Y I ‘ I . I l I IN I f l 4 I H I I 4 HI 1 l ' J l U l 1 l AN EXAMINATION 0‘: JOB LmV’“ "N f )SUQC'ISSlQN INTERN? SN ‘ 35 NE EC' JEAN {ML SER‘M’SCE Mai A? 32 1.116814? 115339 AN EXAMINATION OF JOB LEAVING AND SUCCESSION PATTERNS IN THE MICHIGAN CIVIL SERVICE, JANUARY TO DECEMBER, 1969 BY Catherine Begnoche Smith A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Sociology 1971 ACKNOWLEDGMENTS Thanks go to Dr. Thomas Conner and Mr. Harry Perlstadt for advice at every step of this study; also to Dr. Philip Marcus for helpful criticism in completing this paper. Data for this paper were collected on a grant to the School of Labor and Industrial Relations from the Michigan Department of Civil Service for study of the Civil Service. ii TABLE OF CONTENTS Page INTRODUCTION AND PROPOSITIONS . . . . . . . l MICHIGAN CIVIL SERVICE . . . . . . . . . 8 Static Analysis: Distribution of Males and FemaleS--January , 1969 o o o o o o 0 11 Analysis of Job-Changing Between January and December, 1969 . . . . . . . . . l7 BIBLIOGMPHY O O O O O O O O O O O 0 40 iii LIST OF TABLES Table Page I-A. Female Percentage of Job Category and Hierarchical Characteristics of Job . . . . . . . . . . . . l3 I-B. Female Percentage of Job Category and Range of Pay Level . . . . . . . . 13 II. Actual and Expected Male Incumbents, by JOb Levels 0 O O O O O O O O O 15 III. Percentage Female at Hierarchical Levels by Majority Sex of Occupation . . . . 16 IV. Job-Changing by Sex and Percentage Female in Job Between January and December, 1969 . . . . . . . . . 19 V. Method of Job-Changing by Sex of Worker and Percentage Female in Job . . . . 20 VI. Sex of Appointee by Sex of Incumbent and Percentage Female in Job . . . . 24 VII. Expected and Actual Sex of Appointee by Sex of Incumbent and Percentage Female in JOb O O O O O O O O O O O O 25 VIII. Transition Matrices for High and Low Female Occupations . . . . . . . . 27 IX-A. Sex of Appointee by Sex of Incumbent at Each Level of Low Female Occupations . . . . . . . . . . 29 IX-B. Sex of Appointee by Sex of Incumbent at Each Level of High Female Occupations . . . . . . . . . . 30 iv XI. Expected and Actual Percentage of Females in Newly Created Jobs, by Level, in Low Female Occupations . . . . . . Expected and Actual Percentage of Males in Newly Created Jobs, by Level, in High Female Occupations . . . . . Change in Percentage of Jobs Held by Incumbents by Percentage Female in Occupation and Sex of Incumbent . . . Page 33 34 35 INTRODUCTION AND PROPOSITIONS The purpose of this paper is to determine the job-leaving and succession pattern of a particular bureau- cratic organization, in the hopes of establishing the pre- dictive ability of certain propositions about the processes involved in succession when sex of workers is a variable. Research on labor turnover patterns has shown that formal organizations with paid employees are likely to have fairly constant rates of quitting (Hill, 1951). Since most bureaucratic organizations contain members of several occupations in some hierarchical arrangement, the over-all organizational quit-rate depends upon not only the organ- ization's competitive labor market position but also occu- pational variations in rates of quitting. It has been established empirically that there are stable differences in the job-quitting rates characteristic of various occu- pations (Parnes, 1954 and Ladinsky, 1967). In an attempt to explicate the basis for these occupational differences, Thompson, Avery and Carlson sug- gest that the opportunities for advancement perceived by members of different occupations is a major determinant of job-quitting decisions (Hall, 1969). Workers tend to change jobs in accord with a rational career strategy de- signed to take best advantage of the perceived opportunity structure. But occupations calling for greater skill or education (such as managerial or professional occupations) usually have more opportunity for advancement within a given organization than do less skilled occupations. Con- sequently such high-skilled occupations are likely to have lower rates of quitting than do less skilled jobs (Pavalko, 1971 and Parnes, 1954). Even within the same organization, then, a comparison of different occupations should reveal fairly constant differences in their typical rates of quitting the organization. Once a bureaucracy recognizes that it has a con- tinual problem of job-quitting, it is likely to establish procedures to deal with the problem of appointing replace- ments. Since resignations at any point above the lowest hierarchical level are likely to create a chain of open- ings (White, 1970), the number of workers affected by the quit-rate may be very large. The total job turnover rate is thus a function of the quit-rate since it includes vacancies produced by decisions to promote and transfer workers to fill other vacancies created by resignations. Since the average annual quit—rate ranges between 25 and 35 per cent (Parnes, 1954), organizations must be prepared to decide fairly quickly who will be hired and who will be moved. According to Weber, the only rational criteria for these decisions are based on the relative seniority or achievement of candidates (Weber, 1947). Thus age, sex, race and many other characteristics of workers usually are not rational criteria for promotion or hiring. But while seniority is not difficult to measure,the basis for decisions about achievement may be ambiguous, espe- cially in the case of hiring new workers. Above a rather low hierarchical job-level, achievement tends to become a subjective judgment by superiors (V. Thompson, 1964). In fact, admission to upper hierarchical levels has been described as a process of sponsorship in which evaluators tend to select candidates most like themselves in terms of characteristics like age, sex, race and social background (Caplow, 1954 and Grusky, 1970). Thus each successively higher level of a hierarchy is likely to be increasingly composed of individuals who are homogeneous in these characteristics. The primary concern of this study is to verify propositions relating relating the occupational variables affecting quit-rates and the organizational dispositions about replacement of workers at each level to the distri- bution of females throughout the organization. There is already research about the relationship of sex of worker to the variables affecting quit-rates and succession within organization. For one, analysis of census data shows that there is a distinctive occupational distribution of workers according to sex. Males predominate in mana- gerial, professional and skilled occupations while females are concentrated in clerical, service and unskilled cate- gories (US Census, 1960). Women are usually located in the occupational categories that Thompson et_§1. described as low-ceiling jobs--those with little Opportunity for ad- vancement because of the low level of skills required, little opportunity to exercise discretion and a low ratio of managers to workers (Hall, 1969). Combining the likelihood that women are concen- trated in low-ceiling jobs with the high probability al— ready mentioned that workers in such jobs will quit, we expect to find females quitting more often than males. There is a rational basis for the higher female quit-rates since females are concentrated in occupations whose oppor- tunity structure is limited and little can be gained by the strategy of staying in the organization. Proposition 1: Occupations filled with a high proportion of females have higher quit-rates than those with high proportions of males. In order to establish the next proposition, relat- ing sex and quit-rates within an occupational category we must consider other factors besides career strategy that may affect the decision to quit. In our culture, the dis- tribution of family responsibilities requires that the wife be primarily responsible for the care of home and children, making outside employment of secondary importance to women (Rossi, 1964). This lack of emphasis on em- ployment for money makes it likely that females will devi- ate from the typical career patterns described in the literature (Miller and Form, 1964). In fact the whole concept of career, which implies continuous employment aimed at advancement, may not apply to a large portion of the employed female population. According to Epstein, most women see employment as an activity that is some- times necessary to supplement the family income, not as a central part of their lives (Epstein, 1970). Thus cul- tural pressures Operating upon women may make them more likely than men to quit--or to leave the labor force entirely--even within the same occupation. Even those females who are pursuing a career may perceive a different opportunity structure for them than for males in the same occupation and this may raise the female quit-rate also. PrOposition 2: Within the same occupation, females have higher quit- rates than males. To consider why females might see a different op- portunity structure for themselves than for males in the same occupation, we must consider the existing authority structure in the bureaucracy. To the extent that males already dominate the upper hierarchical levels, bearing in mind that evaluators tend to select candidates like them- selves and that successively higher ranks become more and more homogeneous, we would expect occupations dominated by males to select males as new appointees. Proposition 3: Predominantly male occupations tend to fill vacancies with males; strongly female occupations tend to hire females. Since even occupations numerically dominated by males may have some female occupational members, we might expect that the positions filled by these females are somehow considered not to violate principles of homo- geneity. Thus, when females leave the organization, their positions may be offered to other females without further contradiction of the tendency towards homogeneity. Proposition 4: Females hired in a male-dominated occupation are more likely to replace other females than to replace males. When Caplow's hypothetical statement that suc- cession is a process of sponsorship of candidates similar to the sponsor is recalled, we see that females may face certain restrictions in advancement that males do not en- counter. Caplow describes two mores of American society that greatly complicate a woman's career strategy: it is considered disgraceful for a man to be directly sub- ordinated to a woman (except in family matters) and inti- mate group ties are expected to be composed of either sex but not both (Caplow, 1954). Thus a work-group or sponsor- protégé relationship involving both sexes would be avoided. Since the upper levels of most occupations (even those usually considered female occupations like nursing and library science (Simpson and Simpson in Etzioni, 1969) are usually dominated by men, at the present, the norm dis- couraging mixed-sex sponsorship relationships helps to maintain the sexual homogeneity of managerial levels of most occupations. Proposition 5: Upper levels of occupations tend to be dominated by males. Thus males hired in any occupation are more likely to enter at higher hierarchical levels than do females and this tendency is then maintained by informal adherence to the principles of homogeneity. Since most males will be replaced by other males, the tendency expressed in Proposition 5 will be maintained despite job-turnover. MICHIGAN CIVIL SERVICE The information on turnover in the Michigan Civil Service was copied from Civil Service files containing records of job vacancies during 1969 and decisions about filling them. With permission from the Civil Service, Dr. Thomas Conner and Mr. Harry Perlstadt hired the author and three other students to code certain information from these files: the name of persons incumbent and reason for leaving the job; the job title and number; the location of the vacancy (department, section, county, etc.); and the name, social security number, sex, birth date and method of appointment of the successor. Data on approximately 24,000 vacancies were recorded, although the files were not always complete. This particular paper involved about 20 per cent of the total number of job-vacancies recorded for 1969. Vacancies pertinent to the propositions were selected, keypunched and sorted to provide the data used in the analysis of succession patterns. Information about the sex of incumbents of each level of the relevant occupations was culled from the official Civil Service breakdown of employees by job title, level, sex and race for January, 1969. The characteristic relevant for this paper was sex of worker. The relevant propositions are thus predictions that differences will be found between male and female workers in occupational distribution, in career strategy and in turnover and succession patterns. Specifically, different career strategies are appropriate for certain occupations, so that those employing mostly males are expected to have lower job—leaving rates, for both males and females, since males are presumably concentrated in highly skilled jobs that require continuous labor-force participation. Positions in predominantly male occu- pations are expected to be more advantageously placed in terms of the dimensions Thompson considered critical to mobility. Predominantly female occupations are expected to show less homogeneity (in terms of sex) at upper hier- archical levels than do male occupations due to differ- ences between the sexes with respect to career strategy and because of organizational bases for promotion. The hypothesis that males and females have differ- ent occupational distributions has been empirically sup- ported by census data on employment. These differences might be expected to be less clear-cut in the civil ser— vice, where sex is no longer a formal requirement for the great majority of jobs, but even here there is some 10 indication that informal sex-typing of jobs does occur. Examining the January, 1969 classification of state em- ployees according to occupational category, it becomes obvious that hundreds of classifications are entirely male. The titles of these jobs agree with what census data has shown to be predominantly male jobs. These titles include a variety of managerial, professional and highly technical jobs--for example, "administrative officer," "budget analyst," "land appraiser," "highway planner," “food in- spector," "mechanical engineer," and most state police jobs. Many other categories, including "librarian," and several nursing categories, are almost exclusively femi- nine, just as census data demonstrates. The differen- tiation of occupations, on the basis of sex, seen in the Michigan Civil Service is consistent with that found in the labor market as a whole. On the other hand, there is some evidence to suggest that workers who choose a career that is inconsistent with sex-typing norms are very likely to seek jobs in the civil service, where sex is commonly supposed to be less of a barrier to hiring than in the private sector of the economy (Kosa and Coker, 1965). For example, the 1969 classification shows a higher percentage of female physicians and lawyers in the Michigan Civil Service than is seen in the census data. Given that the distribution of the sexes in Michigan Civil Service categories may be somewhat less ll distinctive than that observed in the general population, it may be important to ask whether propositions based on investigations of non-government employment will be upheld in this case. The civil service may be a critical test of the propositions. If they are verified under civil ser- vice employment they should certainly hold true in private organizations less favorable to female employment. Static Analysis: Distribution of Males and Females--January, 1969 In order to compare and contrast the patterns of male and female employment, certain job categories were selected as being potentially most useful sources of in- formation. These jobs were selected on the basis of sev- eral criteria: they contained at least three hierarchical steps (so that possible career stages could be identified), these levels employed at least thirty workers and, further, this number of workers included at least 10 per cent but no more than 90 per cent females in January of 1969. Fol— lowing these rules, 36 job categories were selected from the total of hundreds of categories. These 36 job clusters accounted for 41 per cent of the 43,182 workers employed by the Michigan Civil Service in January of 1969. The job categories were then divided according to whether more or less than 50 per cent of their employees were female. The 10 categories employing more than 50 per cent female were then labeled "High Female" jobs. 12 The remaining 26 jobs employed between 10 and 50 per cent females and were thus named "Low Female" categories. This division was made in order to allow investigation of the proposition that high and low female occupations should exhibit differences in career pattern of incumbents. As a basis for the comparison of these patterns, certain characteristics of the High Female jobs were compared with those of Low Female categories. Characteristics considered relevant included: number of steps in the job hierarchy, the hierarchical level at which the job begins, the highest level one can reach in this job category and the pay range for the various levels of the job. As Table I shows, High Female and Low Female jobs do exhibit some consistent differences in terms of these variables. Job level is the major indicator of rank in the organizational hierarchy. Civil Service officials explain that a job's level, ranging between 01 and 21, indicates not only rough differences in authority and salary but also, more importantly, indicates the degree of responsibility required of the employee. Job level indi- cates the importance to the organization of decisions made routinely by workers at that level. Jobs that require decisions with far-reaching effects will have high levels attached to them. The level is thus a property of the job, not of the incumbent. Thus the differences in level seen in the comparison of High and Low Female jobs in 13 TABLE I-A.--Female percentage of job category and hier- archical characteristics of job. Job Level Female % > 5 Steps in of Job Hierarchy Starts Ends Base N Below 07 Below 11 Low Female 92% 35% 19% 26 jobs High Female 90% 50% 40% 10 jobs TABLE I-B.--Female percentage of job category and range of pay level. Minimum Maximum Female % Base N < $6400 > $8500 < $11,000 Low Female 31% 23% 23% 26 jobs High Female 50% 10% 50% 10 jobs Table I deserve attention. Fully half of the High Female job categories begin at a level below 07, which is the lowest professional level and the level at which most col- lege graduates enter. Only a third of the Low Female cate- gories can be entered below this level. Further, the Michigan Civil Service defines level 11 as the lowest mana- gerial level. While 40 per cent of the High Female cate— gories end below this point, only a fifth of the Low Female jobs do not extend into the managerial levels. 14 A consideration of pay ranges points out other contrasts between High Female and Low Female job clusters. While the pay scale in one-half of the High Female jobs begins below level 020 (i.e., below $6400 per year), only 30 per cent of the Low Female jobs have such low starting salaries. These differences in pay range are to be ex- pected, since they are mainly reflections of the differ- ences in training required to start below or at level 07. The range of jobs in the Low Female categories not only emphasizes professional training but also, from the job titles, appears to be more technical and more managerial in nature. The High Female jobs, on the other hand, are less likely to require extensive formal training and are more likely to be caretaking or nurturant in nature. Another way of looking at the hierarchical distri- bution of the sexes is to consider not only the charac- teristics of High Female jobs but also to examine the com- position of hierarchy in terms of sex of incumbents. To do this, adjacent levels were grouped so as to produce roughly equivalent numbers of workers, then the expected number of males at each level was calculated, based on the knowledge that males hold 41 per cent of the jobs in the 36 categories. The hypothesis that distribution among levels is independent of sex was rejected following a Chi- square test with a .01 level of significance. Although the distribution is not independent of sex, it is impossible 15 to specify the nature of the relationship between sex and job level at this point. TABLE II.--Actual and expected male incumbents, by job levels. Level E (Male) Actual (Male) Total N 01, 02 41% 34% 2373 03 41 31 3764 04 41 34 2576 05, 06 41 48 2028 07, 08 41 40 2459 09 41 43 2164 10 to 21 41 71 2497 Total 41% of 17861 The finding that males are more likely to be found at upper hierarchical levels, when applied to our division of the occupations under scrutiny into High Female and Low Female jobs, leads us to expect that male-dominated occu- pations will exhibit more homogeneity with respect to sex at upper levels than at lower levels of the occupations. In contrast, we would expect High Female occupations to be less homogeneous with respect to sex at upper levels than are Low Female occupation, since males seem to dominate these levels in all occupations. As Table III shows, this proposition was confirmed by a comparision of the female percentage of each job level in the two occupational cate- gories. As expected, the upper ranks of Low Female 16 TABLE III.--Percentage female at hierarchical levels by majority sex of occupation. Low Female High Female Level % N % N 01 17.8 708 88.0 1613 02 89.0 35 100.0 17 03 57.8 520 70.3 3234 04 55.0 418 67.8 2158 05 40.9 457 69.0 1132 06 21.9 41 57.5 388 07a 35.4 471 66.7 1707 08 36.9 84 54.6 97 09 31.1 440 66.9 1624 10b 34.5 472 51.2 460 11 20.5 488 41.9 241 12 17.0 182 37.2 43 13 11.2 196 30.2 43 14 14.0 56 5.0 19 15 11.7 76 10.0 10 16 11.0 102 [22.2]* 9 17 2.4 41 [50.0] 4 18 0.0 12 [ 0.0] 4 19 4.8 21 [ 0.0] 2 20 no jobs no jobs [ 0.0] l 21 no jobs no jobs [ 0.0] 1 Total 4924 12805 *Brackets indicate small N for cell. a.07 is the first professional level. b.11 is the lowest managerial level. 17 occupations were almost exclusively male, while the highest ranks of the High Female category were not dominated by the majority sex--female--and hence can be considered less homogeneous with respect to sex than the Low Female occu- pations. Females comprise a larger proportion of the upper-level employees in High Female occupations than in Low Female ones. However, it is possible that, since a lower proportion of the jobs in the High Female category are managerial in level (only 2.9 per cent of High Female as compared to 24.2 per cent of all Low Female jobs are at or above level 11) a female who occupies a job in a Low Female occupation may really have as good or better a chance of reaching upper levels than a woman in the High Female category, despite the tendency toward male homo- geneity in Low Female occupations. This examination of the employment data for Jan- uary of 1969 has confirmed that males and females are not randomly distributed throughout the hierarchy and that male occupations become more homogeneous with respect to sex than do female ones at higher job levels. Analysis of Job-Changing Between January and December, 1969 In order to examine the propositions regarding employee turnover and succession patterns, we must examine the pattern of job changes in both High and Low Female categories for the ensuing year. Several of the 36 18 occupations considered in the first section of this paper experienced little turnover during that year, so the final list of job clusters to be studied was shortened. Only those clusters with at least 30 job changes calling for replacement of employees were considered. Only 5 High Female jobs were in the final list--employment and claims interviewers, general clerks, account clerks and execu- tives, clinical social workers and public welfare workers. Of Low Female jobs, 7 were included--employment counselors, driver licensing examiners, account examiners, repro- duction machine workers and supervisors, administrative analysts, management systems analysts, and computer pro- grammers. Sex of incumbent was determined from the names coded and from correlation of name of incumbent with the same name as an appointee in another position, when pos- sible. Less than 1.0 per cent of all cases were rejected because sex was indeterminable. The propositions derived from earlier research pre- dict that females are more likely to leave than males-- presumably reflecting different ideas of careers--and that High Female occupations should have higher quit-rates, since there are fewer high-level jobs for the professional to attain and since there are more semi-skilled jobs (hence, according to other research, more quitting) than in the Low Female category. 19 A preliminary examination of the data gives some support to these propositions. Females did experience more total job-changes than males and the difference between the two was most obvious in the categories of quitting and taking leave of absence. These differences between males and females occurred in both High and Low Female job categories, as Table IV clearly points out. TABLE IV.—-Job-changing by sex and percentage female in job between January and December, 1969. % Female Sex of Remained in Base N in Job Incumbent January Job Low Female Male 82.3% 100% = 805 Female 73.6% 100% = 284 High Female Male 71.0% 100% = 1555 Female 39.0% 100% = 2033 According to Table IV, males are more likely than females to remain in the same job throughout the year, even when the dominant sex of the job category is constant. Also, when sex of incumbent does not vary, we see that there is still a difference between overall job-leaving rates in High and Low Female occupations. This finding lends some credence to the notion that these two occupational types may call for different career strategies, since a strategy 20 is put into practice when the strategist decides to move or to stay. Since the moves counted in this table include not only resignations but also changes in position within the organization, the propositions regarding female pro— pensity to quit require an analysis of the types of job changes involved. We need to know whether female job- changing includes a large component of resignations or if, on the other hand, this component is but one of many equally important modes of job-changing. Table V shows the breakdown of job-changing into major categories. TABLE V.--Method of job-changing by sex of worker and percentage female in job. % Type of Change Female Sex of Same Quit On Base N in Job Worker Job Leave Trans- Pro- Other fer motion Low Female Male 82.3% 5.5 .l 8.7 3.0 .4 100% = 805 Female 73.6 8.7 5.0 9.7 1.3 1.7 100% = 284 High Female Male 71.0 10.9 2.0 12.5 2.3 1.3 100% = 1555 Female 39.0 22.2 7.3 16.2 9.0 6.0 100% = 2033 21 Examination of Table V reveals that the cells rep- resenting female percentages of job changes are almost always larger than the corresponding cells in the male rows, except for promotions in High Female occupations. This trend is especially clear in the case of quitting and taking leave of absence--in three of those four compari- sons, the female percentage is at least twice as large as the male percentage. The only category where the male percentage is larger than the female percentage is the case of promotion in occupations with less than 50 per cent female composition. Thus it appears that females are more likely to change jobs within the organization than are men, by almost every method given, and they are also more likely to leave the organization, either by quitting or by taking a leave of absence. These findings substantiate the propositions suggested by previous re- search, namely, that females are more likely to change jobs in general and to quit the organization in particular than are male workers. Further differences are apparent when High Female jobs are compared to Low Female occupations. Regardless of their sex, incumbents of High Female jobs were more likely to quit, more likely to take a leave of absence and thus more likely to be transferred than are incumbents of Low Female occupations. This contrast makes sense in the light of the analysis of High and Low Female occupations 22 presented already, in which High Female occupations were shown to start at lower hierarchical levels, to have fewer managerial levels open and to receive lower pay. Since Low Female occupations may be characterized as more pro- fessional and managerial than High Female jobs, the lower quit-rate in the former is in accord with the findings of Broom and Jones, that professionals are more likely to have stable careers with relatively few job changes than are less-skilled workers (Broom and Jones, 1969). Also, since only 2.9 per cent of all High Female jobs are above level 10 (as compared to 24.2 per cent of Low Female positions that are above that level) and since that 2.9 per cent is dominated by males, the high quit-rates agree with Vroom and MacCrimmon's hypothesis that the decision to quit an organization is a function of expectation of future Oppor- tunities (Vroom and MacCrimmon, 1968). While no one in a High Female occupation is likely to move higher than level 10, female workers are at a greater disadvantage. The great majority of all female promotions in High Female jobs occurs at levels 01 and 03 in clerical jobs, so the rather high promotion figure presented in Table V does not indi- cate the true state of the opportunity structure. Despite the lower percentages given for promotions in Low Female occupations, there may actually be more opportunity for ad- vancement simply because the ratio of managers to managed is so much higher. The lower quit-rates in Low Female 23 jobs may be seen as an indication that the opportunity structure is perceived as being more promising than is the case in the High Female category. Observation of the different occupational distri- butions of males and females and a recognition of several contrasting characteristics of male and female employment and turnover patterns may lead one to speculate about the processes producing these contrasts. How do males and females become so distinctively distributed throughout the bureaucracy? What effect does the introduction of male workers have upon a strongly female occupation? When males leave such jobs, will they be replaced by other males or not? Not all of these questions can be answered by this study since it is a consideration of an ongoing process at a particular period in time. The last question is the most amenable to examination here. Since both Caplow and Grusky have suggested that there is a tendency towards homogeneity of workers in any given job cluster and that an incumbent is likely to be succeeded by an individual similar to him, we shall hypothesize that both High and Low Female occupations will show a tendency to— wards homogeneity of members (e.g., more likely to hire member of one sex than other). At the same time, if a member of the minority sex is hired, he will be found re- placing an incumbent of the minority sex more frequently than one would expect from chance. 24 To test this hypothesis, a chart was made of the 1902 job changes that occurred in the 12 sample jobs during 1969, where the sex of both incumbent and new appointee could be determined. TABLE VI.--Sex of appointee by sex of incumbent and percentage female in job. Sex of Appointee 1,333.23. N Male Female Low Female Male 67.7% 32.3 100% = 142 Female 52.7 47.3 100% = 74 High Female Male 46.0 54.0 100% = 450 Female 17.3 82.7 100% = 1236 As the proposition predicted, a tendency towards homo- geneity was obvious-~females were hired most frequently in High Female jobs where the incumbent was female. The tendency was not quite as strong in the Low Female cate- gory but, even there, males replaced males two-thirds of the time. The second part of the hypothesis, that females who are hired in Low Female occupations will probably be replacements for other females, is hard to determine from this table. Another table was set up to compare the actual percentage of female appointees in Low Female jobs and of 25 appointees in High Female jobs, compared with expectations based on chance. TABLE VII.--Expected and actual sex of appointee by sex of incumbent and percentage female in job. - Sex of Appointee % Female Sex of Female in Job Incumbent Base N Expected % Actual % Low Female Male 26.1 ‘ 32.3 100% = 142 Female 26.1 47.3 100% = 74 High Female Male 56.7 54.0 100% = 450 Female 56.7 82.7 100% = 1236 As the hypothesis predicted, females replaced themselves far more often than one would expect by chance in both High and Low Female jobs. The effect of incumbents of the Opposite sex seems to be to introduce more random choice when the incumbents are replaced. This suggests that once a member of the minority sex has held a job in an occupation dominated by members of the opposite sex, sex is not so likely to be considered relevant in selecting a successor if that incumbent leaves. It is also possible that certain jobs in an occupation are already considered 26 suitable for either sex and that this table indicates a continuation, not a change, in policy. The next problem to consider is the need to under- stand how turnover affects the pre-existing pattern of sex differences in occupational distribution when all the factors and propositions that have been considered are combined. Since Hill has found support for the hypothesis that turnover (especially that due to quitting) is the result of a quasi-stationary process that reflects a fairly constant organizational policy, we assume that the Michigan Civil Service will probably continue to have a large degree of turnover. If we further assume that the pattern of replacement of workers detailed in Tables VI and VII is unlikely to change drastically, then it is pos- sible to use Markov chain theory to predict the effect of future turnover upon the distribution of the sexes in the organization. We know that the distribution of females and males in High Female occupations is (.567, .433) and in Low Female jobs it is (.260, .740) so these constitute the initial probability vectors for the states of the two chains. (The states are "Female Incumbent" and "Male Incumbent.") We have already established empirically the probability that members of each sex in each category will change jobs and the contingent probability that an ap- pointee of a particular sex will be hired. Thus we can 27 produce a transition matrix describing the probability of going from each state to the next state. The probability of the process going from Female to Female includes two paths--the female stays in the job or she is replaced by another female. The Female to Male process requires that she leave and be replaced by a male. The sum of probabil- ities Female-Female and Female-Male must equal 1.0. TABLE VIII.--Transition matrices for high and low female occupations. HIGH FEMALE: rSEX OF SEX AT TIME 21 no = (.567, .433) INCUMBENT TIME 1 Female Male Female .895 .105 P: Male .157 .843 k ‘4 HI = no P = (.575, .425) LOW FEMALE: 'SEX AT SEX AT TIME 21 ”O = (.250, .740) TIME 1 Female Male Female .860 .140 P: Male .057 .943 L N1 = flo P = (.266, .734) 28 For both High and Low Female occupations, multi— plication of the initial probability vector, no, and the transition matrix produced a new probability vector, “1, that differs by less than .01 from the initial one. This indicates that the process of turnover and replacement of workers does not greatly affect proportions of males and females in High and Low Female jobs. Although the pattern of turnover in the Michigan Civil Service does not appear to have much effect on the distribution of males and females with respect to occu- pational category, it may have an effect upon the dis- tribution of the sexes within each category. Since Table III pointed out that females are likely to be concentrated at lower levels of the hierarchy in both High and Low Fe- male occupations, we can examine the succession data by level for both categories to see if this distribution is affected by turnover and replacement of workers. There were too few cases of turnover at levels 02, 04, and 06 in either category for these to be included. The low Fe— male category also had very few cases of turnover at level 01, 05. or 08. Despite small cell sizes, levels 11 and 12 were used, since they were the only examples of turnover at a managerial level. These two levels were grouped together. 29 TABLE IX-A.--Sex of appointee by sex of incumbent at each level of Low Female occupations Appointee Level Incumbent Male Female Base N 03 Male 75.0% 25.0 100% = 12 Female 40.0 60.0 100% = 5* 07 Male 41.6 58.4 100% = 48 Female 46.1 53.9 100% = 39 09 Male 82.4 17.6 100% = 34 Female 66.7 33.3 100% = 18 10 Male 66.7 33.3 100% = 18 Female 62.5 37.5 100% = 8* 11 and 12 Male 100 0 100% = 22 Female 100 0 100% = 2* *Small N. 30 TABLE IX-B.--Sex of appointee by sex of incumbent at each level of High Female occupations. Appointee Level Incumbent Male Female Base N 01 Male 54.5% 45.5 100% = 55 Female 9.9 90.1 100% = 333 03 Male 40.0 60.0 100% = 30 Female 5.5 94.5 100% = 259 05 Male 50.0 50.0 100% = 6* Female 10.9 89.1 100% = 19* 07 Male 38.1 61.9 100% = 221 Female 26.7 73.3 100% = 431 08 Male 50.0 50.0 100% = 16* Female 33.3 66.7 100% = 12* 09 Male 53.5 46.5 100% = 84 Female 28.5 71.5 100% = 137 10 Male 65.2 34.8 100% = 23 Female 26.5 73.5 100% = 34 11 and 12 Male 71.4 28.6 100% = 14* Female 54.5 45.5 100% = 11* *Small N. 31 This set of tables suggests that the hierarchical distri- bution of males and females that existed in January is likely to remain in December, despite turnover at nearly every level. Just as the proposition predicted and the January distribution suggested, females are most likely to be appointed at or below level 09, in both High and Low Female occupations. As an earlier proposition predicted, High Female occupations will continue to be less homo- geneous with respect to sex than Low Female ones, espec— ially at upper hierarchical levels. The tendency ob- served in Table V1 for incumbents of the majority sex to be replaced by appointees of the same sex was also ob- served in most of the levels, except for the special case of males replacing females at upper levels of High Female occupations and another case that was not predicted--that females frequently replace males at level 07 in Low Female jobs. Since this is the lowest professional level, it is possible that the surge of females entering at this point in 1969 may lead to a gradual increase in the percentage of females at higher professional and managerial levels of these occupations in the future, since the pool of poten- tial promotees will have a greater proportion of female workers. Since the data on job succession indicates that we should not expect much change in the distribution of males and females in either High or Low Female occupations, 32 it is pertinent to ask whether the pattern of appointment of individuals to newly created jobs in those categories also reinforces the existing distribution. The proportion of new jobs is very high in these categories, with the number of Low Female jobs increasing by 18.1 per cent and the number of High Female positions increasing by 35.6 per cent during 1969, so the proposition that new jobs will perpetuate the existing occupational and hierarchical distribution is of major importance in predicting the stability of the existing pattern of dis- tribution. We expect to find that males and females will be hired, on the whole, in proportion to their represen- tation in each occupational type and, in particular, that males will fill jobs created above level 09 while females will be concentrated in jobs at or below that level. As Table X-A shows, females were hired to new jobs in about the proportion one would expect by chance, knowing only the percentage of females in the Low Female occu- pations as a whole. There is also some support for the second part of the proposition, that female appointees will be hired more Often than one would expect by chance at the lower levels of the hierarchy, but this support is not very strong. Even at levels 11 and 12, females were hired to fill 20 per cent of the new jobs. It is possible that, while the filling of new jobs is not greatly affecting the 33 total proportion of females in the Low Female occupations, it may be having some effect in the redistribution of that female proportion throughout the hierarchy. TABLE x—A.--Expected and actual percentage of females in newly created jobs, by level, in Low Female occupations. Female Level Base N Expected % Actual % 03 26.1 33.3* 100% = 9* 05 26.1 21.4* 100% = 14* 07 26.1 47.5 100% = 61 08 26.1 0.0* 100% = 1* 09 26.1 20.6 100% = 58 10 26.1 12.5* 100% = 16* 11 26.1 22.7 100% = 22 12 26.1 20.0* 100% = 10* 13 26.1 0.0* 100% - 6* Total 26.1 28.4 100% =197 *Small N. According to the data shown in Table X-B, males were hired for newly created jobs in High Female occupa- tions, on the whole, about half as often as one would expect knowing that these occupations are 43.3 per cent male before any new jobs were filled. Thus one effect of the very high number of new jobs relative to the 3588 34 TABLE X-B.--Expected and actual percentage of males in newly created jobs, by level, in High Female occupations. Male Level Base N Expected % Actual % 01 43.3 17.9 100% = 396 03 43.3 11.5 100% = 615 05 43.3 100.0* 100% = 1* 07 43.3 29.2 100% = 137 08 43.3 50.0* 100% = 10* 09 43.3 27.7 100% = 54 10 43.3 51.6 100% = 31 11 43.3 90.5 100% = 21 12 43.3 66.7* 100% = 9* 13 43.3 50.0* 100% = 4* Total 43.3 19.9 100% =1278 *Small N. established jobs may be a reduction in the male proportion of High Female occupations. But the distribution of new male appointees throughout the job hierarchy is pretty much as expected--males are more likely to be hired above level 09 than we would expect by chance, even though their overall hiring rate was much lower than expected. Con- sideration of the data on new jobs alone leads one to con- clude that High Female occupations as a whole are becoming 35 more female in makeup and that the hierarchical distri- bution of males and females is accentuated by the pattern of hiring males to fill high level jobs. The much higher rate of increase of positions in High Female as compared to Low Female occupations suggests that High Female occu- pations will employ a greater proportion of employees relative to Low Female jobs by the end of 1969 than it did in January. Table XI shows that this is indeed the case . TABLE XI.--Change in percentage Of jobs held by incumbents by percentage female in occupation and sex of incumbent. % Sex of Percentage of Jobs Female Incumbent January, 1969 December, 1969 Low Female Male 17.2% 14.7 Female 6.1 5.4 High Female Male 33.2 27.2 Female 43.5 52.7 Base N 100% = 4677 jobs 100% - 6152 jobs As predicted, High Female occupations did increase their share of the total jobs in the 12 job categories, going from 76.7 per cent of the January jobs to 79.9 per cent of the December jobs. This increase was entirely due to increases in the female proportion of this category. 36 Although the Low Female occupations contained a lower per- centage of total jobs by the end of the year, the female component of those occupations grew from 26.1 per cent in January to 26.9 per cent in December. Thus if the level of creation of new jobs continued to be as high as that occurring in 1969, we would expect to find that all of these 12 occupational groups, but especially the High Female ones, would continue to become more female in com- position. Since, however, conversation with Michigan Civil Service employees does suggest that 1969 was a year with an exceptionally high rate of new job creation, it is likely that the transition matrix already developed will have predictive power for the next year. The initial probability vector for January of 1970, however, must take into account the effect of the new jobs. Thus, instead of the expected vector for female and male incumbents in High Female jobs (this was predicted to be [.575, .425]), the actual probability vector for January of 1970 is (.659, .341). Also, for Low Female occupations the initial vector for 1970 is now (.268, .732) instead of (.266, .734) which was originally predicted as the probability that incumbents would be (female, male) in January, 1970. Multiplying these initial probability vectors times their respective transition matrices, as on Table VII, we find that the distribution of females and males in High Female occupations is expected to be (.643, .357) in December, 1970. Like- wise, in Low Female occupations it should be (.272, .728), 37 indicating that the Low Female category is not likely to experience much change in the proportions of males and females in future. SUMMARY Not only do females in High Female occupations have the highest quit-rates of all, the male quit-rates in this- category are also significantly higher than their rate in Low Female occupations. Evidently the basis for these high quit-rates must involve the opportunity structure of the occupation itself. Since even males, who have the best chance of reaching the managerial level, have high quit- rates here there is evidently some difference in the pos- sibility of advancement when High and Low Female occupations are compared, as Table I-A suggested. The high quit-rate in High Female occupations is an indication that workers are using a rational career strategy, since such a strategy calls for frequent job changes in occupations with a low ratio of managers to workers. The fact that females have a higher quit-rate than males in both occupational categories supports the idea that there are even more pressures operating on females to quit than just career requirements. Since more than twice as many females as males quit the High Female jobs, one must ask whether females are, for some reason, responding more dramatically than men to occupational pressures or whether the impact of cultural and structural pressures on women is greatest in women whose training and education 38 39 place them in High Female occupations. Perhaps women who take Low Female jobs and perceive a somewhat more advan- tageous opportunity structure there are somehow better able to resist the cultural pressures that induce women to quit in order to raise children and play traditional female roles located in the home. The third proposition, that predominant sex of occupational members is a good predictor of sex of new appointees, was largely supported by the data. The strongest indication of this trend was shown in Table VI, where it appears that incumbents are most likely to be replaced by appointees of the same sex if that sex is the numerically dominant sex of the occupation. On the other hand, once a job has been held by a member of the "minor- ity" sex (of that occupation), the sex of the replacement is hard to predict, at least below managerial levels. Thus, while homogeneity appears to be an important prin- ciple governing the selection of new workers, that prin— ciple is ignored once it is likely to become increasingly irrelevant. In the long run, this should eventually pro- duce considerably less homogeneity in the sub-managerial levels of both categories of occupations. More rational criteria of achievement may be becoming more important than sex of candidate—-once the organization sees that a member of the "minority" sex can function effectively if given the chance, it may be more 40 willing to take a chance and hire workers with less con- sideration of the variable of sex. Analysis of the effects of new jobs upon the dis— tribution of males and females in these occupations showed a slight increase in the percentage of females in both categories, especially at levels below 09. Despite the decreasing importance of homogeneity as a principle for the replacement of workers, proposition 5 was still supported. Males still dominated the upper levels (above 09) of both job categories, although there was not any significant increase in the extent of that domination. The creation and filling of new jobs did not greatly affect the distribution of males and females in Low Female occupations, but in High Female jobs there was some intensification of the trend toward a high proportion of males at managerial levels. Either the de-emphasis on sex as a criterion for appointment has not yet been ex— tended to managerial appointments or else the number of women qualified to hold these positions has not increased fast enough. Since women appear to be able to follow a career strategy only by resisting strong pressures against such moves, pressures that seem to be most successful in affecting the quit-rate of women in High Female jobs, perhaps these pressures are also successful in preventing women in this category from preparing themselves for a real career, including the step into management. BIBLIOGRAPHY BIBLIOGRAPHY Broom, L.,and Jones, F. "Career Mobility in Three Societies." American Sociological Review, 34-5, p. 650. Caplow, Theodore. The Sociology of Work. University of Minnesota Press, 1954, p. 71. Epstein, Cynthia. Women's Place. Berkeley: University of California Press, 1970. Etzioni, Amitai, ed. The Semi-Professions and Their Organization. New York: The Free Press of Glencoe, 1969, pp. 222-4, 230. Grusky, Oscar,and Miller, D. The Sociology of Organiza- tions. New York: The Free Press of Glencoe, 1970, p. 446. Hall, Richard. Occupations and the Social Structure. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1969, pp. 315-8. Hill, J. M. "A Consideration of Labour Turnover as the Resultant of a Quasi-Stationary Process." Human Relations, August, 1951. Kosa, J., and Coker, R. "The Female Physician in Public Health, Conflict and Reconciliation of the Sex and Professional Roles." Sociology and Social Research, 49-3, 1965, p. 294. Ladinsky, Jack. "Occupational Determinants of Geographic Mobility." American Sociological Review, 32-2. Miller, Delbert, and Form, William. Industrial_Sociology. New York: Harper & Row, 1964, pp. 541-5. Parnes, Herbert. Research on Labor Mobility. New York: Social Science Research Council, 1954. 41 42 Pavalko, Ronald. Sociology of Occupations and Professions. Itasca, 111.: F. E. Peacock Publishers, Inc., 1971, p. 7. Rossi, Alice. "Equality Between the Sexes." Daedalus, 93-2, 1964, pp. 607-15. Thompson, Victor. Modern Organizations. New York: Alfred A. Knopf, 1964. U.S. Bureau of the Census. 1960, Vol. 1, Table 202. Vroom, Victor, and MacCrimmon, Kenneth. "Toward a Stochastic Model of Managerial Careers." Administrative Science Quarterly, 13-1, 1968. Weber, Max. The Theory of Social and Economic Organi- zation. New York: The Free Press of Glencoe, 1947. White, Harrison. Chains of Opportunity. Cambridge, Mass.: Harvard University Press, 1970. I ll! 1" |'.|‘l' lr‘l‘l ‘ I [III- I -0 NOV 12 ‘9'.