number of siblings and educational mobility

12
Number of Siblings and Educational Mobility Author(s): Judith Blake Source: American Sociological Review, Vol. 50, No. 1 (Feb., 1985), pp. 84-94 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2095342 . Accessed: 02/10/2013 15:21 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review. http://www.jstor.org This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PM All use subject to JSTOR Terms and Conditions

Upload: judith-blake

Post on 12-Dec-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Number of Siblings and Educational Mobility

Number of Siblings and Educational MobilityAuthor(s): Judith BlakeSource: American Sociological Review, Vol. 50, No. 1 (Feb., 1985), pp. 84-94Published by: American Sociological AssociationStable URL: http://www.jstor.org/stable/2095342 .

Accessed: 02/10/2013 15:21

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access toAmerican Sociological Review.

http://www.jstor.org

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 2: Number of Siblings and Educational Mobility

NUMBER OF SIBLINGS AND EDUCATIONAL MOBILITY*

JUDITH BLAKE University of California, Los Angeles

With regard to educational attainment has American society been relatively open for those from small families and relatively ascriptive for those from large ones? This analysis by separate sibsizes suggests that the influence of the father's education on the son's schooling has, indeed, been conditional on sibsize. Moreover, the interaction itself has differed according to the level of the -son's education. Replication for three age groups, 20-34, 35-49, and 50-64, maintains the inverse relation between sibsize and mobility. In addition, for small and medium sibsizes, we find an increase in mobility over time, but, for large ones, virtually no change in the slopes for the father's education. The educational mobility of American men thus seems to have benefited from a dual set of trends-decreases in the effect (slope) of social origins on educational attainment among small and medium sibsizes, and a compositional shift in the population from large to smaller families.

Ever since Arsene Dumont advanced his hy- pothesis of "social capillarity" sociologists and demographers have been intrigued by the idea that family size may be negatively related to social mobility. Dumont's hypothesis con- cerned a possible inverse relation between men's career mobility and the number of chil- dren they had-a formulation that continues to generate research under the rubric of the "so- cial mobility and fertility hypothesis" (Du- mont, 1890; Westoff, 1953, 1981; Berent, 1952; Bresard, 1950; Westoff et al., 1961; Westoff et al., 1963; Tien, 1965; Perrucci, 1967; Zimmer, 1981a, 1981b). Surprisingly, however, no sys- tematic investigation has addressed the conse- quences for men's generational mobility of the number of siblings in their families of orienta- tion. The pioneering work of Blau and Duncan (1967) and later of Featherman and Hauser (1978), although including number of siblings (sibsize) as an exogenous variable, did not question whether men's dependence on their social origins differed according to the size of family from which they came. Yet, a large body of research concerning the effects on in- dividuals of number of siblings confirms the hypothesis that the more siblings men have, the more diluted are their parents' interper- sonal and material resources, and the less aus-

*Direct all correspondence to: Judith Blake, School of Public Health, University of California, Los Angeles, CA 90024.

The support of the Russell Sage, Foundation and the Fred H. Bixby Foundation is gratefully acknowl- edged. Inas El-Attar and Jennifer Frost Bhat- tacharya assisted with data processing and analysis. The Data Archive Library of the UCLA Institute of Social Science Research made the survey data avail- able. The author is indebted to the two anonymous reviewers for their comments.

picious are the men's life chances (e.g., Blau and Duncan, 1967; Duncan, 1965, 1967; Feath- erman and Hauser, 1978; Sewell et al., 1980; Blake, 1981). It thus seems logical to expect that number of siblings would be related to men's opportunities relative to their fathers' social status. Has there actually been an in- teraction between number of siblings and mo- bility such that American society has been rel- atively open for those from small families and relatively ascriptive for those from large ones?

As a prelude to answering this question, a brief review of some relevant findings of the Blau and Duncan/Featherman and Hauser re- search is useful. Blau and Duncan's analysis of the 1962 survey, Occupational Changes in a Generation, discovered that American men's educational attainment is the principal known determinant of their occupational status, and that the educational and occupational status of men's fathers affects sons' occupational achievement primarily through the sons' edu- cation. Using a replicate survey, Occupational Changes in a Generation 1973, Featherman and Hauser's analysis reaffirmed the im- portance of education as an influence on occu- pational status. Featherman and Hauser also found a greater influence of father's education than father's occupation on the son's educa- tional attainment than was true in the 1962 study. As is well known, the primary focus of both studies was on whether American society is becoming more, or less, ascriptive and rigid. Is it easier or harder in the so-called postin- dustrial society for a man to rise above the educational and occupational status of his fa- ther? The 1973 study concluded that social ori- gins are less influential on men's educational attainment than was the case earlier in the century (Feathermen and Hauser, 1978:241). The authors declined to speculate about why

84 American Sociological Review, 1985, Vol. 50 (February:84-94)

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 3: Number of Siblings and Educational Mobility

SIBLINGS AND EDUCATIONAL MOBILITY 85

American society appears to be less ascriptive as it becomes more industrially "mature."

In attempting to examine the interaction between sibsize and mobility, this paper con- centrates on the effects of social origins on educational attainment as the principal depen- dent variable. The restriction of the dependent variable to education, for the purpose of this initial analysis, is based on the crucial gatekeeping role of educational attainment for occupational status, as found in Blau and Dun- can, as well as in Featherman and Hauser. As already noted, the term "sibsize" refers to the number of an individual's siblings. A person who has no siblings (an only child) is desig- nated as coming from "sibsize 0," a person with one sibling comes from "sibsize 1," and so on.

In this research, the computer tapes from the widely used Occupational Changes in a Gen- eration 1962 and 1973 surveys (OCG 1962 and 1973) provided basic data, together with the tapes from the General Social Survey 1972- 1983 (GSS 72-83). This initial analysis is re- stricted to white males only. Both OCG sur- veys were national probability samples of men in the working ages (age 20-64 in OCG 1962 and 20-65 in OCG 1973), and the GSS is an annual national probability sample of the adult population of all ages and both sexes con- ducted by the National Opinion Research Center of the University of Chicago. The an- nual data from the GSS have been pooled for this analysis.

THE PREVALENCE OF HAVING MANY SIBLINGS AND A POORLY EDUCATED FATHER

Although relatively low fertility per white woman (an average of fewer than four children per woman) has been characteristic of the population since early in the century, most of the adult men in the surveys used here had families of orientation of four or more children, and approximately a quarter of these men came from families of seven or more children (Table 1). This disparity between fertility per woman and the sibsizes of their children is due, of course, to the fact that each new generation is

made up, disproportionately, of the children of the most fertile women. Hence, people who come from small families (1-2 children) are proportionately rare, and those who come from large families (5 children and over) are propor- tionately very prevalent. For example, using United States Census data, Preston has esti- mated that the mean number of children born alive to women 45-59 in 1940 was 2.6, but the mean sibsize of their children was 5.2 (Preston, 1976).

Since there is also a strong negative associa- tion between sibsize and father's educational attainment, large proportions of the men in the three samples studied here came from families in which the father had attained no education, an incomplete grade school education, or only eight years of grade school (Table 2). For example, in OCG 1962, 66.4 percent of the fathers had grade school educations or less (a combined effect of the prevalence of large sib- sizes and the negative relation of sibsize and father's education), whereas only 46.3 percent of the fathers of those men who had one or two siblings were so poorly schooled. Comparable figures for OCG 1973 are 55.5 and 39.3, and for GSS 72-83 they are 48.3 and 33.1

The prevalence of large sibsizes and the con- comitant prevalence of having educationally humble parental origins mean that questions concerning whether the Blau and Duncan status attainment model works differently for different sibsizes cannot be dismissed lightly. If the effect (slope) of father's education on son's education is conditional on sibsize, and particularly if this effect increases substantially as sibsize goes up, then the society has been much more "stratified" for the multitude from large families than for the minority from small ones.

In the present analysis, the effect of father's educational attainment will be used as an indi- cator of the impact of men's social origins on their own schooling. Choice of father's educa- tional attainment was dictated not only by its obvious theoretical relevance to the son's edu- cation, but because in two of the three studies (OCG 1973 and GSS 72-83) the relative im- portance of the father's education compared to the other predictors in the model was clearly

Table 1. Percentage Distribution by Sibsize, White Males, United States, Occupational Changes in a Gener- ation, 1962 and 1973, General Social Survey, 1972-83

Sibsize

Survey 0 1 2 3 4 5 6 + Total (N) Occupational Changes in a Generation, 1962 6.0 12.9 15.3 14.1 11.5 9.6 30.6 100 (18,485) Occupational Changes in a Generation, 1973 6.6 16.3 18.3 15.0 11.2 8.7 23.8 100 (26,963) National Opinion Research Center,

General Social Survey, 1972-83 5.8 15.8 17.4 15.0 11.3 8.8 25.9 100 (6,060)

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 4: Number of Siblings and Educational Mobility

86 AMERICAN SOCIOLOGICAL REVIEW

Table 2. Percentage Distribution by Father's Education and Sibsize Groupings, White Males Age 20 and Over from Intact Families, Occupational Changes in a Generation, 1962 and 1973, General Social Survey, 1972-83a

Sibsize Groupings

Father's Education 0-1 2-3 4-6 7+ Total

OCG 1962 None 1.7 3.9 7.6 11.4 6.2 Grade School 44.6 54.9 66.8 72.2 60.2 High School Incomplete 14.5 12.2 9.6 6.5 10.6 High School Complete 21.5 16.7 9.6 6.5 13.3 College Incomplete 7.6 5.7 3.4 2.0 4.6 College Complete or Higher 10.1 6.6 3.0 1.4 5.1 Total 100.0 100.0 100.0 100.0 100.0 (N) (2,735) (4,388) (4,180) (3,171) (14,474)

OCG 1973 None 1.4 2.3 5.6 10.2 4.3 Grade School 37.9 44.3 59.7 70.4 51.2 High School Incomplete 12.5 13.0 10.6 7.2 11.3 High School Complete 27.0 22.5 14.7 7.8 19.0 College Incomplete 9.7 7.6 4.4 2.6 6.4 College Complete or Higher 11.4 10.3 5.0 1.7 7.7 Total 100.0 100.0 100.0 100.0 100.0 (N) (5,055) (7,439) (5,705) (3,676) (21,875)

GSS 1972-83 None 1.0 1.7 4.1 10.5 3.6 Grade School 32.1 37.4 52.6 65.0 44.7 High School Incomplete 14.6 13.4 10.6 6.3 11.8 High School Complete 27.0 25.8 18.1 11.4 21.7 College Incomplete 10.8 9.2 7.0 3.2 8.0 College Complete or Higher 14.5 12.5 7.6 3.6 10.2 Total 100.0 100.0 100.0 100.0 100.0 (N) (945) (1,444) (1,101) (666) (4,156) a In order to minimize misreporting of father's education (and missing data among those brought up in

fatherless families), only those respondents who grew up in intact families are included here.

outstanding. Moreover, although in OCG 1962 father's occupational status (Duncan index) was relatively more important than father's education for some sibsizes, the standardized coefficients differed little throughout.

SIBSIZE AND EDUCATIONAL MOBILITY

In order to examine the effects of fathers' edu- cational status on men's educational attain- ment among different sibsizes, separate re- gressions for four sibsize groupings (0-1, 2-3, 4-6, and 7 and over) were computed for OCG 1962 and 1973, and for GSS 1972-1983. The purpose of grouping the sibsizes was to allow the reader to compare results for small, medium, and large sibsizes while avoiding cumbersome detail. Since the relation of un- grouped sibsizes to educational attainment is linear, grouping the data is not inappropriate statistically.

Three indicators of son's educational attain- ment were used as the dependent variables: total years of education; total years of graded schooling (0-12); and the proportion of high school graduates going to college. These indi-

cators were used by Featherman and Hauser for their analysis of the OCG samples in Op- portunity and Change. Analyses were also performed using years of grade school (0-8) and years of high school (9-12) as dependent variables. These results will not be presented, but will be referred to when appropriate. Total years of education was based on men age 25 and over, the remaining measures were based on men age 20 and over. None of the analysis is presented in weighted form (although the tabulations were made both with and without weights), because the unweighted results for the white subsamples are virtually identical to the weighted ones. Since using weights has potential disadvantages for the computation of unbiased estimates, the reader should be aware that our results are unweighted (Kish, 1965).

With the exception of proportion of high school graduates going to college, for which a logistic regression was used, ordinary least squares regression was employed throughout. Since my concern was for the effect (slope) of the father's education on the son's education, controls were introduced for major confoun- ders available in all of the data sets-the re- spondent's age, sibsize (within sibsize group-

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 5: Number of Siblings and Educational Mobility

SIBLINGS AND EDUCATIONAL MOBILITY 87

ings), farm background (a dummy variable), and the father's SEI (Duncan index). In order to minimize misreporting of the father's educa- tion, as well as to overcome the problem of missing data on the father's education among those from broken families, the analysis was restricted to men who were reared in intact families. Finally, it is worth noting that this analysis does not attempt to be a full explana- tion of educational attainment. No analysis using these data sets can be anything but lim- ited. For example, intellectual ability is an ob- vious influence on educational attainment but does not appear in the data sets. The reader should bear in mind, however, that even in analyses where ability has been included, plus a barrage of additional variables, the influence of the father's education on the son's educa- tion has maintained its relative position as the most important of parental-background influ- ences (Sewell et al., 1980). In fact, in the Wis- consin data, the father's education was found to be relatively as important as the son's ability in predicting the son's education.

We turn now to our research question, does the effect of father's education on son's educa- tion differ among men from different sibsizes? Beginning with total years of son's educational attainment, Table 3 shows the unstandardized coefficients and standard errors (in par- entheses) for the father's education by sibsize

groupings in OCG 1962 and 1973, and GSS 72-83. Parenthetically, it is worth noting that mean educational attainment declines as sib- size increases. The decrease in achievement as sibsize increases has been discussed in detail by Blau and Duncan (1967) and by Blake (1981). A drop in education with increasing sibsize will be found in all of the tables that follow.

It is evident from the data relating to years of total education in Table 3 that, in all of the studies, the slope for father's education in large families is steeper than the slope in small ones. However, although statistically significant, the interaction is neither so large nor so regular as to allay reasonable skepticism concerning its importance. The question arises, therefore, of whether disaggregating total years of education into components would provide more insight. As we shall see, the answer is affirmative be- cause an additional interaction is operating. We will now consider years of graded school- ing (0-12) and proportions of high school grad- uates going to college.

For years of graded schooling (Table 4), the slopes for father's education rise consistently and markedly as sibsize goes up. In OCG 1962, the regression coefficient (father's education) for sibsizes 0-1 is .065, and among those in sibsizes 7+ it is .197. In OCG 1973, the compa- rable rise is from .081 to .291, and in GSS

Table 3. Effect of Father's Education on Respondent's Total Years of Education by Sibsize Groupings, White Men from Intact Families, Occupational Changes in a Generation, 1962 and 1973, and General Social Survey, 1972-83a

Sibsize Groupings

Total Years of Education 0-1 2-3 4-6 7+ Total

OCG 1962 Beta 0.162 0.204 0.179 0.226 0.197

(0.018) (0.014) (0.014) (0.017) (0.008) Y 13.1 12.0 10.6 9.5 11.2

(3.16) (3.21) (3.18) (3.33) (3.46) N 2,283 3,664 3,694 2,866 12,507

OCG 1973 Beta 0.188 0.210 0.303 0.350 0.261

(0.014) (0.012) (0.014) (0.017) (0.007) Y 13.5 12.7 11.1 9.7 11.9

(3.05) (3.16) (3.55) (3.73) (3.62) N 4,013 5,631 4,736 3,262 17,642

GSS 1972-83 Beta 0.193 0.205 0.274 0.233 0.222

(0.029) (0.023) (0.027) (0.035) (0.014) Y 14.3 13.4 12.0 10.5 12.7

(2.89) (3.04) (3.33) (3.59) (3.45) N 795 1,173 924 603 3,495 a Controls for respondent's age, farm background, sibsize (within the categories in the table), and father's

SEI were included in the regression. Total years of education are calculated for respondents age 25 and over. Betas shown are the unstandardized regression coefficients for father's education. Standard errors and standard deviations are in parentheses beneath the regression coefficients and means for years of education. The differences between the slopes of small (0-1) and large (7+) sibsizes are significant at the .05 level for OCG 1962 and 1973.

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 6: Number of Siblings and Educational Mobility

88 AMERICAN SOCIOLOGICAL REVIEW

Table 4. Effect of Father's Education on Graded Schooling (0-12) by Sibsize Groupings, White Men From Intact Families, Occupational Changes in a Generation, 1962 and 1973, and General Social Survey, 1972-83a

Sibsize Groupings Years of Graded Schooling 0-1 2-3 4-6 7+ Total OCG 1962

Beta 0.065 0.100 0.132 0.197 0.124 (0.009) (0.008) (0.010) (0.013) (0.005)

Y 11.4 10.9 10.1 9.2 10.4 (1.54) (1.88) (2.34) (2.70) (2.30)

N 2,661 4,259 4,025 3,031 13,976 OCG 1973

Beta 0.081 0.102 0.211 0.291 0.166 (0.008) (0.007) (0.010) (0.014) (0.005)

Y 11.5 11.3 10.4 9.4 10.8 (1.70) (1.93) (2.70) (3.23) (2.49)

N 4,590 6,745 5,409 3,537 20,281 GSS 1972-83

Beta 0.027 0.077 0.152 0.163 0.102 (0.009) (0.009) (0.016) (0.025) (0.007)

Y 11.8 11.6 10.9 9.9 11.2 (0.87) (1.31) (1.96) (2.67) (1.83)

N 917 1,384 1,059 649 4,009 a Controls for respondent's age, farm background, sibsize (within the categories in the table), and father's

SEI were included in the regression. The regressions were calculated for respondents age 20 and over. Betas shown are the unstandardized regression coefficients for father's education. All of the differences between the slopes of small (0-1) and large (7+) sibsizes are significant at the .01 level. In addition, seven of the nine comparisons between slopes of adjacent sibsizes are significant at the .05 level.

72-83 the rise is from .027 to .163. Thus, when graded schooling is used as the dependent vari- able, father's education appears to be compar- atively irrelevant among men from small sib- sizes, but coming from a large family greatly potentiates men's dependence on their fathers' backgrounds. For example, in the OCG 1973 data, the difference between having a college- educated father and a grade-school-educated one is 2.33 years of graded schooling for a man with 7+ siblings; whereas for a man with 0-1 siblings, the comparable difference is .65 years of graded schooling.

Similar analyses (not shown), using years of grade school (0-8) and years of high school (9-12) as the dependent variables, indicate that the most marked interaction is at the grade school level (0-8) in each of the data sets. At this level, the slopes for father's education among men from the largest families (7+ sib- lings) are six times the slopes for men from the smallest families in all of the studies. There is a lessening of the interaction at the high school level, apparently because of the pre-existing sharp selection in grade school among men from larger families. This selection will be doc- umented shortly.

Turning now to the proportion of high school graduates going to college (Table 5), we find, for all of the surveys, that the effect of the father's education (the coefficient for the logis- tic regression) is greatest for those from small

families (men with 0-1 or 2-3 siblings) and least for those from large ones (men with 7+ siblings)-just the opposite of the situation at the graded-schooling level. By the time that men have completed high school, those from large families have been drastically selected, whereas those from small families have hardly been selected at all. It is thus not surprising that the father's education has more effect on the college-going chances of those from small families than on those from large ones. The residue of men from large disadvantaged families who have come this far are "sur- vivors," and apparently have attributes that facilitate their further education, but those from small families only begin to be winnowed out between high school and college. We are now in a position to understand why the in- teraction between the effect of father's educa- tion and sibsize was not more marked when total years of education was the dependent variable. This interaction operates in opposite directions for lower and higher levels of edu- cation, and hence most of the interaction effect is cancelled out when all levels of education are combined.

In order to document the selectivity that oc- curs at each educational level by sibsize, we refer the reader to Table 6, which shows the percentages completing various levels of edu- cation by sibsize groupings. The first row, in the case of each study, shows the percentage of

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 7: Number of Siblings and Educational Mobility

SIBLINGS AND EDUCATIONAL MOBILITY 89

Table 5. Effect of Father's Education on College Schooling by Sibsize Groupings, White Men from Intact Families, Occupational Changes in a Generation, 1962 and 1973, and General Social Survey, 1972-83a

Proportion of High School Sibsize Groupings Graduates Going to College 0-1 2-3 4-6 7+ Total OCG 1962

Beta 0.085 0.092 0.058 0.022 0.074 (0.015) (0.012) (0.014) (0.020) (0.007)

Y 0.59 0.49 0.38 0.32 0.47 N 2,169 2,963 1,996 1,054 8,182

OCG 1973 Beta 0.096 0.111 0.094 0.032 0.094

(0.012) (0.010) (0.012) (0.017) (0.006) Y 0.63 0.53 0.40 0.30 0.50 N 4,029 5,410 3,350 1,558 14,347

GSS 1972-1983 Beta 0.139 0.107 0.093 0.080 0.106

(0.026) (0.020) (0.024) (0.032) (0.012) Y 0.69 0.62 0.52 0.42 0.59 N 840 1,177 717 316 3,050 a Controls for respondent's age, farm background, sibsize (within the categories in the table), and father's

SEI were included in the regression. The proportion of high school graduates going to college is based on respondents age 20 and over, and was computed from a logistic regression, with those graduating from high school and not going to college coded 0 and those going to college coded 1. Betas shown are the unstan- dardized regression coefficients for father's education. The differences between the slopes of small (0-1 or 2-3) sibsizes and large (7+) are significant at the .05 level for OCG 1962 and 1973, but not for GSS 1972-83.

all men completing 8 years of schooling by sibsize groupings. Not only do the percentages drop as sibsize increases, but those from sib- sizes 7+ suffer a precipitous drop even at this low level of schooling, and those at sibsizes 4-6 are well below those from small and medium sibsizes. The second row shows the percentage of all men who graduated from high school. Again, not only do these percentages decline as sibsize increases, but they decline very markedly as one goes from sibsizes 2-3 to 4-6 and 7+. For example, in OCG 1962, 79.4 percent of those from sibsizes 0-1 had gradu- ated from high school, but only 32.7 percent of those from sibsizes 7+ had done so. Compara-

ble figures for OCG 1973 are 87.3 and 43.9, and for GSS 72-83 they are 88.7 and 41.7. Finally, if we ask what percentage of the men, by sib- size, ever went to college, we find that in OCG 1962 it is 46.6 for sibsizes 0-1 and 10.4 for sibsizes 7+; for OCG 1973 the respective fig- ures are 54.3 and 12.7; and for GSS 72-83 they are 60.4 and 16.2.

INTERACTIONS BY AGE

The men in the studies analyzed here represent a wide age range, and the studies themselves span a time period of 20 years. The educational experience of the oldest men thus goes back to

Table 6. Percentages Completing Selected Levels of Education by Sibsize Groupings, White Men from Intact Families, Occupational Changes in a Generation, 1962 and 1973, and General Social Survey, 1972-83

Sibsize Groupings

Respondent's Educational Level 0-1 2-3 4-6 7+ Total

OCG 1962 Percent of Men Who Completed Grade School 95.9 93.7 86.2 76.9 88.1 Percent of Men Who Graduated from High School 79.4 67.6 47.3 32.7 55.9 Percent of Men Who Went to College 46.6 32.5 17.2 10.4 25.6

OCG 1973 Percent of Men Who Completed Grade School 97.3 96.1 89.7 80.2 91.9 Percent of Men Who Graduated from High School 87.3 79.5 61.2 43.9 70.1 Percent of Men Who Went to College 54.3 41.4 24.1 12.7 34.7

GSS 1972-83 Percent of Men Who Completed Grade School 98.6 97.2 91.8 80.2 92.9 Percent of Men Who Graduated from High School 88.7 81.7 61.9 41.7 70.5 Percent of Men Who Went to College 60.4 48.3 30.6 16.2 40.2

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 8: Number of Siblings and Educational Mobility

90 AMERICAN SOCIOLOGICAL REVIEW

the early part of the 20th century, and their fathers' education represents the world of the latter part of the 19th century. Has the effect of father's education on son's education within sibsize groupings remained the same over such a time span, or would a breakdown by separate age groupings show that our conclusions are actually conditional on the time period during which these men were reared?

Considering first total years of respondent's education, a regression was run using an in- teraction term, for the respondent's age and the father's education, by sibsize groupings. No interaction was evident in any of the studies.

However, when the same test was applied to years of graded schooling (0-12) and propor- tions going to college among high school grad- uates the results suggested that an interaction was, indeed, present for OCG 1962 and 1973. It thus appeared that a separate analysis by age for OCG 1962 and 1973 would be fruitful.

Accordingly, for both studies, men were di- vided into three broad age categories, 20-34, 35-49, and 50-64. For each age category sepa- rate regressions were run for the four sibsize groupings. The results appear in Table 7. Ob- viously, the results do not invalidate our prior findings. In each age group, the effect of fa-

Table 7. Effect of Father's Education on Years of Graded Schooling (0-12) by Age and Sibsize Groupings, White Men from Intact Families, Occupational Changes in a Generation 1962 and 1973a

Sibsize Groupings

Respondent's Age 0-1 2-3 4-6 7+ Total

OCG 1962 20-34

Beta 0.044 0.084 0.112 0.197 0.097 (0.011) (0.011) (0.017) (0.028) (0.008)

Y 11.7 11.4 10.8 9.8 11.1 (1.14) (1.55) (1.93) (2.54) (1.85)

N 1,186 1,698 1,160 720 4,764 35-49

Beta 0.065 0.087 0.137 0.165 0.114 (0.014) (0.013) (0.016) (0.021) (0.008)

Y 11.4 11.0 10.2 9.5 10.5 (1.53) (1.79) (2.24) (2.57) (2.19)

N 981 1,601 1,591 1,196 5,369 50-64

Beta 0.114 0.162 0.151 0.229 0.175 (0.027) (0.019) (0.017) (0.021) (0.011)

Y 10.7 10.2 9.3 8.4 9.4 (2.08) (2.27) (2.53) (2.73) (2.61)

N 494 960 1,274 1,115 3,843 OCG 1973

20-34 Beta 0.058 0.089 0.166 0.269 0.135

(0.011) (0.009) (0.015) (0.027) (0.007) Y 11.7 11.6 11.1 9.9 11.3

(1.45) (1.59) (2.22) (3.10) (2.02) N 1,984 3,111 1,872 903 7,870

35-49 Beta 0.073 0.092 0.245 0.320 0.171

(0.012) (0.012) (0.017) (0.025) (0.008) Y 11.5 11.2 10.4 9.5 10.8

(1.67) (1.94) (2.68) (3.15) (2.46) N 1,645 2,079 1,731 1,224 6,679

50-64 Beta 0.135 0.150 0.233 0.295 0.211

(0.020) (0.016) (0.019) (0.024) (0.010) Y 11.2 10.8 9.8 8.9 10.1

(2.10) (2.38) (2.99) (3.31) (2.91) N 961 1,555 1,806 1,410 5,732

a Controls for respondent's age, farm background, sibsize (within the categories in the table), and father's SEI were included in the regression. Betas shown are the unstandardized regression coefficients for father's education. The differences between the slopes of small (0-1 or 2-3) sibsizes and large (7+) sibsizes are significant at the .01 level in both of the studies for all the age groups. The difference between the slopes for small (0-1) and medium large (4-6) sibsizes is significant at the .01 level for all age groups except 50-64 in OCG 1962. In addition, within the sibsize groupings, the difference in slopes between younger and older men is significant at the .01 level for sibsizes 2-3 in 1962, and 0-1 and 2-3 in 1973.

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 9: Number of Siblings and Educational Mobility

SIBLINGS AND EDUCATIONAL MOBILITY 91

ther's education increases with an increase in sibsize. The interaction by age (diagnosed by our test regression) consists of a difference in the strength of this change by sibsize groupings among men of different ages. Among older men, coming from a small family does not confer as great a mobility advantage (relative to coming from a large family) as it does for younger men. Comparison of men from small and medium sibsizes (0-1 and 2-3) in both studies shows there has been a decline in the effect of father's education as one moves from the oldest cohort to the youngest. To put it another way, there has been a proportionate increase in educational mobility over time among men from small and medium sibsizes. Among men from large sibsizes (7+), the effect of the father's education has not changed sig- nificantly from older to younger men. For example, in OCG 1962, the coefficients for fa- ther's education among men age 20-34 from sibsizes 0-1 and 2-3 are .044 and .084, but for those age 50-64 the coefficients for these sib- sizes are .114 and .162; whereas among men age 20-34 from large sibsizes the coefficient is .197, and for those 50-64 it is .229.

In addition to assuaging our concern over the possibly invalidating effects of interaction, these separate regressions by age and sibsize thus provide some important additional infor- mation. The results show that the increased upward educational mobility that occurred over time for men generally (an increase noted by Featherman and Hauser) was actually al- most entirely due to the experience of men from small and medium-sized families. The ex- perience of men from large families changed little, if at all. Therefore, the aggregate mobility figures (for men of all sibsizes combined) rep- resent an average of the very different experi- ences of men from different sibsizes. Had these samples been composed of men from small families only, even the oldest men would have had high educational mobility (since the coeffi- cients for men from small families are not large

at any age and thus men would not be con- strained by their father's education) and, in addition, the improvement over time would have been rapid.

Although this paper is concerned primarily with the effect of father's education on son's education, the analysis by age groups suggests a brief consideration of compositional changes in sibsize over time. During the period when, as we have seen, the mobility chances of men from small and medium sibsizes were improv- ing, we would expect (because of long-run de- clines in the birth rate) that increasing propor- tions of men would be from such sibsizes and decreasing proportions from very large ones. Table 8, showing the changing percentage of men, by age, coming from various sibsizes in OCG 1962 and 1973, demonstrates clearly that this expectation is confirmed. As a conse- quence, the educational opportunities of Americans over the time period covered by the experiences of the men in these studies were enhanced in a dual fashion-by improved chances for mobility among men from small and medium sibsizes (a lessening of the slope for father's education) and in addition, a com- positional shift of men from very large sibsizes to the more advantaged small and medium- sized ones.

What of the proportions going to college among high school graduates? Is our overall finding of a greater effect for father's education among men from small and medium sibsizes than among large ones maintained when the data are broken down by age groups? In gen- eral, the answer is yes, and the results for OCG 73 are shown in Table 9, both to illustrate this fact and to document an additional one that is also of interest.

Table 9 shows that the slopes for father's education among small and medium sibsizes are greater than among large sibsizes (4-6 and 7+) by age (although the differences are not statistically significant), reflecting the more drastic selection of men for high school

Table 8. Percentage Distribution by Sibsize and Age Groupings, White Men from Intact Families, Occu- pational Changes in a Generation, 1962 and 1973a

Sibsize Groupings

Respondent's Age 0-1 2-3 4-6 7+ Total N

OCG 1962 20-34 24.9 35.6 24.4 15.1 100.0 4,764 35-49 18.3 29.8 29.6 22.3 100.0 5,369 50-64 12.9 25.0 33.1 29.0 100.0 3,843

OCG 1973 20-34 25.2 39.5 23.8 11.5 100.0 7,870 35-49 24.6 31.2 25.9 18.3 100.0 6,679 50-64 16.8 27.2 31.5 24.5 100.0 5,732 a The distributions in this table are based on the men for whom years of graded schooling were calculated

in Table 7.

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 10: Number of Siblings and Educational Mobility

92 AMERICAN SOCIOLOGICAL REVIEW

Table 9. Effect of Father's Education on the Proportion Going to College among High School Graduates by Age and Sibsize Groupings, White Men from Intact Families, Occupational Changes in a Genera- tion, 1973a

Sibsize Groupings Respondent's Age 0-1 2-3 4-6 7+ Total

OCG 1973 20-34

Beta 0.126 0.142 0.144 0.060 0.131 (0.019) (0.015) (0.020) (0.032) (0.010)

Y 0.633 0.543 0.393 0.282 0.516 N 1,839 2,713 1,432 478 6,462

35-49 Beta 0.083 0.096 0.079 0.035 0.082

(0.019) (0.016) (0.021) (0.029) (0.010) Y 0.649 0.532 0.396 0.324 0.513 N 1,438 1,629 1,045 561 4,673

50-64 Beta 0.089 0.094 0.058 0.013 0.070

(0.024) (0.020) (0.020) (0.030) (0.011) Y 0.565 0.493 0.417 0.293 0.457 N 752 1,068 873 519 3,212

a Controls for respondent's age, farm background, sibsize (within the categories in the table), and father's SEI were included in the regression. Betas shown are the unstandardized regression coefficients for father's education. The difference between the slopes for small and large sibsizes is not significant for any age groupings. Comparing the differences in slopes across age groupings (within sibsizes), the differences between ages 20-34 and 50-64 are significant at the .05 level for sibsizes 4-6 only.

graduation among large-sized families as noted in the discussion of Table 6. Further, the table documents another possible effect of selection on the coefficients for the father's education. Within each sibsize grouping, the effect of fa- ther's education on the chances of a high school graduate going to college is progres- sively greater over time-that is, from older to younger men. Thus, with regard to going to college (given high school graduation), men of all sibsizes suffered some decline in their mo- bility chances over time, a consequence pre- sumably of the greater democratization and prevalence of high school completion. 1 Moreover, the largest increases in the slopes for father's education occurred among men from the two largest sibsize groupings. Ad- ditionally, among large sibsizes, Table 9 shows a slight decline in the proportion of high school graduates who went to college for the youngest cohort.

It should be emphasized that none of the coefficients for father's education shown in Table 9 are large. Also, at this level of educa- tion, the changes over time in the slopes would appear to reflect a progressive shift upward in the level of education at which social origins make a difference. One can argue that it is probably impossible to achieve a major democ- ratization of secondary education and simulta-

neously witness a decline in the effect of social origins on the chances of postsecondary edu- cation as well.

These results by age thus seem to provide both substantiation and specification of the mo- bility processes analyzed by Blau and Duncan and later by Featherman and Hauser. As they have emphasized, among industrialized coun- tries the United States has been unique in its liberal use of education as a mechanism for achieving occupational status. The differential changes in educational mobility by sibsize documented here, coupled with the compo- sitional changes in sibsize over time, would seem to help explain the pervasive belief that, in contrast to Europe, this country has offered and continues to offer unparalleled opportuni- ties to "get ahead."2

SUMMARY AND CONCLUSIONS

This paper has asked whether there has been an interaction in American society between the number of men's siblings and their educational mobility. The question was analyzed using three major American data sets. Father's edu-

I For a discussion of the historical background of this process, and its implications for postsecondary education, see Trow (1966).

2 The relevance of this research to the debate con- cerning whether the United States has actually been more "open" than Europe (Lipset and Bendix, 1959) cannot be explored here. However, students of the progress of this controversy over the past twenty- five years may find some of our results to be of interest.

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 11: Number of Siblings and Educational Mobility

SIBLINGS AND EDUCATIONAL MOBILITY 93

cation was included as the indicator of father's status, controlling for the respondent's age, whether he was brought up on a farm, father's SEI, and sibsize within sibsize groupings. Three indicators of the respondent's educa- tional attainment were used as dependent vari- ables: total years of education; total years of graded schooling (0-12); and proportion of high school graduates going to college.

Our results show that there has been a major interaction between sibsize and the effect of father's education on son's schooling. The in- teraction is minimized when total years of education is analyzed, due to the existence of opposite interaction effects at different levels of the son's education. At the graded-schooling level, men from small sibsizes were virtually impervious to differences in their father's edu- cational attainment, but men from large ones were relatively much more affected. By con- trast, considering the proportion of high school graduates who went to college, the influence of the father's education on men from large sib- sizes was found to be virtually nonexistent, but this influence was significant (although not large) for men from small ones. This result is not surprising given that small proportions of men from large sibsizes graduated from high school, whereas large proportions of men from small ones did so.

We then examined whether the results dis- cussed so far differed according to the ages of the men in the samples. While continuing to validate our initial findings, this analysis pro- vided some interesting additional information. We found that at the graded-schooling level there has been an increase in educational mo- bility over time among men from small and medium sibsizes (0-1 and 2-3), but among men from large sibsizes (7+) the effect of father's education has not changed significantly be- tween older and younger men.

It thus appears that the increased upward educational mobility that occurred over time to men generally (as noted by Featherman and Hauser) was actually due almost entirely to the experience of men from small and medium sib- sizes. Hence, the overall mobility calculations (without disaggregation into sibsize groupings) understate the life chances of men from small sibsizes and overstate those of the men from large ones. On the other hand, over the same time period, we have shown that a compo- sitional change in sibsize also occurred among the men in these samples-a movement from large to smaller sibsizes. Thus, the educational opportunities of American men were improved in two ways between the older and younger generations of men studied here. The effect (slope) of father's education on son's education was diminished over time for those from small

and medium sibsizes, and more and more men came from small and medium sibsizes.

The interaction between the mobility pro- cess and sibsize documented in this paper sug- gests that our thinking about social stratifica- tion in the United States may require some reevaluation. At least with regard to education, the "system" appears to have been remarkably open for those coming from small families and relatively rigid for those coming from large ones. It certainly seems that merely "control- ling" for sibsize at best, or ignoring it al- together at worst, are not viable options for the analysis of social stratification.

REFERENCES

Berent, Jerzy 1952 "Fertility and social mobility." Population

Studies 5:244-60. Blake, Judith

1981 "Family size and the quality of children." Demography 18:321-42.

Blau, Peter and Otis Dudley Duncan 1967 The American Occupational Structure.

New York: Wiley. Bresard, Marcel

1950 "Mobilit6 sociale et dimension de la famille." Population 3:533-66.

Dumont, Arsene 1890 Depopulation et Civilization. Paris: Lec-

rosnier et Babe. Duncan, Beverly

1965 "Family factors and school dropout: 1920- 1960." Ann Arbor: University of Michigan, mimeo.

1967 "Education and social background." American Journal of Sociology 72:363-72.

Featherman, David and Robert Hauser 1978 Opportunity and Change. New York: Aca-

demic Press. Kish, Leslie

1965 Survey Sampling. New York: Wiley. Lipset, Seymour and Reinhard Bendix

1959 Social Mobility in Industrial Society. Berk- eley: University of California Press.

Perruci, Carolyn C. 1967 "Social origins, mobility patterns and fer-

tility." American Sociological Review 32:615-25.

Preston, Samuel H. 1976 "Family sizes of children and family sizes

of women." Demography 13:105-114. Sewell, William H., Robert M. Hauser and Wendy

C. Wolf 1980 "Sex, schooling, and occupational status."

American Journal of Sociology 86:551-83. Tien, H. Y.

1965 Social Mobility and Controlled Fertility. New Haven: College & University Press.

Trow, Martin 1966 "The second transformation of American

secondary education." Pp. 437-49 in Reinhard Bendix and Seymour M. Lipset (eds.), Class, Status, and Power. Second edition. New York: Free Press.

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions

Page 12: Number of Siblings and Educational Mobility

94 AMERICAN SOCIOLOGICAL REVIEW

Westoff, Charles F. 1953 "The changing focus of differential fertility

research." Milbank Memorial Fund Quar- terly 31:24-38.

1981 "Another look at fertility and social mobil- ity." Population Studies 35:132-35.

Westoff, Charles F., Robert Potter, Jr., Philip Sagi and Elliot Mishler

1961 Family Growth in Metropolitan American. Princeton: Princeton University Press.

Westoff, Charles F., Robert Potter, Jr., and Philip Sagi

1963 The Third Child: A Study in the Prediction of Fertility. Princeton: Princeton Univer- sity Press.

Zimmer, Basil G. 1981a "The impact of social mobility on fertility: a

reconsideration." Population Studies 35: 120-31.

1981b "A rejoinder." Population Studies 35:136.

SPATIAL ASSIMILATION AS A SOCIOECONOMIC OUTCOME*

DOUGLAS S. MASSEY NANCY A. DENTON

University of Pennsylvania

Prior research has estimated structural equation models of Hispanic and black spatial assimilation using census tract data in selected cities. While these models appeared to provide an unambiguous picture of ethnic and racial segregation in cities, their estimation from ecological data posed several conceptual and statistical difficulties. This paper replicates the earlier analyses using individual-level data from the 1970 Census. Prior findings were reconfirmed, thereby validating the theory of spatial assimilation and reinforcing earlier substantive conclusions. Results also indicate that, to the extent that ecological biases affect the pattern of results, errors of substantive interpretation are conservative in nature. Thus, models estimated using census tract data for 1980 (when appropriate micro information will not be available) are not likely to yield erroneous conclusions.

A recent series of papers elaborated a theory of spatial assimilation that combined the status attainment perspective with an ecological model (Massey and Blakeslee, 1983; Massey and Mullan, 1984; Massey and Bitterman, 1985). The theory argued that an important outcome of socioeconomic advancement for minorities is residential integration within mainstream society. A host of variables im- portant to people's social and economic well- being are determined by residential location. For example, health, quality of education, ac- cess to employment, exposure to crime and, of course, social prestige all depend in part on where one lives. As social status rises, there- fore, minorities attempt to convert their socio- economic achievements into an improved spa- tial position, which usually implies assimilation with majority members.

In prior work, structural equation models based on this view were estimated using census tract data on Hispanics and blacks in selected SMSAs. The associated path models con- firmed the basic tenets of the theory. As the SES of minority members rose, so did the probability of residential contact with Anglos. Blacks and Puerto Ricans, however, were much less able than Hispanics to convert their status attainments into Anglo contact. The black disadvantage seemed to reflect the prejudice of whites who systematically avoided areas of black settlement (Massey and Blakes- lee, 1983; Massey and Mullan, 1984). How- ever, the relative inability of Puerto Ricans to achieve spatial assimilation appeared to stem from two other factors: their very low SES, which was insufficient to allow much assimila- tion; and their black heritage, which drew them strongly towards residence near and among non-Hispanic blacks, thereby eliciting avoidance by Anglos (Massey and Bitterman, 1985).

While the structural equation models seemed to provide unambiguous results, their estima- tion using ecological data was problematic. For example, the theory of spatial assimilation is based, in part, on status attainment theory, which is framed at the individual level (Blau and Duncan, 1967; Duncan et al., 1972); yet the

* Direct all correspondence to: Douglas S. Mas- sey, Population Studies Center, University of Penn- sylvania, 3718 Locust Walk CR, Philadelphia, PA 19104.

The authors would like to thank Professor Frank D. Bean for making a usable extract of the 1970 Neighborhood Characteristics PUS available to us, and two anonymous reviewers for their helpful comments. This research was supported by NICHD grant number HD-18594.

American Sociological Review, 1985, Vol. 50 (February:94-106)

This content downloaded from 130.15.241.167 on Wed, 2 Oct 2013 15:21:18 PMAll use subject to JSTOR Terms and Conditions