Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Sex Segregation. Extent and Recent Trends The most common method of assessing the extent of sex segregation compares the distributions of women and men across a set of occupational categories. The difference in the distributions of the sexes across occu- pational categories can be summarized by the index of segregation (see note 2, Chapter 1, for the formula), which was developed by Duncan and Duncan (19551. Its value rep- resents the minimum proportion of persons of either sex who would have to change to an occupation in which their sex is under- represented in order for the occupational distributions of the two groups to be iden- tical. Its value is O in the case of complete integration, in which the occupational dis- tributions of men and women are identical, and 100 when every occupation is either entirely female or entirely male. For ex- ample, in 1981 the index of sex segregation computed over 11 major occupational cat- egories was 41 among whites and 39 among nonwhites (see Table 2-1), indicating that at least 40 percent of all women or men would have to change to an occupational category dominated by the other sex for their broad distributions to be identical (and for the pro- portion female or male in each category to 18 be equal to the proportion female or male in the total labor force). In interpreting the value of the index of segregation, one must bear in mind that its magnitude is unaffected by the type of oc- cupational shifts workers would need to make. Shifts from a sex-typical occupation to a closely related sex-atypical occupation- for example, from elementary school teacher, which is 84 percent female, to school ad- ministrator, which is predominantly male (U. S. Department of Labor, Bureau of La- bor Statistics, 1981cWare considerably more probable in the short run than shifts to oc- cupations requiring vastly different skills that are performed under different working con- ditions. Given the occupational structure, however, to achieve total integration both women and men would have to move to occupations that are atypical for their sex. As we noted in the previous chapter, our use of this measure of segregation does not imply that we believe complete integration of all occupations is an appropriate policy goal. We do, however, believe that job seg- regation should be substantially reduced. The index of segregation is influenced by the sizes of more and less segregated oc
SEX SEGREGATION: EXTENT AND RECENT TRENDS 19 TABLE 2-1 Occupational Segregation Indices Across Major Census Categories for Sex and Race, 1940-1981 1940 1950 1960 1970 198 Occupational segregation by sex among: Whites Blacks and others Occupational segregation by race among: Men Women 46 58 43 62 43 50 36 52 44 52 35 45 44 49 30 tan 41 39 24 17 NOTE: Indices are calculated for occupational distributions across 11 major census categories. The data from 1940 to 1960 are classified according to the 1940 census detailed occupational classification; the 1970 data are classified according to the 1960 census detailed occupational classification: and the 1981 data Arc ~l~cciG-A a^~^rtli-rs to the 1970 census detailed occupational classification. ~^~ TV- ~, bOURCJELS: For date from 1940 to 1970, Treiman and Terrell (1975b:167), Copyright it, Russell Sage Foundation, 1975. Repunted by permission of the publisher, Russell Sage Foundation. The indices for 1981 were computed from data published in U.S. Department of Labor, Bureau of Labor Statistics (1982a). cupations. If the most sex-typed occupations employed relatively few workers and the most integrated occupations employed most of the work force, the index would be fairly low. Alternatively, a few large, highly seg- regaled occupations could dominate a large number of small, integrated occupations to yield a large index. This feature of the index is desirable because it represents the actual occupational structure workers encounter. When one compares segregation levels over time or across populations with differing oc- cupational structures, however, differences in the values of the index will confound dif- ferences in the amount of segregation within occupations with differences in the sizes of occupations. (Blau and Hendricks. 1979 and Bianchi and Rytina, 1984, decompose the total index into components representing these aspects; we discuss their findings be- low.) To get a feeling for how much segregation is associated with a particular value of the index, it is helpful to compare different types of segregation. In Table 2-1 segregation in- dices are computed for 11 major census oc- cupational categories by both race and sex for each decade since 1940. Although the amounts of race and sex segregation across these broad occupational categories were similar in 1940, by 1981 the drop in the race segregation index was substantial (from 43 to 24 among men and from 62 to 17 among women), while the index of sex segregation decreased much less (from 46 to 41 among whites and Tom 58 to 39 among blacks and other races). One can also evaluate the mag- nitude of the index in the context of typical levels for other industrial countries. Using 14 broad occupational categories, Boos (1985) computed indices for 12 societies. The value for the United States, 47, fell toward the high end of the distribution, which ranged Dom a low of 27 for Japan to a high of 60 for Sweden. Of course, these values are a func- tion of the number of occupational cate- gones (which differed slightly across the countries), and we present these results only as a gauge for assessing the magnitude of a single index. The magnitude of the index changes as the number of occupational categories in- creases. Using 1981 Current Population Survey data, Jacobs (1983) calculates the in- dex as 40 on the basis of the 10 major census occupational categories and as 62.7 on the basis of 426 three-digit census occupations. The magnitude of the latter index is com- parable to the values shown in Table 2-4, also based on detailed census occupations.
20 WOMEN,S WOW MEN'S WOW The index of segregation can also be cal- culated for subsets of occupations, to inves- tigate how subsets compare with each other or to the whole. A. Beller (1984) computed the index for 1981 data for 262 occupations as well as for the subset of 59 professional occupations classified at the same level of detail. As we would expect, the inclex for the professional occupations was smaller than that for the full range of occupations (51 and 62, respectively), indicating less segregation across professional occupations than across all occupations. CURRENT EXTENT OF SEX SEGREGATION In 1980, 48 percent of all women worked in occupations that were at least 80 percent female (Rytina and Bianchi, 1984~. These include many clerical occupations (bank tell- ers, bookkeepers, cashiers, data-entry clerks, receptionists, secretaries, typists, and tele- phone operators) and service occupations (chambermaids, waitresses, practical nurs- es, chilc] care workers, hairdressers, ant! pri- vate household workers) as well as opera- tives in apparel manufacturing. Men were even more likely to work in occupations dominated by members of their own sex: 71 percent were employed in occupations that were at least 80 percent male, such as sci- entific, technical, ant! professional occupa- tions (engineers, chemists, dentists, phar- macists, and physicians), skilled crafts (carpenters, electricians, painters, plumbers, machinists, and auto and heavy equipment mechanics), operatives (meat cutters, grind- ing machine operators, forklift operators, welders, deliverymen, and truck drivers), and laborers (construction laborers, freight handlers, and gardeners). These proportions are slightly lower for black women and men (Malveaux, 1982b). Based on data for 312 detailed occupa- tions, Table 2-2 shows employment in the 10 largest occupations for women and men, and their percentage female in 1980. Of the largest 10 occupations for women, 9 were more than 70 percent and 7 were more than 80 percent female, compared with the total civilian experienced labor force, which was 42.5 percent female. Of the 10 largest oc- cupations for men, all were at least 70 per- cent mate and 7 were more than 80 percent male. Only one occupation-managers, not elsewhere classified was common to both lists. As we noted in Chapter 1, even measures of segregation based on detailed occupa- tional categories underestimate actual levels of segregation in employment because they do not measure the segregation of the sexes at the level of the establishment. As we not- ed furler, sex segregation can occur within occupations when the sexes have the same occupation, but at different ranks, within an establishment. For example, Halaby (1979b) provides evidence of rank segregation among managerial employees in a utility firm, and Norwooc! (1982) notes that among assem- blers and machine tool operators in the mo- tor vehicle parts industry, women were dis- proportionately concentrated in class C, the lowest-paid class. Occupations can also be more segregated across establishments than they are in the aggregate. Blau's (1977) investigation of office work- ers in three northeastern standard metro- politan statistical areas documented intra- occupational sex segregation across firms (i.e., the segregation of female and male workers in the same occupations in different firms). She assessed the amount of segre- gation for several occupations that were rel- atively sex-integrated in each city by com- paring the actual index of segregation for an occupation with the expected index given the size of the pool of qualified female and male workers and the percentage of women in the occupation in each firm. In most oc- cupations, the difference between the ex- pected and actual was considerable. Inter- estingly, it was smallest among computer programmers, an occupation that had grown twentyfold during the 1960s. Blau also found
SEX SEGREGATION: EXTENT AND RECENT TRENDS MULE ~0 Employment in the 10 Largest Occupations for Mien and Women, 1980 21 Ten Largest Occupations for Men 1970-1980 Change in Detailed 1980 Occupational Number Percentage FemalePercentage Title and Code of Men 1980 1970 Female 1. Managers, N.E.C. (019) 3,824,609 26.9 15.3 11.6 2. Truckduvers, heavy (804) 1,852,443 2.3 1.5 0.8 3. Janitors and cleaners(453) 1,631,534 23.4 13.1 10.3 4. Supervisors, production (633) 1,605,489 15.0 9.9 5.1 5. Carpenters (567) 1,275,666 1.6 1.1 0.5 6. Supervisor, sales (243) 1,137,045 28.2 17.0 11.2 7. Laborers (889) . 1,128,789 19.4 16.5 2.9 8. Sales representatives (259) 1,070,206 14.9 7.0 7.9 9. Fanners (473) 1,032,759 9.8 4.7 5.1 10. Auto mechanics (505) 948,358 1.3 1.4 - 0.1 Ten Largest Occupations for Women 1970-1980 Change in Detailed 1980 Occupational Number Percentage Female Percentage Title and Code of Women 1980 1970 Female 1. Secretaries (313) 3,949,973 98.8 97.8 1.0 2. Teachers, elementary school (156) 1,749,547 75.4 83.9 -8.5 3. Bookkeepers (337) 1,700,843 89.7 80.9 8.8 4. Cashiers (276) 1,565,502 83.5 84.2 - 0.7 5. Office clerks(379) 1,425,083 82.1 75.3 6.8 6. Managers, N.E.C. (019) 1,407,898 26.9 15.3 11.6 7. Waitresses and waiters (435) 1,325,928 88.0 90.8 - 2.8 8. Salesworkers (274) 1,234,929 72.7 70.4 2.3 9. Registered nurses (095) 1,232,544 95.9 97.3 - 1.4 10. Nursing aides (447) 1,209,757 87.8 87.0 0.8 SOURCE: Rytina and Bianchi (1984). that firms tended to have consistent patterns of sex segregation across occupations. If a firm employed more men than expected in one occupation, it was likely to do so in other occupations, and such firms tended to pay workers of both sexes higher wages. In another study of segregation at the es tablishment level, BielLy ant] Baron (1984) found an astonishing amount of job segre gation. Using data for 393 finns that the California State Employment Service col lected between 1959 and 1979, they found that 30 firms employed workers of only one sex. In an aclditional 201 firms, women and men shared none of the same job titles. ~us, Industries, too, stiffer both in their pro 231 of 393 firms were totally segregated (in- Density to employ women and in their levels dices of 1001. Only 16 establishments had of occupational sex segregation. The distri segregation indices below 60, and closer ex- buttons of the sexes across eight broad in amination of these relatively integrated firms revealed that in very few did women and men work side by side at the same jobs. For example, one integrated establishment em- ployed apartment house managers, each of whom resided in the builcling he or she man- aged. In another, women worked during the day shift, while men in the same job worked at night. Studies of specific occupations (travel agents by Mennerick, 1975; retail clerks by Talbert and Bose, 1977) or estab- lishments (HarIan and O'Farrell, 1982) con- firm patterns of considerable segregation by sex at the firm or fob level.
22 WOMEN'S WORK MEN'S WORK dustrial categories, shown in Table 2-3, dif- fer considerably. In general, women are concentrated in personal and professional services; finance, insurance, and real estate; communications; and retail trade. In con- trast, they make up less than 10 percent of workers in Togging, fisheries, horticulture, construction, metals and mining, and rail- roads (U. S. Department of Labor, Bureau of Labor Statistics, 1981c:Table 30~. This is not surprising given that industries have dif- ferent propensities to employ workers in particular occupations that we know to be sex-segregated. For example, financial firms employ many clerical workers, most of whom are women, and construction firms employ many laborers, most of whom are men. Sex segregation across industries occurs, how- ever, in amounts greater than would be ex- pected from their occupational distributions alone (Blau, 1977; Sto~zenberg, 1982~. For example, in 1970, 49.4 percent of all assem- blers, who usually work in manufacturing, were women. In electrical machinery man- ufacture, women constituted 74.2 percent of assemblers; in motor vehicle manufac- ture, they constituted only 17.2 percent (U.S. Department of Commerce, Bureau of the Census, 19721. Several researchers have concluded that women tend to be concen trated in economically peripheral industries (Kohen, 1975; Bridges, 1980), while men work disproportionately in the "core" sector ofthe economy (Beck et al., 1980), but there is disagreement regarding this finding and the definition of core and peripheral sectors. We stress that sex segregation at both the firm and the industry level limits the em- ployment opportunities of women. Some firms consistently exhibit more segregation than would be expected from the occupa- tional mix they hire, and more firms do this than would be expecter! by chance. It is hard to escape the conclusion that discriminatory practices of one sort or another are probably occurring. Such segregation appears, from Me few studies available, to be quite exten- sive, and it is not measured by occupational segregation alone. Clearly we need more data and more studies at the establishment level. The next section examines trends in segregation by sex and necessarily relies on occupational-level data. RECENT TRENDS IN OCCUPATIONAL SEX SEGREGATION Summary measures indicating current levels of segregation are primarily of interest as data points that reveal trends over time. TABLE 2-3 Sex Distribution Over Major Industrial Categories for Nonagricultural Industries, October 1984 Women Men Number Percentage Percentage Number Percentage Industry Division (in thousands) distribution Female (in thousands) distribution Mining123 .3 12.2889 1.7 Construction439 1.0 9.54,206 7.9 Manufacturing6,461 15.1 32.413,396 25.2 Transportation and public utilities1,434 3.4 27.23,838 7.2 Wholesale trade1,605 3.8 28.54,032 7.6 Retail trade8,573 20.1 51.97,961 15.0 Finance, insurance, and real estate3,462 8.1 60.72,240 4.2 Services12,587 29.4 59.98,440 15.9 Government8,061 18.9 49.78,152 15.3 Total42,745 100.0 44.653,154 100.0 SOURCE: U.S. Department of Labor, Bureau of Labor Statistics (1985a:Tables B-2 and B-3).
SEX SEGREGATION: EXTENT AND RECENT TRENDS . . . After decades of considerable stability, there has been some reduction in segregation over the past 10-20 years. Whether the over- whelming impression is one of change or stability, however, depends partly on whether one looks at the overall picture, which reflects the experiences of more than 100 million workers, or at certain occupa- tions or subgroups in the labor force. Among the latter, increased integration has taken place. We begin by examining two summary measures that necessarily mask change with- in specific occupations; we then turn to data on the experience of young people; we con- clude by examining changes within selected occupations. Ibe concentration of workers in occupa- tions that are at least 80 percent male or female has increased slightly over the last three decades (Blau, 1977, Waite, 1981~. Ike trend, however, is sensitive to the definition of a sex-dominated occupation and may be an artifact of the growing number of occu- pations that the census distinguishes. Using as a criterion the overrepresentation of either sex by at least 5 percentage points relative to its representation in the labor force, A. Belier (1984) observed a decline cluring the 1970s in men's concentration in some tra- ditionally male occupations. Beller's finding of a decrease in the pro- portion of men in male-dominated occupa- tions for the 1970s is corroborated by Rytina and Bianchi (19841. They also found a de- crease in the proportion of women in female- dominated occupations. The occupational data from the 1980 census and all earlier censuses are especially difficult to compare because of sweeping changes macle in the 1980 census occupational classification scheme. Using data for a sample of 120,000 individuals in the experienced civilian labor force whose occupations were "double cod- ed" with both the 1970 and the 1980 detailed] occupational codes, Bianchi and Rydna (1984) were able to recode 1970 data into 1980 cat- egories and then compare the sex compo- sition of occupations in the two census years. 23 This procedure aHowed them to use virtually all occupations representing the entire 1970 and 1980 labor force. Using a 20 percentage point spread around the proportion female in the labor force (taken as 40 percent), they defined male-intensive occupations as those that were no more than 20 percent female and female-intensive occupations as those that were at least 60 percent female. The proportion of men who were in male-inten- sive occupations fell from 72.3 percent in 1970 to 52.9 percent in 1980, and the pro- portion of women who were in female-in- tensive occupations fell from 73.6 percent in 1970 to 63.3 percent in 1980. The pro- portion of men employed in female-inten- sive occupations did not change, while the proportion of women employed in male-in- tensive occupations actually fell, Tom 9.4 to 6.1 percent, but the proportion of both men and women working in the sex-neutral oc- cupations rose substantially (Bianchi and Rytina, unpublished data, 19841. Indices of occupational sex segregation for the labor force as a whole show remarkable stability over most of this century as well as a decline during the 1970s. The index of segregation computed for three-digit occu- pational classifications for each decennial census has fluctuated between 65 and 69 between 1900 and 1970 (Gross, 1968; Blau and Hendricks, 1979) and declined to about 60 in 1980 (A. Belier, 1984; Bianchi and Rytina, 1984~. Ibe index increased slightly between 1950 and 1960 and then dropped slightly between 1960 and 1970. According to Blau and Hen- dricks (1979), the increase during the 1950s stemmed primarily from the growth of pre- dominantly female clencal and professional ~ Attempts to determine the extent of occupational segregation in the nineteenth century (Oppenheimer, 1970; Sorkin, 1973; Williams, 1979), although plagued by problems of the comparability of data, suggest some movement toward desegregation between 1870 and 1920, probably due to the emergence of new occupa- tions that had not yet been sex-typed.
24 WOMEN'S WORK, MEN'S WORK occupations, while the decline during the 1960s was due largely to increased integra- tion of occupations, which was the conse- quence of men's movement into tradition- aDy female professions such as elementary school teacher, librarian, nurse, and social worker, rather than to an increase in wom- en's representation in male-dominated oc- cupations. Since the number of job openings generated by occupational growth and turn- over sets limits on the amount of desegre- gation, Blau and Hendricks compared the observed decline in the index with the amount that would have occurred had all positions that became available during the period! been filled randomly with respect to sex.2 This simulation, summarized in Table 24, indicates that sex segregation would have dropped by almost 25 percent during each of the two decades had the allocation of workers to new jobs been sex neutral. In light of this, the actual decline of 3 points (4.5 percent) between 1960 and 1970 is ex- tremely modest. Recent research (A. Belier, 1984; Jacobs, 1983; Bianchi and Rytina, 1984) suggests that more rapid change has occurred during the 1970s. The segregation indices that Belier computed for 262 detailed census occu- pations3 declined by 6.6 points between 1972 2 Blau and Hendricks (1979) operationalized sex-ran- dom hiring to mean that new positions are filled ac- cording to the sex ratio that prevailed in the pool of new labor force entrants and individuals released Dom declining occupations. Lacking data on the magnitude of replacement, they assumed no change in occupa- tional sex composition due to turnover, thereby ignor- ing the potential contribution to integration that sex- blind replacements of job turnover would produce and thus underestimating the amount of integration pos- sible. They also note, however, that failing to consider occupational entry requirements may yield an over- estimate of the amount of integration that could occur in filling new positions. 3 In order to construct a consistent data series for the period 1972-1981, Beller included only those oc- cupabons that had at least 25 respondents repre- senting occupations with at least 40,000 incumbents and 1981, from 68.3 to 61.7. To put these values in some context, Belier computed in- dices for the same 262 occupations in 1960 and 1970 using census data. During that dec- ade the index declined from 68.7 to 65.9, a decline of only 2.8 points.4 Between 1972 and 1981 the index of segregation declined at an annual rate nearly three times that for the 1960s (BeDer, 19841. Of the decline of 6.6 points between 1972 and 1981, 18 per- cent was due to changes in the sizes of more and less segregated occupations; the re- maining 82 percent represents changes in the sex composition of the occupations and reflects increased integration of occupa- tions. Using data Tom the 1970 arid 1980 censuses for virtually all occupations, Bian- chi and Rytina (1984) obtained similar re- sults. The indices of segregation they cal- culated declined by 8.4 points (from 67.7 to 59.3) between 1970 and 1980, with 76 per- cent of the decline due uniquely to shifts in sex composition within occupations. Jacobs (1983) used Current Population Survey data to compare sex segregation for 1971 and 1981 across both broad and narrow occupational categories as well as for over 10,000 occu- pation-by-industry categories. Jacobs's re- sults for 426 detailed occupations closely re- semble those of Beller for 262 occupations and those of Bianchi and Rytina for the com- plete set of occupations. Of particular in- terest is the decline during the 1970s of over 13 percent (from 80.3 to 69.6) in the seg in both the 1974 and l9f7 Current Population Surveys (CPS) (Annual Demographic Files). Beller used CPS data for 1971-1974 and 1977. In addition, the 1972, 1977, and 1981 indices were based on Bureau of Labor Statistics annual averages of monthly Current Popu- lation Surveys. The Current Population Survey and the Bureau of Labor Statistics annual averages yield slightly different results. Their comparability is discussed in A. Beller (1984). 4 The values Beller obtained differ from those of Blau and Hendricks and Bianchi and Rytina (shown in Table 2~) because each used different data and occupational categories. Only comparisons within the individual studies are appropriate.
SEX SEGREGATION: EXTENT AND RECENT TRENDS 25 TABLE 2~ Actual and Predicted Segregation Indices, 1950-1980, and Percentage Decline Predicted If Hiring During Previous Actual Decade Were Sex-Neutrala - Decadal Percentage Decadal Percentage Year Index Change Decline Index Change Decline 1950 73b 1960 74b or 1 0.0 56 -17 23.4 1970 71b -3 4.2 56 -18 24.0 1970 67.7c 1~0 59.3C -8.4 12.4 47.8 - 19.9 29 4 . a Each value reflects the amount of change that would have occurred over the previous decade, relative to the actual level of segregation at the decade's beginning. Thus, had hiring been sex-neutral between 1960 and 1970, the segregation index in 1970 would have declined by 18 points Dom 74 to 56. iIndices are computed for 183 detailed occupational categories in all three decennial censuses. Large residual categories such as "other operatives,' which are necessary to account for the entire labor force, were eliminated. The occupations included employed 66 70 percent of the labor force in the three census years. CIndices are computed for all occupational categories in the 1980 census, with 1970 census data recoded to the 1980 categories. SOURCE: 1950-1970: computed Dom Blau and Hendricks (1979:Table 3 and text). 1970-1980: computed from Bianchi and Rytina (1984:Table 7~. regation index computed for over 10,000 de- tailed occupation-by-industry categories.5 These three major studies of sex segre- gation in the 1970s (A. Belier, 1984; Jacobs, 1983; Bianchi and Rytina, 1984) all agree Mat sex segregation declined substantially during the decade, although earlier studies (e.g., Lloyd and Niemi, 1979) failed to find a substantial decline. Most of the decline, firer more, was found to be due to the greater integration of occupations, not to changes in the size of the predominantly male or predominantly female occupations. Nevertheless, change was less rapid than it would have been had all hiring during the decade been sex-neutral. Bianchi and Rytina (1984) replicated for the 1970s the exercise Blau and Hendricks (1979) carried out for the 1950s and 1960s, comparing actual and potential declines in occupational segrega 51~he decline was greatest in the New England and Pacific and Mountain states, which showed the lowest values in 1981, and smallest in the Mid-Atlantic and South Central states, the latter of which showed the highest level of occupational segregation of any of the regions in 1981. lion. The 12.4 percent decline in the seg- regation index actuary observed represent- ed less than half of the 29.4 percent decline that would have occurred had aD new hires been independent of sex (see Table 241. Changes in Sex Segregation Among Population Subgroups Given the large amount of stability built into the occupational structure (Blau ant! Hendricks, 1979; Tolbert, 1982; Treiman and Hartmann, 1981), the potential for change in sex segregation should be greatest for new entrants into the labor force and among those who are young enough to train for or shift to sex-atypical occupations. Bureau of Labor Statistics data for 1981 support this expec- tation. Younger workers showed slightly less segregation across 44 two-digit occupations. The index of segregation for all workers was 53.5, but for workers ages 20-24, it was 51.1. Women ages 20-24 were more likely than women of other ages to work as engineers, engineering and science technicians, other salaried professionals, managers, and administrators; and they were unfderrepre- sentecI among retail salespersons, operatives
26 WOMEN'S WORK, MEN'S WORK (especially in nondurable goods, a predom- inantly female occupation), and in most ser- vice occupations (see Table 2-51. The oc- cupational distribution of women ages 25-34 was closer to that for older women. lacobs's (1983) results for 426 detailed occupations, while also revealing slightly less segregation among younger workers, dyer in showing the greatest decline in the segregation index Ed the least segregation among women ages 25-34; those ages 16-24 were slightly more segregated. A. Beller (1984) also found that workers who had been in the labor force no more than 10 years were less segregated than the remainder of the labor force in both 1971 and 1977, and that the gap has been wid- ening. She identified two sources of chance: the 1971 entry cohort became less se~re- gated as it aged and the cohort entering in 1977 was less segregated than the 1971 co- hort had been at entry. An earlier study by Eleller (1982a) may explain some of this change. She found that equal opportunity legislation enhanced the likelihood of get- ting into a sex-atypical occupation more for new entrants into the labor market than for any other group. According to Jacobs (1983), the segrega- tion index decliner! by about the same amount among whites and blacks, but other groups (primarily Hispanics and Asian Americans) showed the most decline. Their sex segregation index dropped from 75.6 to 64.6 between 1971 and 1981. Beller (1984), who distinguished only whites and non- whites, observed larger declines among the latter, although the index for professional occupations dropped more for whites than nonwhites, indicating that much of the in- crease in integration by sex for nonwhites occurred at the lower end of the occupa- tional distribution. Changes in Sex Segregation Among Occupational Subgroups Of course, the decline in segregation was far from uniform across occupational cate gories, much less within detailed occupa- tions. For example, using census data through 1970, Scott and Semyonov (1983) report that three major occupational cate- gories operatives, farm managers, and managers became more male-dominated, while clerical occupations became more fe- maTe-dominated; occupations that moved toward parity were professional and sales, and, since 1960, domestic service, crafts, and labor. Rytina and Bianchi (1984) report that managers have become much more ~n- tegrated since 1970; in 1980, managers were 31 percent female, a very substantial in- crease of 12 percentage points since 1970. Jacobs (1983) and A. Beller (1984) examine patterns of change within detailed occupa- tions since 1970. Jacobs's analysis of the 1971 and 1981 Current Population Survey data showed that among nonfarm occupational categories the index of segregation cleclined most for professiorlal occupations (by almost 27 percent: 16.5 points). Beller (1984) concluded that the decline observed in the index of sex segregation dur- ing the 1970s was due, in addition to in- creased integration of some occupations, to declines in the sizes of two heavily female occupations-private household maids and servants and sewers and stitchers; each ac- counted for more than a one-point decline in the segregation index. Three other oc- cupations dominated by one sex (telephone operator, private household child care work- er, and delivery and route worker) also con- tributed to the dropping index because they declined in size. A smaller proportion of the female labor force worked as retail sales clerks, typists, and cooks, while women en- tered three rapidly growing male occupa- tions: accountant, bank officer and financial manager, and janitor. Beller also showed Mat the observed decline in the index masked some changes in the occupational structure that actually contributed to greater segre- gation. Several female-dominated occupa- tions have grown rapidly (i.e., registered nurse and office manager), and some have simultaneously become more female (com
SEX SEGREGATION: EXTENT AND RECENT TRENDS - TABLE 2-5 Percentage Female in Detailed Occupational Groups by Age, TweIve-Month Annual Averages, December 1981 Occupation All Workers Ages 20-24 Ages 25-34 Total 43 (10D,397) 47 (14,122) 42 (28,180) Professional, technical, and kindred workers 45 (16,419) 53 (1,687) 47 (5,906) Engineers 4 (1,537) 13 (132) 6 (447) Physicians, dentists, and related practitioners 14 (828) 52 (23) 21 (240) Other health professions 86 (2,297) 83 (336) 84 (911) Teachers, except college and university 70 (3,197) 78 (226) 71 (1,176) Engineering and science technicians 18 (1,141) 23 (226) 18 (427) Other salaried professionals 36 (6,668) 47 (713) 39 (2,482) Other professional and self-employed workers 27 (751) 40 (32) 27 (223) Managers and administrators, except farm 27 (11,540) 42 (754) 29 (3,051) Manufacturing, salaried 15 (1,566) 36 (58) 20 (374) Other industries, salaried 30 (8,011) 44 (640) 32 (2,292) Retail, self-employed 35 (870) 29 (24) 31 (154) Other independently self-employed 16 (1,093) 15 (32) 15 (231) Sales 45 (6,425) 51 (854) 39 (1,626) Retail 63 (3,262) 57 (583) 56 (667) Other 26 (3,162) 39 (271) 27 (958) Clerical 80 (18,564) 82 (3,352) 80 (5,212) Bookkeepers 91 (1,961) 89 (251) 92 (S15) Office machine operators 73 (966) 74 (231) 73 (349) Stenographers, typists, secretaries 98 (5,022) 98 (928) 99 (1,463) Other clerical 70 (10,615) 74 (1,942) 70 (2,885) Craft and kindred workers 6 (12,662) 6 (1,656) 6 (3,879) Carpenters 1 (1,122) 3 (177) 2 (395) Other construction crafts 1 (2,593) 2 (376) 2 (808) Foremen, not elsewhere classified 11 (1,816) 15 (115) 11 (471) Machinists and job setters 4 (668) 4 (97) 5 (199) Other metal 4 (626) 4 (65) 6 (180) Mechanics, auto 0.6 (1,249) 0.4 (243) 0.7 (408) Other mechanic 3 (2,159) 3 (266) 3 (692) Other craft 17 (2,430) 20 (317) 17 (726) Operatives, except transport 40 (10,540) 33 (1,841) 35 (3,002) Mine workers 2 (357) 2 (go) 2 (134) Motor vehicle equipment 19 (452) 17 (52) 19 (148) Other durable goods 36 (4,153) 30 (736) 33 (1,233) Nondurable goods 58 (3,339) 52 (543) 52 (928) All other 30 (2,240) 22 (419) 26 (560) Transport equipment operatives 9 (3,476) 6 (480) 9 (1,029) Dnvers, delivery 10 (2,966) 7 (382) 10 (862) All others 5 (511) 5 (98) 4 (166) Nonfarm laborers 11 (4,583) 10 (1,037) 12 (1,035) Construction 2 (797) 1 (203) 3 (203) Manufacturing 15 (986) 13 (230) 13 (254) Allother 13 (2,800) 12 (605) 15 (577) Private household workers 96 (1,047) 93 (8 ~97 (152) Service workers, except private household 59 (12,391) 59 (2,054) 60 (2,776) Cleaning 39 (2,489) 30 (320) 37 (441) Food 66 (4,682) 62 (926) 68 (840) Health 89 (l,995) 86 (385) 86 (S61) Personal 76 (1,766) 78 (252) 81 (476) Protective 10 (1,459) 13 (171) 10 (459) Farmers, farm manager 11 (1,485) 7 (81) 11 (252) Fawn laborers, foremen 25 (1,264) 15 (239) 25 (261) Paid labor 16 (1,010) 14 (211) 16 (223) Unpaid family members 65 (254) 29 (28) 84 (38) NOTE: Numbers in parentheses are numbers of workers; they represent actual sample sizes and include both men and women. SOURCE: Unpublished data, U.S. Department of Labor, Bureau of Labor Statistics (1981~.
28 WOMEN'S WORK, MEN'S WORK TABLE 2-6 Sources of Employment Growth for Women, 1970-1980 Panel A Occupations in Which the Percentage Female Increased 20 Points or More, 1970-1980 Occupation Number of New Female Jobs Percentage Female 1970 Percentage Female 1980 Executive, administrative, and managerial occupations Management-related occupations, N.E.C. Professional and specialty occupations Inhalation therapists Foreign language teachers Recreation workers Public relations specialists Technicians and related support occupations Broadcast equipment operators Sales occupations Advertising and related sales occupations Sales occupations, other business services Administrative support occupations, including clerical Computer operators Production coordinators Samplers Insurance adjusters, examiners, and investigators Protective service occupations 12,006 24,963 2,432 6,308 37,199 24,040 33,526 126,439 192,037 85,479 449 70,483 12,238 Service occupations, except protective and household Bartenders95,480 Food counter, fountain, and related occupations88,063 Guides13,676 Farming, forestry, and fishing occupations Animal caretakers, except farm Graders and sorters, agricultural products Precision production, craft, and repair occupations Engravers, metal Machine operators, assemblers, and inspectors Typesetters and compositors Miscellaneous printing machine operators Total, experienced civilian labor force 16 years and over 20.1 28.6 34.2 45.4 26.6 22.1 20.5 8.4 33.9 20.2 20.4 29.6 22.2 21.2 56.8 32.9 26,781 3,246 4,074 24,779 17,903 13,957,618 30.7 52.0 15.7 16.8 23.8 38.0 53.5 56.5 59.4 67.6 48.8 44.0 41.6 37.4 59.1 44.4 44.8 60.0 42.3 44.3 81.1 57.2 59.0 78.6 38.1 55.7 52.9 42.6 puter and peripheral equipment operator and miscellaneous clerical worker). Accord- ing to Rytina and Bianchi (1984), women's participation increased most between 1970 and 1980 in those occupations that were be- tween 20 and 60 percent female in 1970. Some of these occupations became more fe- male-intensive (those more than 40 percent female), while others became more inte- grated (those less than 40 percent female). Women's participation also increased to a lesser degree in some occupations that were 80-90 percent male but failed to grow in those that were 90-100 percent male. Among all male-dominated occupations, women's representation increased more rapidly between 1972 and 1981 than during the 1960s (A. Belier, 19841. Prior to 1970, their representation increased in only one- fourth of the occupations in which men were overrepresented by at least 5 percentage points. However, between 1972 and 1981, their representation increased in more than half of those occupations as well as in most
SEX SEGREGATION: EXTENT AND RECENT TRENDS TABLE 2-6 Sources of Employment Growth for Women, 1970-1980 (continue]J Panel B Ten Detailed Occupations Providing Largest Number of New Jobs for Women, 197(~-1980 29 Occupation Secretaries Managers and administrators, N. E. C., salaried General office clerks Cashiers Registered nurses Teachers, elementary school Assemblers Child care workers, except private household Nursing aides Machine operators, not specified Number of New Female Jobs Percentage Female 1970 Percentage Female 1980 1,145,033 900,308 800,124 756,132 491,031 482,892 418,955 405,284 382,383 332,929 SOURCE: U.S. Department of Commerce, Bureau of the Census (1984a). male white-colIar occupations. According to Belier, in managerial and administrative oc- cupations, the increases ire the proportion female were large. More than 90 percent of these occupations became more female by 1981, although only 10 percent became more female during the 1960s. As noted above, Rytina and Bianchi (1984) corroborate the increased representation of women in man- agement. Male craft, operative, and laborer occupations remained highly segregated (Belier, 1984~; women's representation did not increase significantly in these occupa- tions through 1981. Women's increased representation in a wider range of occupations is displayed in Pane! A of Table 2-6, which presents the proportions of women workers in all detailed occupations in which women's representa- tion increased by 20 percentage points or more between 1970 and 1980. Fifteen of the 21 occupations listed in Pane} A shifted from predominantly (over 60 percent) male to welI- integrated occupations (less than 60 percent of either gender). Among these are man- agers, public relations specialists, broadcast equipment operators, protective service oc- cupations, bartenders, animal caretakers, and typesetters and compositors. Two of the 21 occupations that experienced substantial 97.8 lo.6 75.3 84.2 97.3 83.9 45.7 92.5 87.0 35.6 98.8 26.9 82.1 83.5 95.9 75.4 49.5 93.2 87.8 33.5 growth in their proportion female had only a slight majority female in 1970 but became heavily female-dominate`] by 1980: food counter, fountain, and related occupations, and graders and sorters of agricultural prod- ucts. Table 2-7 shows the 26 female-dominated occupations in which the representation of men increased 1 percentage point or more. In several occupations where few men have ventured, slow change is occurring, includ- ing registerer! nurses, prekindergarten and kindergarten teachers, cooks in private households, and textile and sewing machine operators. More dramatic shills have oc- curred in the categories of chief communi- cations operators, and hand engraving and printing occupations. The movement of men into female-dom- inated occupations and women into male- dominated occupations has contnbuted to the decline in sex segregation Luring the 1970s. As noted above, the decline was slowed by the growing numbers of women in large, heavily female-dominatecl occu- pations. All the occupations listed in Pane] A of Table 2-6 accounted for only 6.5 per- cent of the growth in female employment between 1970 and 1980. Pane] B of Table 2-6 lists the 10 occupations that provided
30 WOMEN'S WORK MEN'S WORK TABLE 2-7 Female-Dominated Occupations in Which the Percentage Male Increased One Point or More, 1970-1980 Percentage Percentage Percentage Percentage Occupation Male 1970 Male 1980 Occupation Male 1970 Male 1980 Professional specialty occu- pations Registered nurses Dieticians Speech therapists Teachers, prekindergar- ten and kingergarten Dancers Administrative support oc- cupations Chief communications operators Stenographers Interviewers Order clerks File clerks Billing, poshog, and calculating machine operators Mail preparing and paper handling machine operators Telephone operators Data entry keyers Private household occupa- tions Launderers and ironers 2.7 8.0 7.4 2.1 8.7 18.2 6.3 18.6 22.6 18.6 9.9 21.8 6.0 6.3 4.6 4.1 10.1 10.9 3.6 25.4 65.6 9.1 22.6 32.6 20.0 13.0 37.5 9.0 7.6 23.8 Cooks, private household Private household clean- ers and servants Service occupations, except protective and house- hold Waiters and waitresses Kitchen workers, food preparation Maids and housemen Hairdressers and cosme- tologists Public transportation 5.7 13.5 4.1 5.4 9.2 12.0 8.2 21.8 24.2 10.0 attendants 18.7 Precision production, craft, and repair occupations Electrical and electronic equipment assemblers Textile sewing machine operators Solderers and brazers Hand engraving and printing operations Total, experienced civilian labor force, 16 years and over 22.3 3.1 18.3 18.4 62.0 12.2 21.9 24.2 5.9 22.0 68.3 57.4 SOURCE: U.S. Department of Commerce, Bureau of the Census (1984a). the largest number of new jobs for women during the same period, accounting for ap- proximately 44 percent of the net increase in female employment. Seven of these oc- cupations are heavily female-dominated (over 75 percent female). The occupational cate- gory "secretaries," which is 98. 8 percent fe- male, alone created more new jobs than all occupations in Panel A combined. Some fe- male-dominated occupations have become more so; bookkeepers were 77. 7 percent fe- male in 1950 and 93 percent female 30 years later. Other clerical occupations that have become even more female-intensive since 1970 include biding clerks, cashiers, file clerks, keypunch operators, receptionists, legal secretaries, typists, and teacher's aisles (U.S. Department of Labor, Bureau of La- bor Statistics, 1981c; U.S. Department of Commerce, Bureau of the Census, 1973b, 1984a). While the general tendency for white women was to move out of female-domi- nated occupations, black women were less likely than white women to have done the same. Nevertheless their occupational sta- tus improved substantially as they moved to white collar jobs from lower-paid service alla laborer jobs. Many black women moved Dom lower-paying female-dominated occupa
SEX SEGREGATION: EXTENT AND RECENT TRENDS lions, particularly private household worker and to a smaller degree laborer (A. Beller, 1984), to clerical and other service occu- pations that were also female-dominated. In 1940, 70 percent of black women workers were private household workers; by 1981, just 6 percent worked in this occupational category, and fewer than 2 percent between the ages of 18 and 34 held such jobs (Mal- veaux, 1982b). Between 1973 and 1981, the proportion of black women in clerical oc- cupations increased from under 25 percent to almost 30 percent (U.S. Department of Labor, Women's Bureau, 1983; U.S. De- partment of Labor, Bureau of Labor Statis- tics, 1982b); in 1940 only 1 percent had held clerical jobs (Treiman and Terrell, 1975b). Occupations in which black women are to- day overrepresented include postal clerk, cashier, telephone operator, and duplicating machine operator. In contrast, black women are underrepresented among receptionists, bank tellers, and secretaries. Malveaux (1982b) notes that the clerical jobs in which black women are overrepresented have a behind-the-scenes character. While these changes among black women do not con- tribute to a reduction in the total amount of sex segregation, they represent an improve- ment in their position in the labor market and help to explain the sharp drop in the index of occupational race segregation among women shown in Table 2-1. Between 1977 and 1981, Hispanic women increased their representation in female-dominated white- colIar (primarily clerical) occupations, while their representation in female-dominated blue-collar jobs declined (Malveaux, 1982b). Is Resegregation Occurring? 31 grated occupations become resegregated, with members of one sex replaced by mem- bers of the other.6 Bank tellers and secre- tanes exemplify onginaDy male jobs in which women replaced men (Davies, 1975, 19821. Men have been hypothesized to leave for- merly male occupations when large num- bers of women are hired because of the ac- companying prestige Toss (Touhey, 1974) or declining real wages (Nieva and Gutek, 1981; Strober, 19841. As with secretaries and bank tellers, the shift from men to women may occur as the occupation is being restructured to provide, for example, less advancement to higher-level management, and becoming less attractive to men. Evidence regarding the prevalence of re- segregation is limited. Strober and her col- leagues (Strober and Lanford, 1981; Tyack and Strober, 1981) have traced the changing sex composition of the teaching profession, but (lo not attribute it to tipping. Pane} A of Table 2-6 includes a few occupations that shifted from being predominantly male to predominantly female. Insurance adjusters, examiners, and investigators, for example, were 29.6 percent female in 1970 and 60.0 percent female in 1980. Animal caretakers, except farm, changed from 30.7 percent fe- male in 1970 to 59.0 percent female in 1980. Shaeffer and Axe} (1978) point out that ma- chine operators in banks and technical em- ployees in insurance companies are both be- coming predominantly female, and Nieva and Gutek (1981) have suggested that com- puter programming may follow the pattern of bank tellers. When the occupation emerged 20 years ago, it was male-clomi- nated; in 1970, computer and peripheral ma- chine operators were 29.1 percent female. Ten years later, women's representation had Lee relative stability of the aggregate lev e! of sex segregation over time, coupled with several examples of large sex shifts in oc cupations, has led some observers to spec-6 The process is similar to residential "succession," . ,. . .In which segregated neighborhoods that are becoming u ate t tat integration or occupations Is a tem porary, unstable phenomenon. Perhaps, after reaching some "tipping point," inte integrated are eventually abandoned by the original residents to new residents of a different race or eth- nicity.
32 WOMEN'S WORK MEN'S WORK increased to 59.8 percent (although the du- ties have also changed), and Beller and Han (1984) conclude that the projected growth of this occupation will contribute to in- creased segregation. Greenbaum (1976, 1979) has argued, however, that that occu- pation was only briefly integrated, and, rath- er than tipping, it has split into two sex- segregated specialties: the computer oper- ator and some computer programming jobs are female-dominated, while higher-level programming and systems analyst jobs are male-dominated. Affirmative action needs to be thorough to counteract a potential tendency to reseg- regation. O'Farrell and Harlan (1982) point out that pressures to hire women may result in their concentration in and ultimately re- placement of men in formerly male-domi- nated entry-level jobs. Unless these jobs are on ladders that lead to positions that men continue to occupy, resegregation is likely. Resegregation can go in either direction. In one case, Kelley (1982) found that affirma- tive-action hiring in a manufacturing plant between 1972 and 1976 in general meant that white men supplanted white women in job classifications previously dominated by women. Some empirical evidence exists regarding a related issue: whether employers hire women in occupations that are declining in size or importance, usually because of tech- nological change. In at least half of the 53 nontraditional occupations in which women had made substantial gains between 1960 and 1970, their progress was due to the slow or negative growth of male employment (Reubens and Reubens, 19791. It has been alleged, for example, that AT&T hired wom- en for for overly male positions they planned to eliminate. As central office work was sim- plified by computers in that organization, women were moved into these jobs and en- countered little male resistance. Two stud- ies of AT&T (Hacker, 1979, Northrup and Larson, 1979) concluded that without care- fid planning, technological change could lead to a smaller number of newly segregated jobs. Feldberg and Glenn (1980) note sev- eral examples, in addition to the AT&T case, which suggest that women are hired ex- pressly as a transitional labor force in some instances associated with the introduction of electronic data processing. Whether some of the newly integrated occupations will remain integrated or whether substantial resegregation will occur cannot, of course, be predicted with any certainty. The next section presents scenar- ios of a variety of changes and their possible effect on the aggregate index of segregation. OCCUPATIONAL SEX SEGREGATION PROJECTED THROUGH 1990 The index of occupational segregation by sex declined by approximately 10 percent during the 1970s, but in 1981 it was still about 60. Can the changes that occurred during the 1970s be expected to continue, and, if so, at what rate? Are changes in the occupational structure likely to retard or ac- celerate further desegregation? As Table 2- 8 shows, the Bureau of Labor Statistics (BLS) (Carey, 1981) projects substantial growth in many heavily and historically female occu- pational categories, such as professional and practical nurses, nurse's aides, secretaries, bookkeepers, typists, and waitresses and waiters. These occupations are included in the 20 occupations in which employment growth, in absolute numbers, is expected to be greatest until 1990. If the proportions of these occupations that are female remain approximately constant, their growth will represent a demand for an additional 3.3 million female workers. Three of the occu- pations of largest predicted growth are cur- rently predominantly male but have expe- rienced recent growth in the participation of women: janitors and sextons, accountants and auditors, and guards and doorkeepers. Several other predominantly male occupa- tions that have not experienced substantial
SEX SEGREGATION: EXTENT AND RECENT TRENDS _ 33 TABLE 2-8 Twenty Occupations With the Largest Projected Absolute Growth, 1978-1990 Occupation Percentage Female 1980a Growth in Employment 1978-1990 (in thousands) Percentage Growth 1978-1990 Janitors and sextons Nurses' aides and orderlies Sales clerks Cashiers Waiters/waitresses General clerks, office Professional nurses Food preparation and service workers, fast food restaurants Secretaries Truck drivers Kitchen helpers Elementary school teachers Typists Accountants and auditors Helpers, trades Blue-collar workers, supervisors Booldceepers, hand Licensed practical nurses Guards and doorkeepers Automotive mechanics 17.3 87.5 71.1 86.6 89.1 80.1 96.5 66.9 99.1 2.2 66.9 83.7 96.9 36.2 NA 10.8 90.5 97.3 12.4 .6 671.2 594.0 590.7 S45.o 531.9 529.8 515.8 491.9 487.8 437.6 300.6 272.8 262.1 254.2 232.5 222.1 219.7 215.6 209.9 205.3 26.0 54.6 21.3 36.4 34.6 23.4 50.3 68.8 21.0 26.2 39.0 21.4 26.4 32.7 25.0 17.4 23.7 43.9 35.5 24.3 NA = not available. Approximate, due to the use of different occupational classifications in sources. SOURCES: Carey (1981:48) and U.S. Department of Labor, Bureau of Labor Statistics (1981c:Table 23~. growth in their proportion female (truck drivers, automotive mechanics, and helpers in the trades) are also expected to grow dur- ing the 1980s. Although the occupations projected to grow the most in absolute terms are nearly ad preclominantly male or female, several of the occupations that are expecter! to grow at the most rapid rate, shown in Table 2-9, are somewhat more integrated, particularly those that reflect advances in technology, such as computer programmers and com- puter systems analysts. Several others as- sociated with new technology, such as data processing machine repairers and office ma- ch~ne and cash register servicers, are now more than 90 percent male, but they may provide likely opportunities for women. Many of the other rapidly growing occu- pations reflect the continued tendency for the service and health sectors to grow; some of those occupations are fairly well inte- grated, while others are not. Some observ- ers suggest that as the United States econ- omy continues to restructure itself toward services of various kinds, sex-neutral occu- pations can be expected to grow in impor- tance. Others believe the growth of occu- cations associated with high technology may be overestimated by the Bureau of Labor Statistics. And recently its projections of substantial growth in the female-intensive clerical occupations have been questioned for underestimating the extent to which cler- ical work may be affected by automation. While there are several reasons for hypoth- esizing continued reduction in sex segre- gation associated with this predicted occu- pational growth, available data do not yet support them.
34 WOMEN'S WORK, MEN'S WORK TABLE 2-9 Twenty Occupations With the Largest Projected Growth Rates, 1978-1990 ~ _ Occupation Data processing machine mechanics Paralegal personnel Computer systems analysts Computer operators Office machine and cash register services Computer programmers Aermastronautic engineers Food preparation and service workers, fast food restaurants Employment interviewers r . ax preparers Corrections officials and jailers Architects Dented hygienists Physical therapists Dental assistants Peripheral electronic data processing equipment operators Child care attendants Veterinarians Travel agents and accommodations appraisers Nurses' Odes and orderlies Percentage Female 1980¢ 7.4 NA 25.1 63.2 5.6 28.4 1.2 66.9 48.7 NA 5.? 5.0 5iA 67.3 97.9 63.2 86.7 NA NA 84.3 NA = not available. Approximate, due to the use of different occupational classifications in sources. SOURCES: Carey (1981:Table 2~; Rytina (1982:Table 1). At issue in projecting the extent of oc- ~ cupational sex segregation are questions of Me number of new jobs created and the relative rates of growth in sex-neutral as op posed to sex-segregated occupations, as well as the rate of change of the sex composition within these occupations. Using the Bureau of Labor Statistics oc- cupational employment projections for 1990, Beller and Han (1984) project the index of sex segregation under various assumptions. Ike first set of projections, for the labor force as a whole, assumes first that the occupa- tional desegregation of the 1970s will con- tinue throughout the 1980s at a linear rate; the model is then permitted to take a logistic form. Ibe rationale for the assumption of linearity is that since it is easier for women to enter growing occupations than stagnant Growth in Employment 1978-1990 (in thousands) 93 38 199 148 40 150 41 492 35 18 57 40 31 18 70 26 20 17 25 594 Percentage Growth 1978-1990 147.6 132.4 107.8 87.9 80.8 73.6 70.4 68.8 66.6 64.5 60.3 60.2 57.9 57.6 57.5 57.3 56.3 56.1 55.6 54.6 or declining ones, the proportion of men in an occupation is a function of the initial pro- portion of men and the growth rate of the occupation. The logistic motley is employed for greater accuracy at the extremes, i.e., for occupations with very high degrees of sex segregation. Ibe results based on the linear mode] project a decline in the index of sex segregation of 1.7 points, from 61.7 in 1981 to 60. 0 in 1990, if it is assumed that the change in sex composition over time is the same for all occupations; and a decline of 1.3 points, to 60.4, assuming that the sex composition of each occupation is a Function of time. Using the logistic mode] for indi- vidual occupations, Belier and Han project a decline in the segregation index Dom 61. 7 in 1981 to 56.1 in 1990. Standardized to the 1981 occupational distribution (rather than
SEX SEGREGATION: EXTENT AND RECENT TRENDS - that projected for 1990 by the BLS), the drop in the index is slightly greater, indicating that the direction of the projected change in occupational distribution is toward more sex segregation, although the magnitude is small. In other words, the logistically pro- jected decline in the sex segregation index is likely to be partially offset by changes in the sizes of occupations. To project the index of sex segregation under varying assumptions, Beller and Han examine occupational segregation by work experience cohort for four different scenar- ios. Their most conservative projection as- sumes that there will be no further changes in the sex composition within each occu- pation as it ages, although as the labor force ages, less segregated cohorts replace older, more segregated ones. On the basis of these assumptions only a slight decline in the in- dex of sex segregation is projected: from 64.2 in 1977 to 62.1 in 1990. The latter figure is slightly above the actual 1981 index, reflect- ing the trend toward a more sex-segregated occupational distribution projectec! by the BLS. Beller and Han argue that the decline of 2.1 points in the index of sex segregation can be taken as a lower bound; they expect a decline by 1990 of at least that much. On the basis of the assumption that the rate of change in the sex composition of occupations for the entering cohort will be the same be- tween 1977 and 1990 as it was between 1971 and 1977 (a period of considerable change) they project an index of 57.3 in 1990. This decline of 6.9 percentage points comes clos- est to the logistic projection. In what they term their most optimistic scenario, they assume that affirmative action, attitudes, and other factors will continue to change at the same rate as during the 1970s, so that all cohorts experience declining sex segrega- tion between 1977 and 1990. The index de- clines 11.7 points to 50. 0 on the basis of this assumption, if the rate of change between 1977 and 1990 is half that between 1971 and 1977; it declines nearly 20 points to 42.2 if the rate of change between 1977 and 1990 is double what it was from 1971 to 1977, figures they consider to be an upper bound. Beller and Han argue that the rate of oc- cupational desegregation during the 1970s is too great to be maintained during the 1980s because the female labor force is unlikely to grow rapidly enough; all their projections imply higher female labor force participation rates and higher growth in the female share of the labor force than the BLS projects. Hence, they do not believe that the lower levels of occupational segregation they pro- ject for 1990 are likely to occur. Despite these limitations, their results are instruc- tive in that they set upper limits on the amount of desegregation likely to occur dur- ing the 1980s. They point out that the di- rection of public policy can affect the amount of fixture change. SUMMARY AND CONCLUSION The amount of occupational segregation by sex continues to be substantial. In 1981, the index of occupational segregation by sex was 62, indicating that more than 60 percent of all women or men would have to move to occupations dominated by the opposite sex for segregation across occupations to be entirely eliminated. Additional segregation occurs across industries and firms. Men and women are disproportionally distributed across firms and industries even when the occupational mix they employ is taken into account. For example, even in integrated occupations, like payroll accounting clerk or assembler, some firms and industries tend to hire more women and others more men. In one study (Bielby and Baron, 1984), 231 of 391 California firms were totally sex-seg- regated; men and women worked in none of the same job categories. The current situation is of greatest inter- est in the context of recent trends. Decen- nial census data since 1940 show a small decline in the total amount of occupational sex segregation among whites and a larger
36 decline among other races. These two trends have produced a convergence in levels of occupational sex segregation between whites and nonwhites. Since World War II occu- pational segregation by race has declined much more rapidly than by sex. One com- ponent of this improvement has been black women's movement out of service occupa- tions into clerical occupations. But within a sex-segregated occupational structure, race segregation persists. For example, black women are now overrepresented among postal clerks and telephone operators rela- tive to their proportion in the labor force. The sex segregation index dropped more during the 1970s than during previous dec- ades, and the decline was most pronounced among younger workers. During the past decade men became slightly more likely to work in a few heavily female occupations, such as office machine operator or telephone operator, and women's representation has increased in several predominantly male oc- cupations, including attorney, bank official, computer programmer, baker, bus driver, and bartender. Their numbers remain small in some of the occupations that women en- tered or increased their representation in during the 1970s (for example, coal miner, engineer), but their participation rate has increased markedly. Women's representa- tion also increased among several predom- inantly female occupations that grew during WOMEN'S WORK, MEN'S WORK the 1970s, including bookkeepers, billing clerks, cashiers, and keypunch operators. Although relatively substantial change oc- curred in the index of occupational sex seg- regation in the 1970s, the most likely pro- jections for 1990 suggest that the rate of change throughout the 1980s will be much slower. The index fell by approximately 10 percent in the 1970s, from 68.3 in 1972 to 61.7 in 1981, according to Beller (1984), and from 67.7 in 1970 to 59.3 in 1980, according to Bianchi and Rytina (19841. In contrast, various likely projections of the job segre- gation index range from 56.0 to 60.0 in 1990. Only slight further declines are anticipated, primarily because occupations that are pre- dominantly male or female are expected to grow more than those that are relatively in- tegrated. And, of course, we do not have information that would permit us to estimate probable changes in job segregation at the establishment level. The next two chapters provide a basis for assessing the likelihood of additional change. Chapter 3 examines the evidence for several explanations that have been offered for sex segregation in employment and consequent- ly offers some guidance for developing pol- icies for reducing segregation. Chapter 4 re- views a variety of attempts to reduce segregation in employment, education, and training, assesses their effectiveness, and provides further policy guidance.