CLASS DIVISIONS AND STATUS RANKINGS
The Social Psychology of American Stratification
None of the studies we have reviewed can justify the conclusion that American workers lack class consciousness. On the contrary, past studies have consistently overlooked available evidence that Americans are indeed class conscious but that this consciousness is not translated into successful protest because of the opposition of a healthy and vigilant capitalist class. Past explanations of working-class protest failed not because they included social psychological explanations but because they did so without any independent evidence of the workers’ actual psychology. The lack of class consciousness was inferred from the behavior; class consciousness itself was never studied.
Since we cannot infer class consciousness from the outcome of class conflict, we must study workers’ ideas directly. Two images of class are found in the research literature on class perceptions (Ossowski, 1963). In the radical vision, class divides society into two conflicting camps that contend for control: workers and bosses, labor and capital, proletariat and bourgeoisie; in this dichotomous image, classes are bounded, identifiable collectivities, each one having a common interest in the struggle over control of society. In the conservative vision, class sorts out positions in society along a many-runged ladder of economic success and social prestige; in this continuous image, classes are merely relative rankings along the ladder: upper class, lower class, upper-middle class, “the Toyota set,” “the BMW set,” “Brahmins,” and the dregs “from the other side of the tracks.” People are busy climbing up (or slipping down) these social class ladders, but there is no collective conflict organized around the control of society.
The dichotomous image of class best accounts for conflicts.1 Conflict, as our parents all taught us, requires at least two parties. Just as there can be no conflict in social isolation, there can be no class in isolation; in our view, a class exists only insofar as it enters into antagonistic relations with another class. Masters and slaves, lords and serfs, bourgeoisie and proletariat (Marx and Engels , 1976: 482–85) are defined by the social relations of domination and subordination. The very concept of slaves as a social category cannot exist unless there are masters; similarly, feudal lords are defined by the existence of serfs; and, at least in Marx’s original meaning, the bourgeoisie and the proletariat exist only as an interrelated system. E. P. Thompson (1963:10–11) reminds us in a deservedly famous statement from his classic The Making of the English Working Class:
There is today an ever-present temptation to suppose that class is a thing. This was not Marx’s meaning, in his own historical writing, yet the error vitiates much latter-day “Marxist” writing. “It,” the working class, is assumed to have a real existence, whith can be defined almost mathematically—so many men who stand in a certain relation to the means of production. . . . If we remember that class is a relationship and not a thing, we cannot think in this way.
In recent years, new classes have been added to neo-Marxist analyses in order to explain better the conflicts in twentieth-century capitalism. For example, the division between supervisors and subordinates has been added to Marx’s original distinction between owners of capital and wage laborers. We discuss this and other neo-Marxist analyses in some detail in Chapter 4, but the important point now is that the new analyses retain the concept of class as a social relation between dominant and subordinate groups. Supervisors control subordinates: it is this power relation that puts them in different classes. Supervisors also tend to have higher incomes and more social prestige than subordinates, but it is not these status distinctions that separate them into different classes. Class requires a power relation.
Continuous rankings, in contrast, minimize the awareness of class conflict.2 The ladder image is essentially classless; while economic and social differences persist and are widely recognized, no sharp division separates workers from their bosses. Inequality is described by degree of access to socially valued resources, typically income or prestige. People can be ranked in such a hierarchy—sociologists even assign quantitative scores indicating precise social standing3—but there is no class relation between positions on the scale.
We will, following Dahrendorf, call all such distinctions based on continuous rankings status distinctions to keep them separate from class divisions based on the social relations of production.
Class is always a category for the purposes of the analysis of the dynamics of social conflict. . . . [But] wherever classes are defined by factors which permit the construction of a hierarchical continuum, they are wrongly defined; i.e., the term has been applied wrongly. Status, ranking by others, self-ranking, style of life, similar social conditions, and income level are all factors which define social strata but not social classes. (1959:76)
The class versus status distinction is basic to an understanding of the social psychology literature on the American working class: ours is a study of class perceptions.
In the United States, the ladder of success is the more widely accepted image. It is found in both popular usage and academic inquiry (Ossowski, 1963). Schools and mass media use the ladder image as if it were the only appropriate picture of American society. “Making it” has been a cultural preoccupation, and “making it” has always been evaluated by a detailed scale of social success. Horatio Alger provided a common fantasy of sudden leaps up the economic scale, but most versions of the message were concerned with the finer distinctions of a one- or two-car garage or a “title on the door.”
Academic research in the United States (but not, for the most part, elsewhere) adopted this same status-ranking model. Robert Nisbet, for instance, is typical in arguing that “the very forces which dissolved the class lines of pre-industrial society acted, in the long run, to prevent any new classes from becoming fixed. . . . The differences between the extremes of wealth and poverty is very great, today as always, but the scale is more continuous” (Nisbet, 1959:14; emphasis added).
The U.S. government allocated substantial resources to develop the status-ranking model. The 1947 NORC (National Opinion Research Center) prestige study (Reiss et al., 1961) made the first ambitious attempt to calculate exact occupational prestige scores based on the evaluations of a national sample of Americans.4 Later studies followed (Hodge, Siegel, and Rossi, 1964; Siegel, 1971; and, in England, Goldthorpe and Hope, 1969), even more ambitious in scope and coverage. These occupational ratings and the generalization of their results to other occupations by Duncan (1961) and Featherman et al. (1975) have provided us with an occupational scale of appropriately detailed gradation (see Figure 3.1). In theory, no gaps exist in these prestige scales. Since the scores were based on the subjective judgments of the American people, the continuous status ladder appeared to be the universal American image.
Prestige rankings proved to enjoy an impressive consensus. Blacks and whites, college graduates and high school dropouts, men and women: all rank occupations in approximately the same order. Between 1947 and 1961, occupational prestige rankings changed very little, despite years of rapid economic growth (Hodge, Siegel, and Rossi, 1964).
In fact, prestige rankings appear to be similar in all industrial societies (Inkeles and Rossi, 1956; Hodge, Treiman, and Rossi, 1966; Treiman, 1977).5 Both the supposedly class-conscious Europeans and the more individualistic Americans rank occupational prestige in much the same way. But nobody noted the implication of these cross-national similarities for the question of American exceptionalism. In none of these countries do any gaps divide one class of occupations from another. Thus there seems to be no special affinity for continuous images among Americans that would explain their supposed lack of class consciousness.
The “robustness of the occupational prestige consensus” (Rossi, 1976) established the status-ranking model as the central fixture of American sociology. The occupational scales became the basis of the status attainment research industry that flourished in the 1960s and 1970s (e.g., Blau and Duncan, 1967; Duncan, Featherman, and Duncan, 1972; Sewell, Hauser, and Featherman, 1976; Hauser and Featherman, 1977; Featherman and Hauser, 1976; Jencks et al., 1972 and 1979). Here again the American cultural fixation on “making it” was paramount, translated now into the less romantic terms of raising oneself a couple of points on the Duncan socioeconomic index (SEI).
In the 1970s, dissent arose about the accuracy of the status hierarchy as a description of American society (e.g., Coser, 1975; Horan, 1978). Critics of the status quo attacked it as a distorted representation that ignored the real class divisions in the system (e.g., Wright and Perrone, 1977). But few challenged the popular belief in the ladder imagery among the American people. Conservatives who defend status rankings as an accurate model of the stratification system and radicals who attack it as an ideological myth both assume that the ladder imagery exhausts the content of Americans’ perceptions of themselves; class divisions, however real, are not perceived to be real: “Class consciousness among manual workers is a transitional phenomenon . . . a clearly defined working class no longer exists, if it ever did” (Wilensky, 1966:12).
The evidence to justify this dismissal of class perceptions is embarrassingly thin. Class divisions have usually been ignored in research rather than proved irrelevant. The mistake made in interpreting the prestige studies is assuming that the perception of prestige ladders necessarily inhibits the consciousness of class divisions—as if Americans can have only one image of their society. The two models have been erroneously presented as mutually exclusive.
For example, as can be seen in Figure 3.1, there is an overlap in prestige scores between lower-status white-collar jobs and skilled blue-collar crafts. Machinists and electricians are ranked above reporters, building managers, and shipping clerks; beauty operators are ranked above file clerks. This overlap has led the prestige studies to dismiss the idea that the blue-collar-white-collar difference might be a class barrier: “The cleavage between white-collar, blue-collar, and farm occupations—if it exists at all—is based not so much upon matters of societal evaluations as perhaps upon the character of dress and work in the three groups” (Hodge, Siegel, and Rossi, 1966:327). In other words, since these particular “societal evaluations” can be shown to reflect a continuous status image of society rather than a bounded class model, it has been assumed that America uses a continuous status image in all “societal evaluations.”6
The Left itself has assumed that status ladders and class divisions are mutually exclusive images of society. Early in this century the Hungarian Marxist George Lukács asserted explicitly that “status consciousness . . . masks class consciousness; in fact, it prevents it from emerging at all” ( 1971:58). In the 1960s and 1970s, Lukács had a great influence on Left interpretations of the American working class. For example, Howard Wachtel, in his 1974 essay “Class Consciousness and Stratification,” blames workers’ concerns about status ladders for their abandonment of class consciousness:
With the development of monopoly capitalism, labor has become stratified with workers acquiring sharp status differences, a fact which mitigates their identification with a class. The impact of this stratification process on consciousness is to divert workers’ consciousness from a class orientation and replace it with an identification with one’s strata in society. (1974:96,106)7
None of this work ever asked whether status ladders actually do mask class consciousness, as Lukács presumed. The possibility was ignored that status and class models may both coexist in the popular imagination. Each way of looking at the world—as ranging along a continuous status ladder or as divided into opposed classes—may be useful in its own context and does not necessarily interfere with the use of the other model in other contexts. Why insist on a single vision?
Ralf Dahrendorf has argued that each model is useful for different purposes:
In a sociological context, neither of these models can be conceived as exclusively valid or applicable. They constitute complementary, rather than alternative, aspects of the structure of total societies as well as of every element of this structure. We have to choose between them only for the explanation of specific problems; but in the conceptual arsenal of sociological analyses they exist side by side. Whatever criticism one may have of the advocates of one or the other of these models can therefore be directed only against claims for the exclusive validity of either. Strictly speaking, both models are “valid” or rather, useful and necessary for sociological analysis. (1959:183)
That is, class divisions are useful for understanding conflict in society: strikes, revolutions, absenteeism, and the like. Status ladders, or “social strata,” are useful for understanding such phenomena as lifestyles, evaluations of worth, and mobility aspirations.8
The occupational prestige studies clearly fall into Dahrendorf’s definition of studies of social strata, not social classes. They are popular rankings of general standing and thus naturally permit the “construction of a hierarchical continuum.” The prestige research supplied response categories that were themselves steps along an obvious continuum (“poor,” “below average,” “average,” “good,” and “excellent”), the sorts of graded criteria that Dahrendorf had in mind as “factors which define social strata but not social classes.” It is not surprising, therefore, that people reacted by utilizing continuous criteria in their judgments and ignoring class divisions. The rankings are hierarchies of social strata, not perceptions of class position.
Thus, in arguing that Americans do perceive class divisions, we find no fault with the long tradition of prestige research. Prestige differences are recognized also—as are age, racial, ethnic, and gender differences. There is no need to quibble with the results of the prestige studies. Improvements may be possible, but no amount of tinkering with that research should ever be expected to test class perceptions. The investigators asked for status rankings, and within acceptable limits that is what they got.
But to grant that class divisions were irrelevant to this particular ranking task is not to say that class is never perceived. Different tasks and different situations may elicit different images of society. If we are to study class perceptions, we must ask for class perceptions. That research has a quite different tradition.
Subjective Class Identification
The alternative, class-oriented tradition of social psychological research began with Richard Centers’s (1949) national surveys that asked Americans, “If you were asked to use one of these four names for your social class, which one would you say you belonged in: the middle class, lower class, working class, or upper class?” Centers found that a slight majority (51 percent) chose the working-class label. These working-class identifiers more often voted Democratic and endorsed more radical attitudes than did middle-class identifiers (1949:118–30). Centers somewhat grandly declared that his results would “convincingly dispel any doubt that Americans were class conscious” (1949:76).
Centers was reacting to studies published by Fortune magazine (1940), which announced that almost 80 percent of the American population considered itself “middle class.”9 For Fortune, this was reassuring news: American capitalism had not created the class divisions that Marxists had expected. Instead, most Americans saw themselves as comfortably placed in the predominant middle. The middle class represented the “American way of life,” was its main beneficiary and its guarantor of future stability.
Centers recognized the ideological bias in the research and revised the question to produce quite opposite results. Fortune had provided respondents with only three possible choices—lower, middle, and upper class. Faced with what they saw as the stigma of the lower class and the pretentiousness of the upper class, most Americans readily chose the safe and respectable middle-class option. Centers had only to add one more choice, “working class,” and the percentage of middle-class identifiers dropped to 43 percent; in fact, the middle class was outnumbered by the 51 percent working-class identifiers (1949:77).10
Thus began a tradition of manipulating the wording of the question to produce whatever results the researcher desired. The later addition of an “upper-middle-class” option (Tucker, 1966 and 1968; Hodge and Treiman, 1968; Jackman and Jackman, 1983) restored a middle-class majority (62–66 percent). Open-ended questions that denied respondents any clues about the categories of interest likewise produced middle-class majorities. Debate raged about which was the “valid” measure of class identification, and as American sociology assumed a more conservative mold, Centers’s research seemed largely discredited in mainstream sociology.11
Centers also lost his natural defenders on the Left. Once establishment social science adopted survey research as its own favored technique, radical critics responded by dismissing not just its conservative uses but survey research in general.12 Centers’s work was tainted by this association. It did not help that his principal monograph had an unorthodox—indeed, naive—conceptual discussion of class. And since the Left, too, had its doubts about the class consciousness of American workers, Centers’s claims undoubtedly appeared as extravagant to many on the Left as they did to established social science.
The fact that Americans would choose a class label in response to a forced-choice question could not, by itself, demonstrate that they were class conscious. Critics were quick to point out that class consciousness encompasses more than the mere choice of a class label—especially if that choice is limited by a fixed number of possible responses.13
Even Centers’s name for the question, “class identification,” stretches the meaning of the responses. Identification suggests something more central to a person’s identity than is implied by a casual choice among four possible alternatives. Identification usually connotes an affective attachment to a class (see Landecker, 1963); there is nothing in the responses to Centers’s question that suggests any affect. We will, therefore, refer to the Centers question as the class-placement question, not the class-identification question.
To this day, there are many sociologists who will reject out of hand any research using the class-placement question. Part of this hostility may be the result of Centers’s own tendency to claim too much for his results, making his research especially vulnerable to the many critics eager to dismiss any mention of class divisions in the United States. Those critics, for a time, had their way, as not only class consciousness but class structure faded from attention in the research literature.
What is to be gained from rescuing Centers’s work? We are convinced that there is much of value in the traditional class-placement approach—indeed, most of the interesting questions have still not been asked more than a third of a century after Centers’s original analysis. We recognize the loneliness of our position, however. Both conservative and radical schools now have a well-developed resistance to research using the Centers question; both will be skeptical and tend to evaluate new work in line with the failings of past interpretations.
There are some hopeful signs, nevertheless, that the time has arrived for a fresh consideration. Class divisions are now taken seriously in much of American sociology. In the 1970s new research borrowed the familiar tools of academic social science—including survey research—to demonstrate the continued importance of class divisions. These studies have understandably concentrated on the “hard” data of income, both personal and national, thus protecting themselves from most charges of biased measurement. There is no such safety for studies of “soft” phenomena like class perceptions. Measurement methods are crucial here, and inasmuch as methodological problems provoked the major accusations against Centers’s research, we are working in more difficult territory. Still, now that class divisions have been demonstrated to have significant effects on Americans’ income, it seems reasonable to ask whether they are also important to Americans’ perceptions of their society.
If the responses to Centers’s question did not indicate class consciousness, what did they signify? What did his respondents mean when they chose working-class or middle-class labels? One possibility was to reinterpret the question within the predominant status-ranking model of American stratification. The reinterpretation was promoted most skillfully by Hodge and Treiman’s (1968) statistical analysis of a similar class-placement question. They found several joint determinants of class self-placements; income, education, and occupational prestige were the most important. Even so, only 20 percent of the variance in class placements could be explained by the stratification variables. The weakness of the relationships and the joint importance of quite different dimensions of stratification argued that there could not be a very clear class consciousness in the United States.
The final support for Hodge and Treiman’s stratification model was their discovery of the importance of social contacts for people’s self-placements: the status of friends, relatives, and neighbors helped determine class placements even among Americans with similar occupations, income, and education. Such social networks confused the picture even further for Americans trying to choose a class label. Besides, the importance of friends to the choice hardly suggested a conflict model of class divisions but emphasized instead the style-of-life concerns that were central to the status-ranking tradition.
Subsequent reanalysis by Jackman and Jackman (1973) suggested that social positions weren’t nearly so obscure as Hodge and Treiman had intimated. The several dimensions of social position that predict class placements are themselves closely related. Thus, education helps determine occupational status, and both of these determine income, and all of these determine social contacts.14 These linkages persuaded Jackman and Jackman to reject the “pluralist” model of Hodge and Treiman in favor of a model that emphasizes the importance of social position in “constraining” people’s social lives.
But the Jackmans themselves never adopted a true class model. According to them, Americans derive their social positions from “general socioeconomic prestige and income” (1973:580); that is, from status rankings, not from class divisions. The Jackmans attempted to evaluate a true Marxian model by testing whether capital ownership or self-employment made people feel they were middle class. They found that neither of these Marxist divisions had any effect on how Americans placed themselves in the class structure. The significant status effects and nonsignificant ownership effects argue that self-placements are status judgments rather than class perceptions.
The Jackmans’ research strategy was a major advance over earlier analysis because it showed how to test what stratification models people use in making their self-placements. If class divisions predict self-placements, then we can accept the existence of class perceptions. If only status rankings predict self-placements, then there is no evidence for class perceptions, and Americans have indeed fully adopted status-oriented criteria of social position. The logic is sound, and we accept their research strategy—although by extending it, we end up questioning their anticlass conclusions.
The problem the Jackmans faced was the inadequate means available for testing the importance of Marxian class divisions. They relied only on measures of capital ownership or self-employment—as was conventional at the time. Unfortunately, a 1975 expansion of the Jackmans’ original analysis into a major survey came too soon to incorporate much of the resurgent interest in class. It is evident from their 1983 description of these results that the Jackmans still don’t take the idea of class divisions in America seriously. Their rewording of the class-placement alternatives into “poor,” “working class,” “middle class,” “upper-middle class,” and “upper class” reveals their status ladder orientation. Their analysis makes a halfhearted attempt to investigate the differences between blue-collar and white-collar workers, but the cryptic description of the results (no numbers are reported) and their dismissal of a statistically significant collar-color effect betray their lack of interest in a class perspective.
The Marxist analysis of contemporary class structure has now progressed beyond such simplistic notions of class division. Class structure cannot be reduced to the simple legal ownership or nonownership of productive property; nor is collar color an adequate measure of the social relations of production. The role of authority in an enterprise (Dahrendorf, 1959; Poulantzas, 1974; Wright, 1980, 1985) and the design of the work process (Poulantzas, 1974; Braverman, 1974) are better recognized now as central in the capital accumulation process and thus in the exploitation of the working class. This more complex but more accurate analysis of contemporary class structure requires that we raise again the question of whether self-placements reflect a true class division in American society.
1. This discussion (and much of this chapter) draws on the work of the Polish sociologist Stanislaw Ossowski, Class Structure in the Social Consciousness: “The dichotomic presentation may serve to underline the antagonistic relations existing in the society, relations where one side is “on top,” the other “at the bottom,” where one exploits the other, where one rules and the other obeys” (1963:30).
2. Again, from Ossowski: “It is possible to reduce the sharpness of the class stratification in a social-status system, not by trying to weaken the picture of social inequalities, but by stressing the continuity in the system . . . the conception of a continuum of social statuses does away with the classes themselves, without heeding the scale of inequalities” (1963:96).
3. Even the fashionable statistics in American social science during the 1960s and 1970s reinforced the use of the status-ranking model. The more sophisticated research methods at that time required interval-level or at least ordinal-level scales. Statistics based on the general linear model are neatly congruent with a conception of the stratification system as a continuous, linear hierarchy. Methodologists’ dislike of categorical concepts was well expressed in a remark of George Bohrnstedt (1974:130): “Most of the variables of interest in the social sciences are continuous and measurable in theory at least at the interval level.” So much for Marx.
4. The NORC study and its successors asked large national samples to rate dozens, and eventually hundreds, of occupations on their “general standing.” The respondents had to choose a rating from five possible levels: “excellent,” “good,” “average,” “below average,” and “poor.” National prestige scores were calculated from the average responses for each occupation being rated.
5. However, Parkin (1971), Penn (1975), and Yanowitch (1977) present evidence that socialist societies rank the industrial working class higher in prestige than do capitalist societies.
6. The existence of an overlap between blue-collar and white-collar occupations does not even imply that this class division is irrelevant for prestige judgments. The white-collar-blue-collar gap may be one among many considerations Americans use in judging the prestige of an occupation, and these other considerations (e.g., average income) may blur the class division without eliminating it. Otis Dudley Duncan (1961) warned against too quickly dismissing the class divisions on the basis of the prestige scores alone. He insisted on a direct quantitative comparison of the class dichotomy against the continuous status scales to measure their relative importance. When Duncan (1966) compared the effect of white-collar work, average education, and average income on the perceived prestige of the various occupations, he found that white-collar work conveys no special prestige in and of itself; it is only the greater average incomes and better schooling that cause higher prestige judgments (see also, Glenn, 1975).
7. Paul Sweezy (1967:38) has made a similar argument. See also Lopreato and Hazelrigg (1972:124) for the argument that status consciousness promotes individual action at the expense of collective movements.
8. See also 1959:159. Dahrendorf’s distinction between class and status avoids many of the conceptual fallacies of American sociology; e.g., Werner Landecker (1960) tried to discover class boundaries by observing where income, educational, and occupational prestige hierarchies coincide. Edward Laumann (1966) attempted to discover subjective class boundaries by testing for the statistical significance of the perceived social distance between pairs of occupations. Such studies are wrongly conceived because they seek to find class divisions not between conflict groups based on the organization of production but within status ladders that rank people and positions along a scale of desirability.
9. See also Hadley Cantril’s (1943:78) conclusion that “the overwhelming majority of the American people identify themselves with some category of the great middle class,” and a similar conclusion by George Gallup (Gallup and Rae, 1940). More recently, Ben Wattenberg (1974:51) has proclaimed a “massive majority middle class” in the United States, and Nathan Keyfitz (1976:30) has claimed that “in survey after survey most Americans place themselves in the middle class.” This is yet another example of Richard Hamilton’s (1975) point about how “restraining myths” live on despite ample discontinuing evidence.
10. Centers also rearranged the order of presentation (middle, lower, working, upper) to discourage a status-ranking orientation.
11. Gordon’s verdict is perhaps representative: “a dubious methodology [that] has obfuscated the delineation of cleavages in politico-economic attitudes in American life” (1958: 201). However, the work of Jackman and Jackman (1973, 1983) has kept the question alive in American sociology. Kluegel et al. (1977) have now shown that the Centers class placement question has substantial reliability and validity.
12. For the reservations of a practitioner of survey research, see Portes (1971a:243).
13. Several sociologists have attempted to dissect the concept of class consciousness into various levels. Such armchair analyses have not been noticeably effective, and there is no agreement on what levels constitute a full class consciousness: e.g., both Mann (1973) and Lopreato and Hazelrigg (1972:126) identify five levels, but Mann’s first level incorporates the first four of Lopreato and Hazelrigg’s, while they subsume under their highest level what Mann requires four levels to describe. We believe also that these so-called higher levels of class consciousness have origins different from those of the direct experience that determines class perceptions (see Chapter 5).
14. Subsequent research questioned whether social contacts had much importance at all (Jackman and Jackman, 1983:188–89). As the Jackmans put it, “It appears that frequent interaction with his superiors on the job does not make the chauffeur middle class.”