WW” : IUIJWIIHHJ! W I! ill! \ l 1mm WW —'(D_\ MO )| (JD—ta \ WI! Ill}!III!IIIfill/lIll/IIIIIIIIIIIIII/”Milli/11W " L m R A R y ,uv‘l's 3 1293 10062 9256 Michigan State University This is to certify that the thesis entitled MINORITY-GROUP STATUS AND THE FERTILITY OF BLACK AMERICANS,1965: A RE— PLICATION AND EXTENSION presented by RAMADAN SENUSSI BEL-HAG has been accepted towards fulfillment of the requirements for MASTER . SOCIOLOGY degree m JO! professor ZWWU Date AUGUST 8.1979 0-7639 OVERDUE FINES ARE 25¢ PER DAY PER ITEM Return to book drop to remove this checkout from your record. MINORITY-GROUP STATUS AND THE FERTILITY 0F BLACK AMERICANS, 1965: A REPLICATION AND EXTENSION By Ramadan Senussi Bel-Hag A THESIS Submitted to Michigan State University In partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Sociology l979 ABSTRACT MINORITY-GROUP STATUS AND THE FERTILITY 0F BLACK AMERICANS, 1965: A REPLICATION AND EXTENSION By Ramadan Senussi Bel-Hag The explanation of the relationship between minority-group status and fertility has been dominated by two major perspectives: the characteristics hypothesis and the minority-status hypothesis. Using 1960 data Goldscheider and Uhlenberg (1969) found support for the strong form of the minority status hypothesis; Johnson (1979) on the other hand found support for the weak form of characteristics hypothesis by using data from the 1970 National Fertility Study. Furthermore, work by Sly (1970) and Ritchey (1975) suggested that the effect of race and education on fertility might interact with region. A decade of profound changes in the socio—economic and political statuses of blacks intervened. Conse- quently the present study by using data from the 1965 National Fertility Study examined the two hypotheses at the midpoint of that decade. A three-way of interaction was found between race, education and region. ACKNOWLEDGEMENT I wish to express sincere appreciation to the members of my Guidance Committee for their interest and constructive criticisms. These members were Professors Nan E. Johnson, Chairperson, J. Allan Beegle and Harry K. Schwarzweller. A special acknowledgement of deep appreciation and gratitude must be extended to Dr. Johnson. At no time during the preparation of the thesis was she too busy to provide help for my problems. Without the most generous offering of her valuable time and suggestions, this thesis could never have been completed. Finally, my deepest gratitude to my family for their unquestioning faith and understanding during the preparation of this thesis. TABLE OF CONTENTS PAGE INTRODUCTION ........................... 1 BACKGROUND ............................ 5 PROCEDURES ............................ l0 FINDINGS ............................. 13 CONCLUSION ............................ 28 LIST OF REFERENCES ........................ 32 APPENDIX A ............................ 33 Correlation Coefficients for the Nation APPENDIX B .................. . .......... .34 Correlation Coefficients for the South APPENDIX C ............................ 35 Correlation Coefficients for the Non-South LIST OF TABLES TABLE PAGE Black-White Distribution and Change, Inside and Outside Metropolitan Areas: 1960—1970 ................ 3 Contrast Vector Coefficients and Group-Mean Comparisons for Decomposition of Joint Race-Education Effect on Fertility. . . 14 Means and Standard Deviations for the Two Populations (Blacks and Whites) .......................... 15 Means, Standard Deviations, and Sizes for Ten Race by Education Groups (The Nation) ................. 16 Means, Standard Deviations, and Sizes for Ten Race by Education Groups (The South) ................. 17 Means, Standard Deviations, and Sizes for Ten Race by Education Groups (The Non-South) ............... 19 Hierarchical Regression of Children Ever Born Upon Compositional Variables, Race, and Race-Education Interaction (The Nation) ......................... 21 Hierarchical Regression of Children Ever Born Upon Compositional Variables, Race, and Race-Education Interaction (The South) .......................... 22 Hierarchical Regression of Children Ever Born Upon Compositional Variables, Race, and Race-Education Interaction (The Non-South) ........................ 23 ii 10. 11. 12. LIST OF TABLES, continued TABLE - PAGE Analysis of Variance in Fertility Jointly Explained by Race and Education (The Nation) .................. 25 Analysis of Variance in Fertility Jointly Explained by Race and Education (The South) ................... 26 Analysis of Variance in Fertility Jointly Explained by Race and Education (The Non-South) ................. 27 LIST OF FIGURES FIGURE PAGE 1. Four Possible Interactive Relationships Between Race and Education Upon Fertility .................... 7 iv INTRODUCTION The last century witnessed great changes in the black population of the United States. Most obvious has been the change in legal status resulting from the Emancipation Proclamation. Moreover, from a socio- logical viewpoint, important changes have occurred in the living condi- tions and social position of blacks. Although slavery was abolished during the Civil War, assimilation and acceptance of blacks did not immediately follow; nor were their political and legal rights clearly established. For a relatively long time after Emancipation, most blacks remained isolated in the rural South, and their life styles and cultural practices were little altered. Industrialization associated with World War I initiated the out- migration of blacks from the South, and in the last fifty years the geographical location and social characteristics of the black population have changed greatly. In 1860, 92 percent of the black population lived in the South, and by 1910 approximately 90 percent still lived there. However, by 1950, the percentage of blacks living in the South dropped to 68 percent and, in 1960, it had decreased to 60 percent. It continued to drop to 53 percent by 1970 (U.S. Bureau of the Census, 1970). By 1970, alsojthe black population was more urbanized than the white, for three out of every five blacks in the United States lived in central cities of a major metropolitan area (U.S. Bureau of the Census, 1970). During the decade of 1960-1970, the black population in metropolitan areas increased 1 2 by four million (U.S. Bureau of the Census, 1971). Table 1 is an illus- tration of the distributional change of black and white populations inside and outside metropolitan areas: 1960 and 1970. The educational and occupational characteristics of blacks have changed greatly. By 1960, manufacturing replaced agriculture as the modal industrial category for employed blacks (U.S. Bureau of the Census, 1960). Between 1960-1970, total employment of blacks and other races increased 22 percent, but their employment in professional, technical, and clerical occupations doubled. Changes in educational attainment also occurred. Blacks reaching age 18 just after World War I had attained a median of 5 years of schooling. In 1970, 56 percent of all young adult blacks 25 to 29 years old had completed high school compared with 38 percent in 1960. By 1970 about 17 percent had at least one year of college. In comparison, approximately 78 percent of the whites had a high school education and about one-third had received some college training (U.S. Bureau of the Census, 1970). These changes in the social characteristics and residence of blacks should have implications for their demographic rates. Demographers might predict that as long as blacks remained in the rural South, fertility would remain high but that the rapid migration of blacks away from farms and to the North with concomitant increments in educational attainment, occupational status and earning power would produce lower fertility rates (Farley, 1966, p. 189). Nevertheless, black fertility rates began declining in the 18805, well before the large-scale exodus of blacks from the rural South. In fact, evidence presented by Farley (1966) suggested that fertility rates of blacks had declined fifty percent by the 19305. Table 1. Black-White Distribution and Change, Inside and Outside Metropolitan Areas: 1960-1970 (numbers in millions) Population Black White 1960 1970 1960 1970 United States ....... 18.9 22.7 158.8 177.6 Metropolitan areas ..... 12.8 16.8 106.4 121.3 Central cities ....... 9.9 13.1 50.1 49.5 Outside central cities. . . 2.8 3.7 56.3 71.8 Outside metropolitan areas. 6.1 5.8 52.5 56.4 Change, 1960-1970 Black white Number Percent Number Percent United States ....... 3.8 20 18.8 12 Metropolitan areas ..... 4.1 32 14.9 14 Central cities ....... 3.2 33 .6 -1 Outside central cities. . . .8 29 15.5 28 Outside metropolitan areas. -.3 -4 3.9 7 Source: U.S. Bureau of the Census, "The Social and Economic Status of Negroes in the United States, 1970." Current Population Reports, series P-23, No. 38. 4 Moreover, after the Depression of the 19305, rural-to-urban migra- tion became a significant form of geographic mobility for blacks and attended their improvements in educational attainments and occupational levels. As the black population made these advances, demographic tran- sition theory would predict that their fertility would become controlled and that moderate sized families would become the norm. Oddly, black fertility rates climbed, and the percentage of black women having large families increased rather than decreased in the 19405 and early 19505. That black fertility rates fell when one would have expected them to be relatively constant and rose when one would have expected them to decline is an enigma which exemplifies the need for greater theoretical under- standing of minority-group fertility. BACKGROUND One explanation of black-white differences in fertility has been called the "characteristics hypothesis." The characteristics hypothesis assumes that differences in socio-economic and demographic composition completely explain black-white fertility differentials. For example, Lunde (1965) argued that the decrease in black fertility since 1957 can be explained in terms of their advancement in health, urbanization and educational attainment. An implication of this approach is that as the socio-economic and demographic characteristics of two groups (e.g., blacks and whites) become similar, so will their level of fertility. Thus, Farley (1966) concluded that "lower fertility rates (among blacks) may be indicative of an increasing involvement of blacks in (the) American society...as status differences disappear so may fertility dif- ferences" (Farley, 1966, p. 203). Johnson (1979) wrote that the characteristics hypothesis could be cast in two forms. For example, if race were regressed upon education, one could correlate the residual (unexplained) variance in race with a measure of fertility. If the resultant semipartial correlation be zero, the "strong form" of the characteristics hypothesis would be supported, since race and fertility would be unrelated at every level of education. Another way of viewing these relationships is that the regression line of fertility upon education would be the same for blacks and whites (Diagram A, Figure 1). Thus, the "strong form" of the characteristics hypothesis holds that when the differences in the socio-economic and demographic characteristics between blacks and whites are rendered similar through statistical controls, race will have no net effect on fertility at each level of education. 5 6 The weak form of the characteristics hypothesis argues that if the differences in the socio-economic and demographic characteristics between blacks and whites are rendered similar through social change, the racial differences in fertility between blacks and whites will disappear first among the highly educated, since it is this class which is thought to gain first access to more favorable social and demographic statuses. For example, using the 1970 National Fertility Study (NFS II) data, Johnson (1979) re-examined the interrelationship between race, education and fertility and concluded that support was found for the "weak form" of the characteristics hypothesis (Diagram 8, Figure 1). The explanation for the disappearance of fertility differences between blacks and whites among those having attended college may be the greater integration of these blacks into educational, political, and economic institutions (Johnson, 1979, pp. 15-16). Goldscheider and Uhlenberg (1969) argued that a lack of potential for social mobility could reduce incentive for rational fertility controls and thus account for higher black than white fertility at lower education levels. Another perspective, known as the "minority-group status hypothe- sis," holds that race has an independent effect upon fertility beyond the effects of relevant compositional factors. A5 blacks begin to compete more generally with whites for upward social mobility, they are thought to encounter greater barriers to achievement and to experience greater insecurities (Goldscheider and Uhlenberg, 1969). The personal insecuri- ties are thought to lead to greater deferments or limitations of child- bearing for upwardly mobile blacks than upwardly mobile whites. Goldscheider and Uhlenberg (1969) argued that this approach illuminates the lower fertility of Jews and Japanese-Americans, as well. An Figure 1. "Four possible interactive relationships between race and education upon fertility."1 5 High 5 High 5. “ <9 '0 'U :21? :21? .23. SI- 32 8.8 W as as .,_> .,..> . F0 PC) 35 IS ' s. S- I Q) d) 14.. LL. I Low Low . ow High ow High Education Education Diagram A. Strong form of the Diagram B. Weak form of the characteristics hypothesis. characteristics hypothesis. : High High Q) s: 3- cu ‘U L r—A '5 .P: PA .25. 'PC 00 £5- Vn 00 >,,_ 1 ‘43 Iv :2 : as l :0) . 20>) . 8 '2 I 35 LL LOW l L8 LOW Low High Low High Education Education Diagram C. Strong form of the Diagram D. Weak form of the minority-status hypothesis. minority-status hypothesis. Note. B = black; W = white. (Source: (Johnson, 1979). 8 implication of this perspective is that even when the social, economic, and demographic characteristics of blacks and whites are similar, black fertility will remain distinctive. ’ Johnson (1979) argued that the minority-status hypothesis can also assume a “strong form" and a "weak form." The strong form would predict racial differences in fertility at every level of education: higher black than white fertility at lower education levels, lower black than white fertility at higher education levels. Support for the strong form of the minority-status hypothesis was found in Goldscheider and Uhlen- berg's (1969) analysis of 1960 census data (Diagram C, Figure 1). In contrast, support for the weak form of the minority-status hypothesis would show no fertility differences between blacks and whites at lower educational levels but lower black than white fertility at higher educa- tional levels (Diagram 0, Figure 1). Recent work suggested that the minority status effect upon fertility may vary by ecological setting. For example, Sly (1970) used 1960 census data to test: (1) the relationship between minority-group status and fertility, (2) the relationship between each of the socio-economic factors (education, income and occupation) and fertility, and (3) the interaction effect of each of the compositional factors (education, income and occupation) and minority-group status (race) on fertility. Sly (1970) claimed that an analysis of variance technique could serve this purpose. By comparing the white-non-white differences in fertility by the socio- economic factors mentioned above, Sly (1970) first found support for the minority-status hypothesis, since significant effects were found due to race, education and their interaction. When the South was excluded from the analysis, non-whites showed lower fertility than whites regardless of the level of education. In other words, the interaction term between the 9 socio-economic factors, race and fertility disappeared when the South was excluded. Therefore, Sly (1970) argued that this non-Southern fertility pattern could be explained by the characteristics hypothesis. Ritchey (1975) made use of the 1970 census data to construct an aggregate measure of racial inequity for each U.S. state. Ritchey (1975) hypothesized that race and education interact to affect fertility and this interaction is positively related to racial inequity. Blacks would have higher fertility than whites at the lower educational level and this difference in fertility would decrease as racial inequity decreased. At higher educational levels, blacks would have lower fertility than their white counterparts and this would be most evident in areas of higher racial inequity. The findings strongly suggested that an interactive relation- ship among education, race and fertility still existed in the 1970 census public-use sample. Consequently, Ritchey's findings challenged Sly's (1970) earlier conclusion that minority-status effect upon fertility operates only in the South (Johnson, 1979, p. 6). The purpose of the present study is to re-examine the previous inquiries into the race-fertility relationship. The results of Johnson's (1979) study supported the weak form of the characteristics hypothesis, while Goldscheider and Uhlenberg's (1969) findings corroborated the strong form of the minority-status hypothesis. However, these outcomes are not necessarily inconsistent. Goldscheider and Uhlenberg's (1969) data were from the 1960 census, whereas Johnson's data came from the 1970 National Fertility Study (NFS II). A decade of profound changes in the social, economic, and political statuses of blacks intervened. Conse- quently, one task of the present study is to explore the interrelation- ships of race, education, and fertility at the midpoint of that decade. Inasmuch as Johnson's statistical methodology shall be employed, the 10 present effort may be viewed as a partial replication of her work. Inas- much as these interrelationships shall be examined separately for the South and the non-South, the current study may be viewed as an extension of previous inquiries. PROCEDURES The data used in the analysis were obtained from the 1965 National Fertility Study (NFS I). A probability sample of 5,617 currently married women under 55 years old and living in the conterminous United States were interviewed. A comprehensive discussion of the sample design is available in Ryder and Westoff (1971). Since the present study was interested in the investigation of black and white fertility, all women whose race was neither black nor white were excluded from the analysis. Also omitted were those having invalid codes for dependent or independent variables; those whose husband's were students, were in the armed forces, had never worked, or were unemployed; and women wed more than once. The effective sample size thus became 4,533, of whom 3,679 were white and 854 were black. In order to explore the interrelationship between education, race and fertility, it is necessary to hold constant the effect of social, economic and demographic factors that are related to fertility and that are known to differentiate blacks and whites. Black women are more likely to be working, to reside in urban places, to prefer a non-Catholic religion, to experience divorce or widowhood, and to be young. Each of these characteristics has been associated with lower fertility in past research. Thus, female labor force participation, urban residence, religion, marital instability and age would minimize the degree to which black fertility exceeds white fertility. These variables were scored in 11 the following way: Wife's working status was coded (1) working and (2) not-working. Place of residence was measured (0) rural and (1) urban. Wife's religious preference was (0) Catholic and (1) non- Catholic. Marital instability was controlled in two ways. First, analysis was limited to women who were currently married and in the first union. Secondly, duration of marriage was recorded as (1) 0-4 years; (2) 5-9; (3) 10-14; (4) 15-19; (5) 20-24; (6) 25-29; and (7) 30+ years. Finally, age was coded in century months by subtracting the date of birth from the date of interview. Factors which are thought to maximize black-white differences in fertility are: occupational status, income and age at first marriage. Black men and women are likely to hold low status jobs, to be poor and to get married at a younger age. Each of these characteristics has been associated with higher fertility in past research. Therefore, these variables must be controlled and were measured as follows: Occupation was shown by husband's occupation and was coded by an ordinal variable: (1) professional, technical, and kindred workers; (2) managers, officials and proprietors (except farm); (3) clerical, sales and kindred; (4) craftsmen, foremen and kindred; (5) operatives, kindred and service workers; (6) private household workers and laborers (except farm); (7) farmers and farm managers; (8) farm laborers and foremen. Income was operationalized as combined income of husband and wife (family income) and was coded (0) under $2,000; (1) $2,000 to 2,999; (2) $3,000 to 3,999; (3) $4,000 to 4,999; (4) $5,000 to 5,999; (5) $6,000 to 6,999; (6) $7,000 to 7,999; (7) $8,000 to 9,999; (8) $10,000 to 14,999; (9) $15,000 or more. Age at first marriage was coded in century months by subtracting the date of marriage from the date of the interview. 12 After these several social, demographic and economic variables have been controlled statistically, it will be possible to examine the inter- relationship among race, education and fertility. Race was categorized as (1) white and (2) black. The respondent's years of completed schooling was measured by five categories: (1) elementary school 0-8 years; (2) high school 1-3 years; (3) high school 4 years; (4) college 1-3 years; (5) college 4 or more years. Region was measured as: (0) South and (1) non-South. Fertility was measured as the total number of children ever born alive to respondent. Since the characteristics hypothesis gives priority to the several compositional factors in determining the black-white differences in fertility, it was first necessary to determine whether the main effect of race and the interactive effect of race and education upon fertility were negligible after social, economic, and demographic variables had been controlled. For this purpose, a hierarchical multiple regression model with the following three inclusion levels was deemed most appropriate: (1) the social, economic and demographic variables; (2) race; (3) the race-education interaction. Since the weak form of the characteristics hypothesis and the strong and the weak forms of the minority-status hypotheses all predict a race- education interactive effect on fertility, such an interactive effect must be decomposed before one of these hypotheses can be confirmed. For this purpose, an analysis of variance technique was chosen. Since race was measured by two categories and education by five, the variance in fertility jointly accounted for by race and education would yield nine degrees of freedom: one for race, four for education, and four for their interaction. By associating each degree of freedom with an orthogonal 13 vector (Table 2), it would be possible to describe the joint variance completely by nine contrasts of group means. Thus, the joint effects of race and education on fertility could be separated into four effects produced by the educational-group differences (Table 2, rows 1-4) and five effects created by the racial differences in natality within each education group (Table 2, rows 5-9). FINDINGS Table 3 provides the general characteristics of the effective sample. It shows that blacks had borne an average of (3.438) children which is higher than the mean number of children ever born for whites (2.591). Furthermore, blacks had an earlier mean age at first marriage, less edu- cation, less prestigious occupations, lower family incomes and lower rates of wife's employment as compared to the whites. These racial differences in fertility can be partially accounted for by the differences in socio- economic and demographic characteristics of these two populations. Table 4 gives the mean number of children ever born for five educa- tional groups for U.S. blacks and whites. Within every educational category except four or more years of college blacks had higher fertility than whites. Excess black fertility ranged from 1.472 (4.970-3.498) more children than whites for those with an elementary school education to .041 more children than whites for women having completed 1-3 years of college. Among women having four or more years of college, whites had an average of .657 (2.119-l.462) more children than did blacks. Table 5 provides the mean number of children ever born by educational groups for the South. Within each educational category blacks had higher fertility than whites with the exception of the last category. Among women who had an elementary school education blacks had an average of 1.174 (5.114-3.400) more children than whites. This difference decreased l4 map_;z .co mxum_m o o o o F- _ o o o o ”mamas e .Foogum gap: mmumgz .wu mxonpm o o o o o o P- _ o o ”mcmmx mip .poosum sow: mauve: .cu mxumpm o o o o o o o o P- P Hates» w-_ .Foogum 3Lapcosmpm memo» mace go e .mmwppou .30 v- e- F F F F F F F P mmmppou .mcmmz e pmmmp um cusp mmw3 memo» nip .wmo_pou .30 o o m- m- _ F F P _ _ .mamaa m-3 .mmmp_oo anew mam; memo» e .Foogum saw: .36 o o o o N- N- _ F _ F .mtm33 e ._oo;am now; can“ mam; memo» mip .poogom saw: o.mu o o o o o o P- _- _ 3 .mamma m-3 .Foo;um samacmsapu 3 m 3 m 3 m 3 m 53 mm mangUwcewou .mca egos .mg» nip .mg» q .mL3 mip Foogom gouom> ummcucou co e .oamppou .mamppoo .Foogum gum: .Poosum saw: accucmswpm —=.3uw_muemw co uommem cowumosuwimomc ucwom co cowuwmoasoowu Low mcomwcmasoo camsiaaogm ecu mucmmowwemoo Lopum> “magpcou= .N m—nmh 14a .Amump .comcgoov ”mugzomp =.ou cmgmqsoo= u .woo .ASTFppcwe some mums: 3 a .33_P_Saac cams SumFm mm weave; .co mxua_m F- P o o o o o o o o "memos wees so 3 .mmmppou mmuwcz .mo meMFm o o F- F o o o o c o "mamma m-3 .ammp_ou 3 m 3 m 3 m 3 m n3 mm mucmwowmwwou .mg3 wees .mg» m.— .mgz e .mp3 m-_ Foocom Louuw> ummcpcou go q .mmo_Pou .mmmppou .Poogum 3mm: .Foogom sac: Aguacmempm umzcwucou .N mpam» 15 Table 3. Means and standard deviations for the two populations (Blacks and Whites).a Blacks Whites Variables Means Standard Dev. Means Standard Dev. Fertility 3.438 2.875 2.591 1.834 Duration of Marriage 3.135 1.836 3.350 1.709 Age at First Marriage 243.817 49.792 249.809 44.840 Age 400.559 116.324 419.434 111.182 Place of Residence .770 .421 .769 .422 Region of Residence .355 .479 .706 .455 Education 2.311 1.067 2.858 1.899 Religion .932 .252 .711 .454 Husband Occupation 4.298 1.657 3.197 1.863 Family Income 4.002 2.209 5.878 2.176 Working Status 1.584 .493 1.682 .466 3Data for this and subsequent tables were from the 1965 National Fertility Study. 16 Table 4. Means, standard deviations, and sizes for ten race-by- education groups. (The Nation) "ataxitz‘dre" 33:32:23.. ~ Elementary school, 0-8 yrs.: Blacks 4.970 3.648 236 Whites 3.498 2.412 466 High school, 1-3 yrs.: Blacks 3.416 2.560 303 Whites 2.870 1.831 832 High school, 4 yrs.: Blacks 2.554 2.081 269 Whites 2.374 1.636 1855 College, 1-3 yrs.: Blacks 2.215 1.663 65 Whites 2.174 1.558 476 College, 4 or more yrs.: Blacks 1.462 1.253 39 Whites 2.119 1.605 311 17 Table 5. Means, standard deviations, and sizes for ten race-by- education groups. (The South) "2.32:2?” 32:23:52.. ~ Elementary school, 0-8 yrs.: Blacks 5.114 3.698 201 Whites 3.400 2.464 215 High school, 1-3 yrs.: Blacks 3.670 2.732 191 Whites 2.472 1.632 263 High school, 4 yrs.: Blacks 2.824 2.180 136 Whites 2.156 1.589 475 College, 1-3 yrs.: Blacks 2.281 1.571 32 Whites 1.920 1.351 113 College, 4 or more yrs.: Blacks 1.536 1.374 28 Whites 1.770 1.309 74 18 to only .361 (2.281-1.920) more children for blacks than whites among women having completed 1-3 years of college. For women having completed four or more years of college whites had an average of .234 (1.770-1.536) more children than blacks. Table 6 shows that outside the South, the relationships between race, education and fertility were different from those inside the South. While blacks with an elementary school education had an average of .561 (4.143-3.582) more children than did whites, blacks with higher levels of education had fewer children than did whites. The excess of white over black fertility ranged from .073 (3.055-2.982) children for women with 1-3 years of high school to .955 children for women having completed at least four years of college. Since the racial difference in fertility declined gradually with the rise in educational levels and within each education group, it seemed that education had depressed the black and white fertility progressively and was stronger for blacks than whites for the nation as a whole and the South and non—South. Also, it looked like race and education had a significant interaction effect upon fertility. The characteristics hypothesis assumes that differences in socio- economic and demographic composition completely explain black-white ferti- lity differentials, such that when these compositional factors have been controlled, race will not exert a significant effect on fertility. In order to test whether the differences in mean fertility between blacks and whites that varied by education level might be explained solely by the compositional differences between blacks and whites, I regressed the number of children ever born upon ten compositional variables as a first inclusion level. The second level of inclusion was race and the third 19 Table 6. Means, standard deviations, and sizes for ten race-by- education groups. (Non-South) "static?” 32:23:33. ~ Elementary school, 0-8 yrs.: Blacks 4.143 3.273 35 Whites 3.582 2.367 251 High school, 1-3 yrs.: Blacks 2.982 2.181 112 Whites 3.055 1.892 569 High school, 4 yrs.: Blacks 2.278 1.944 133 Whites 2.449 1.645 1380 College, 1-3 yrs.: Blacks 2.152 1.770 33 Whites 2.253 1.610 363 College, 4 or more yrs.: Blacks 1.273 .905 11 Whites 2.228 1.675 237 20 was the race-education multiplicative term. These procedures were applied first for the nation, then for the South and the non-South regions separately (Tables 7,8,9). When the number of children ever born was regressed upon the twelve independent variables for the nation (Table 7), 24.9% of the variance was explained (p<.001). Of the ten compositional variables, education had the strongest relationship with the dependent variable (Beta=.4l4, p<.001). The direct relationship between education and fertility was unanticipated. One possible explanation for this anomaly is that the strong correlation (r=.87, Appendix A) between the main effect of education and the interactive effect between race and education may have produced an unstable regression coeffi- cient estimate for education (Althauser, 1971). As such, the sign of the beta coefficient for education may have been reversed from negative to positive. Duration of marriage bore the weakest relationship to the dependent variable (Beta = .029), and its effect was not significant. After the differences among the blacks and whites in the socio-economic and demo- graphic characteristics had been controlled, race showed a significant increment in explained variance in the number of children ever born (F=148.710, p<.001). Blacks had more children than whites. Moreover, the race-education interaction had a significant non-additive effect upon fertility in the nation (Table 7), in the South (Table 8), and in the non- South (Table 9). Since the race-education multiplicative term had been the last variable added to the regression equation, the probability that it would explain significant portions of variance in fertility had been reduced. Therefore, it appeared unlikely that the interactive effect of race and education on fertility was spurious. 21 Table 7. Hierarchical regression of children ever born upon compositional variables, race, and race-education interaction (The Nation). Independent Simple 2 a b Variables r R b Beta F SR F HR 1. Duration of .336 .113 .035 .029 .156 marriage .182 .153 -.042 -.046 7.494 2. Family income 3. Age at first -.216 .179 -.013 -.293 72.421* marriage 4. Working status .134 .193 .617 .139 101.765* 5. Religion -.038 .201 -.559 -.116 70.072* 6. Age .234 .205 .006 .325 17.204* 7. Husband occupation .134 .208 .047 .042 7.786* 8. Place of residence -.139 .211 -.340 -.068 23.474* 9. Education -.l7O .212 .481 .414 49.184* 10. Region of residence-.053 .212 .201 .046 9.946* F]_10=118.367* 11. Race .158 .238 2.149 .401 159.925* F11=148.710* 12. Race x education -.127 .249 -.499 -.494 61.783* F12=61.783* Note - N = 4,399; intercept = .761. The overall F ratio calculated by F R is 121.085, which is significant at P<.001 at 12 and 4386 degrees of freedom. aThis is tBe standard regression F-ratio. It tests the significance of the change in R due to the addition of variable i after all other independent variables are in the equation: F = A rzy (i - 1, 2, ... K)/1 . SR 117- R4 y 1, 2; ... i, ... K)/(N - K - 1) 21a Table 7, continued. bThis is the hierarchical regression F-ratio. It tests the signi- ficance of the change in R2 due to the addition of variables to the restricted model to obtain the full model: 2 2 FHR ‘ (R F ' R R)/(KF ' KR) (l - RZF)/(N — KF - 1) *p<.001. 22 Table 8. Hierarchical regression of children ever born upon compositional variables, race, and race-education interaction (The South). Independent Simple 2 Variables r R b Beta FSR FHR 1. Duration of .352 .124 .441 .334 7.620 marraige 2. Family income -.264 .195 .052 .052 3.119 3. Age at first -.224 .218 .007 .140 6.739 marriage 4. Education -.331 .228 .378 .174 6.424 5. Working status .120 .233 .539 .106 22.256* 6. Husband .167 .236 .047 .036 2.125 occupation 7. Place of resi- -.190 .238 .358 .070 9.125* dence 8. Religion .008 .238 .590 .075 11.797* 9. Age .248 .238 .000 .019 .024 F]_9=54.323* 10. Race .272 .291 .299 .448 78.407* F10=117.OOO* 11. Race x education -.151 .300 .429 .308 18.090* F11=18.09O* Note - N = 1573; intercept = .628. 60.665, which is significant at P<.OOl at 11 and 1561 degrees of freedo *p<.001. The overall F ratio calculated by FHR is Table 9. 23 Hierarchical regression of children ever born upon compositional variables, race, and race-education interaction (The non-South). Independent Simple 2 Variables r R b Beta SR FHR 1. Duration of .323 .105 .216 .192 4.201 marriage 2. Age at first -.203 .142 .017 .413 85.816* marriage 3. Working status .150 .171 .658 .164 87.335* 4. Religion -.086 .185 .521 .130 57.572* 5. Age .228 .195 .010 .583 33.417* 6. Family income -.097 .201 .032 .034 3.015 7. Husband .102 .203 .043 .043 5.146 occupation 8. Place of resi- —.O77 .204 .258 .051 8.669* dence 9. Education -.118 .205 .362 .399 10.647* F]_9=80.660* 10. Race .013 .209 .395 .227 21.590* F10=13.959* 11. Race x education -.120 .212 .381 .453 12.466* F11=12.466* Note - N = 2826; intercept = 1.712. is 68.976, which is significant at P<.OOl at 11 and 2814 degrees of free *P<.001. The overall F-ratio calculated by Eggm 24 An analysis of variance was employed to decompose the variance in fertility jointly explained by race and education into nine uncorrelated segments, each associated with a contrast in group means. Data for the United States as a whole (Table 10) showed that each successive level of educational attainment was associated with lower fertility than that for preceding levels. Moreover, among those who never graduated from high school, blacks had significantly higher fertility than did whites. Among those graduating from high school or attending college for one-to-three years, black and white fertility did not differ (F=1.956 and 0.025, respectively). However, among those graduating from college, blacks had substantially fewer children than did whites (F=3.835; p<0.05). The interactive pattern of race, education, and fertility traced the inter- relationship predicted by the strong form of the minority-status hypo- thesis (see Figure 1, Diagram C). These relationships were reexamined separately for the South (Table 11) and the non-South (Table 12). For both areas, it was found that increases in educational attainment brought declines in fertility with one exception: persons outside the South with only one-to-three years of high school had borne the same number of children as had those not entering high school. Among persons who had only elementary - school educations, who had attended but not finished high school, and who had graduated from high school, Southern blacks had significantly higher fertility than did Southern whites. Outside the South, higher black than white fertility among elementary - school dropouts approached statistical significance (F=2.934; p<0.09); but no differences in natality were found among blacks and whites who had attended or graduated from high school. Among those attending but not graduating from college, black and white Table 10. and education (The Nation). Analysis of variance in fertility jointly explained by race . Sum of Degrees of Mean Source Of Variance Squares Freedom Square F Between Contrasts 2204.907 2_ 244.990 62.721 1. Elementary cf. 173.740 1 173.740 44.484*** High school, (1-3) years 2. Elementary and high 888.408 1 888.408 227.466*** school, (1-3) years cf. High school, (4) years 3. Less than college, (1-3) 484.383 1 484.383 124.021*** years cf. college, (1-3) years 4. Less than college, (4) or 229.859 1 229.859 58.853*** more years cf. college (4) or more years 5. Elementary school: 339.674 1 339.674 86.970*** Blacks cf. whites 6. High school, (1-3) years: 66.130 1 66.130 16.932*** Blacks vs. whites 7. High school, (4) years: 7.639 1 7.639 1.956 Blacks vs. whites 8. College, (1-3) years: .096 l .096 .025 Blacks cf. whites 9. College, (4) or more years: 14.978 1 14.978 3.835* Blacks vs. whites Within Contrasts 18,911.251 4842 3.906 Total 21,116.158 4851 *p<0.05 ***p<0.001 Table 11. and education (The South). Analysis of variance in fertility jointly explained by race Sum of Degrees of Mean Source Of Variance Squares Freedom Square Between Contrasts 1731.617 9_ 192.402 39.138 1. Elementary cf. 308.188 1 308.188 62.698*** High school, (1-3) years 2. Elementary and high 550.986 1 550.986 112.092*** school, (1-3) years cf. High school, (4) years 3. Less than college, (1-3) 227.507 1 227.507 46.284*** years cf. college, (1-3) years 4. Less than college, (4) or 129.113 1 129.113 26.267*** more years cf. college (4) or more years 5. Elementary school: 305.337 1 305.337 62.118*** Blacks cf. whites 6. High school, (1-3) years: 158.978 1 158.978 32.342*** Blacks cf. whites 7. High school, (4) years: 47.142 1 47.142 9.591*** Blacks cf. whites 8. College, (1-3) years: 3.248 1 3.248 .661 Blacks cf. whites 9. College, (4) or more years: 1.118 1 1.118 .227 Blacks cf. whites Within Contrasts 8444.771 1718 4.916 Total 10,176.388 1727 ***p<0.001 Table 12. 27 Analysis of variance in fertility jointly explained by race and education (The non-South). . Sum of Degrees of Mean Source Of Variance Squares Freedom Square F Between Contrasts 603.310 2_ 67.034 20.332 1. Elementary cf. .170 l .170 .052 High school, (1-3) years 2. Elementary and high 270.182 1 270.182 81.943*** school, (1-3) years cf. High school, (4) years 3. Less than college, (1-3) 214.217 1 214.217 64.969*** years cf. college, (1-3) years 4. Less than college, (4) or 95.152 1 95.152 28.858*** more years cf. college (4) or more years 5. Elementary school: 9.674 1 9.674 2.934** Blacks cf. whites 6. High school, (1-3) years: .490 1 .490 .149 Blacks cf. whites 7. High school, (4) years: 3.521 1 3.521 1.068 Blacks cf. whites 8. College, (1-3) years: .314 1 .314 .095 Blacks cf. whites 9. College, (4) or more years: 9.590 1 9.590 2.908** Blacks cf. whites Within Contrasts 10,267.494 3114 3.297 Total 10,870.804 3123 **p<0.09 ***p<0.001 28 fertility was similar both for the South (F=0.661) and for the non-South (F=0.095). Thus, the higher black than white fertility reported at the national level for persons never graduating from college (Table 4) resulted primarily from residence in the South. Among persons having finished at least four years of college, black and white fertility did not differ significantly in the South (F=O.227); but outside that region, the lower black than white fertility approached statistical significance (F=2.908; p<0.09). As such, the pattern of lower fertility for black than for white college graduates observed in national data (Table 4) occurred mainly among women outside the South. CONCLUSION Two complementary explanations of black-white fertility differences were examined in the present study. The characteristics hypothesis holds that race may affect fertility only indirectly by determining one's social, economic and demographic characteristics. Thus, it argues that once the socio-economic and demographic characteristics of blacks and whites become similar, so will their fertility. On the other hand, the minority-status hypothesis holds that race has an independent effect beyond relevant compositional factors, since minority group status itself represents a unique barrier to upward social mobility. From this latter perspective, black-white differences will remain, even after their social, economic and demographic characteristics have become similar. Each of these interpretations might take one of two forms, known as the strong form and the weak form. The strong form of the characteristics hypothesis argues that when the differences in the socio-economic and demographic characteristics between blacks and whites are rendered similar through statistical controls, race will have no net effect on fertility 29 at any level of education. In contrast, the strong form of the minority- status hypothesis predicts racial differences in fertility at every level of education, higher black than white fertility at lower educational levels and lower black than white fertility at the higher educational levels. The weak form of the characteristics hypothesis argues that if the differences in the socio-economic and demographic characteristics between blacks and whites are rendered similar through social change, the racial differences in fertility between blacks and whites will disappear first among the highly educated, while higher black than white fertility remains for a while among those having little education. 0n the contrary, the weak form of the minority-status hypothesis predicts no fertility differences between blacks and whites at lower educational levels but lower black than white fertility at higher educational levels. Support for the strong form of the minority-status hypothesis was found in Goldscheider and Uhlenberg's (1969) analysis of 1960 census data. In contrast, the results of Johnson's (1979) study supported the weak form of the characteristics hypothesis by using data from the 1970 National Fertility Study. A decade of profound changes in the socio- economic and political statuses of blacks intervened. Consequently, the present study tried to explore the interrelationship among race, educa- tion and fertility at the midpoint of that decade. The findings indicated that social, economic and demographic factors did account for a significant proportion of variance in the number of children ever born. After the differences among blacks and whites in the compositional characteristics had been removed, race and the race-educa- tion multiplicative term added significant increments to explained variance. By using orthogonal contrast vectors, the total variance in fertility explained jointly by race and education was then broken down 30 into its uncorrelated components. Black women had higher fertility than white women among those who had elementary school education or one to three years of high school. There were no significant differences in average number of live births to black and white women who had completed four years of high school or one-to-three years of college. Among college graduates, black women had significantly fewer children than white women. That these interrelationships, in the national data for 1965 supported the strong form of the minority-status hypothesis was consistent with Goldscheider and Uhlenberg's earlier analyses of the 1960 U.S. census. While lower white than black fertility at low levels of education and higher white than black fertility at high levels of education were observed in national data for 1960 (Goldscheider and Uhlenberg, 1969), work by Sly (1970) and Ritchey (1975) suggested that this race-education effect on fertility might interact with region. For example, contingency tables of mean fertility by education and race of women in the 1960 census suggested that this strong form of the minority-status hypothesis might exist only for the South, since nonwhites had lower fertility than whites in every educational category for other regions (Sly, 1970). Sly's (1970) analysis of variance showed that education and race did interact with fertility when the national data was used but that this interaction vanished when the South was excluded. He thus concluded that the minority-status hypothesis interpreted nonwhite-white fertility differ- entials in the South but that the characteristics hypothesis provided a better explanation for those differentials in other regions. To investigate the interrelationship among race, education, region, and fertility in 1965, the current study undertook analyses of variance in fertility separately for the South and the non-South. In the South, 31 significantly higher black than white fertility was found among women who had never attended college, but no racial differential was found for women who had ever attended college. Therefore, the interaction of race and education on fertility in the South suggested the weak form of the characteristics hypothesis. These results are inconsistent with those of Sly, who reported that in 1960, the South provided the only empirical evidence for a minority-status effect. Yet as Sly did not decompose the interactive term found in the South between race, education, and fertility, his conclusion was premature, since the weak form of the characteristics hypothesis also posits an interactive effect among the three variables. Future research may profitably explore Sly's findings by using the statistical technique employed in the present analysis. Outside the South, no important white-black fertility differences were uncovered except for those who had elementary - school educations, (where blacks had somewhat higher fertility than whites) and for those who had college degrees (where blacks had somewhat lower fertility than whites). The black-white fertility differences for these two education categories approached statistical significance (p<0.09). Since no racial differences in natality had been observed in the South among those at the highest education level, the lower black than white fertility observed in the national data for college graduates arose primarily in non-Southern regions. Since blacks were thought to encounter fewer structural barriers to upward mobility in non-Southern regions in 1965, new sociological explanations should be sought for the lower black than white fertility occurring among college graduates in these areas. LIST OF REFERENCES Althauser, Robert P. 1971 "Hulticollinearity and Nonadditive Regression Models." Pp. 453-472 in Hubert M. Blalock, Jr. (ed.), Causal Models in the Social Sciences, Chicago: Aldine-Atherton. Farley, Reynolds 1966 "Recent Changes in Negro Fertility." Demography 3 (1): 183-203. Goldscheider, Calvin, and Peter R. Uhlenberg 1969 "Minority Group Status and Fertility." American Journal of Sociology 74 (January): 361-372. Johnson, Nan E. 1979 . "Minority Group Status and the Fertility of Black Americans, 1970: New Look." American Journal of Sociology 84(6): 1386-1400. Kerlinger, Fred N., and Elazar N. Pedhazur 1973 Multiple Regression in Behavioral Research. New York: Holt, Rinehart, and Winston, Inc. Lunde, Anders 1965 "White - Non-White Fertility Differences in the United States." Health, Education and Welfare Indicators (September): 1-16. Ritchey, P. Neal 1975 "The Effects of Minority Group Status on Fertility: A Re-Examination of Concepts." Population Studies 29 (July): 249-257. Rosenberg, Morris 1968 The Logic of Survey Analysis. New York: Rasik Books, Inc. Ryder, Norman 8., and Charles F. Westoff 1971 Reproduction in the United States 1965. Princeton: Princeton University Press. Sly, David F. 1970 "Minority-Group Status and Fertility: An Extension of Goldscheider and Uhlenberg." American Journal of Sociology 76 (November): 443-459. 32 LIST OF REFERENCES, continued U.S. Bureau of the Census 1960 1970 1971 Census of Population: 1960, PC(1)1B. Washington, D.C.: U.S. Government Printing Office. Census of Population: 1970. Vol. 1, General Population Characteristics, Part 1, Section 1. Washington, D.C.: U.S. Government Printing Office. The Social and Economic Status of Negroes in the United States, 1970. Current Population Reports, Series P-23, N 38. Washington, D.C.: U.S. Government Printing Office. 32a APPENDIX A Correlation Coefficients for the Nation x a H I w u w o u m < XpWPwugmm cmm. mmo.- Fmp. me3.- mwo. Sew. Noo.- mNF. mo_.- mep. me3.- ANF.- eeeceesee x seem .3 mmo.- emm.- omm. _o~. «N..- mmm.- 3°C.- weo.- Fmo.- Pmo.- amp. seem .x m3~.- mmo.- Neo.- cmo.- mmo. eeo.- emo.- mmo.- Peo.- emp. msaeem mecxcez .e moe.- mp_.- mmm. mpm. mam. APP. mc_. mmo. Nm3.- cease? seesea .H coo. LmN.- mm..- 430.- meo.- cc..- P_o.- emp. eeeeeaseee ceases: .: NPO. ceN.- SF..- NNo. ~o_.- Leo. mmo.- eeeae_em .e o~_. m__. emo.- mc_. m~3.- 033.- eeeceesem .L “W amp. ~_o. mm_. emo.- mmo.- encasemec co eecmem .m Fmo.- No_. oo_.- mm3.- eeeeecmec Le scape .o mmm. com. emm. em< .o oeo.- e_~.- canvases Sweet as em< .m emm. wmo_ccms co coPngza .< .cowumz an» to; mpcmwuwemmou comumpmcgou .< xwvcmaa< APPENDIX 8 Correlation Coefficients for the South a H I w u m o o m < aumeume wee. ec_.- mmm. _ep.- owe. mm“. _eN. wcp.- m_m. emN.- pm3.- eeeeeesee x seem .¥ meo.- emm.- emu. me_. wo~.- w~o.- omo.- Loo. mmo.- New. some .4 om~.- w_o.- emo.- _N_.- ___.- 080.- mmo.- Neo.- cup. escape meexcez .H mNe.- Nmo.- new. Nmm. «mo. mm_. moo. ee~.- cease? aPeseL .z Pmo. Noe.- _wo.- meo.- mpp.- emo.- ASP. eeeceaseee oceans: .e u“ o_o. emo.- see. Pmo.- mmo. moo. eewmwpem .e owm. em~.- mmm. mm~.- 3mm.- eeeeeesem .m meo.- m3_. m~3.- om3.- eeeeewmec Le scape .9 _mm. com. mew. em< .u L_P.- eNN.- emewccee emcee an em< .m Nmm. mmmwggme mo compmgao .< .cuzom 63p Low mucmwuwmemou cowumpmcgou .m xwucwna< APPENDIX C Correlation Coefficients for the Non-South w H I pom. mmo.i mNH. omo.i moH.i mN~.i 35 w omH.i NvH. oeo.i mom.: 3 mac. oep. wmo.i emo.i 330.: m nmm. mmo.i mHo.i owH. mom.i omo. o 330. meH. Hmo.i HvH. one. Nmo.i mmo. o m < AuHHHume aeo.- mHH. mNH.- ONH.- compasses x some .3 NoH.- mmo.- «mo.- mpo. seem .e ooH.- emo.- mmo.- amp. aspecm mewxcez .3 meH. NeH. NHH. Hmo.- eeece_ apesea .1 Neo.- meH.- Noo.- mop. eeveeaseee scenes: .9 ope. emo.- eeo. emo.- eewmepem .a emo.- mmp. omo.- mHH.- eepaecseu .u mmo.- «mo. meo.- Heo.- eceeeemec co scape .o cam. mew. emu. em< .u mmo.- moN.- emewcces unset Se em< .m mmm. mmmwcgme Ho comgmgzo .< .cuzomicoz on» com mucmmuHmmmou :oHumngcou no ancmaq< ll".- iICHIan STATE UNIV. LIBRARIES llHIWUIl"“HAWAIIAN1|“IIIHWIIHWIW 31293100629256