Pin Money, Homemaking, and Class Conflict
Women have held an uncertain place in the class consciousness debates. The traditional stereotype has questioned women’s commitment to class struggle; it has regarded their employment as a source of “pin money” and concluded that their grievances are either borne patiently (because temporary) or avoided by withdrawal. Women have long been reputed to be poor candidates for union organization and, once organized, to have dubious staying power during strikes. Their class affiliations have generally been thought to derive from their husbands’ or fathers’ occupational positions. All of these unsubstantiated but widely held impressions add up to the stereotype of a docile worker.
The facts have shown otherwise. When tested, women have played no less heroic a role than men in American labor struggles. From the early organization of the New York garment workers with their 1909 strike oath (“If I turn traitor to the cause I now pledge, may this hand wither from the arm I now raise,” Levine, 1924:154; Laslett, 1970:121), through the Wobblies’ great 1912 victory in the Lawrence textile mills (Dubofsky, 1969), to the Women’s Emergency Brigade defense of the 1937 General Motors sit-down strike (Fine, 1969) and, most recently, the expansion of public sector unionism, women have disproved the notion of their docility. And well before women achieved prominence in most other fields, radicals such as Mary Harris (“Mother Jones”) and Elizabeth Gurley Flynn were organizing workers, men and women alike, against industrial capitalism; today Crystal Lee Jordan (“Norma Rae”) and Karen Silkwood may have won similar places in the popular culture. (On the other side of the class divide, women protagonists—though less common—have proved no less militant: Katherine Graham’s Washington Post faced down the newspaper crafts where her archrival, the New York Times, had flinched.)
We must remember that for women, too, class consciousness can develop out of class conflict and does not necessarily precede it. One of the leaders in the Women’s Emergency Brigade has described the conversion to militancy among the Michigan housewives: “A new type of woman was born in the strike. Women who only yesterday were horrified at unionism, who felt inferior to the task of organizing, speaking, leading, have as if overnight, become the spearhead in the battle of unionism” (Fine, 1969: 201). “Women . . . horrified at unionism” are hardly the models of class consciousness that we would expect to play the decisive role in the critical labor conflict of the Depression, to insert themselves between police and their husbands barricaded in the Chevrolet engine plant. And yet these same women, when faced with a situation testing their class allegiance, took the decisive step that propelled the class struggle forward, taking them along with it. Their class allegiance may have been a necessary precondition, but a politically elaborated ideology was not.
The image of docile women as an obstacle to working-class consciousness survives in spite of these dramatic events. Reductionist fallacies again tend to blame working-class women for their own oppression. Gender differences in class conflict are routinely attributed to different psychologies rather than to the different situations that confront men and women workers.
It is a fact that women have always had lower unionization rates than men; in 1977 only 11 percent of female private-sector employees were unionized, compared with 27 percent of men (Freeman and Medoff, 1984: 27).1 These low rates are routinely attributed to women’s weak class consciousness. Women are said to be unreceptive to union organization because they lack a commitment to their jobs, because they are too submissive to challenge their bosses, or because labor conflict would appear unfeminine. “Let’s face the fact,” one (male) AFL-CIO unionist declared in 1959, “women are, in the main, unorganizable. They are more emotional than men and they simply lack the necessary staying power to build effective unions” (quoted in Foner, 1980:419). His words echoed Gompers’s pessimistic advice to New England telephone operators 40 years earlier: “You’re only girls and such strikes have an awful record” (Foner, 1980:111). Fortunately, the “girls” didn’t listen. They struck—and promptly won union recognition and wage increases.
Strictly structural factors may account for the low unionization rates. Women are segregated into the least-skilled, lowest-paid jobs in the economy, precisely those jobs that are most difficult to organize, whether held by men or women. Meredith Tax, in her study of the women’s labor movement in the late nineteenth century, points to just these structural factors:
In fact, they were unorganized because they had just become workers; because they had so much work to do at home that they could hardly move; because their husbands, boyfriends, and fathers did not let them go to meetings; because they earned so little that they could not afford to take risks; and because no one would organize them. And when anyone tried, women often showed that despite all these barriers they were raring to go. (1980:32)
Today, we can examine the gender differences in unionization rates more quantitatively; the results point to the same structural differences in the position of women workers. By far the largest part of the reason for women’s lower unionization is the type of jobs they are segregated into. Clerical and service workers are less unionized than skilled and semiskilled workers; trade and service industries are less unionized than transportation and construction. This is true for both men and women in those jobs—although men have the good fortune to be more often skilled craftsmen and operatives working in transportation and construction (and therefore unionized), while women are stuck as clerks and service workers in retail trade and personal services (and therefore nonunion). When we compare men and women working in the same occupation in the same-sized firm in the same industry, the 16 percent difference in unionization rates reduces to just 6 percent (Freeman and Medoff, 1984:28).2
Although we have hard evidence to show that these structural factors are important, there is little evidence pointing to gender differences in class consciousness. In fact, what attitudinal evidence we do have suggests exactly the opposite. Attitude surveys document that women are more favorable to unions than men (see Table 8.1). In a 1982 ABC News/Washington Post poll (ICPSR, 1983), more women (62 percent) than men (52 percent) reported that they would join a union if they had not already done so.3 Women are also more sympathetic with labor than with business, disapprove more of the air traffic controller (PATCO) firings, and agree more often that workers are better off in unions. In our GSS samples, 74 percent of working women say they have some or a great deal of confidence in organized labor; only 64 percent of working men report that confidence.
These direct investigations of union attitudes do not find the psychological differences that are commonly inferred from different behavior. Instead, they indicate that the different behavior (low unionization) results from the different structural position of women (segregation into low-skill positions), not from different attitudes (hostility toward unions or weaker class consciousness). The structural obstacles that women workers face are imposing enough; they hardly need the additional burden of a gratuitous stigma that they are lacking in class consciousness.4
Still, suspicions persist about women’s class consciousness: that women do not see themselves divided into sharply defined classes, that their work experiences are less central to their class perceptions than is true of men. This is a variant of the “pin money” thesis: since women do not have the same lifetime commitment to work, it figures less strongly in their perceptions of their own class position.
NOTE: a* = statistically significant difference, p < .05
Although women’s self-perceptions of class do not differ in the aggregate from men’s (51 percent of white women place themselves in the middle class; 52 percent of white men), the process by which women arrive at their class placement is different. We have already seen (in Chapter 4) that women’s jobs do not determine their class placements as much as men’s jobs determine men’s placements.
One explanation might be that women depend on their husbands’ jobs for their class placements so that their own jobs have less influence. This was the point of view of the early literature on family status (see, for example, Parsons, 1943; Centers, 1949:35). The assumption was—and it was little more than an assumption—that the entire family’s status was determined by the husband’s achievements; once she was married, a woman’s employment was irrelevant.
Fortunately, women’s class perceptions have now received considerable research attention.5 The first results challenged the traditional stereotype; in placing themselves in the class structure, working wives appeared to use their own occupations and education, although their husbands’ occupations and education were also important (Ritter and Hargens, 1975; Hiller and Philliber, 1978; Van Velsor and Beeghley, 1979). In contrast, the men ignore their wives’ attainments (Felson and Knoke, 1974; Van Velsor and Beeghley, 1979).6
Still later research by Jackman and Jackman (1983) has now challenged the revisionist interpretation and reasserted the traditional model. In their data, women seem to use only the status of their husbands’ occupations in choosing a class label; their own jobs do not appear to alter their class placements. The Jackmans attribute the earlier revisionist results to methodological problems. But their research has problems of its own: their analysis includes women who work only part time (31 percent of the sample). Our data, as one might expect, show that women with part-time employment do rely entirely on their husbands’ jobs in choosing a class label.
But the main problem with all the earlier research is that it does not take into account two very important factors. First, women’s class perceptions differ from men’s because women tend to adopt a maximizing strategy: if either a woman or her husband holds a mangerial position, she claims a middle-class position. Second, women’s jobs differ from men’s jobs in systematic ways that tend to obscure class divisions. This second explanation says nothing about women or men themselves; it is their jobs that differ, not their psychology. For both men and women in typically “female” jobs, the manager-worker class division does not produce much difference in class perception. Because this second explanation focuses on the characteristics of work, not of “consciousness,” it parallels the explanations given above for women’s low unionization rates: the gender differences in class conflict result from the different structural positions in which women workers find themselves more than from different reactions to a similar structural position.
To compare men’s and women’s class perceptions, we must first broaden the research design. The past interest in working wives crowded out two groups that ought not to be ignored: single women and housewives. Excluding these two groups removes the majority of women from the research.7
SOURCE: General Social Surveys.
The role of the spouse’s class position
We divide the GSS sample of women into eight categories whose class placements we would like to estimate. The design is most easily understood in Table 8.2, which reports the sample sizes for each of the eight categories. A comparable design for men also has eight categories, although only six are directly comparable.8
The class placements in the eight cells are estimated after adjustments for differences in family income, age (since the married-not married differences compare respondents of quite different ages), the education of the respondent and (if married) of the spouse. The adjusted middle-class placements of the eight categories are reported in Table 8.3. Inspection of this table reveals a major reason previous estimates of the effects of women’s work were so low: managerial positions do not matter for women who are married to managerial husbands. If the husband is himself managerial, then it makes little difference whether the wife is employed in a working-class or a managerial position; she already sees herself as quite middle class. Similarly, the husband’s job makes little difference to a woman who is herself managerial; again, she already sees herself as quite middle class. This is the maximizing strategy suggested above: either a woman’s own position or her husband’s position as a manager is sufficient to justify a middle-class placement.
In contrast to the women’s maximizing strategy, men follow a markedly egocentric strategy in relating objective class position to perceived class placement: they simply ignore their wives’ jobs. Men with managerial wives are no more likely to see themselves as middle class than men with working-class wives. Thus, the men do indeed fit the traditional family stereotypes (developed by men, incidentally); their class perceptions depend almost entirely on their own jobs.
Three other comparisons are important to notice in these results: the effects of marriage, of the wife’s not working, and of both wife and husband in managerial positions. First, the traditional model suggests that marriage completely alters the social significance of a woman’s job: while single, her job may determine her class placement; once she is married, it is her husband’s job that is important. The results disprove this model. Not only single women but women married to working-class husbands use their own jobs in placing themselves in the class structure. The effect of managerial position for single women is + 16 percent; the effect for women with working-class husbands is +19 percent—about the same within the limits of our surveys.9
SOURCE: General Social Surveys.
NOTE: Adjusted percentages are calculated after controls for respondent’s and spouse’s education, family income, and age (see Table 8.D).
Second, more housewives see themselves as middle class than do equivalently placed women who have working-class jobs. The experience of a subordinate role at work seems to reinforce a woman’s recognition of herself as working class. The housewife role is more ambiguous in class terms.
Finally, women who have managerial jobs themselves and who are married to husbands with managerial jobs—a stereotypically dual-career household that we would expect to be most middle class—are, in fact, less middle class than couples with only one managerial spouse. This curious anomaly occurs in both the GSS and the Election Sample, and despite many additional tests we have not been able to explain it.10 It remains a mystery.
To summarize these results: the relevance of husbands’ managerial jobs accounts for a large part of the difference between men’s and women’s class perceptions. A man ignores the occupation of his wife even if she is a manager; he uses only his own occupation to place himself (an egocentric strategy). A woman will use her husband’s managerial position to place herself in the middle class and ignore her own working-class job (a maximizing strategy). If we exclude from the comparison men and women married to managerial spouses, then the gender difference in class perception is much less noticeable. In the GSS, a managerial position increases women’s middle-class placement by an average of 17.9 percent and men’s by an average of 25.9 percent. In the Election Sample, the difference is almost erased: the increase is 15.5 percent for women and 18.8 percent for men.
The gender composition of men’s and women’s occupations
Thus far, we have explained part of the difference between men’s and women’s class perceptions by their different responses to being married to a managerial spouse. The second part of the explanation has nothing to do with the differences between men and women but is a result of the difference between men’s and women’s jobs. It is well known that women are largely segregated in a narrow range of occupations that are almost entirely held by women: in nursing, teaching, clerical work, food service, and textile industries, for example (Oppenheimer, 1970). It may be that these kinds of work do not imply class membership as clearly as do the craft and administrative positions that men more often occupy. The division between managerial and nonmanagerial positions may be quite distinct among men’s jobs but is more blurred and gradual among women’s jobs (Glenn and Feldberg, 1977). If it is the jobs and not the people holding them that determine the perceived sharpness of the manager-worker gap, then we would expect that those (few) women in typically “male” occupations will perceive their class position much as men do: that is, with a sharp division between managers and workers. Conversely, those (few) men in typically “female” occupations will perceive their class position much as women do: that is, with a weak division between managers and workers.
We can test for the effects of job characteristics by scoring every occupation for its gender composition. Using census reports (U.S. Bureau of the Census, 1963:1–10, 1973b:1–11), we calculate the percentage of males in each occupational category; these scores are then assigned to each individual according to his or her occupation. Of course, men end up on the average with much higher scores (84.0 in the GSS) than women (34.4), but there is a range within both sexes, so that we can look at the effects of sex composition of occupations among women and among men. Do women in the more “male” occupations perceive their class position any differently from women in thoroughly “female” occupations?
Gender composition, by itself, is not related to middle- or working-class placement; male jobs are not seen as any more middle class than female jobs.11 But we are interested in whether the managerial-class division is more decisive for class placements among male jobs than among female jobs. To answer this question, we estimate the difference between managers and workers at each level of occupational “maleness.” The results provide good support for a sharper class division among typically male jobs than among typically female jobs.
Figure 8.1 displays the results of this analysis. Among women, the difference between managers and workers is enormous for those women in typically male occupations. But this difference shrinks as we look at women in more typically female occupations. For example, in jobs of 90 percent male composition, women managers are estimated to be 30.5 percent more middle class than are workers. But in jobs of only 10 percent male composition, the estimated self-placements differ by only 6.0 percent. Thus, part of the reason the managerial effect is weaker for women is that they are segregated into female occupations where the manager-worker difference is not a very sharp gap.
FIGURE 8.1. Class perceptions by sex composition of occupation and sex of worker
SOURCE: General Social Surveys.
What strengthens this interpretation is that the same phenomenon is observed among men. For men too, the management-labor gap is more important among male occupations than among female occupations. In the GSS, for men in occupations of 90 percent male composition, managers are estimated to be 37.6 percent more middle class than workers; for men in occupations of only 10 percent male composition, the class placements differ by only 8.6 percent. Since men are more often in male jobs, the manager-worker division seems quite important; that is, men are more often found on the right side of Figure 8.1, where the gap in class perceptions is large, and women more often on the left side where the gap is small. But if we hold constant the “maleness” of the occupation, there is no difference in the size of the managerial division. Thus, the gender composition of jobs explains all the remaining difference between men’s and women’s class perceptions.12
What is it about “female” jobs that makes them less relevant for class perception? We might find some answers in the growing body of feminist psychology that studies gender differences in levels of moral development (Gilligan, 1982). Because women are socialized to preserve the family and to take primary responsibility for child care, they attend more to interpersonal relations. This attention to affiliative ties and the needs of others is replicated in the expectations set for women working within the paid labor force in such roles as domestics, secretaries, sales clerks, and teachers. In those positions, as well as in those where tending to the personal needs of others is not explicitly included in the job definition (as in textile and light factory work), interpersonal attachments remain more important to women; thus, they may face a greater barrier to identifying major divisions between themselves and their bosses.
But despite the internal conflicts and the importance of interpersonal attachments, class divisions are perceived. In Ellen Goodman’s (1979) study of how people confront and move through changes in their lives, one respondent was a Black woman who achieved a managerial position after spending 30 years as a secretary. She describes the strong feelings of attachment to her bosses that evolved over the years but never loses sight of the differences in their positions—even acknowledging that secretaries can be “office maids”:
You can have a personal attachment for somebody you’re working for. At least I did. I had worked with two really fine men, which is one of the hazards of a secretarial job. You may work for truly fine people. You get so identified with them that you don’t know that you have a career that’s your own.
If you work for a person, as I did, who has some respect for your mind, you’re not really an office maid. The personal attachment is there. It’s one of the reasons why you like coming to work. And yet, that makes it very difficult to leave. So when I thought about leaving, there was something sad in it for me and yet something I wanted. (1979:49–50)
We suspect that the interpersonal attachments of women’s jobs may explain their lack of class relevance. We do not have the data to test such an explanation; more information on interpersonal relationships at work would be helpful. But with more research attention now focusing on the effects of gender segregation at work, perhaps such data will eventually become available.
We have made some progress in explaining why women seem to perceive their work less in class terms than men do. Two factors are important: the first individual, the second more structural. Women married to managerial husbands do not use their own jobs in determining their class placement; men married to managerial wives do. It is mainly among this subset of men and women that men are the more “class conscious.” For the remaining types, who are the bulk of the labor force, the differences between men and women are accounted for largely by the differences between men’s and women’s jobs, not by anything intrinsic to men and women themselves. The perceived class division between managers and workers is relatively weak in “female” jobs, whereas there is a clearly perceived class division between managers and workers in “male” occupations. Since men are more often in male occupations, men appear more class conscious on the average, but women in these male occupations are just as class conscious as the men are. Conversely, men in largely female occupations do not see as great a gap between managers and workers; they have as difficult a time as women do in translating their “objective” class position into a class perception. But because there are relatively few men in these female jobs, they don’t affect the overall averages much. Since women are concentrated in just these occupations where the manager-worker distinction is not seen as a major class division, women as a group seem to be less class conscious.
SOURCE: American Election Surveys.
SOURCE: American Election Surveys.
NOTE: Adjusted percentages are calculated after controls for respondents and spouse’s education, family income, and age (see Table 8.D.).
1. The restriction to private-sector employees exaggerates the gender difference. Statistics for all employed workers in 1985 report 13 percent unionization for women and 22 percent unionization for men (U.S. Bureau of Labor Statistics, 1986).
2. Another study of 1976 unionization rates (Antos, Chandler, and Mellow, 1980) showed that the 15.4 percent lower female unionization reduces to a 12.7 percent difference when regional and personal factors such as part-time employment are controlled, and to just 5.5 percent after applying even crude controls (seven and nine categories) for occupation and industry.
3. The slightly different wording of this question in the 1977 Quality of Employment survey (Freeman and Medoff, 1984) produces lower levels of agreement but the same gender difference favoring women. If current union members are added back into the totals as favoring joining a union, the difference reduces to women 67 percent, men 62 percent in the ABC News/Washington Post poll—a statistically nonsignificant five percentage points.
4. Nancy Seifer (1973:40–41) records a different psychological disadvantage that women may suffer from: she suggests that “perhaps more than any other single factor it is women’s lack of self-confidence which mitigates against organizing.” She cites a remark by a (female) labor official that “long years of discrimination have convinced many of them the effort will be fruitless.”
5. As recently as 1973, Joan Acker quite correctly complained that American sociology had too often ignored women; e.g., Centers’s original research investigated only men’s class identification. His admonition (1949:35) that women’s class psychology must be studied next went largely unheeded during more than a quarter-century of American research. The early exemplars of class-identification research, the studies by Hodge and Treiman (1968) and by Jackman and Jackman (1973), subsumed most women’s class placements under their husband’s occupational attainments.
6. There are many methodological problems in this research. The Ritter and Hargens (1975) and Hiller and Philliber (1978) studies of the relative importance of husbands’ and wives’ jobs fail to include any controls for educational attainments, controls that in our data eliminate the significant occupation effects they report (see also Jackman and Jackman, 1983). Nor do the data from the studies always support their conclusions. The Van Velsor and Beeghley analysis, the most thorough of the group, shows a nonsignificant coefficient for the effect of husband’s occupation on the wife’s class placement, and yet it claims (1979:775) that “the data . . . indicate the importance of both husband’s and respondent’s own occupational characteristics” (emphasis added). The Felson and Knoke (1974) analysis depends on an inappropriate partitioning of variance that precludes a direct assessment of the effects of husband’s and wife’s occupation and thus renders the results equivocal. Nevertheless, these authors discount the one direct test of wife’s status that they do make—a test inconveniently yielding positive results that contradict most of their main conclusions.
7. It also ignores some important research questions. For instance, as Acker (1973:177) has pointed out, even the traditional stratification models accepted the importance of a single woman’s job (“women determine their own social status only when they are not attached to a man”). But according to the traditional models, marriage will reduce or eliminate the status relevance of a woman’s job and education. A theoretically interesting comparison, therefore, is between the importance to class placements of single women’s jobs and married women’s jobs. The traditional stereotype of women would predict that single women’s jobs are more important to single women’s class perceptions than married women’s jobs are to theirs. As yet, this comparison has not been made because the sample is restricted to married women.
8. To simplify the analysis somewhat, we have dropped cases of single women who are not working or married women whose husbands are not working. Since these are usually students or retirees (or wives of students or retirees), we have also restricted the sample to women between the ages of 25 and 60, when such omissions are inconsequential. Nor have we adjusted the estimated class placements for occupational prestige, since to be consistent we would then have to adjust for the effects of the respondent’s and spouse’s occupational prestige; the analysis would become so complex and the problems of multicolinearity so pervasive that the results would be impossible to interpret meaningfully. Most of the job effects are captured by the managerial-nonmanagerial differences, and the omission of occupational prestige reduces the overall fit only slightly. Managers are defined as all the census classifications of managers and professionals except technicians.
9. The GSS and Election Sample are quite consistent, although several of the effects are stronger in the GSS. The Election Sample results are presented in the chapter appendix (Tables 8.A-8.D).
10. Before introducing controls for the other variables, the dual-manager cell is the most middle-class cell, although not much above the single-manager couples (see the appendix tables). We thought the controls for income might be lowering the middle-class placements in this cell especially; we thought it might be an interaction effect of college educations; we thought it might result from an age effect or age interaction with managerial position. Tests for all of these did not eliminate the anomaly.
11. Male jobs, of course, tend to pay better than female jobs, so the extra income results in more middle-class placements. But if we control for income, there is no association of gender composition of the occupation with class placements.
12. The Election Sample data are not quite as clear, but again women perceive class differences between managerial and nonmanagerial jobs as readily as men do when we control for sex composition. But for women there is a slight deviation from the standard pattern. The women in the Election Sample still perceive a difference between managers and workers even in typically female occupations—perhaps less difference than do women in male occupations but still a difference. Nevertheless, the central point is still supported: the gender composition of the job determines the importance of the manager-worker division.