985b; Goldthorpe and Payne 1986).

Moreover, alongside unemployment, tendencies have appeared in most Western societies—though in widely varying form and degree—toward what is aptly described by a term arising from the French experience: la précarisation du travail (see Michon 1981). A growth has occurred of diversed forms of employment of a "nonstandard" kind—part-time work, temporary work, home-work, labor-only subcontracting, etc.—that creates a "secondary" labor force that is highly exposed to both market fluctuations and managerial authority. One could say that this is a labor force that is highly flexible and disponible , chiefly because its members lack the protection that "primary" workers possess, whether through organization, legislation, or mere custom and practice (Berger and Piore 1980; Goldthorpe 1984). Recognition of an emerging dualism in this sense need not entail an acceptance of current theories—Marxist or otherwise—of dual or segmented labor markets that are systematically differentiated by levels of skill, pay, or mobility. But it does underline the error of supposing that at the stage of advanced industrialism upgrading proceeds uniformly in all aspects of employment alike.

2.2. Changes in Class Structure

As was previously shown, a close connection exist for both liberal and Marxist theorist between the course of long-term change in the structure of employment in Western in societies and the development of their class structure. Thus, it might be expected that insofar as the accounts offered of changes in the structure of employment are called into question, analyses made at the level of class structure will also be undermined. However, while the structure of employment relations—together with the structure of property relations—can be regarded as forming, so to speak, the matrix of the class structure, the actual derivation of class structure is good deal less straightforward than seems often to be supposed.

To start with the Marxist case, it has already been observed that to reject the idea of a comprehensive and progressive degrading of labor necessitated by the logic of capitalism is not to deny that the degrading of specific types of work, over specific periods, is a feature of modern economies. Thus, there would appear ample evidence to support the arguments advanced by Marxist (and others) that a wide range of nonmanual—especially clerical, sales, and personal service—occupations are now little differentiated in their associated employment relations and conditions from rank-and-file manual occupations. But what has then to be challenged in the Marxist position is the further claim—or simple assumption—that this degrading of nonmanual work can be equated with a process of proletarianization through which the numerical strength of the present-day working class is maintained, if not indeed enlarged.

What is most obviously neglected here is the question of the appropriate unit of class structure and thus of class analysis. In tradition of such analysis, Marxist and non-Marxist, the appropriate unit has been seen as the family rather than the individual (see Goldthorpe 1983). And it must then be pointed out that for so long as this view is maintained, the implications for class structure of recent changes in the character of nonmanual employment cannot be reckoned as in any way so dramatic as Marxist theorists would make out. In all modern societies the expansion of lower nonmanual occupations has largely occured via the increased employment of women, and women now predominate in these occupations. But where the significance of these changes for the class structure has been considered by the authors in question, they have concentrated on the fact that growing numbers of women in nonmanual occupations are married to men who are manual workers and have taken this as further evidence of the consolidation of the working class as the differentiation between nonmanual and manual employment steadily weakens. As Braverman puts it (1974, 353), not only are clerical workers and factory workers increasingly recruited from similar origins but increasingly "they are merged within the same living family." However, what is thus missed is the significance of the further fact that while many women in lower-level nonmanual work are married to manual workers, generally more would appear to be married to men in higher-level nonmanual positions, and that is these latter women who create major difficulties for claim of proletarianization.^[5] Why, one may ask, should women who perform deskilled nonmanual work but who are married to professionals, administrators, or managers, and thus share substantially in their standards and styles of living, be regarded as proletarian in the same way as women in similar jobs who are married to manual workers?^[6]

It can of course be contended that in the case of societies where most married women at some time enter paid employment, the family should no longer be taken as the unit of class analysis, and that rather than the class position of married women being derived from that of their husbands, it should be seen as following from their own employment. This argument has found much favor with radical feminists who are concerned to demonstrate the inherent sexism of the conventional view (e.g., Delphy 1981; Stanworth 1984). But what its exponents have so far failed to show is that any analytical or explanatory advantage is to be gained from adopting it: that is, that when married women are systematically attributed class locations by reference to their own employment, this makes more intelligible their own class identifications or, say, their life-styles and patterns of sociability or their political involvements. Indeed, insofar as the question of married women's class identity has been seriously investigated, the results would suggest that wives themselves (and not just "sexist" sociologists) do derive their class positions from their husbands' employment rather than from their own (Jackman and Jackman 1983, 140–52; Hernes and Knudsen 1985). Instead, then, of the expansion of degraded, and largely feminized, nonmanual work leading to mass proletarianization and thus, presumably, to an increased potential for radical sociopolitical action, it would seem that the performance of such work by many married who, however, still take their class identity from husbands in superior positions could be a major stabilizing influence within modern class structure.^[7]

Furthermore, the simple equation of the degrading of nonmanual work with proletarianization again runs into difficulties—even if attention is concentrated on men—through the neglect of another, quite different issue: the differentiation of apparently similar jobs according to the characteristics of the individuals who occupy them. Although the distinction between "positions" and "persons" is crucial to class analysis, it needed not prevent one from recognizing that employing organizations can, and do, channel different kinds of employees into essentially similar work-task and roles—but in the context of what are, or are envisaged as being, quite different work histories. Studies of clerical work in particular have shown that while women workers are overwhelmingly expected to remain in low-grade employment, men who are at any one time found in clerical jobs can be divided into two categories: those who, usually rather late in their working lives, have moved into such jobs from manual ones and have few further prospects, and those—generally younger and better educated—who are spending some time in clerical jobs early in careers that they, and their employers, see as leading eventually to administrative or managerial positions (Goldthorpe 1980; Stewart, Prandy, and Blackburn 1980). Even if men in both categories are engaged in deskilled, routinalized work, it still does not follow that either case exemplifies mass proletarianization: these men either were rank-and-file wageworkers previously or are, for the most part, passing through jobs that, for them, are staging-posts on the way to the higher levels of bureaucratic structures.

In sum, one could say, claims of extensive proletarianization fail because workers in the widening range of degraded or low-grade nonmanual occupations are precisely not "an immense mass," as Braverman would have it. On the contrary, they constitute a workforce that is highly differentiated—by sex, age, and qualifications—in ways that clearly delimit the influence of the growth of this workforce on the shape of the class structure.

In turning, then, to the position of liberal theorist, an awareness that the structure of employment does not "map" in any direct way onto the class structure must lead one to the question of how serious for their understanding of class structural change is their failure to recognize the widespread deterioration of labor market conditions within Western capitalism after the ending of the "long boom." It might in fact be maintained that their disregard of rising levels of unemployment is not in this respect of major consequences since the unemployed do not themselves form a class; or again, that the growth of a secondary labor force of men and women in nonstandard jobs will have little effect on the class structure insofar as women at least are themselves secondary earners within families and households and have other sources of economic support—and of social identity—than those provided by their work. Such arguments are not without force, and it must be said that whatever weaknesses may exist in the liberal account, the tendency that it emphasizes for the working class to contract and the salariat to expand remains rather more in accord with available evidence than the largely contrary tendencies that Marxist theories envisage. Nonetheless, radically changed circumstances in labor markets still reveal several ways in which liberal conceptions of the emerging class structure are gravely inadequate.

For example, although the unemployed may not constitute a class, it is increasingly apparent that a substantial increase in the numbers of long-term unemployed—as can be found, say, in the United Kingdom—results in a polarization of the working class in terms of incomes, life-chances, and life-styles.^[8] Thus, contrary to liberal expectations, the overall range of economic and social inequality has significantly widened rather than narrowed. Moreover, while workers in nonstandard jobs are often secondary earners, attracted into the labor market in part by the availability of such jobs (married women working part-time, juveniles or semiretired persons in temporary work, etc.), this is by no means always the case. Most obviously, in depressed labor markets nonstandard jobs may be taken simply because nothing better is available. To the extent, then, that such employment represents the main source of familie's economic support, the formation of an "underclass" is further promoted.

Finally, it should be noted that the more difficult economic conditions that have followed the long boom may be associated with a growth in the numbers of the self-employed and small employers. The processes at work here are various and complex: self-employment may represent an attempt to escape unemployment, some nonstandard jobs in effect impose self-employment on those taking them, small employers often play key roles in organizing the secondary labor force as, say, in subcontracting, and so on. However, what is clear is that liberal—and also, of course, Marxist—expectations that the importance of such "independents" would steadily diminish within modern economies have not been borne out, and the very idea that the petty bourgeoisie is a declining or transitional class now itself appears anachronistic in many Western societies (Berger and Piore 1980; Bechhofer and Elliot 1981; Scase 1982).^[9]

2.3. Changes in Social Mobility

As regards class mobility, it was earlier remarked that liberal and Marxist theorists differ fundamentally on its nature and extent. For liberals, the mobility of individuals between different class positions is a process central to the "social metabolism" of industrial nations; for Marxists, mobility occurs to only a negligible degree, except in the form of the collective—and downward—movement associated with the degrading of labor and proletarianization. As thus posed, the issue is not difficult to decide: there is an overwhelming body of evidence that is consistent with the liberal view but renders the Marxist position untenable.

For example, Carchedi's claim that "generally speaking, a member of the working class remains a member of the working class" is clearly falsified by findings from mobility research, which show that within the populations of advanced Western industrial societies some 20–30 percent of the sons of manual wage earners are regularly found in quite different (that is, salaried professional, administrative, and managerial or self-employed) class positions—and that this proportion has been steadily rising. Mobility research also shows that about 20 percent of men of such origins who started their own working lives as manual wage earners are subsequently able to move out of such employment.^[10] It may furthermore be observed that the idea of the mass proletarianization of lower-level nonmanual workers is also undermined by the tendency, previously referred to, for younger males to be promoted from routine clerical and sales jobs as part of a more-or-less planned career path. In addition, there is also considerable upward mobility from this kind of work that is of a more contingent kind—associated, say, with changes of employer. It may be estimated that in present-day Western societies men who begin their working lives in routine clerical and sales occupations have better than a one-in-three chance of being promoted to professional, administrative, or managerial positions; and this probability rises to around one-in-two for men who are of nonmanual class origins. The relevance of a point I have made elsewhere is thus brought home: "It is work that is degraded, and individuals who are proletarianized; and where high mobility prevails, the former process in no way entails the latter" (Goldthorpe 1985b, 189 [my translation], emphasis in original; cf. Gagliani 1981).

However, while there is no shortage of evidence to support the liberal claim that in industrial societies the amount of class mobility is substantial, this is not to say that liberal accounts of either mobility patterns or trends—or of the sources of change in these—need be accepted as they stand. Indeed, by reference again to the findings of mobility research, one can show that these accounts are seriously mistaken. The basic flaw is that liberal theorists have taken the clear evidence of increasing rates of upward mobility, which—as they themselves have emphasized—must be expected to follow from the expansion of the higher levels of the class structure, and have then interpreted this evidence as being indicative also of the increasing openness or social fluidity that they see as required by the logic of industrialism and that they believe is created through an emphasis on achievement rather than ascription in social selection. They have, moreover, felt supported in such an interpretation by further evidence of a "tightening bond" between educational and occupational attainment—or, in other words, by what might be regarded as evidence of the increasing sway of the meritocratic principle. However, what is here neglected is the possibility that rising rates of upward social mobility are not merely favored by the changing shape of the class structure but are attributable almost entirely to such structural shifts, and that little if any change need therefore be supposed in openness or fluidity. In fact, it is this latter interpretation, rather than the one advanced by liberal authors, that the research findings bear out. Contrary to liberal expectations that advancing industrialism should cause the association between individuals' class origins and their eventual class destinations to weaken steadily, this association has proved to be patterned in a remarkably stable way—once the effects of structural shifts are allowed for—both over time and cross-nationally. That is to say, while actually observed mobility rates vary quite widely within the experience of industrial nations, this variation is to a predominant extent structurally conditioned, and a large commonality would seem to prevail in underlying "mobility regimes" or "patterns of social fluidity" (Featherman, Lancaster Jones, and Hauser 1975; Grusky and Hauser 1984; Erikson and Goldthorpe 1985b, 1987). Some instances can be cited of fluidity showing an increase over time but, rather than exemplifying any inherent tendency of industrialism, these appear to be better interpreted as episodes arising out of the specific historical circumstances of particular nations (Simkus 1981; Goldthorpe and Portocarero 1981; Erikson 1983; Erikson, Goldthorpe, and Portocarero 1983).

Once, then, the idea of a general movement toward greater openness is rejected, other aspects of the liberal account of class mobility within industrial societies also come into question. If the observed increases in mobility are overwhelmingly the result of structural change, there seems no longer any good reason for believing that societies either have in the past or will in the future become more mobile pari passu with their economic development. For while economic development is of course associated with structural changes of a kind that will clearly affect class mobility, the historical record would suggest that this association is complex and is far more likely to produce fluctuations in mobility rates rather than any unidirectional change (Goldthorpe 1985a).

Again, it must follow that, in the absence of greater fluidity, the increase in intergenerational upward mobility into the expanding higher levels of the class structure that liberals have especially stressed will be accompanied by a decrease in downward mobility from these levels. And indeed the empirical findings show that what I have elsewhere referred to as the "service class" of modern industrial societies, that is, the class of salaried professionals, administrators, and managers (Goldthorpe 1982; see also Renner 1953; Dahrendorf 1964), is not only growing steadily but is also increasing in the intergenerational stability of its constituent families.^[11] Moreover, declining downward mobility also implies less diversity in the social origins of those found at the lower levels of the class structure, and this trend—along with the declining outflow from the agricultural sector—has affected the composition of the working classes of modern industrial societies in a way quite overlooked in liberal scenarios: that is to say, they have become increasingly self-recruiting.^[12] At the same time, the tightening bond between education and occupation—the supposed consequence of meritocracy—means that decisive mobility away from the working class is now being more often achieved before entry into employment rather than in the course of the individual's working life. And from these two developments together, then, it may be expected that the collectivity that actually occupies working-class positions at any one point in time will comprise a substantial, and growing, core of those who are in fact both "hereditary" and prospective "lifetime" members (Goldthorpe 1980, 1985b).

Finally, the idea that the greater importance of education in determining employment chances is indicative of the prevalence of meritocratic social selection is itself thrown into doubt if no evidence of an accompanying increase in fluidity can be produced. In this case, the alternative hypothesis is suggested that education is simply substituting for the previous determinants of processes of mobility or immobility without leading to any significant change in outcomes, so that, as one commentator has put it, these processes become in effect ones "in which ascriptive forces find ways of expressing themselves as 'achievement' " (Halsey 1977, 184; see also Parkin 1974; Halsey, Heath, and Ridge 1980).

In sum, then, while trends in rates and patterns of class mobility in modern industrial societies may well have had a stabilizing effect, as liberals have argued, their account of how this effect has been produced can only be accepted to a very limited extent: that is, insofar as it points to increased opportunities for upward movement as a consequence of class structural change. There is no evidence that modern societies have become generally more fluid, whether through the application of more meritocratic criteria of social selection or otherwise, or that these societies have experienced progressive class decomposition. Recruitment to expanding service classes has been necessarily broad-based, but these classes are now in the process of consolidation; and working classes, although contracting, are becoming increasingly homogeneous at least in terms of their members' social origins.

Therefore, even if a secular decline can be traced in the propensity for collective identities and collective action to develop on a class basis (which might be disputed; see Bottomore 1982; Korpi 1983; Heath, Jowell, and Curtice 1985), there is little reason to attribute this decline, in the manner of Blau and Duncan, to class formation being inhibited by the increasing openness of Western industrial societies. If an explanation in structural terms is to be given—rather than, say, one primarily in terms of the organization and strategies of working-class movements (Esping-Andersen 1985; Przeworski 1985)—it would far more plausibly refer to the divisions created within contemporary working classes by developments that, as earlier noted, liberal scenarios conspicuously overlook: the growth of nonstandard forms of employment, the dualizing of labor markets, and the return of unemployment on a mass scale.

3. Social Theories, Historicism, and the Historical Record

Why, then, should it be that when reviewed against the findings of systematic empirical research, liberal and Marxist theories alike appear so inadequate as a basis for understanding long-term trends of change in the class stratification of Western societies? The answer I would give at the most general level is the following: that these theories fail to do justice to the social processes that they address because in their conception and application there is a persisting historicist strain. They are theories that stem from an ultimate ambition of achieving some cognitive grasp on the course of historical development—which can then be used for normative and political purposes: that is to say, to show that certain political beliefs, values, and commitments have an "objective" superiority in being those that the movement of history favors, and that others can be "correctly" dismissed as historically outmoded.

The grave logical—and also moral—defects of such historicist endeavors have been evident enough at least from the time of Popper's devastating attacks on them (1944–45, 1945), which focused on the claims of classical Marxism and social-evolutionary theory. Since then historicism has to some extent been forced underground; but, as I have sought to show elsewhere (Goldthorpe 1971, 1979; see also Goldthorpe 1972), it has retained a powerful appeal for many social scientists—especially for those who would aspire to the more elevated status of "intellectual" and, moreover, within the liberal camp just as much as within the Marxist camp.

Thus, not only may one find Carchedi (1977, 17) speaking in more or less orthodox Marxist fashion of "dialectical determination" as "the mechanism connecting one stage of the development of capitalism to another" but, likewise, Kerr and his associates ([1960] 1973, chaps. 1 and 2) explicitly taking Marx's quest for the "laws of motion" of capitalist society as their model in their attempts to establish a developmental "logic of industrialism" (see also Kerr 1983, chap. 1). Other theorists have expressed themselves more guardedly—but chiefly, it would seem, in order to try to obtain the political advantages of a historicist position while protecting themselves against the more obvious objections to such a stance. For example, Wright (1985, 114–18) disavows the idea of "iron laws of history" and sees stages of class relations as no more than a sequence of "historical possibilities," yet still wishes to claim—without any attempt at explaining why this should be so—that "the overall trajectory of historical development" is of an inherently "progressive" kind. Or again, Blau and Duncan, while in large part pursuing purely descriptive and analytical goals, still base their ultimate interpretation and evaluation of long-term changes in the process of stratification on the central tenet of the "neoevolutionism" of Talcott Parsons, namely, that "a fundamental trend toward expanding universalism characterizes industrial society" (Blau and Duncan 1967, 429; see also Parsons 1951, 480–535; Parsons 1964, 1966).

Following Popper, it could be argued that theories of a historicist cast must in the end always disappoint when matched against the historical record since they are flawed by a basic misconception: that is, their exponents fail to recognize that, in any science, propositions can be either theoretical or historical but not both at the same time. Thus, their hope of "theorizing" the movement of history is vain in principle. However, of present purposes it may still be instructive to consider those deficiences that would appear in a more immediate way to undermine the analyses of social change that are conducted from essentially historicist positions.

To begin with, one cannot overlook the prevalence of what is in fact rather blatant wishful thinking. Marxists clearly want to be able to show progressive proletarianization, or liberals a steadily increasing equality of opportunity, so that they may thus fortify the faithful, confound their political enemies, and sustain their own beliefs and values. Logically, within a historicist program, political commitment should follow from a correct appreciation of the course on which historical development is set; but, psychologically, commitment seems often to dictate what view of history it is that must be represented as correct. Such an outlook is obviously then not one conducive to the recognition—and still less to the searching out—of evidence contrary to the developmental scenarios that are favored. For much more is at stake than an intellectual construction: historicism makes it difficult to accept Popper's advice that we should be ready to let our ideas die for us.

Further, though, there are also deficiencies to be noted that derive directly from the structure of historicist argument itself. Most serious here is the inadequate attention that is given to what might be called microsociological foundations, that is, to the way in which it is envisaged that the large-scale and long-term trends of change that are postulated actually come about at the level of social action (Nisbet 1969, 233). It is characteristic of historicist theories of social change—and indeed essential to their author's project of "theorizing" history—that they should concentrate on identifying one central source of dynamism that is, so to speak, "built into" the overall process of historical development and drives it along from stage to stage. In Marxist theories this dynamism is usually seen as resulting from the contradictions that recurrently arise between the forces and the relations of production; in liberal theories this dynamism stems from the ever-renewed and reshaped requirements of technological change and economic rationality. In other words, the focus of interests is on what are taken to be successive "system exigencies" to which determinate responses must follow; and the way in which these exigencies are then experienced and accommodated by social actors be of no more than minor concern. For if a historicist theory is taken as valid, then the patterns of social action that are involved in its realization are mere epiphenomena. If the analysis moves to the level of social action at all, it need do so only for illustrative and, perhaps, political purposes. Thus, for Marxists, the progressive degrading jobs by employers and their managers is a necessary response as technological advance proceeds within the constraints of the capitalist mode of production—and serves to reveal its inhumane character; for liberals, the emergence of more meritocratic procedures of social selection is dictated by the universalist logic of modern industrial society—and shows how this offers ever-wider opportunities for the expression of talent.

However, the difficulty here is of course that if the theory is not valid, then there is no reason whatever for highlighting the patterns of action and the underlying values and motivations that are its supposed vehicles. They may not in fact be those that prevail in the determination of social change; if they are present at all, they may be opposed, cross-cut, or overwhelmed by a host of other factors of which the theory takes no account. Thus, while economic constraints may lead some employers to degrade jobs, other employers may survive by linking technological innovation to either the reduction or the upgrading of their work forces, or by opting for strategies that aim more at reshaping employment relations than job content—and that carry quite different implications for the composition of the work force and the class structure. Likewise, while in modern societies pressures may indeed exist for selection for education and employment to be based increasingly on achievement, the interests and factors favoring ascription are not thereby annulled, and the emergence of a more open and fluid society is in no way guaranteed. In sum, in the pursuit of the key dynamic inherent in the movement of history—which one may regard as not merely elusive but illusory—theorists who yield to historicist temptations become captive to remarkably limited and oversimplified views of the processes of social change, and, in turn, they seek to sustain what are at best one-sided and at worst wildly misleading accounts of its actual patterns.^[13]

If, then, we are to move toward a more satisfactory theoritical treatment of social change—in class stratification or any other aspect of social structure or process—I would maintain that a prerequisite is to accept, once and for all, that historical and theoretical propositions should not be confounded. This would mean two things. First, we should, as social scientists, treat the historical record with respect. We should recognize it as being independent of our theories of social change, and not suppose that these theories can in any way transcend it. There is no empirical realm to which they can refer other than this record, and without having established it, as best we can, in an area of interest to us, we do not know what it is that we would wish our theories to account for. As Nisbet has argued, "Between the study of change … and history there is quite evidently an unbreakable relationship, when we come down from the empyrean heights of abstractions, wholes and universals…. Whatever the demands of a social theory, the first demands to be served are those of the social reality we find alone in the historical record" (1969, 303–4).

Second, though, we must recognize that simply by "tracing the path" followed by historical development we can explain nothing; there is no theory immanent in the course of history that we can discover and then, so to speak, apply back to history. Whether taken as comprising a series of events or empirical regularities à la longue durée, the historical record is, from the point of view of the social scientist, and an explanandum —or rather a vast source of explananda —but not an explanans (see Gellner 1964, 20). This means, therefore, that the theories through which we may seek to gain an understanding of social change will not be in themselves historical (or evolutionary or developmental) theories; more specifically, one could say that they will not be theories that embody propositions about historical—as opposed to analytical—time. In the end, I would suggest, they will have to be, like theories addressed to questions of social order and stability, ones that are couched in terms of human action, individual and collective, and its intended and unintended consequences.^[14] In other words, we must learn to take seriously the idea that "men make their own history." This does not of course prevent us from recognizing that they do not make it "just as they please"; but what is precluded is any idea, overt or covert, that in the last analysis history is made "behind men's backs" in accordance with some immanent plan or ultimate goal.

References

Åberg, Rune. 1984. Arbetsförhallanden. In Välfärd i förändring, ed. Robert Erickson and Rune Aberg, 102–15. Stockholm: Prisma.

Åberg, Rune. 1984b. Teoriarna om arbets degrading och arbetsmarknadens dualisering: ett försök till empirisk prövning. Sociologisk Forskning 2.

Bagnasco, Arnaldo, and Carlo Trigilia. 1984. Società e politica nelle Aree di Piccola Impresa . Dorsoduro, Italy: Arsenale Editrice.

Bechhofer, Frank, and Brian Elliot, eds. 1981. The petite bourgeoisie . London: Macmillan.

Bell, Daniel. 1967. Notes on the post-industrial society. The Public Interest 6:24–35.

Bell, Daniel. 1973. The coming of post-industrial society . New York: Basic Books.

Berger, Suzanne, and Michael Piore. 1980. Dualism and discontinuity in industrial societies . Cambridge: Cambridge University Press.

Blau, P. M., and O. D. Duncan. 1967. The American occupational structure . New York: Wiley.

Bottomore, Tom. 1982. The political role of the working class in Western Europe. In Social class and the division of labor, ed. Anthony Giddens and Gavin Mackenzie, 265–75. Cambridge: Cambridge University Press.

Boudon, Raymond. 1977. Effets pervers et ordre social . Paris: Presses Universitaires de France.

Braverman, Harry. 1974. Labor and monopoly capitalism . New York: Monthly Review Press.

Brusco, Sebastiano. 1982. The Emilian model: Productive decentralisation and social integration. Cambridge Journal of Economics 6:167–84.

Carchedi, G. 1977. On the economic identification of classes . London: Routledge.

Clark, Colin. 1957. The conditions of economic progress . 3d ed. London: Macmillan.

Crompton, Rosemary, and Gareth Jones. 1984. White-collar proletariat . London: Macmillan.

Crozier Michel. 1971. The world of the office worker . Chicago: University of Chicago Press.

Dahrendorff, Ralf. 1964. Recent changes in the class structure of European societies. Daedalus Winter: 225–70.

Delphy, Christine. 1981. Women in stratification studies. In Doing feminist research, ed. Helen Roberts 114–28. London: Routledge.

Duncan, O. D. 1968. Social stratification and mobility. In Indicators of social change, ed. Eleanor B. Sheldon and Wilbert E. Moore, 675–719. New York: Russell Sage.

Elster, Jon. 1982. Marxism, functionalism and game theory. Theory and Society 11:453–82.

Elster, Jon. 1985. Making sense of Marx . Cambridge: Cambridge University Press.

Erikson, Robert. 1983. Changes in social mobility in industrial nations: The case of Sweden. Research in Social Stratification and Mobility 2:165–95.

Erikson, Robert, and John H. Goldthorpe. 1985a. Are American rates of social mobility exceptionally high? European Sociological Review 1:1–22.

Erikson, Robert, and John H. Goldthorpe. 1985b. Commonality and variation in social fluidity in industrial nations: Some preliminary results. CASMIN Working Paper 4.1, University of Mannheim.

Erikson, Robert, and John H. Goldthorpe. 1987. Commonality and variation in social fluidity in industrial nations. European Sociological Review 3:54–77, 145–66.

Erikson, Robert, John H. Goldthorpe, and Lucienne Portocarero. 1983. Intergenerational class mobility and the convergence thesis. British Journal of Sociology 34:303–43.

Esping-Andersen, Gøsta. 1985. Politics against markets . Princeton: Princeton University Press.

Featherman, David L., F. Lancaster Jones, and Robert M. Hauser. 1975. Assumptions of social mobility research in the U.S.: The case of occupational status. Social Science Research 4:329–60.

Fuchs, Victor. 1968. The service economy . New York: National Bureau of Economic Research.

Gagliani, Giorgio. 1981. How many working classes? American Journal of Sociology 87:259–85.

Gellner, Ernest. 1964. Thought and change . London: Weidenfeld and Nicolson.

Gershuny, Jay. 1983. Social innovation and the division of Labour . Oxford: Oxford University Press.

Giddens, Anthony and Gavin Mackenzie, eds. 1982. Social class and the division of labour . Cambridge: Cambridge University Press.

Goldthorpe, John H. 1971. Theories of industrial society. Archives européennes de sociologie 12:263–88.

Goldthorpe, John H. 1972. Class, status and party in modern Britain. Archives européennes de sociologie 13:342–72.

Goldthorpe, John H. 1979. Intellectuals and the working class in modern Britain. Fuller Bequest Lecture, University of Essex.

Goldthorpe, John H. 1980. Social mobility and class structure in modern Britain . Oxford: Clarendon Press.

Goldthorpe, John H. 1982. On the service class: Its formation and future. In Social class and the division of labour, ed. Anthony Giddens and Gavin Mackenzie 162–85. Cambridge: Cambridge University Press.

Goldthorpe, John H. 1983. Women and class analysis: In defense of the conventional view. Sociology 17:465–88.

Goldthorpe, John H. 1984. The end of convergence: Corporatist and dualist tendencies in modern Western societies. In Order and conflict in contemporary capitalism, ed. John H. Goldthorpe, 315–43. Oxford: Clarendon Press.

Goldthorpe, John H. 1985a. On economic development and social mobility. British Journal of Sociology 36:549–73.

Goldthorpe, John H. 1985b. Soziale Mobilität und Klassenbildung: Zur Erneuerung einer Tradition soziologischer Forschung. In Die Analyse Sozialer Ungleichheit, ed.

Hermann Strasser and John H. Goldthorpe, 174–204. Opladen, W. Ger.: Westdeutscher Verlag.

Goldthorpe, John H., and Clive Payne. 1986. Trends in intergenerational class mobility in England and Wales, 1972–1983. Sociology 20:1–24.

Goldthorpe, John H., and Lucienne Portocarero. 1981. La mobilité sociale en France, 1953–1970: Nouvel examen. Revue française de sociologie 22:151–66.

Grusky, David B., and Robert M. Hauser. 1984. Comparative social mobility revisited: Models of convergence and divergence in 16 countries. American Sociological Review 49:19–38.

Halfpenny, P. J., A. J. Laite, and I. Noble. 1985. The Blackburn and Darwen labour market . Manchester: University of Manchester, Centre for Applied Social Research.

Halsey, A. H. 1977. Towards meritocracy? The case of Britain. In Power and ideology in education, ed. Jerome Karabel and A. H. Halsey, 173–86. New York: Oxford University Press.

Halsey, A. H., Anthony Heath, and J. M. Ridge. 1980. Origins and destinations . Oxford: Clarendon Press.

Hamnett, C. 1984. Housing the two nations: Socio-tenurial polarisation in England and Wales, 1961–81. Urban Studies 43:389–405.

Heath, Anthony, Roger Jowell, and John Curtice. 1985. How Britain votes . Oxford: Pergamon Press.

Hechter, Michael, ed. 1983. The microfoundations of macrosociology . Philadelphia: Temple University Press.

Hernes, Gudmund, and Knud Knudsen. 1985. Gender and class identification in Norway. Paper presented to the Social Stratification and Mobility Research Committee of the International Sociological Association, Harvard University, Cambridge, Mass.

Jackman, Mary R., and Robert W. Jackman. 1983. Class awareness in the United States . Berkeley: University of California Press.

Kerr, Clark. 1983. The future of industrial societies . Cambridge: Harvard University Press.

Kerr, Clark, John T. Dunlop, Frederick H. Harbison, and Charles A. Myers. [1960] 1973. Industrialism and industrial man . Rev. ed. Cambridge: Harvard University Press.

Korpi, Walter. 1983. The democratic class struggle . London: Routledge.

Kuznets, Simon. 1966. Modern economic growth . New Haven: Yale University Press.

Lockwood, David. 1981. The weakest link in the chain? Some comments on the Marxist theory of action. Research in the Sociology of Work 1:435–81.

Marshall, Gordon, Howard Newby, David Rose, and Carolyn Vogler. 1988. Social Class in Modern Britain . London: Hutchinson.

Michon, François. 1981. Dualism and the French labour market: Business strategy, non-standard job forms, and secondary jobs. In The dynamics of labour market segmentation, ed. Frank Wilkinson, 81–97. London: Academic Press.

Nisbet, Robert. 1969. Social change and history . New York: Oxford University Press.

Olson, Mancur. 1982. The rise and decline of nations . New Haven: Yale University Press.

Pahl, Ray. 1984. Divisions of labour . Oxford: Blackwells.

Parkin, Frank, ed. 1974. The social analysis of class structure . London: Tavistock.

Parsons, Talcott. 1951. The social system . Glencoe, Ill.: Free Press.

Parsons, Talcott. 1964. Evolutionary universals in society. American Sociological Review 29:339–57.

Parsons, Talcott. 1966. Societies: Evolutionary and comparative perspectives . Englewood Cliffs, N.J.: Prentice-Hall.

Penn, Roger, and Hilda Scattergood. 1985. Deskilling or enskilling: An empirical investigation of recent theories of the labour process. British Journal of Sociology 36:611–26.

Popper, Karl. 1944–45. The poverty of historicism. Economica 11:86–103, 119–37, 12:69–89.

Popper, Karl. 1945. The open society and its enemies . London: Routledge.

Poulantzas, Nicos. 1974. Les classes sociales dans le capitalisme aujourd'hui . Paris: Seuil.

Przeworski, Adam. 1985. Capitalism and social democracy . Cambridge: Cambridge University Press.

Renner, Karl. 1953. Wandlungen der Modernen Gesellschaft: Zwei Abhandlungen über die Probleme der Nachkriegszeit . Vienna: Wiener Volksbuchhandlung.

Scase, Richard. 1982. The petty bourgeoisie and modern capitalism. In Social class and the division of labour, ed. Anthony Giddens and Gavin Mackenzie, 148–61. Cambridge: Cambridge University Press.

Simkus, Albert A. 1981. Changes in occupational inheritance under socialism: Hungary, 1930–1973. Research in Social Stratification and Mobility 1:171–203.

Singlemann, Joachim, and H. L. Browning. 1980. Industrial transformation and occupational change in the U.S., 1960–70. Social Forces 59:246–64.

Singlemann, Joachim, and Marta Tienda. 1985. The process of occupational change in a service society: The case of the United States, 1960–1980. In New approaches to economic life, ed. Bryan Roberts, Ruth Finnegan, and Duncan Gallie, 48–67. Manchester: Manchester University Press.

Stanworth, Michelle. 1984. Women and class analysis: A reply to John Goldthorpe. Sociology 18:159–70.

Stewart, A., K. Prandy, and R. M. Blackburn. 1980. Social stratification and occupations . London: Macmillan.

Treiman, Donald J. 1970. Industrialisation and social stratification. In Social stratification: Research and theory for the 1970s, ed. E. O. Laumann, 207–34. Indianapolis: Bobbs Merrill.

Wright, Erik Olin. 1976. Class boundaries in advanced capitalist societies. New Left Review 98:3–42.

Wright, Erik Olin. 1978. Class, crisis and the state . London: New Left Books.

Wright, Erik Olin. 1985. Classes . London: Verso.

Wright, Erik Olin, and Joachim Singlemann. 1982. Proletarianisation in the changing American class structure. American Journal of Sociology 88 (supplement): 176–209.


Notes:

[1] It will also be apparent that, contrary to the mood ("argument" would be too strong a word) of much recent "methodological" writing in sociology, I adhere to the—essentially Popperian—view that if theories are to have any informational and hence explanatory value, they must be open, either directly on through propositions derivable from them, to criticism and to possible refutation in the light of empirical inquiry. I am aware of the philosophical difficulties that arise from a naive—or even with a sophisticated—"falsificationist" position, most notably, the possibility that the empirical results that apparently disconfirm a proposition may themselves be called into question via the concepts and theories that guided and conditioned their production. However, it seems to me that in the actual practice of any kind of scientific endeavor such difficulties can be largely circumvented, for example, by the proponents of a theory being themselves prepared to state what empirical results, produced in what way, would lead them to regard their theory as no longer valid. I believe that most of the authors whose work I discuss in this paper do, at least in practice, take a view of the relationship between theory and research that is broadly compatible with my own, and none explicitly adopts a position that would render his or her theoretical arguments immune to empirically based criticism.

[2] Poulantzas (1974) represents what must be regarded as a deviant Marxist view in this respect. His conceptualization of modern class structures would apparently require him to recognize, along with liberal theorists, a numerically declining working class, together with what he would regard as an expanding "new petty bourgeoisie." It is difficult to resist the conclusion that criticism by other Marxists of the position adopted by Poulantzas has been as much inspired by the uncongenial nature of its political implications as by preexisting theoretical disagreements. Thus, for example, Wright complains (1976, 23) "It is hard to imagine a viable socialist movement developing in an advanced capitalist society in which less than one in five people are workers." For an incisive critique of recent Marxist debates on class structure and action. see Lockwood (1981).

[3] This point could scarcely be better demonstrated than by the very extensive literature generated by British industrial sociologist under the affliction of "Bravermania." See, purely for purposes of illustration of the generally prevailing incoherence of theory and evidence, the recent contributions of Crompton and Jones (1984) and Penn and Scattergood (1985) and the earlier debates to which these authors refer. It must, moreover, be recognized, as argued elsewhere, that even quite indisputable evidence of degrading derived from particular case-studies can be of little value in defending the degrading thesis against the upgrading thesis in so far as the argument is about class structures . Supporters of the latter thesis do not seek to deny that deskilling or other forms of degrading occur (although they might wish to see these as integral to the development of industrial, rather than of specifically capitalist, societies); but they would still maintain that the net result of technological and organisational change over the economy as a whole is an increase in skill levels and in proportion of the work-force in salaried or bureaucratic conditions of employment . In other words, macro-sociological arguments can only be adequately discussed on the basis of macro-sociological data. (Goldthorpe and Payne 1986, 23)

[4] It is remarkable that in Bell's The Coming of Post-Industrial society —subtitled A Ventured in Social Forecasting and published in 1973—the index contains only three references to unemployment, and each of these turns out to be quite trivial.

[5] Thus, for example, data relating to married women aged 20–64 living in England and Wales (taken from the sample of the British general election study in 1983) reveal that of all such women whose present or last employment was in routine clerical associated work (N = 301), 45 percent were married to men in professional, administrative, and managerial positions, as against only 27 percent married to manual wageworkers. The husbands of the remaining 28 percent were either themselves in routine nonmanual work or otherwise small proprietors, self-employed artisans, foreman, and supervisors, or technicians.

[6] Braverman appears (1974, 130) to sense the weakness in his argument here in citing Crozier's (1971) observation that "the proletarianization of white-collar employees does not have the same meaning at all if it is women, and not heads of family, who constitute the majority of the group." But rather than offering any serious response, he merely embarks on a diversionary attack on Crozier for being "pleased to use women as that category of the labor force for which any job is good enough."

[7] Support is lent to this suggestion by the finding in a recent British study that the association between a married woman's political partisanship and her class position, as assessed by reference to her own employment, is less strong than that between her partisanship and her husband's class position, and that once the latter variable is controlled for, introducing a woman's own class into the analysis does not in fact make any further significant contribution to accounting for her partisanship (Marshall, et al. 1988, chap. 5). It is of course always possible that a larger sample would be able to show a significant—if slight—effect, but such an effect might again be eliminated if a woman's education were also considered. The results reported by both Jackman and Jackman and Hernes and Knudsen would suggest that education has a far greater influence than employment on a woman's sociopolitical orientations.

[8] The theme of working-class polarization from several British studies in recent years. See, for example, Hamnett 1984; Pahl 1984; Halfpenny, Laite, and Noble 1985; Goldthorpe and Payne 1986.

[9] As Berger and Piore's work well brings out, the Italian case is a particularly important one in this respect. See also, for example, Brusco 1982; Bagnasco and Trigilia 1984.

[10] I draw here on the results of work currently being undertaken, in collaboration with Robert Erikson, as part of the CASMIN (Comparative Analysis of Social Mobility in Industrial Nations) project, based at the Institut für Sozialwissenschaften of the University of Mannheim and supported by the Stiftung Volkswagenwerk. The data from national inquiries conducted in the early and mid-1970s have been systematically recoded in order to produce information of a generally high standard of comparability on rates and patterns of class mobility across a range of industrial societies. Reference in this paper is chiefly to findings for France, the Federal Republic of Germany, Great Britain, Sweden, and the United States. (For more detailed accounts, see Erikson and Goldthorpe, 1985a, 1985b, 1987).

[11] Results from the CASMIN project would indicate that for the men born into service-class families in the first two decades of the twentieth century, there was typically around a three-in-five probability that they would themselves "succeed" to service-class positions, while for men born into such families over the next two decades, this probability increased to around two-in-three.

[12] CASMIN results show that by the mid-1970s the proportion of the industrial working class—understood as manual wageworkers—who were at least "second-generation" was as high as 75 percent in Great Britain, around 65 percent in the Federal Republic of Germany, and 50 percent in France, Sweden, and the United States. It was in these last three nations, however, that the proportion of second-generation workers had most rapidly increased in the recent past.

[13] From this point of view, a welcome recent development is the growing interest in providing macrosociology with better "micro" foundations through the revival of rational choice theory. Especially encouraging is the fact that such an approach must begin from an acceptance of methodoligical individualism, which is in itself a powerful prophylactic against lapses into historicism, and that this approach has been taken up both by theorists with evident liberal commitments (for example, Boudon 1977; Olson 1982; Hechter 1983) and by others who would still apparently wish to be regarded as marxisant at least (for example, Elster 1982, 1985; Przeworski 1985).

[14] In this respect (as in others) sociologists interested in problems of long-term social change could, I believe, find useful guidance in the work of the "new" economic historians, whom they appear so far to have largely neglected.