CHAPTER 4

WHO IS WORKING CLASS?

In 1848, Marx and Engels predicted that capitalism would create a growing polarization between capitalists and workers:

Our epoch, the epoch of the bourgeoisie, possesses . . . this distinctive feature: it has simplified the class antagonisms. Society as a whole is more and more splitting up into two great hostile camps, into two great classes directly facing each other: Bourgeoisie and Proletariat. (Marx and Engels [1848], 1976: 485)

In one sense, they were correct. The capitalists have become ever bigger: in 1950 the largest 200 corporations in the United States owned 48 percent of all manufacturing assets; in 1980 this had grown to 60 percent (U.S. Bureau of the Census, 1983:535; see also Baran and Sweezy, 1966). Meanwhile, the old middle class (the “petty bourgeoisie”) declined. At the time of the Declaration of Independence, some two-thirds of Americans were economically independent: self-employed artisans, shopkeepers, and farmers. In 1985, roughly 8 percent of Americans were self-employed.

But Marx’s prediction was wrong about the proletariat. True, the formerly self-employed became employees, but as the old middle class declined, a new middle class emerged as a buffer to absorb class conflict and stabilize capitalist society. As firms grew, an army of managers, professionals, and white-collar employees took over some of the managerial functions previously reserved for capitalists alone. These salaried officials work for owners of productive property, just as blue-collar workers do, but earn generous incomes and enjoy substantial prestige. And—what is crucial for a class analysis—the new middle class also shares in some of the power that capital has exercised over workers.

The Problem of the Middle Class

Conservative critics of Marx have interpreted the growth of the middle class as evidence disproving Marx’s prediction about class polarization.¹ For example, Ben Wattenberg (1974:51) claims that America is now united in a “massive majority middle class”: that is, no proletariat, no class conflict.² “Postindustrial” theorists have taken a somewhat more sophisticated view: classes still exist but have changed; Marx’s nineteenth-century division between capital and labor is no longer the crucial one. According to John Kenneth Galbraith’s (1967) New Industrial State, middle-class technocrats dominate society because their knowledge has supplanted industrialists’ capital as the crucial resource. In a similar vein, Daniel Bell (1973) identifies the explosion of middle-class professions and services as the harbinger of a new class order. John Naisbitt’s (1982) Megatrends helped popularize the idea that advanced technology creates a new class system—a system in which the struggle between capital and labor becomes irrelevant.

Marxist interpretations of the middle class

Twentieth-century Marxists responded more slowly to the growth of the middle class. Much of their thought was hobbled by a misplaced loyalty to Marx’s emphasis on the division between owners of productive property and hired labor. Many Marxists seemed to regard the new middle class as only a bourgeois idea designed to obscure the division between capital and labor.³ They reasoned that managers, engineers, and professionals—like factory workers—sell their labor to an employer; therefore, they are—like factory workers—part of the working class.⁴ Such an analysis homogenizes 90 percent of America into an exploited working class opposed to a handful of capitalists who own the principal means of production.

This analysis will not do as a description of contemporary America. It provides no insight into the changing nature of class conflict in the twentieth century and cannot explain the durable grip of capitalism on the state and economy. An adequate analysis of contemporary class divisions cannot focus exclusively on the legal ownership of productive property. That does not explain much of the current class conflict, and it certainly does not describe the class divisions perceived by Americans.

What is needed is a class analysis that recognizes a position for the new middle class without abandoning the importance of conflict between capital and labor.⁵ Such an analysis is now available and is essential for our understanding of how Americans perceive the working class. Earlier work on class perceptions was confused about the middle class and the role it plays in modern class conflicts. The middle class is not merely an arbitrary range along some status scale; it is a genuine class with interests in opposition to the working class. We believe that Americans recognize this opposition and use a true class model in identifying the division between the working and middle classes.

Our objective in this chapter is to demonstrate Americans’ perception of this class division. The task requires two steps: first, we must identify the conflicts that divide the middle class from the working class; second, we must demonstrate that Americans define the middle class in a way that takes these conflicts into account. Our real interest is the second step, but to accomplish it, we must first understand the role of the middle class in contemporary conflicts between labor and capital. Once we recognize the distinctive class position of the middle class, we can ask how Americans perceive that role and therefore how they identify the working class in today’s society.

The three dimensions of control

In the last decade, many conflict-oriented sociologists sought to rescue the concept of class without denying the reality of the growth of the new middle class. One of the first of these attempts, by the Greek Marxist, Nicos Poulantzas, remains the best. Poulantzas studied law at the University of Athens, went to Paris during the intellectual ferment of the 1960s, and first attracted notice as part of the circle around the Marxist philosopher Louis Althusser (see Jessop, 1985). Today, Poulantzas is better known for his work on politics, but his authoritative, Classes in Contemporary Capitalism (1974), provides the best starting point for an analysis of the middle class. Poulantzas places the middle class within the larger context of the struggles between capital and labor by identifying three divisions within the class structure that keep workers subordinate (1974:14).⁶ He begins with the familiar economic division separating owners of productive property from productive labor. He gives the economic division primacy because the overriding conflict in capitalist society remains the economic exploitation of labor by capital.⁷ But Poulantzas recognizes that capital’s ability to get work out of labor does not depend on its economic power alone. Capital also directly supervises at the work site to ensure maximum effort from labor. This authority defines a political division between workers and their bosses. Third, capital plans how work is to be organized and tries to shape the workplace and eventually even the whole society in support of its need to accumulate capital. Poulantzas interprets this planning power as an ideological division separating the manual labor of workers from the mental labor of professionals and managers. Together, these three aspects of power—ownership, authority, and mental labor—determine the class divisions in contemporary society. It is the joint influence of all three social relations that subordinates the working class to capital.

Middle-class Americans who are not themselves capitalists share in the exercise of these types of control. Supervisors direct workers; engineers design factories; social workers help regulate the poor. The power that people in these positions exercise separates them from working-class Americans who do not have such power. Neither working class nor capitalist, such positions can best be described as middle class.

The Poulantzas analysis of the middle class avoids the conventional emphasis on affluence and status; instead, the central focus is on power and exploitation. For Poulantzas, the middle class is a true class, not just an intermediate stratum straddling the center of a continuous status scale. Being middle class requires people to enter into specific social relations with capitalists on the one hand and workers on the other. In particular, the middle class dominates labor and is itself subordinate to capital. It is this simultaneous dominance and subordination that puts it in the “middle.” Its relations with capital and labor—economic, political, and ideological relations—have developed gradually over the last century out of the conflicts between the two polar classes. It is this history that has shaped the roles played by the middle class in the functioning of modern capitalism.

The middle class as a class

Four principles distinguish Poulantzas’s theory of the middle class from other class approaches: domination of the working class, subordination to capital, historical development, and a functional role within capitalism. Later works may incorporate some of these, but no other theory is faithful to the full set.⁸ They are worth specifying in some detail, as they will recur in our analyses of each of the three dimensions of class domination.

1. The middle class stands in opposition to labor. Its members are part of the control apparatus over workers, either the direct supervisory control exerted by line officers or the more indirect control that plans the institutions and environments that help keep workers in their place. Supervisors are the direct agents of capital in obtaining work from the workers. Engineers, teachers, lawyers, doctors, nurses, advertisers, welfare workers, and state officials act to control the lives and opportunities of workers. They stand in opposition to workers not merely because they engage in creative work while manual workers do not but because they have taken over the design and planning activities that workers once had (Poulantzas, 1974:246). The Ehrenreichs have emphasized the opposition most clearly:

Thus the relationship between the [middle class] and the working class is objectively antagonistic. The functions and interests of the two classes are not merely different; they are mutually contradictory. True, both groups are forced to sell their labor power to the capitalist class; both are necessary to the productive process under capitalism; and they share an antagonistic relation to the capitalist class. . . . But these commonalities should not distract us from the fact that the professional-managerial workers exist, as a mass grouping in monopoly capitalist society, only by virtue of the expropriation of the skills and culture once indigenous to the working class. (Ehrenreich and Ehrenreich, 1979:17)

2. Whatever the capability of the middle class to dominate workers, its power is constrained by the ultimate power of capital (Poulantzas, 1974: 270). The great mistake of the postindustrial theorists has been to discount the power of capital.⁹ Middle-class subordination is often hidden: supervisors, mental labor, and the petty bourgeoisie appear to have a great deal of freedom in pursuing their goals. But the middle class exercises power only so long as it helps support the accumulation of capital. Only on the rare occasions when the middle class confronts the interests of capital directly, do we observe the ultimate power of capital. Managers who do not make profits get sacked. Professional organizations working for “the public interest” atrophy at the expense of those working within the corporate establishment.

Recognizing this power, the middle class usually aligns itself with capital.¹⁰ The general foreman that Studs Terkel (1974:184) interviewed understood where his loyalties lay:

Prior to going on supervision, you think hourly. But when you become management, you have to look out for the company’s best interests. You always have to present a management attitude. I view a management attitude as, number one, a neat-appearing-type foreman. You don’t want to come in as sloppy, dirty. You want to come in looking like a foreman.

The capitalist orientation carries over outside the factory as well. The Lynds’ observations about the Muncie middle class (what they called “the business class”) are typical:

On the other hand, the business class, in the main, either embraces or huddles toward the [Ball family, the town’s leading capitalists] because they know that the system through which they earn their salaries, receive dividends, buy new Buicks, and send their children to college depends upon the enterprise of men like these. The [Balls] symbolize security to the Middletown business class. (Lynd and Lynd, 1937:94)

3. The analysis of the middle class is necessarily historical; the middle class has developed within and because of the class struggle between capital and labor. No static analysis can hope to understand the position of the middle class. We must know the origins of middle-class positions and trace their growth and decline. Only this historical picture can detect the larger social forces that govern middle-class strength. Thus Braverman’s (1974) case for the distinct class position of engineers rests on a historical analysis of corporate efforts to get more work out of their employees. Similarly, the analysis of social workers as a control mechanism (Piven and Cloward, 1971) rests on the historical correlation of the creation and expansion of welfare with riots and other threats to capitalist stability.

4. Positions are defined on the middle-class side of the divide not so much through a microanalysis of the content of their work (there is no need for a new wave of time-and-motion studies to determine how much work is spent in creative endeavor and how much in mere execution, or in supervision and in taking orders) as in the context of the entire class system and how it supports the demands of capitalist accumulation. This is especially important in considering mental labor. Many jobs that involve autonomous work—like those of skilled workers, for instance—are not middle class because their mental labor does not control other workers. Engineers and quality-control specialists control workers; machinists do not. Both require creative thought, but engineers are middle class; machinists, working class. Or, to take a different sort of example, the positions of advertising specialists depend on the peculiarly capitalist—in fact, monopoly capitalist—need to create and expand demand, giving them an orientation to the maintenance of monopoly capitalism which is fundamentally different from that of workers.

The perception of the middle-class–working-class division

Do Americans see the working class as divided from the middle class, or is the working class just a rough range along a status scale? Conflict occurs between identifiable groups (Dahrendorf, 1959:179–93): Blacks and whites, Republicans and Democrats, Americans and Russians. If Americans recognize a true class division, then we can expect class conflict to occur along this line.

We believe that most Americans do in fact recognize a division between the working and middle classes, if only implicitly. The division is defined by the three dimensions of control: economic, political, and ideological. In this sense, popular practice has anticipated social theory in the understanding of class relations. Moreover, we believe that Americans put more weight on these three dimensions of class division than they do on the conventional scales of occupational prestige. What matters is not how much prestige one’s job has but how much power it confers.¹¹ To be middle class in America is to own productive property, or to have supervisory authority, or to perform mental labor at the expense of manual workers.

In what follows we consider each dimension separately, first analyzing how it subordinates workers and then demonstrating that Americans use it in assigning themselves working-class or middle-class labels.

Authority

The most obvious form of control over the working class is direct supervisory authority. The most tried and trusted of capital’s methods for subordinating labor remains intensified supervision. Management’s first instinct when labor trouble arises is to increase the scope and intensity of supervision.

For example, even after General Motors had precisely designed its new Lordstown, Ohio, Vega plant to minimize the possibility of worker disruption, the company had to resort to intensified direct supervision to reach its original goals (Rothschild, 1973). In October 1971, shortly before the great wildcat strike of March 1972, GM installed a completely new management at Lordstown. The new managers came from a separate branch of the company, the General Motors Assembly Division (GMAD). They quickly introduced stiffer, more militaristic discipline.

GMAD is charged, at Lordstown and throughout the corporation, with a more general rearrangement of factory discipline. At Lordstown this rearrangement took the form of layoffs, increased severity by foremen, the assigning of extra tasks and extra penalties for failure to perform these tasks—the sort of changes that have earned the division a national reputation for ruthlessness. (Rothschild, 1973:113)

These details are classic examples of the exercise of supervisory authority. The methods differ little from those of nineteenth-century mill owners, except perhaps that GM used a larger cadre of supervisory personnel to carry them out. GM also had more sophisticated control mechanisms in its repertoire, but it chose intensified supervision to solve its problems.

The GMAD intensification of discipline is a characteristic extreme expression of modern Fordist attempts to increase auto productivity. Just as factory planning is cheaper than mechanical planning, so managerial discipline is cheaper than inspection or time study or other similar corporation techniques. Managers are trained to identify and eliminate waste moments. And beyond such training, the managers learn (for free) a lasting attitude of tough-mindedness, to be shared by executives and plant managers and middle managers and general supervisors and foremen on the line. (Rothschild, 1973:115)

The exercise of this sort of authority naturally produces class conflict of the most ordinary and divisive kind. Both sides accept conflict as the natural state of industrial life. General Motors almost boasts of the strikes that GMAD provokes; as its vice-president for industrial relations explains: “Ten consolidations [by GMAD] have produced eight strikes. It should be apparent, to employ an understatement, that these consolidations are difficult to accomplish without conflict” (Rothschild, 1973:121). Old-fashioned supervisory discipline caused these strikes. Although Lordstown became notorious for the monotony and the impersonality built into its factory design, intensified authority was the immediate cause of the strike there.

Workers’ grievances at Lordstown concerned not only the speeding up and intensification of jobs, but also the disciplinary character of plant management—where workers must ask, and wait, to leave their jobs for one or two minutes; must ask, and wait for, permission to get married on a Saturday; must show a doctor’s note if they stay home when they get sick; or a note from the funeral director when they go to their father’s burial; or a garage bill if they arrive at work late because their car broke down. (Rothschild, 1973:115–16)

In the twentieth century, salaried supervisors issue the orders that workers must execute; it is these supervisors who come into direct conflict with workers. Most supervisors are not owners; they are employees like their subordinates. In the past, capitalist entrepreneurs both owned their enterprises and supervised the workers within these enterprises. As firms have grown into giant corporations, the two functions of ownership and control have become increasingly differentiated: shareholders own, supervisors control.

Dahrendorf

The German sociologist and Social Democrat Ralf Dahrendorf saw these organizational changes as requiring a change in class theory. Dahrendorf (1959) wanted to update Marx without dropping Marx’s emphasis on class conflict. He therefore substituted authority for ownership as the principal class division in contemporary society.¹²

In his new class theory, Dahrendorf separates from the working class all those managerial positions that exercise authority in the work place. Supervisors, middle and top management are in a class position antagonistic to working-class interests. This categorization excludes many white-collar workers, such as the army of clerical workers, who are often claimed for the middle class. Since clerks exercise no formal authority, they belong with manual workers in the contemporary working class.

Dahrendorf’s theory is a breakthrough, but he overstates his case in claiming that authority has totally supplanted capital as the basic class division. Dahrendorf overlooks the subordination of managers to capital, the relation that makes them a true middle class.¹³ Sporadic conflicts break out between capital and management; for example, capital threatens management with unfriendly takeovers, and management responds with its own array of defenses.

As in all class conflicts, the outcomes of these contests are not all predetermined, but the balance of power remains clear. Over the long run, managers are not free to run their enterprises against the interests of capital; if they try, they are eventually sacked. Even in the 1960s, when corporate chiefs had reasonably secure tenure, the most common cause for executives’ dismissal was low profits for capital (James and Soref, 1981). Maurice Zeitlin (1974:1093–94) relates a revealing incident about Anaconda Copper. When it suffered heavy losses after Salvador Allende’s nationalization of its Chilean mines, the company’s board of directors immediately replaced the chief executive officer with the vice-chairman of the Chase Manhattan Bank—a complete neophyte to the copper business but a representative of big New York capital—and the new boss proceeded to dismiss half of Anaconda’s former management. Incidents like this one aptly illustrate the relative power of capital and managerial expertise; capital’s power is less visible, but in a crisis it overwhelms managerial authority.¹⁴

Capital retains control over management by an array of positive incentives that usually obviate the need for dismissals. Bonuses, profit sharing, and stock options closely tie managers’ incomes to their firms’ profitability. These plans yield great rewards for managers, but they also ensure managers’ subordination to capital’s interest in profits.

Empirical research

Research on the actual characteristics of positions of authority blossomed in the late 1970s and early 1980s. We now know (Wright and Perrone, 1977; Robinson and Kelley, 1979; Kalleberg and Griffin, 1980) that supervisors earn higher incomes than other employees of the same educational training and occupational status. It is not their ability or background that earns them higher pay; it is the class position of boss. (Supervisors’ incomes are below those of similarly qualified owners, however—a reminder that supervisors are a middle, not a dominant, class.) Being a boss is also more personally fulfilling; supervisors’ jobs are more challenging and allow more personal growth (Kalleberg and Griffin, 1980). Bosses are also more likely to be Republicans (Robinson and Kelley, 1979).

Perceptions of supervisors’ class position

We are interested, however, in whether Americans recognize authority as one of the criteria of middle-class position. Is the division between bosses and subordinates perceived as a class division? Robinson and Kelley (1979) report that male (but not female) supervisors are more likely than subordinates to identify themselves as middle class in both the United States and Britain. Beyond that research, we have a lot of anecdotal evidence that authority has great subjective impact on workers. The interviews of Studs Terkel capture well the feelings of everyday Americans. One steelworker describes the division between workers and bosses this way:

This one foreman I’ve got, he’s a kid. He’s a college graduate. He thinks he’s better than everybody else. He was chewing me out and I was saying, “Yeah, yeah, yeah.” He said, “What do you mean, yeah, yeah, yeah. Yes, Sir.” I told him, “Who the hell are you, Hitler? What is this ‘Yes sir’ bullshit? I came here to work, I didn’t come here to crawl. There’s a fuckin’ difference.” One word led to another and I lost. (Terkel, 1974:xxxiii)

On the other side of the authority divide, bosses also recognize the importance of the gap that separates them from the working class. This general foreman is well aware of the difference, in spite of his lip service to American egalitarianism:

There’s a few on the line you can associate with. I haven’t as yet. When you get familiarity it causes—the more you get to know somebody, it’s hard to distinguish between boss and friend. This isn’t good for my profession. But I don’t think we ever change much. Like I like to say, “We put our pants on the same way.” We work together, we live together. But they always gotta realize you’re the boss. (Terkel, 1974:183)¹⁵

Workers return this ambivalent class distrust, as a spot-welder attests:

Oh yeah, the foreman’s got somebody knuckling down on him, putting the screws to him. But a foreman is still free to go to the bathroom, go get a cup of coffee. He doesn’t face the penalties. When I first went in there, I kind of envied foremen. Now, I wouldn’t have a foreman’s job. I wouldn’t give ’em the time of day. (Terkel, 1974:161)

The point for the welder is not that the foreman is free from hassles—a foreman still has a boss “knuckling down on him”—but that a class line still separates him from workers: the foreman has control of his time and, more important, control of the welder.

General Social Surveys results

A statistical analysis of the class placement question (Chapter 3) provides a more systematic test of how Americans perceive authority in the class structure. For this analysis we use mainly the General Social Surveys (GSS), a series of national surveys from 1972 to 1984 (see the chapter appendix).

In the GSS, 47 percent of full-time employed men and 34 percent of full-time employed women report some supervisory responsibility.¹⁶ These estimates are based on responses to the overly broad question, ‘Tn your job, do you supervise anyone who is directly responsible to you?” The general frame of reference (“anyone”) does not specify whether the people being supervised are direct subordinates, clients, students, or even people totally outside the employing organization, and that ambiguity probably explains the high reported levels of supervisory responsibility. This crude measure, however, is the best available in these data.¹⁷

We are interested in whether supervisory authority confers middle-class position, in the view of our American sample. In a simple comparison it is easy to demonstrate that supervisors more often see themselves as middle class than do nonsupervisors: of the male supervisors, 63 percent accept middle-class labels; only 40 percent of men without supervisory authority so classify themselves—a difference of 23 percentage points. The difference is only 8 percentage points among women (50 percent versus 42 percent), but again, more supervisors than nonsupervisors claim middle-class position.

This difference is not surprising; neither conservative nor Marxist interpreters of American society would expect the results to be otherwise. The important question is why supervisors see themselves as middle class. Supervisors differ from other workers in many ways: they earn more money, have usually had more education, enjoy higher prestige, and so on. But these factors—income, education, and occupational prestige—are primarily status advantages they enjoy, not a division between dominant and subordinate classes. What we want to test is whether supervisors more often see themselves as middle class because of their greater authority (that is, their class position) or because of the higher esteem and income their work is given (that is, their status ranking). If we can show that their authority determines (at least in part) their middle-class self-placement, then we can conclude that Americans perceive more than status ranks; they recognize class divisions as well.

We can answer this question because we can make separate estimates of how class position and status rank affect class perceptions. Since there are ranges of status among supervisors and among nonsupervisors, we can compare the class placements of supervisors and nonsupervisors with those of similar status levels; similarly, we can measure the association of higher status with class perceptions within each class.

Like most statistics, the analysis is best understood in graphic form. We will consider three possible outcomes of an analysis of the simultaneous class and prestige effects on class perceptions (see Figure 4.1). Their interpretations, however, are decisively different.

In Figure 4.1a there is no class effect on the self-placements; “class” perceptions are entirely a function of occupational prestige. Bosses would still see themselves as middle class more often than workers, but only because they have higher prestige, not because there is a sharp class division separating them from subordinates.

In Figure 4.1b, the reverse is true. The measure of continuous occupational prestige explains nothing beyond the supervisor-worker class dichotomy. In this case we would be justified in interpreting the class placements as reflecting the perception of a true class division, not merely of a crude status ranking.

Figure 4.1c illustrates both class and occupational prestige effects. High-prestige supervisors (corporate executives) more often consider themselves middle class than do lower-prestige supervisors (building superintendents); high-prestige workers (architects) more often than low-prestige workers (receptionists). But at equal levels of occupational prestige, those wielding authority (building superintendents) are substantially more likely to see themselves as middle class than those without authority (receptionists). In this case, we would infer that the class placements reflect the perception of both class divisions and status ranks.

Besides enjoying higher occupational prestige, supervisors also earn more money and are better educated than other workers. Earlier research and our own data tell us that income and schooling both increase middle-class placements (Hodge and Treiman, 1968; Jackman and Jackman, 1973 and 1983). The supervisors’ more frequent middle-class placements may be due to their greater income and education rather than to the power they exert over subordinates. A fair test of class perceptions contrasts the class self-placements of supervisors and workers who are otherwise alike: that is, who earn the same salaries, have had the same schooling, and hold jobs of similar prestige. Thus, we must compare supervisors and workers only after matching them on all status factors.¹⁸

In Figure 4.2 we compare middle-class placements of equivalent supervisors and workers.¹⁹ For men, supervisory power alone increases middle-class placements by an average of 10 percentage points.²⁰ This is not a great difference, but we must remember that we are comparing supervisors and workers who are exactly the same on all other class and status characteristics. As we will see below, other class divisions have a greater impact on class perceptions. Nevertheless, as Figure 4.2 makes clear, the authority division determines class perceptions as much as occupational prestige does. For workers at average occupational prestige (42—about the prestige of mail carriers or tool-and-die makers), gaining an authority position would increase the frequency of middle-class placement from 47 percent to 57 percent, somewhat more than moving to the highest prestige occupation possible (a prestige score of 82 would, without any supervisory authority, have a 54 percent predicted rate of middle-class placement). In fact, the effect of occupational prestige on class perception is less than we might expect by chance alone. What matters is whether one gives orders to other workers or not.²¹

For women (these results are not shown in the figure), authority is not important. This pattern—that men put more weight on their class position than do women—will hold true throughout the remaining analyses of class self-placements.²² We address the reasons for this gender difference in more detail in Chapter 8; suffice it to say here that the difference results from differences in the nature of “women’s work” and from the confounding effects of the husbands’ class positions for the wives in the sample.

Mental Labor: The Managerial Class

The specialization of mental labor within the firm

The new middle class includes more than supervisory managers.²³ Many jobs that have no direct supervisory authority nevertheless permit substantial control over workers in particular and over society in general. The engineers who design factories determine the work lives of factory employees as much as do the plant managers who supervise those factories. The marketing manager has more say over what products the worker manufactures than does the production supervisor. The personnel officer and the labor relations specialist may not directly supervise any workers, but they have great influence over how workers are organized and treated. The oversight of the accountant and the quality-control inspector threatens workers as much as the authority of their supervisors. To include only direct supervisory management in the middle class and to exclude the engineer, personnel officer, marketing chief, and accountant is to create an artificial division between “line” and “staff” management. Both have control over workers and stand in much the same class position vis-a-vis both the working class and capitalists.

A class division between line and staff is especially absurd from a historical perspective. The specialized staff tasks of work planning, quality control, timekeeping, personnel, and the like, originally belonged to supervisory management. In early factories, line managers exercised all these powers. If we did not classify staff specialists in the middle class with supervisors, we would be compelled to conclude that these control activities passed out of the hands of management into those of the working class, a new stratum of white-collar workers. In fact, the opposite was the case, since the functional departments provided more control over labor than could be gained by simple line hierarchies.

The rationale for using staff specialists is often couched in terms that have little to do with class relations. Engineers introduce new technologies in order to increase technical efficiency, not to subordinate workers. But the new technology almost always has the consequence of reducing the latitude of the workers’ independent action and extending the control of capital over workers’ time. Workers realize this. A steel worker interviewed by Sandy Carter (1979:112) is quite explicit: “As far as I’m concerned I got no use for the intellectual—the so-called expert, who sits around all day dreaming up new ways to control my life.”

Taylorism. Harry Braverman’s (1974) classic study, Labor and Monoply Capital, traces the growth of these “so-called experts” in eliminating workers’ control over production (see also Clawson, 1980). Some of his best evidence is the popularity of the time-and-motion studies and “scientific management” championed by Frederick Winslow Taylor in the early twentieth century. Taylor was quite explicit about his objective of controlling workers more thoroughly.²⁴ Taylor developed scientific management in order to force workers to work harder. He advocated the routinization of every task and the detailed calculation of piece rates, not because they were inherently more efficient but because they were weapons in the struggle against the great evil of “soldiering,” the deliberate limitation of output by workers.

But Taylor’s schemes required management to understand better the details of the production processes. Workers could often resist employers’ demands because they knew better how to manufacture the product (Gutman, 1973; Montgomery, 1979). Experienced workers could claim that it was impossible to produce at a faster rate, and their capitalist bosses had insufficient knowledge to overcome such objections. Taylor insisted, therefore, that to get control over workers, management had to eliminate workers’ mental skills. Bosses must separate design and execution: “All possible brain work should be removed from the shop and centered in the planning or lay-out department” (1903:98–99). Formalized, written instructions, dictated by the employers, weakened the informal control of workers over work and, for a while, overcame some (but certainly not all) working-class resistance to management’s demands. Employers were thus able to squeeze more profit out of their workers than would otherwise have been possible.

But—and this is crucial for understanding the modern middle class—the functions of work design were not taken over directly by capitalists themselves; they were entrusted to a new legion of managerial employees who also worked for capitalist wages. The result was the expansion of the corps of managerial, engineering, and professional positions during the twentieth century. The separation of design and execution created two classes of wage labor: the bulk of the workers who have little (and decreasing) say in the design of work, and the middle-class managers and professionals who have assumed those responsibilities. Thus the “new” middle class owes its very position to the destruction of workers’ control, and this history makes the mental-manual division an antagonistic class division. It results from the employers’ need to extract more labor from workers. Workers and middle-class managers have opposed interests in the organization of production. Two case studies make this historical conflict clear.

The steel industry. Katherine Stone’s (1975) history of the steel industry also shows how the separation of mental and manual labor fit into a larger strategy of increasing control over the labor process. In the early days, skilled workers controlled the steel production process: they determined the pace of work, divided the labor among members of a crew, and sometimes even hired and paid their own unskilled labor. They lost these powers at the end of the nineteenth century. After management eliminated unions in the 1892 Homestead strike, the steel companies pursued a two-pronged effort to subordinate workers (Stone, 1975:55). First, supervisory authority was intensified; supervisors were forbidden to perform manual labor themselves; and training courses instilled a management orientation into first-line supervisors. In Poulantzas’s terms, corporations reinforced the political relations of production.

But second, and equally important, the corporations took the knowledge of production processes away from the skilled workers and relocated it in management; that is, they reinforced the ideological relations of production. Training courses for narrow specialized skills replaced union-controlled apprenticeships. Labor-saving machinery eliminated skilled jobs. New dispatching systems and flow charts kept managers informed of the progress of production. These procedures, which are standard elements of managerial practice today, were innovations in the steel industry then. Because of them, the skilled workers lost control of the flow of production. The new organization required not only closer line supervision but a new cadre of staff officials. Just as the supervisory authority of the skilled workers was transferred to first-line supervisors, their overall knowledge about production was transferred to the staff managers. To fill the latter positions, employers began hiring a new class of white-collar employees, who became the bottom rung of the management hierarchy (Stone, 1975: 60). What is important here is the similar roles played by supervisory authority and mental labor in the class struggle within the steel industry.

GM’s Lordstown plant. Mental labor played a similar supportive role at the Lordstown, Ohio, Chevrolet plant (Rothschild, 1973). General Motors designed the plant to incorporate the most advanced technology of the 1970s: computer controllers replaced human supervision and inspection; layout planning reduced workers’ free time; automation replaced skilled work (machining and welding) with unskilled work (handing materials to robots). All these changes were the product of mental labor devising means to extract more labor out of the workers’ time. As we emphasized in the section on authority, GM later applied more traditional supervisory pressures to increase production still further. The important point is the complementarity of supervision and mental labor in subordinating the workers. Both methods, the original plant design and the direct hierarchical pressure from bosses, were aimed at the same objective: securing maximum labor from the worker’s hours on the line. In fact, the GM management hoped the control obtained through planning would be superior to strictly supervisory methods. GM’s ideal system put workers “largely under the direction of machines,” not of human supervisors.

GM prefers the control exerted by the staff specialists because it is more impersonal and more pervasive. It does not depend on having a supervisor constantly looking over someone’s shoulder. Nor does it suffer from the human temptation to soften discipline in an attempt to secure a more friendly compliance. Its impersonality makes the control more difficult to resist. As Terkel’s steelworker puts it, “Who you gonna sock? You can’t sock General Motors, you can’t sock anybody in Washington, you can’t sock a system” (1974:xxxii). To include supervisory managers but not these staff specialists in the middle class is to ignore how the modern corporation seeks to control its labor.

What is important for a class analysis is that this specialization in mental labor involves a social relation between planners and executers: it distinguishes the mental labor of the engineer from the mental labor of the skilled craftsperson. Both the engineer and the skilled mechanic enjoy substantial autonomy deriving from the lack of routinization in their work. They differ, however, because the engineer plans work primarily for other workers; the mechanic plans primarily for his or her own work. In the engineer’s case, therefore, the mental labor defines a class relation because it entails control over others; in the skilled craftsperson’s case, the mental labor does not alter the worker’s class position. The engineer is therefore middle class; the skilled craftsperson, working class.

Mental labor outside the factory

The issue becomes more complex outside the factory gates, where many professionals are in positions of substantial control over workers’ lives but are not part of a productive enterprise. The class interpretation of these professionals is especially controversial because beyond controlling workers, they are engaged in providing many services; indeed, their control function is often unspoken if not vigorously denied. Thus social workers channel many benefits to those in need, but they are also responsible for ensuring that the unemployed and the poor do not sink so low that they disrupt the smooth working of the system. Indeed, the growth of welfare can be largely traced to this need to prevent disruptions (Piven and Cloward, 1971). Teachers instruct children (and adults) in countless skills, but schools are also the major screening device in modern society for allocating people to good and bad jobs, a function that gives them enormous power over workers and their children. Doctors make people healthy, or at least try to, but in doing so they exercise a control over people’s lives that often extends far beyond what any boss could hope to achieve.

If the middle class is defined by the control it exerts over other people, then, it necessarily incorporates the social worker, teacher, and doctor as well as the first-line supervisor and plant manager. What the social worker, teacher, and doctor share with the engineer, accountant, and personnel officer is a specialization in mental labor: they all plan, design, and analyze, but their plans, designs, and analyses are largely executed by others.

Recognizing a class distinction between mental and manual labor helps to interpret the failure of the New Left that we described in Chapter 2 (Ehrenreich and Ehrenreich, 1979). One of the obstacles to the extension of radical protest outside campus walls was the students’ reluctance to recognize a class division between themselves and the workers they were trying to reach. Students too readily accepted the naive assumption that because they were not capitalists, they too were working class, thus conveniently reinforcing their self-definition as the revolutionary vanguard—a vanguard that workers were bound to follow because of their common class position. This attitude only reproduced the mental-manual division in society: the student New Left, primarily from middle-class origins, would be responsible for designing the protest that workers would execute. The Ehrenreichs suggest in retrospect that the New Left’s recognition of a class division separating students from workers would have cautioned a more deliberate and respectful effort to bridge that division.

Classifying jobs as mental labor

Unlike authority and self-employment, no one survey question will suffice to draw the line dividing mental from manual labor. Ideally, we would have a class analysis of each major occupation in the economy, a renewed sociology of occupations but one focusing on the conflict and macrosocietal implications of the division of labor.²⁵ Lacking this, we are somewhat reassured to find that the major theorists of mental labor all sketch roughly the same division: mental labor is defined as all those occupations²⁶ in the census classifications of professionals and managers—the only exception being technicians, who, Braverman (1974) astutely argues, are really only a new kind of skilled worker.

This definition draws the line between mental and manual labor somewhat higher than the conventional white-collar-blue-collar line often used to denote a working-class-middle-class division (Weber, 1921; Lockwood, 1958; Parkin, 1971; Giddens, 1973; Gagliani, 1981). The problem with the white-collar-blue-collar line is that the routinized white-collar positions such as clerical and retail sales are as separated from the design and planning of office work as are blue-collar workers from the mental labor of factory work (Gorz, 1967; Braverman, 1974:293–374; Glenn and Feldberg, 1977).²⁷ It is for this reason that we categorized the lower white-collar workers in Figure 1.2 as working class. The issue is controversial since if these workers are included in the middle class, then the growth rate of the new middle class is so greatly exaggerated that it becomes the majority class in contemporary America. Even without them, the new middle class now constitutes over 20 percent of the labor force.

Empirical research

Several studies of American social structure have documented the actual separation of managers and professionals from the rest of the working class. Even before much of the theoretical and historical work had been done, Richard Hamilton (1972) found that the principal class division in American politics located the lower-paid white-collar workers within the working class. In national opinion polls, the decisive division in political attitudes separates both blue-collar and lower-white-collar workers from the “upper-middle” class (especially from the white Protestants who predominate in this class) that forms the core of conservative Republican politics in America (Hamilton, 1972:359–61).

Vanneman (1977) used cluster-analysis procedures to sort occupations into classes according to their patterns of residential isolation, father-son mobility, and friendship choices. For each analysis, the middle-class-working-class line was drawn at a place that separated professionals and managers from manual, clerical, and technical workers. For example, the residential distribution of technicians, bookkeepers, and other clerical workers more closely resembles the residential pattern of manual workers than that of professionals or managers. (However, retail sales workers cluster with other middle-class occupations, not with the working class, as in the Braverman and Ehrenreich analyses. A cluster analysis based on father-son mobility patterns shows that clerical workers have parental origins more similar to those of the working class than of the middle class.²⁸

A persuasive rationale has accumulated, therefore, that in the contemporary United States a true class division separates the mass of workers from the professionals and managers who control work and help stabilize capitalist society. The division is not just status superiority but is based on class interests derived from antagonistic functions in the maintenance of capitalist society. The question remains whether this manager-worker class division is perceived by Americans themselves. Are working-class and middle-class labels used, at least in part, to distinguish between the two classes who engage in mental and manual labor?

The perception of the mental-manual division

Workers’ comments. There is substantial anecdotal evidence that workers do recognize the antagonism between mental labor and the working class: for instance, the steelworker (quoted above) who “has no use for the so-called expert who sits around all day dreaming up new ways to control my life.” The middle class also recognizes the power inherent in mental labor. As one advertiser told Studs Terkel: “My career choice in advertising, which I’ve drifted into, is connected with the fantasy of power. I have a sense of slowly increasing power, but the limits are very frustrating” (1974:77).

Perhaps the best example of the popular perception of mental labor as an antagonistic class comes from E. Wight Bakke’s (1940) study of the Depression unemployed. An American mechanic interviewed by Bakke anticipated Poulantzas, Braverman and the Ehrenreichs by more than a third of a century. The mechanic is worth quoting in detail because he notes most of the themes we have considered. He begins with pro forma protestations about the uncertainty of class structure in the United States, then zeroes in on the mental-manual division.

You know it’s hard to tell just what class divisions there are in America, and I don’t know just how to say it. It seems to me it’s something like this, that when you’ve a job where there’s some call for planning, some call for figuring out things—I think that’s the word, figuring out things—in your head, you feel that you’re in a different class from the fellow who handles things.

For the mechanic, this mental-manual division is clearly a class division, not just a status differential. Indeed, he discriminates between the categorical class division separating manual and mental labor and the status differentials within each category:

Now, I think that’s the big class division in America, a division that comes right in the experience of men, something that’s real, something that they see every day. And it usually works out that the “figuring-things-out” group is the same as the bosses and employers and the ones that tell you what to do. Now, of course, within the figuring-out group, and in the handling-things group, there’s a lot of divisions too, but those aren’t real class divisions.

His quick list of positions in the dominant class includes precisely those that Braverman and the Ehrenreichs have identified:

[The fellow who figures out things] doesn’t feel out of place when he’s associating with men like professors, lawyers, and doctors—and oh say—bankers. He doesn’t feel that he’s out of place. He associates with them and he’s sure of himself.

The mechanic is equally clear about what does not constitute a class division. First, he rejects the simple owner-employee distinction of the older Marxists:

Now that fellow, I’ll tell you right now, that fellow feels that he is just about next to the top notch, even though he is paid a salary just like others are paid wages, but just the same, he feels that he is important.

Even his “next to the top notch” description is apt, implicitly reserving the apex of the class structure for the genuine capitalist. He also rejects a simple money image (cf. Goldthorpe et al., 1969) of the class structure:

Now, sometimes, you know, a man who’s a real skilled artisan will be getting more money than that fellow, but it isn’t always the money that makes the difference; it’s the fact that you’re figuring out things or you ain’t.

The mechanic recognizes the association of a college education and mental labor, although the importance of formal schooling to class distinctions is clearly subordinated to the actual role of mental labor.

If you’ve had a long experience of doing that sort of thing [figuring things out], you get a confidence and an assurance that just naturally makes you feel a bit superior. Some men get that by going to college. Sometimes it ain’t so; sometimes they don’t do much figuring out in college, but at least they think that they’ve done it; but it isn’t necessary to have gone to college if your training has been that of figuring out things. You feel pretty much the same way. (Bakke, 1940:89–90)

Bakke’s mechanic is an unusually astute analyst, more perceptive than the best social scientists of his time, and quite as articulate. We do not claim that all American workers can provide such a clear analysis. We do suggest that this mechanic’s image of the class structure is implicit in most Americans’ class perceptions. Not everybody can spontaneously describe that image in such detail, but most workers would recognize it as an accurate representation of the American society they know and work in.

The GSS results. In the GSS, 38 percent of the men and 30 percent of the women fall into the mental labor (that is, professional-managerial) class. The differences in class self-placement between these managers and other workers is striking—much greater than the differences between supervisors and nonsupervisors. For men, 68 percent of the mental laborers and only 23 percent of the workers identified themselves as middle class—a difference of 45 percent. Among women, the difference is 29 percent, since 62 percent of the mental laborers and only 33 percent of the workers identify with the middle class.

Of course, people in professional-managerial positions also earn higher incomes, have more prestigious occupational titles, and have usually acquired considerably more education. Is it their greater status rather than the power conferred by their mental labor that makes them middle class?²⁹ When we compare mental and manual labor for workers of equivalent education, income, occupational prestige, and other class positions, the results are striking—much more so than for direct supervisory authority. For both men and women, the mental-manual class division is the single most important determinant of class perception. The graphs in Figure 4.3 convincingly demonstrate the role of mental labor in determining class perceptions. The gap between mental and manual labor is substantial at every level of occupational prestige. For men, the average difference is 25 percentage points. The entire range of the occupational prestige scale has a smaller effect (11 percent) on class perceptions. The crucial job characteristic for class perceptions is class position, not status rank. For women, the difference between mental and manual labor is somewhat smaller—an average of 12 percentage points—but still noticeable across the whole range of occupational prestige.³⁰

The comparison of the mental-labor class effect with the occupational prestige effect demonstrates that the difference between managers and workers cannot be explained by the prestige differences between them. Without detailed analysis, we cannot tell whether it is the managers’ dominant class position or their higher social status that places them more often in the middle class, but the results in Figure 4.3 eliminate the social status explanation. Occupational prestige is too unimportant in class perceptions to begin to explain the gap between the placements of mental and manual labor. If it is not the higher prestige, income, or education characteristic of mental labor that makes these managers see themselves more often as middle class, it must be their greater power—that is, their class position.

Ownership of Production: Self-Employment

The last class division to consider is the economic division between the self-employed and wage labor. The self-employed are the old middle class: the shopkeepers, artisans, independent professionals, and small farmers, whose numbers have dwindled (see Figure 1.2) but who are nevertheless still with us. This division has historic priority; even Marx acknowledged the intermediate position of the “petty bourgeoisie.” Their class position must be distinguished from workers because their labor does not contribute to the profits of others. Capitalists make money from the labor of workers; the self-employed receive the full value of their own labor.

This economic difference generates different class interests: the petty bourgeoisie do not share the same interest in the transformation of capitalism as workers do. As a result, the self-employed generally act as a conservative force within capitalist societies (Lenin [1920], 1975 [vol. 3]:310); in fact, the old middle class has acquired a particularly nasty reputation for supporting reactionary regimes. In Chile it was the independent truck drivers who helped bring down the socialist Allende and install General Pinochet. And the self-employed in Germany, especially the rural Protestant self-employed, helped bring the Nazis to power (Hamilton, 1982).

The conservatism of the old middle class loomed large in many explanations of American exceptionalism. Their great numbers in early U.S. history provide a convenient explanation for the weakness of the American Left. Tom Bottomore’s analysis is typical:

In the USA, in contrast with the European countries, the ownership of property was quite widely diffused in the early part of the nineteenth century. . . . America was, predominantly, a society of small farmers, small traders, and small businessmen; the closest approach there has been to a “property-owning democracy.” (1966:48)³¹

With the proportion of self-employed now down to 8 percent of the labor force, the old middle class can hardly be an explanation for American exceptionalism any longer.³² Yet its decline has not brought an intensification of class conflict. This suggests that either the old middle class was not an important cause of American exceptionalism or that its role in muffling class conflicts has been assumed by someone else—in particular, by the new middle class of managers and professionals (see Bottomore, 1966: 50).

Today, the self-employed are important more for their theoretical significance than for their numbers. Unlike the self-employed, the working class must labor for some employer; all other forms of capitalist domination derive from this economic subordination (for Poulantzas, it is “determinant in the last instance”). The self-employed escape this subordination, and when they are themselves employers, they benefit from it. If workers do not recognize this division, they cannot understand the principal way in which they are kept subordinate.

The genius of capitalism, as compared to such earlier methods of organizing workers as slavery and feudalism, is that employers trap labor into “voluntary” subordination. Capitalism transfers to workers the responsibility for better performance. Because employers can so easily dismiss their workers and hire new ones, they can pick and choose the best producers. To survive economically, much less to prosper, workers must exert themselves; slaves and serfs were under no such compulsion. This economic subordination can be seen most clearly in times when it is most threatening: absenteeism, turnover, and strikes all decline during periods of high unemployment. Workers afraid for their jobs are more submissive.

Capital’s economic power underlies the other forms of power. Both the giving of orders and the reorganization of work depend ultimately on the fact that the job belongs to the company and not to the worker. Michael Burawoy, who generally discounts the importance of economic power, reports an incident from his own experience in a machine shop: “When I was resisting new inspection controls, a foreman came up to me, shook his fist angrily, and reminded me that a few days ago hundreds of auto workers had been laid off” (1979:130). The foreman was using the company’s economic power to reinforce his supervisory authority and to impose new planning controls on workers.³³

The economic division based on ownership is more than a means to an end; it is the end itself. Labor supports this compulsion by producing more value than it is paid in wages. Owners earn higher incomes than either supervisors or workers who have the same occupation and the same education and experience (Wright and Perrone, 1977; Robinson and Kelley, 1979; Kalleberg and Griffin, 1980). It is for the purpose of economic exploitation that the other forms of control are devised.

But Burawoy claims that workers do not understand their economic exploitation. For him (1979:28–29), the characteristic feature of capitalism is its ability to obscure economic exploitation in a way that slavery and feudalism never did. What must be studied, he says, is the way capitalists mystify economic relations and thus “manufacture consent” among workers so that they willingly participate in their own economic exploitation.

We believe that Burawoy assumes too much. There is, in fact, considerable evidence that workers are not “mystified.” They do recognize their exploitation; they merely lack the power to do anything about it. The evidence comes in two forms: directly in comments about the value of their labor to their employers, and indirectly in their recognition of a distinct class position for the petty bourgeoisie who escape this exploitation. Most of this section concentrates on the second kind of evidence, the perception of the class position of the self-employed—this provides the broadest-based evidence that workers recognize the importance of the ownership division. But the occasional comments of workers also suggest that they understand how the economic relations benefit capital at their expense.

Petty bourgeois aspirations

The dream of self-employment still captivates workers’ imaginations, however unrealistic the possibility may be.³⁴ The postwar automobile workers studied by Eli Chinoy (1955:86) entertained “widespread interest in small business.” A former farmworker confirmed this fascination with self-employment among the working class:

All farm workers I know, they’re always talking: “If I had my own place, I’d know how to run it. I’d be there all the time. My kids would help me.” This is one thing that all Chicano families talked about. We worked the land all our lives, so if we ever owned a piece of land, we knew that we could make it. (Terkel, 1980:168)

Such petty bourgeois illusions are said to explain American exceptionalism in that they divert workers’ consciousness from the possibilities of collective action. The problem with such explanations (see Chapter 11) is the scarcity of evidence that workers’ personal desires for getting ahead do in fact reduce their class consciousness. The opposite seems equally plausible: those workers who most fervently wish to escape the working class may be precisely those who most resent their capitalist bosses. One survey (Schlozman and Verba, 1979:160) showed aspirations for self-employment more closely related to personal unhappiness than to belief in the American Dream. Self-employment may be attractive precisely because it is an escape from working-class subordination. Workers may both aspire to self-employment and be class conscious. To one machine operator, for example, the appeal of small business was precisely the possibility of escape from class domination:

The main thing is to be independent and give your own orders and not have to take them from anybody else. That’s the reason the fellows in the shop all want to start their own business. Then the profits are all for yourself. When you’re in the shop there’s nothing in it for yourself. When you put in a screw or a head on a motor, there’s nothing for yourself in it. So you just do what you have to in order to get along. A fellow would rather do it for himself. If you expend the energy, it’s for your own benefit then. (Chinoy, 1955:86)

Not only does self-employment mean “giving your own orders” (Poulantzas’s political division); it also means “the profits are all for yourself” (Poulantzas’s economic division). For the autoworker the dream of self-employment is one expression of his class consciousness, not a denial of it. The same resentment of capitalists’ domination that prompts the petty bourgeois aspirations may also lead to militant class conflict. The direction that the resentment takes is probably a function of what seems most possible for the worker at the moment. To many American workers, self-employment—however remote a possibility—seems to offer a more realistic chance to escape from working-class subordination than does a socialist transformation. But our point is that petty bourgeois aspirations may be compatible with class conflict, given a good opportunity for class protest. Most of the American exceptionalism literature has not considered such a possibility.

Artisan origins of working-class movements

History teaches us that petty bourgeois origins have often contributed to successful working-class resistance to capital. The nineteenth-century Lynn, Massachusetts, shoemakers studied by Alan Dawley built one of the first radical unions in America on the basis of their previous artisan self-employment.

Factory workers in the shoe industry were able to organize because most of them had been [self-employed] shoemakers in prefactory days. This gave shoe-workers a common identity through a continuity of shared ideas and experiences. . . . Artisan protest inspired factory protest. Artisan organization engendered organization among factory workers. The Mutual Benefit Society of Journeymen Cordwainers and the Mechanics Association left a legacy that became the Knights of St. Crispin [the first shoemakers’ union]. The legacy contained not only the experience of organizing but also the stubborn conviction that a worker had as much right to organize as anybody else. (1976:176–77)

In Lynn there was continuity between the artisans’ resentment of their loss of independence and the growth of working-class consciousness.³⁵ The two sentiments were compatible and perhaps mutually reinforcing. Both the working-class consciousness and the artisan independence rejected the capitalist exploitation of their labor. Similarly, the autoworkers’ dreams of escape into small business should be interpreted not as an endorsement of capitalism but as a rejection of it.

But does the dream of going into business for oneself really represent an escape from the working class? Do Americans still equate being self-employed with being middle-class? Or is self-employment no longer important—a meaningless vestige of an outdated Marxism? The next section explores the issue by comparing the class self-placements of the self-employed and the salaried.

The GSS results

The General Social Surveys include the customary measure of ownership of the means of production: whether the person is self-employed or an employee. Hodge and Treiman (1968) and Jackman and Jackman (1973) used this measure (without success) in their studies of class placements. More recently, Wright and Perrone (1977) and Robinson and Kelley (1979) incorporated a self-employment variable in constructing class categories to study income inequalities.

In the GSS, only 12 percent of the men are self-employed and 8 percent of the women. These percentages are far too small to account for many of the middle-class placements in U.S. society. Nevertheless, the difference between the self-employed and wage laborers still addresses the important theoretical question about the role of ownership of productive property in the perception of class position. If the self-employed more often see themselves as middle class than do wage laborers (when we control for other influences), we can infer that the conventional Marxian emphasis on ownership of the means of production still plays some role in the popular perceptions of class position.

For men, self-employment does have an effect on class perceptions. In Table 4.1, men are 7 percentage points more likely to see themselves as middle class than are employees with the same income, education, and occupational prestige and in the same authority and mental-labor class positions. The difference is modest, but these are the first results that have found a middle-class ownership effect (cf. Hodge and Treiman, 1968). The difference for women is smaller and not greater than chance expectations.³⁶ Even for the men, the self-employment effect is smaller than the effects for authority and for mental labor. It might help if the surveys had more information about the nature of self-employment: for instance, how much property is owned and whether employees are hired. But with the available data we can conclude, somewhat tentatively, that American men do perceive self-employment as a class division that separates the middle class from the working class.

Conclusion

This chapter has reviewed evidence that mental labor, authority, and ownership each contribute separately to a middle-class placement.³⁷ American men use class labels, in part, to reflect actual position in the class structure: to exercise authority, to participate in the design of work, or to be in business for oneself is to be middle class; the working class enjoys none of these advantages.

American men perceive these divisions as class divisions; at work, it is their class position that determines middle-class placement, not just the higher status levels associated with managers, supervisors, and the self-employed. Even among men with exactly the same levels of occupational prestige, income, and education, mental labor is more middle class than manual labor; supervisors are more middle class than nonsupervisors; the self-employed are more middle class than wage laborers. Class divisions do determine class perceptions.

In fact, mental labor, authority, and self-employment have more noticeable impacts on class perception than does occupational prestige. For men, the estimate of prestige effects is less than what we might expect by chance (see chapter appendix, Table 4.B); thus, this analysis gives us no reason to believe that occupational status makes much difference at all to class perceptions. What counts is class position: managerial control, ownership of the means of production, and authority in the enterprise. For women, the results are less clear; mental labor determines their class perceptions but authority and self-employment do not. Even the mental-labor effect is weaker than for men. The explanation of this gender difference is complex, and we defer a more detailed analysis to Chapter 8.

Appendix

The statistical analyses of the class-placement question use data from two sources. The principal analyses are based on the 12 General Social Surveys (GSS) completed between 1972 and 1985. In the GSS the question was worded as follows:

If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class? (Davis and Smith, 1985:207)

We also analyzed data from the American National Election Studies between 1966 and 1978. For the longitudinal analysis (Chapter 6), we add the 1952 through 1964 election studies. In the Election Sample, the question was worded:

There’s been some talk these days about different social classes. Most people say they belong either to the middle class or to the working class. Do you ever think of yourself as belonging in one of these classes? [if yes] Which one? [if no] Well, if you had to make a choice, would you call yourself middle class or working class? (ICPSR, 1975:242)

Among the four choices offered in the GSS—lower, working, middle, and upper class—very few respondents chose the lower (1.6 percent) and upper (1.8 percent) extremes; thus, this question offers a basic dichotomy. We collapsed the two additional categories with the working and middle categories for our analyses. More drastic tinkering (e.g., adding an “upper-middle” category: Tucker, 1966; Hodge and Treiman, 1968; Jackman and Jackman, 1973 and 1983) changes the nature of the question and should be expected to yield quite different results (see Hamilton, 1966b).

Our rationale interprets the traditional “class identification” question as an exercise in cognitive judgment—respondents are asked to place a particular individual within a class structure. We infer the cognitive rules used in such placements from the pattern of results. Thus, we can use the statistical relationships between people’s social position and their class self-placements as an index of the criteria used to assign people to classes. The important questions are which variables determine the self-placements (class or status) and how the social context (e.g., time, race, gender, or national culture) affects the relative strengths of those associations. Self-placements offer the advantage that respondents have full knowledge of the person they are classifying; no relevant characteristics are unknown to the respondents.

Our use of the self-placement question assumes that self-perceptions of class position follow the same rules as perceptions of the class positions of others. The equation of self-perceptions with other-perceptions is supported by a long tradition of psychological research (e.g., Bern, 1972). It is also an integral part of the symbolic interactionist approach and was one of the great insights of George Herbert Mead (1962).

Of course, since a respondent is classifying only a single person, we cannot develop a complete picture of each person’s image of the class structure. But by aggregating responses across a great many interviews, we can develop a complete picture of how class labels are applied in the entire society. Do foremen typically place themselves in the middle class? Do clerical workers place themselves in the working class?

Samples

Together these surveys include data on the class perceptions of more than 20,000 Americans. There is no great virtue in these large numbers: for estimating national averages a sample of 2,000 is almost as good as a sample of 20,000. But we can use so many cases to get reliable estimates for subgroups that happen to be of special theoretical interest: line foremen and payroll clerks, Polish- and Scottish-Americans, women working in predominantly male occupations and men in predominantly female occupations.

All these national samples combine data from several surveys, and some of the surveys include much larger samples than the others (e.g., the 1972 Election data), though the larger samples represent populations that are virtually the same size as in other years. To correct for the fluctuating sample sizes, respondents were assigned weights so that each survey was weighted equally in the analysis (to the harmonic mean of sample sizes, see Winer, 1971). In addition, many of the samples included internal weights reflecting over- or under-sampling of certain parts of the population. These were included in the weighting of each respondent by multiplying the weight for that survey year by the internal weight (which usually had to be adjusted downward to reflect the true sample sizes).

All analyses have been computed for the white nonfarm labor force. Only labor-force members are included because of the different interpretations given to occupation and even to income outside the labor force. Farmers are excluded because of their ambiguous position in a dichotomous, industrial class structure. Blacks require a separate analysis since there is substantial evidence (see Chapter 10; Jackman and Jackman, 1973 and 1983; Goyder and Pineo, 1974; Cannon, 1980; Cannon and Vanneman, 1986) that racial oppression shapes the perception of class. Separate analyses are computed for men and women because there are important differences (see Chapter 8) in the perceived class structure of men’s and women’s jobs.

Statistical analysis

The statistical technique used to test most of the empirical questions throughout this book is probit analysis (Hanushek and Jackson, 1977), a first cousin of the more familiar multiple regression using ordinary least squares. Probit analysis is more appropriate to the task of assessing dichotomous outcomes such as choices between “middle” and “working” class labels. Most of the estimated coefficients are presented in the tables appended to the appropriate chapters. The conclusions drawn from these statistical tests are presented in graphs; the coefficients are transformed into simple descriptive statistics such as percentages and included in the body of the text.

Variables

The three main class variables—authority, mental labor, and self-employment—are described in the text. The election surveys included no direct measure of authority, but we devised an alternative measure for a separate analysis described below.

Occupational prestige. Occupational prestige was recoded from the 1970 U.S. Census occupational codes as reported in the General Social Surveys codebook (Davis and Smith, 1985:448). In the Election Sample, occupational prestige was recoded from modified 1960 U.S. Census occupation and industry codes (ICPSR, 1968: 216–34). Both codes are derived from a direct measure of subjective occupational prestige (Siegel, 1971) and should therefore represent the best contrast to objectively defined class distinctions.

Education. Education is scored as years of school completed except that precise data are not available for the college-educated categories in the Election Sample. All Election Sample respondents with incomplete college were coded as having 14 years of school and respondents with graduate degrees were coded as having 18 years.

Income. The income measure is based on total family income from all sources. The surveys reported income according to different categorization schemes in different survey years. To achieve comparability, each income category in each year was recoded to the dollar value of the midpoint of that category. The top category was open-ended (e.g., $25,000 and over) so Pareto estimates were used (Shryock and Siegel, 1975:366). Because each set of surveys spans several yeras of high inflation, the income data are converted to real 1985 dollars (1978 dollars for the Election Sample) by adjusting for the relevant consumer price index. The logarithms of these values are used in all analyses rather than the actual dollar values, because we assume that income effects are likely to be proportional (i.e., the increase from $10,000 to $20,000 is equivalent to the doubling of $20,000 to $40,000, not to an increase from $20,000 to $30,000).

Authority in the Election Sample

There is no direct measure of authority in most of the election surveys. However we devised an alternative measure of authority from the Dictionary of Occupational Titles (DOT) codes (U.S. Department of Labor, 1965). One of the DOT codes rates the nature of a worker’s interaction with other people, especially subordinates, on the job. In the past the DOT “People Code” has been interpreted as an index of “job complexity” (Kohn and Schooler, 1969; Miller, 1971), but closer inspection shows that the “people” referred to are almost all subordinates, not coworkers or superiors. For this reason, we feel the code is best interpreted as a measure of authority.

The code clearly implies authority in its higher categories: mentoring, the counseling functions performed by doctors, lawyers, and the clergy; negotiating, the executive management functions of “formulating policies and programs and/or arriving jointly at decisions, conclusions, or solutions”; and instruction, the exercise of authority over students. Two other categories also imply authority: supervising, “determining or interpreting work procedures for a group of workers, assigning specific duties to them”; and speaking-signaling, “giving assignments and/or directions to helpers or assistants.”

Of the three remaining categories in the DOT People Code, serving (0.8 percent) clearly does not imply authority. Persuading (7.6 percent), however, occupies a special position. It is defined as “influencing others in favor of a product, service, or point of view.” The attempt to influence, as well as the dominant interpersonal style typical of sales work, gives this classification something of the character of authority without genuinely fulfilling the condition of legitimized power. A separate measure was therefore created to isolate this type of relationship. The remaining small (0.2 percent) category, diverting, was combined with persuading because of a similar requirement to please and convince others.

The DOT code does not measure authority on the job as directly as does the supervision measure used in the GSS. But the DOT captures more of the richness of official occupational responsibilities; it reflects qualitative rather than quantitative distinctions. Further, the DOT measure does not depend on the worker’s own subjective definition of authority (see Wright and Perrone, 1977:36). In sum, we feel that the DOT code has compensating advantages and disadvantages for measuring authority. Taken together with the GSS supervision code, the two different measures should provide a more rigorous test of the importance of authority in class perceptions.

For the men in the 1966 to 1972 election surveys, we recoded the occupations recorded in the original surveys according to the DOT codes. The five authority categories taken together represent 35 percent of the Election Sample, and when compared with the nonauthority categories, they do confer more middle-class position (see Table 4.C). On the average, the authority positions are 9 percent more middle class than the nonauthority positions. As in the GSS results, the authority effect is modest but more than we would expect by chance alone.

Each of the authority subcategories is more middle class than the nonauthority category. Mentoring (0.692), negotiating (0.472), and instructing (0.423) are all well above the nonauthority occupations (0.000). However, supervising (0.226) and speaking-signaling (0.128), although more middle class than the comparison, are less different than what we might expect by chance alone.

Notes

1. Many Marxists also accepted the stabilizing impact of the new middle class; see, e. g., Bottomore (1966:48), who explains the continuation of American exceptionalism based on the growth of the new white-collar middle class.

2. For Wattenberg, the middle class is just a statistical range whose boundaries can be manipulated to suit any purpose. To justify the “massive majority” label, he stuffs assembly-line workers and corporate executives into the same class receptacle.

3. The importance given to the new middle class by revisionist Social Democrats such as Eduard Bernstein ([1899] 1961) may account for orthodox Marxism’s rejection of the concept.

4. Some Marxists (e.g., Mallet [1963], 1975) even anointed the new technicians and professionals as a “new working-class” vanguard who would, in the twentieth century, finally fulfill Marx’s revolutionary prophecy (see Low-Beer, 1978).

5. We should first settle a point about terminology. Our use of the term “middle class” does not follow that of most other class analysts. “Middle” connotes an intermediate position on a ranking and thus suggests a range within a status model rather than a bounded category engaged in class conflict (cf. Lopreato and Hazelrigg, 1972:143). To avoid any status-ranking connotation, class theorists have replaced the term “middle class” with such inventions as the “new petty bourgeoisie” (Poulantzas, 1974 and 1977); the “PMC,” for professional-managerial class (Ehrenreich and Ehrenreich, 1979); and “contradictory class locations” (Wright, 1980). We do not quarrel with these, but we think more is to be gained by staying with the familiar “middle class” designation; doing so builds a bridge between a rigorous class analysis and the popular perceptions of class. “Middle class” is the term in popular usage; it is, of course, imprecise because people often use it in ways that have nothing to do with twentieth-century class conflict. But our research shows that there is a genuine class content in the popular perception of the middle class. It seems to us most profitable to grasp the class content already signified by “middle class” and “working class” and to provide more rigorous definitions for terms already accessible to most people.

6. Our emphasis on Poulantzas’s three-dimensional analysis differs somewhat from Erik Wright’s interpretation which places greater stress on Poulantzas’s distinction between productive and unproductive labor (see Wright, 1976, and, especially, 1985). The three dimensions reflect the influence of Althusser and, thus, may seem unfashionably structuralist today. In fact, Poulantzas combines a structural appreciation of how the new middle class contributes to capitalist accumulation with a practical concern for its political role in the struggles between capital and labor. Although Poulantzas would have denied it, the three-dimensional approach also has a decidedly Weberian ring to it (cf. Weber [1921], 1978:926). As Frank Parkin (1979:23) has remarked, “Inside every neo-Marxist there seems to be a Weberian struggling to get out.”

7. The importance attached to the capital-labor conflict separates Poulantzas’s analysis from other revisionist approaches to the middle class, such as Dahrendorf’s (1959) exclusive emphasis on authority and Wright’s (1985) multidimensional scheme that weights each division equally as reflecting an independent mode of production. We do not take up Poulantzas’s other economic distinction between productive and unproductive labor. Although we find this distinction important for fitting the working class within the larger process of capital accumulation, we have been unable to find its reflection in class perceptions.

8. These principles can be compared with Erik Wright’s (1985:34) list of “conceptual constraints.” Both sets emphasize the relational and antagonistic nature of middle-class positions. Wright includes in his constraints the consequences of class divisions for class formation, class struggle, class consciousness, and the history of social change. We prefer to consider such consequences as empirically testable propositions rather than as criteria of class definition. We find Wright’s treatment especially deficient on the historical origins of the middle class: his reliance on John Roemer’s (1982) abstract model of economic exploitation often leads Wright into a totally ahistorical analysis. As a result, his scheme neglects the subordination of the middle class to capital.

9. Daniel Bell (1973) ignores capital almost completely. Galbraith (1967) at least addresses the conflict between capital and mental labor (“technocrats”); he was wrong, however, about who was stronger.

10. Poulantzas labeled the new middle class a “new petty bourgeoisie,” in part because of the procapitalist positions it took on most issues. American data (Hamilton, 1972:202) confirm the political similarity of the old and new middle classes. But see Therborn (1982:33) for a dissent on conflating the new and old middle classes.

11. Most sociologists have assumed just the opposite, without benefit of (or even interest in) any empirical evidence; e.g., Daniel Bell (1973:72) assumed that the new middle class was primarily concerned to maintain the status of its largely petty-bourgeois origins and its “clean cuff occupations.”

12. In fact, Dahrendorf argues, it has always been the exercise of authority that separated the dominant class from workers. The original overlap that gave early capitalists both authority and ownership of the means of production led to Marx’s confusion in identifying ownership as the differentiating characteristic of class relations. Only now that ownership has been separated from authority can we see that it is the exercise of authority that determines capitalist class conflict.

13. It is their emphasis on managers’ subordination to capital that distinguishes the neo-Marxian theories (Poulantzas, 1974; Wright, 1976 and 1980; Carchedi, 1977) from Dahrendorf’s. Nonetheless, the neo-Marxian approaches owe an often unacknowledged debt to Dahrendorf for first emphasizing the importance of authority (Parkin, 1979:23).

14. Managerialism theories (Berle and Means, 1932; Dahrendorf, 1959; Galbraith, 1967) flourished in the prosperity of the immediate postwar period when corporate affluence made the profit constraints on managers largely invisible. As firms encountered losses in the stagnation of the 1970s, the subordination of managers to capital became more readily apparent.

15. In fact, the class position of first-line supervisors is somewhat ambiguous. They often are not the “lowest level of management” but are on a rung that is unconnected to the rest of the management ladder. They come from working-class—not middle-class—backgrounds; they are not college educated; and they rarely get promoted to higher levels of management.

16. All means and standard deviations for variables used in this chapter are presented in the chapter appendix (Table 4.A).

17. A similar question has been used in other empirical studies of authority (Fox et al., 1977; Wright and Perrone, 1977; Robinson and Kelley, 1979; Wright et al., 1982). Besides not specifying who is being supervised, the GSS question does not specify what type of authority is being exercised: e.g., general oversight, task assignment, hiring and firing (see Wolf and Fligstein, 1979; Wright et al., 1982). The measure therefore subsumes many different types of authority, and it is likely that the respondents were not consistent in interpreting the meaning of the question. While we can lament the fact that so central a concept is so poorly measured in empirical research, we must nevertheless use what data are available to estimate the results that might be achieved with more valid measures.

18. Controls are included for years of schooling, family income in the preceding year, and occupational prestige (see the appendix). We also control for the other two class dimensions as well, i.e., class placements are computed for both supervisors and nonsupervisors who have average probabilities of being self-employed and engaging in mental labor. For clarity of presentation, we discuss the relation between class divisions and class perceptions in separate sections. However, all multivariate analyses in this chapter have included all three dimensions of class position.

19. The plotted points represent the class placements for occupations at each prestige level averaged across nine prestige points and adjusted for the effects of the other class and status variables. We have also drawn the best-fitting curves, normal ogives, elongated S-shaped curves that are estimated by probit analysis. These curves are close to the results reported in the appendix in Table 4.B; the only difference is that in the plot we have calculated separate prestige curves for supervisors and workers (i.e., we have introduced an interaction term for prestige and authority). This interaction term is not statistically significant.

20. This and other percentages cited in the text are calculated from the probit equations in the appendix, Table 4.B. We substitute the mean values for all the control variables and calculate predicted scores for supervisors and nonsupervisors; these scores are then transformed into percentages according to the cumulative normal distribution.

21. As with all the analyses in this chapter, we attempted to replicate these findings with the Election Sample—a second, independent sample based on a different series of surveys. Using a quite different measure of authority, that analysis confirmed its effect on class perceptions (see the discussion in the chapter appendix). With the GSS data, we also attempted to explore the effects of different levels of authority (see Lopreato, 1968; Fox et al., 1977; and Jackman and Jackman, 1983:118). Respondents who supervised others were asked whether any of the people they supervised were themselves responsible for supervising others. Then a three-category scale was constructed: supervisors with one level of authority are 9 percent more middle class than workers without authority (all else being held constant). Supervisors with two levels of authority are only very slightly more middle class (3 percent) than the first-line supervisors; this slight difference is not greater than we might expect by chance. Thus, authority itself seems to have a truly dichotomous effect, as Dahrendorf suggested (1959: 171).

22. It may be also that we need a more stringent test of authority. Wolf and Fligstein (1979) show that women lack higher levels of authority (e.g., to determine pay: 37 percent of men, 14 percent of women; to hire and fire: 28 percent of men, 9 percent of women) even more than they lack supervisory authority (61 percent of men, 38 percent of women).

23. Several Marxist interpretations would disagree (e.g., Wright and Perrone, 1977; Reich, 1978:180).

24. Taylor also sold the new methods as more technically efficient: he claimed that scientific management could figure out the one best way of producing something. In fact, it rarely did, and skilled craftworkers easily outperformed Taylor’s cookbook methods.

25. We have reservations about the practice of Wright et al. (1982) in defining a similar mental-labor division (what they call “semiautonomous labor”) on the basis of survey responses to questions on the extent of work autonomy. This seems to concentrate too much on the microspecifics of an individual’s working conditions and thus to ignore the social relations between mental labor and the working class. In particular, the concept of semiautonomous labor does not distinguish between the autonomy retained by skilled craftspersons and the autonomy enjoyed by mental labor, which capital has created at the expense of the working class. It is only this latter type of autonomy that merits a separate class location.

26. The use of census occupational classifications to create a class distinction has drawn sharp criticism from some class theorists. Wright (1979) insists that occupational data measure the technical relations of production and thus must be kept distinct from class divisions determined by the social relations of production. Carpenters, e.g., may be either employers, supervisors, or workers. This critique is a thoughtful corrective to the functionalist or even atheoretical use of occupational data in the past, but we think it overstates the case in at least two ways. Occupational classifications were not created solely to measure the technical relations of production; in fact, they do include social relations as well (e.g., between store managers and store clerks, or between construction foremen and construction workers). Moreover, the technical content of work for many occupations (social work, engineering, etc.) is very much the product of the historical development of class conflicts. To say that these positions are defined solely by the technical relations of production is to ignore their special role as agents of control over the working class.

27. Poulantzas is more inclusive in his definition, classifying clerical work, retail sales, and some service work as mental labor. But he acknowledges that these categories have an “objectively proletarian polarization” (1974:316–27) compared with the managerial-professional cadres isolated by Braverman and the Ehrenreichs. Although professionals are also subordinate to capital (see Derber, 1982), they retain control over others’ lives in a way that clerical workers do not.

28. Breiger (1981) reports more recent results that argue for the more conventional manual-nonmanual division. The most complete set of analyses (Pomer, 1981) explains the discrepancy. Clerical workers’ origins are similar to those of blue-collar workers, but their children’s occupations are more similar to those of white-collar children.

29. Mental labor is also often associated with supervisory authority: 60 percent of the mental-labor class also directly supervises subordinates; and only 30 percent of the manual-labor class has any supervisory power. But it is the close association between mental labor and occupational prestige that raises the most methodological questions. This multicolinearity makes it difficult to separate statistically the effects of mental labor and prestige on class perceptions. Fortunately, the overlap between mental labor and prestige is not so great that our results are endangered (see appendix, Figure 4.A). Our sample sizes provide sufficient numbers of low-prestige mental laborers and high-prestige manual laborers for separate estimates of the two effects on class perceptions.

30. The Election Sample confirms the substantial effect of mental labor on men’s class perceptions (see appendix, Table 4.B). However, for women the difference between mental and manual labor is small and not statistically significant.

31. Cf. labor historian Selig Perlman (1928:157): “The enormous strength of private property in America, at once obvious to any observer, goes back to the all-important fact that, by and large, this country was occupied and settled by laboring pioneers, creating property for themselves as they went along and holding it in small parcels.” See also de Tocqueville [1835, 1840], 1954 (vol. 2): 267, and Karabel, 1979:211.

32. The old middle class may still have a conservative influence well beyond its meager numbers. The self-employed are disproportionately active in politics. Realtors, shopkeepers, morticians, and local attorneys still dominate the rhythms of local politics in hundreds of towns and small cities across America. These are people whose economic struggles have made them conservative individualists and the backbone of antilabor sentiment.

33. Burawoy was a Ph.D. candidate at the University of Chicago at the time he worked in the machine shop, so his employer’s economic power over him was negligible. Moreover, the recessionary period he describes as rather ineffective in increasing the subordination of workers was, by a fluke, a time when the factory was expanding employment and demanding substantial overtime: i.e., a period of labor shortage for that firm. Those circumstances are not likely to provide a convincing demonstration of capital’s economic power.

34. Aspirations for self-employment have fallen considerably since the first half of the century: from 71 percent of employees in 1939 to 34 percent in 1976 (Schlozman and Verba, 1979:156).

35. Craig Calhoun (1982:123–26) identifies similar artisan origins of early nineteenth-century English radicals, but he argues that an artisan background prevented the English workers from developing into a class-conscious proletariat; instead, they were trapped in a backward-looking “reactionary radicalism” that was anticapitalist but not prosocialist in even a rudimentary form. It sought only to restore the privileges of the earlier artisan community and was thus distinct from the later radicalism of the urban factory workers who were the real “making of the English working class” (cf. Thompson, 1963). The distinction between the two types of anticapitalist sentiment is an issue we will pick up again in Chapter 12. For now, the important issue is whether the American worker was anticapitalist at all, or whether—as Sombart maintained (1906:20)—“he loved it.” The artisan origins of the Massachusetts and English working-class protests suggest that petty bourgeois aspirations need not be precapitalist but may contribute to anticapitalist sentiments.

36. The Election Sample also included a measure of self-employment, with an intermediate category for people who are both self-employed and wage laborers: 14 percent of the men were fully self-employed; another 1.6 percent were partially self-employed. In this sample, self-employment has little effect on class placements once the other class and status variables are controlled (see appendix, Table 4.B). We suspect a measurement difference between the two surveys. A separate analysis of personal income in the 1976 and 1978 election surveys found no effect of self-employment; in the GSS, a similar analysis found the well-established relationship between self-employment and income. Thus the Election Sample measure of self-employment is unrelated to either middle-class placements or personal income; the GSS measure is related to both. Since the self-employment effect on earnings is well established, we put more trust in the GSS results showing a self-employment effect on class perceptions.

37. We have investigated these three dimensions as independent effects on class perceptions because we feel this best reflects Poulantzas’s analysis of the middle class. Other Marxist analyses have instead combined the dimensions to create various class categories (see Wright and Perrone, 1977; Wright 1985). We feel this categorical approach is overly complex and theoretically unnecessary. Moreover, we did not find any statistically significant interactions among the three class effects (see McNamee and Vanneman, 1983, for an earlier report of this analysis).