AN ECONOMIC ANALYSIS OF FERTILITY DIFFERENTIALS AMONG RURAL-FARM COWVIUNITIES IN THE ERRED HATES IN 1960 “um Ior the Degree of DII. D. MICHIGAN STATE UNIVERSITY Asoka B. Andarawewg 1954 This is to certify that the thesis entitled AN ECONOMIC ANALYSIé‘OT FERTILITY DIFFERENTIALS AMONG RURAL-FARM COMMUNITIES IN THE UNITED STATES IN 1960 presented by ASOKA B. ANDARAWE'A has been accepted towards fulfillment of the requirements for M degree in Mn 1 Economics Majo professor ( Datew 1964 0-169 LIBRARY Michigan State University ABSTRACT AN ECONOMIC ANALYSIS OF FERTILITY DIFFERENTIALS AMONG RURAL- FARM COMMUN IT IES IN THE UNITED STATES IN 1960 by Asoka B. Andarawewa The purpose of this study was to delineate and examine some of the factors that affect fertility differ— entials among rural-farm communities in the United States. Data from the 1960 Census of Population were available on a magnetic tape, capable of computer processing, which en- abled the use of multiple regression techniques to examine the relationships between fertility rates and a set of pre- dominantly economic variables. The index of fertility used in the study was the number of children ever born to ever married females age 15-44 per 1000 ever married females age 15-44 per county in 1960. Fertility rates of white females per county were analyzed for each division, for each region and for the Conterminous United States. County fertility rates of non—white females were analyzed for each division of the Southern region and for the Southern region. Asoka B. Andarawewa The relative importance of the factors differed within each geographical area studied. Education (median years of school completed by males and females age 25 and over) was the most important determinant of white fertility rates in the East North Central, West North Central, East South Central, West South Central and Pacific divisions, the North Central and Southern regions, and in the nation. It was second most important determinant in the South At- lantic division. In all cases the relationship between fertility rates and education was negative, thus emphasiz- ing the influence of education on the attitudes of couples towards family limitation. Other important factors were median family income, median female income, urban influence, the percent of farm laborers and farm foremen in the labor force and married females age 15—24. In most of the divisions and in each region, female income and family income were negatively related to white fertility rates. The relative prevalence of farm laborers and farm foremen tended to raise fertility rates in a county. The relative prevalence of married fe- males in the age group 15-24 depressed fertility rates. Urban influence tended to lower fertility rates. In all the divisions except in New England, in each region and in Asoka B. Andarawewa the nation, the cultural influence of urban concentrations on fertility rates was a function of both the distance of the community from the city and the size of the city itself. Education, family income, and married females age 15-24 were the most important determinants of non-white fertility rates. Urban influence was an unimportant fac- tor except in the East South Central division. The effects of some of the variables were different among the divisions and among the regions. Among the divi— sions, the chief differences in the effects of female em— ployment in the East South Central division, of education in the Mountain division, of family income in the West South Central division and of married females in the West North Central, West South Central and Mountain divisions. Among the divisions in the Southern region, the effect of most variables on non-white fertility rates studied were differ- ent in the West South Central division. Among the regions, the effect of most of the variables on white fertility rates were different in the North Central region as compared to the Southern and‘Western regions. I In summary, high education, high income, and the relative prevalence of married females in the younger age Asoka B. Andarawewa groups, tended to lower fertility rates of both white and non-white females. Urban influence also was an important factor tending to lower fertility rates among white females. AN ECONOMIC ANALYSIS OF FERTILITY DIFFERENTIALS ‘AMONG RURAL-FARM COMMUNITIES IN THE UNITED STATES IN 1960 by Asoka B. Andarawewa A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1964 Cr LXI-8““ 9/3/24 ACKNOWLEDGMENTS The author wishes to express his thanks to Dr. Dale E. Hathaway, who supervised this study. His criticisms and guidance were of invaluable.help. Thanks are also due to Dr. Carl K. Eicher, Dr. A. Allan Schmid and Dr. Lester V. Mandersheid of the Department of Agricultural Economics. Dr. Robert Barrett and Dr. Gerald J. Massey of the Depart- ment of Philosophy, Dr. Subbiah Kannappen of the Department of Economics, and Dr. W. Keith Bryant, presently of the university of Minnesota for their suggestions, criticisms and help, at various stages of my studies. My graduate career at Michigan State University was guided by Dr. Glenn L. Johnson. To him, I owe my intellec- tual training and the present confidence in my ability to make a substantial contribution towards the progress of my country. My wife, Nancy, deserves special thanks, not only for the help on my thesis, but also for her patience and courage under the most trying circumstances. ii Finally, I should note my debt to Dr. L. L. Boger, Hoad, Department of Agricultural Economics, for the generous financial assistance I received, during my stay in this country. iii TABLE OF CONTENTS Page ACKNOWLEDGMENTS. . . . . . . . . . . . . . . . . . . . ii LIST OF TABLES . . . . . . . . . . . . . . . . . . . . Vii LIST OF APPENDICES . . . . . . . . . . . . . . . . . . x Chapter I. FERTILITY DIFFERENTIALS IN THE RURAL-FARM SECTOR 1 Introduction The Empirical Nature of the Problem The Organization of the Study II. THE ANALYSIS OF FERTILITY: A REVIEW OF LITERATURE 9 Introduction Economic Analysis of Fertility III. THE CONCEPTUAL FRAMEWORK: A DISCUSSION OF THE HYPOTI'ESES O O O O O O O O O O O O C O O I O O 24 An Index of Fertility The Conceptual Framework The Discussion of the Hypotheses Urban influence Age distribution of married females Educational level Occupation Female labor force participation rate and Female personal income Total family income Color Regional differences iv Chapter Page IV. THE STATISTICAL FRAMEWORK: THE SOURCES OF DATA AND THE METHOD OF ANALYSIS . . . . . . . . . . 51 The Sources of Data The Equations Estimated The Regression Models: "White Fertility" Equations "White fertility" equation (1) "White fertility" equation (2) "White fertility" equation (3) "Non-White Fertility" Equations Constant Terms The Beta Coefficients Simple Correlation Analysis Statistical Hypotheses V. THE RESULTS OF THE ANALYSIS. . . . . . . . . . . 69 Introduction The Northeast Region The New England Division The Middle Atlantic Division The Northeast Region The North Central Region The East North Central Division The West North Central Division The North Central Division The Southern Region The South Atlantic Division The East South Central Division The West South Central Division The Southern Region The Western Region The Mountain Division The Pacific Division The Western Region The "Non-White Fertility Rates" Analysis The South Atlantic Division The East South Central Division Chapter Page The West South Central Division The Southern Region The Conterminous United States Summary VI. COMPARISON OF THE RESULTS AMONG DIVISIONS AND REGIONS . C O O O I C O C O O O O C O C C C O O 145 Introduction Comparison Among the Divisions "White Fertility" Equations "Non—White Fertility" Equations Comparison Among Regions VII. SUMMARY AND CONCLUSIONS. . . . . . . . . . . . . 158 Summary of Findings and Their Implications for Policy Evaluation of the Study IJTERATURE CITED . . . . . . . . . . . . . . . . . . . 169 APPENDIX 0 O O O O O Q 0 O O O O O O O O O O O O O O O l 7 3 vi LIST OF TABLES Table Page 4.1 Expected Results of the Analysis of the Factors Influencing the White and Non-White Fertility Rates of Rural Farm Females in a County. . . . 66 5.1 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. New England Division . . . . . . . . 73 5.2 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Middle-Atlantic Division . . . . . . 78 5.3 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Northeast Region . . . . . . . . . . 82 5.4 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East North Central Division. . . . . 87 5.5 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. West North Central Division. . . . . 91 5.6 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. North Central Division . . . . . . . 95 5.7 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. South Atlantic Division. . . . . . . 100 5.8 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East South Central Division. . . . . 104 vii Table 5.9 Some of the Factors Influencing the Number of Children Ever Born to Married White‘Females in 1960. west South Central Division. . . . . 5.10 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Southern Region. . . . . . . . . . 5.11 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Mountain Division. . . . . . . . . 5.12 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Pacific Division . . . . . . . . 5.13 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Western Region . . . . . . . . 5.14 Some of the Factors Influencing the Number of Children Ever Born to Married Non-White Females in 1960. South Atlantic Division. 5.15 Some of the Factors Influencing the Number of Children Ever Born to Married Non-White Females in 1960. East South Central Division. 5.16 Some of the Factors Influencing the Number of Children Ever Born to Married Non—White Females in 1960. West South Central Division. 5.17 Some of the Factors Influencing the Number of Children Ever Born to Married Non-White Females in 1960. Southern Region. . 5.18 Some of the Factors Influencing the NUmber of Children Ever Born to Married White Females in 1960. Conterminous United States . viii Page 108 111 115 117 120 123 126 129 132 136 Table Page 5.19 Summary of the Results of the Analysis of Fertility Rates, by Division, Region and for the Conterminous United States, in Rural-Farm Counties, 1960. . . . . . . . . . . 140 6.1 Summary of the Results of the Multiple Compari- son Tests Among Divisions of the Conterminous United States. . . . . . . . . . . . . . . . . 146 6.2 Summary of the Results of the Multiple Compari- son Tests Among Regions of the Conterminous United States. . . . . . . . . . . . . . . . . 154 ix LIST OF APPENDICES The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. New England Division. . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. New England Division. . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. New England Division. . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Middle Atlantic Division. . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Middle Atlantic Division. . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Middle Atlantic Division. . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Northeast Region. . . . . . . . . . . . Page 174 175 .176 177 178 179 180 Table 10 11 12 13 14 15 The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Northeast Region. . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Northeast Region. . . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East North Central Division . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East North Central Division . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East North Central Division . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. West North Central Division . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. West North Central Division . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. west North Central Division . . . . . . xi Page 181 182 183 184 185 186 187 188 Table 16 17 18 19 20 21 22 23 The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. NOrth Central Region. . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. North Central Region. . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. North Central Region. . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. South Atlantic Division . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. South Atlantic Division . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. South Atlantic Division . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East South Central Division . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East South Central Division . . . . . . xii Page 189 190 191 192 193 194 195 196 Table 24 25 26 27 28 29 30 31 The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East South Central Division . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. wast South Central Division . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. west South Central Division . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. West South Central Division . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Southern Region . . . . . . . . . ... . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Southern Region . . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Southern Region . . . . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Mountain Division . . . . . . . . . . . . xiii Page 197 198 199 200 201 202 203 204 Table 32 33 34 35 36 37 38 39 The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Mountain Division . . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Mountain Division . . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Pacific Division. . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Pacific Division. . . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Pacific Division. . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Western Region. . . . . . . . . . . . The Result of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. wester Region . . . . . . . . . . . . . The Results of the Analysis of Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Western Region. . . . . . . . . . . . xiv Page 205 206 207 208 209 210 211 212 Table 40 41 42 43 44 45 46 47 The Results of the Analysis Influencing the Number of Born to Married Non—White South Atlantic Division . The Results of the Analysis Influencing the Number of Born to Married Non-White South Atlantic Division . The Results of the Analysis Influencing the Number of Born to Married Non-White South Atlantic Division . The Results of the Analysis Influencing the Number of Born to Married Non—White of Factors Children Ever Females in 1960. of Factors Children Ever Females in 1960. of Factors Children Ever Females in 1960. O O O O O O O of Factors Children Ever Females in 1960. East South Central Division . . . The Results of the Analysis Influencing the Number of Born to Married Non—White Females in 1960. of Factors Children Ever East South Central Division . . . . . . The Results of the Analysis Influencing the Number of Born to Married Non4White Females in 1960. of Factors Children Ever East South Central Division . . . The Results of the Analysis Influencing the Number of Born to Married Non-White Females in 1960. of Factors Children Ever West South Central Division . . . . . The Results of the Analysis Influencing the Number of Born to Married Non-White Females in 1960. of Factors Children Ever West South Central Division . . . . . . XV Page 213 214 215 216 217 218 219 220 "( b\ “I (n D" CHAPTER I FERTILITY DIFFERENTIALS IN THE RURAL-FARM SECTOR Introduction During the past few decades, there has grown a marked interest in population problems both in the United States and in the rest of the world. Since werld War II, the attempts at economic development by the underdeveloped countries, which have high rates of population growth, have made population problems the focus of intensive study. Sociologists and demographers have carried the burden of demographic research tillrecently. Their studies have analyzed trends and differentials among geographic divi- sions by socio—economic status, residence and color. Hewever, the area of population study is becoming in- creasingly inter-disciplinary. Economists, in their analyses, have long considered Population as an exogenous variable. But the recent empha— sis placed on economic development has stirred the interest 0f the economists in problems of population growth. A con- siderable number of studies have been conducted on the relation between population growth and economic develop: ment.1 These studies have stressed the impact of pOpulae tion growth on development, especially the barriers to development presented by high rates of population growth. For example, the study by Coale and Hoover made use of economic models to estimate the rate of growth of Nation- al Income, the rate of growth of investment and the dis— poSition of labor supply assuming different population projections.2 But the influence of economic factors on popula- tion growth has received scant attention, especially the study of the influence of economic factors such as em- ployment, income, education, etc., on fertility rates. Occasionally studies have attempted to relate a variable, e.g., urban influence, to fertility. But these studies have been such that the relevance of their results was limited. The relationship of economic variables to 1H. Leibenstein, Economic Backwardness and Econ- gmic Growth (New YOrk: John Wiley and Sons, Inc., 1957). A. J. Coale and E. M. Hoover, Population Growth and Economic Development of Low-Income Countries; A Case §£udy of India (Princeton: Princeton University Press, 1958). . 2Ibid, pp. 227—294. a1 «U nu. a.» in 'r on ... |¢a V i a? s a Cu . e I I ., I: In]. I I: v I. . .nr. I [ill -w'}.ll fertility differentials;-not among countries but within a country stratified broadly by census or political divi- sions, by color and by residence, has-been relatively unexplored. No extensive study has been undertaken on the differentials in fertility arising from differences in cultural and economic factors among rural-farm com- munities. The main reason for the lack of such studies, other than the economist's tendency to view the field of population study as outside the scope of his disci- pline, is the lack of adequate data. Either adequate measures of variables were absent or they have been available in units inappropriate to meaningful analysis. For a considerable period, data on many variables were available at the state level and not below. This is true even in the United States, where the coverage of the census is extensive, refinements and replacements of census concepts are continually made to improve the collectipn of data. It is pertinent to note that devel- opment of research on fertility in the united States has been dependent upon the development of the census 0f population. Better data were available for an analysis of fertility differentials in the 1960 Census than have been‘ previously available. In the 1960 Census of Population, a host of economic, demographic and sociological charac- teristics of the population were gathered for a 25 per- cent sample of households in the United States. These data were available for population groups classified by residence at the county level. Moreover, they were avail- able for the first time on electronic computer tapes which facilitated analysis on a scale hitherto impossible. These data enabled the study of factors which affect fertility rates among rural-farm communities. The primary purpose of this study was to examine the relationship of several factors on fertility differ- entials among rural-farm women in the United States in 1960. The study was undertaken for the census divisions, regions, and for the counterminous United States as a whole. Previous studies have concentrated either on the computation of trends and differentials by occupation, residence, or socio-economic status, or on explaining how these trends have occurred. Studies relating econ- omic variables to fertility have concentrated on one indicator of economic conditions, namely, income or urban influence, or were confined to a limited geographic area. The major hypothesis of the study was that economic ' factors were responsible for different fertility rates within a geographic area, and that there were differences in the importance of these factors among the geographic areas. In considering other factors along with urban in- fluence, this study attempted to analyze the effects of a group of factors and to clarify the relations between these factors and the resulting fertility differentials. A majority of the variables used in the analysis were related to the characteristics of the population in the community. Other variables were related to the loca- tion of the community with respect to other communities. The Empirical Nature of the Problem The problem studied pertained to inter-community fertility differentials within the rural-farm sector. The purpose of the study was to identify and evaluate some of the factors that affected these differentials. Prior to the actual study, the operational definition of a rural4farm community has to be clarified. A rural community may be defined as a group of peOple living in a limited area who are engaged primarily in agricultural olrlh Ur. 6 occupations. Data based on such a concept were not avail- able. The concept of a rural community in this study closely follows the census classification. The United States Bureau of Census has experimented with new definitions and concepts to measure the transformation that has taken place in the residential distribution of the population. Since the first census of 1790, there has been a continuous change in the residential complexion of the population. Peterson has re- capitulated the development of the differentiations used in each Census.3 At the beginning of the nineteenth century, according to the Census definition, the rural population were mostly farmers, who lived in the countryside in unin- corporated places. There were few ambiguities in the opera- tional definition of a rural community. Today the census classifies the population in a county on the basis of resi- dence in an urban place, or in a rural area. Those people living in a rural area were further classified into rural- non-farm and rural-farm on the basis of their residence on a place defined as a farm. Residents in a rural area, to be classified as rural—farm, had to live on a place of 10 acres or more from which the sale of farm products amounted -___¥ 3W: Peterson, ngulation (New York: MCMillan Company, 1961), Table 8-2, p. 188. The principal source for this table was Leon E. Truesdell, "The Development of the Urban- Rural Classification in the united States," United States Bureau of Census, Current ngulation Reports, Series P-23, Nb. 1, 1949. to $50 or more in 1959 or on places of less than 10 acres from*which the sale of products amounted to $250 or more in 1959.4 It is possible, under this definition, that residents in rural areas, though their main occupation was non;farm, were classified as rural-farm by meeting the above requirements. In this study the rural-farm resi- dents in a county, as defined by the census, were taken to be the rural community. The operational variable used was the fertility rate of rural-farm women. Fertility has to be distin- guished from fecundity. Fecundity is the physiological capacity to bear children and fertility is the realiza- tion of this potential, the actual birth performance as measured by the number of offspring. The fertility rate measures the number of children ever born per 1,000 ever— married women, age 15-44. Data on children ever born were enumerated in the census of 1960. A detailed dis- cussion of this variable is included in Chapter III. 4For a detailed discussion of the bases of the rural-non-farm, rural—farm division, see U.S., Bureau of Census, United States Census of ngulation: 1960, united States Summary, General Social and Economic Char- acteristics, P. C. (1), 1C, pp. vii-viii. (Hereafter this publication will be referred to as P. C. (1), 1C). The Organization of the Study Much of the research on fertility has been con- fined to the analysis of trends and differentials by socio-economic groups, residence, and color, using data from the united States Censuses. Few studies have con- centrated on the explanation of such trends and differ? entials. Little attention has been paid to the study of the effects of economic variables on differentials in fertility rates. Chapter II contains a review of the literature analyzing fertility differentials. In Chapter III, the hypotheses tested in this study are discussed. Chapter IV discusses the statistical methods used in the study, In Chapter V, the results of the analyses of fertility rates are discussed by each divi- sion, region, and for the conterminous United States. Chapter VI compares the results among divisions and re- gions, and Chapter VII summarizes the results of the study. The Appendix contains detailed statistical results. CHAPTER II THE ANALYSIS OF FERTILITY: A REVIEW'OF LITERATURE Introduction A considerable amount of literature is available on the subject of fertility. These studies can be grouped into three broad categories. The first set contains stud- ies which measure trends and differentials in fertility rates. The United States Census Reports have been the basis of such studies. Data on ratios of children to white women are available by states from published census data for each census year since 1800, and by counties, townships, and cities up to 1840 and since 1890, a part of the necessary data being omitted from the censuses of l Rural-urban fertility ratios are available 1850 to 1880. since 1910, and rural-farm and rural-non-farm ratios since 1930. Beginning with the census of 1890, census enumera- tion included women by age by marital status for states and other areas, and hence, ratios of children to married 1P. K. Whelpton, Needed Population ResearCh (Lan- caster: Science Press Printing Company, 1938), p. 41. 9 10 women can be secured for this period. Also, the number of children ever born to ever married women was enumer- ated for the censuses of 1890, 1900, and 1910. It was enumerated for a 5 percent sample of married women in 1940, and for a 3-1/3 percent sample in 1950. Since 1890, the inclusion, in the census, of children ever born, provides a better index of fertility than crude birth rates or other general measures. The above ratios computed from census data make it possible to show trends of fertility for all white women in the nation and each state during a period of 160 years, and for all women and for married women by race, nativity, and residence in the various political units for shorter periods. The availability of such data has led to the build-up of a large body of know- ledge regarding fertility trends and differentials in this country. The literature includes studies of trends and differentials in fertility between states, cities, counties, and rural areas in different parts of the na- tion: between large cities, small cities, rural—non-farm, and rural-farm groups in the same part of the nation, and :9 hf is *1 A. u *u L"! (5) 11 Between racial and nativity groups.2 These studies also have related trends and differ- entials in fertility rates to occupational groups, socio— economic status, and educational level.3 The phenomenon of differential fertility according to occupational or socio-economic status has sometimes been described as a transitional phase of declining fertility. The theory postulates that family limitation practices find their first acceptance and extensive use among the so-called "upper" classes in urban areas. Later it spreads down- wards to the so-called "middle" classes, and finally to the lower classes. The spread of knowledge and acceptance of family limitation practices is accompanied by a decline 2W} S. Thompson and P. K. Whelpton, Population Trends in the United States (New York: McGraw-Hill, 1933). W. H. Grabill, C. V. Kiser, and P. K. Whelpton, The Fertility of American WOmen (New Yerk: John Wiley and Sons, 1958). T.CL WOofter, "Trends in Rural and Urban Fertility Rates," Rural Sociology (March 1959), XIII, No. l. I 3P. K. Whelpton, "Geographic and Economic Differen- tials in Fertility," Annals of the American Academylof Pol- itical and Social Science, CLXXXVIII, (November, 1936). C. V. Kiser, "Differential Fertility in the Uhited States," Demographic and Economic Change in Developed Coun- tries (Princeton: National Bureau of Economic Research, 1960), pp. 77-112. 12 in fertility. In the meantime, such knowledge spreads outwards from the urban areas to the rural areas, and the process presumably follows a similar pattern, the declines in fertility beginning in the upper classes and spreading to the lower classes. If fertility de- clines do proceed on the pattern described above, first an enlargement, and later contractions of group differ- ences in fertility can be expected. In the United States, the theory has been consistent, to a lesser degree with the data since 1800, and is in close agreement with the data since 1900.4 The second set of studies relate to the ways in W which the trends and differentials have occurred, includ- ing psychological factors and voluntary control measures. On the one hand, fertility trends and differentials are attributed to biological factors such as the decline in the physical strength of the people vis a vis the pre- vious generation. On the other hand, emphasis was placed on the planned actions of human individuals, such as early or late marriage and voluntary birth control measures. Studies conducted by public health authorities and by private groups prove conclusively that conscious control k 4Grabil, Kiser, and Whelpton, op. cit., chaps. 2, 31 and S. 13 of fertility is more important than biological or physi- ological factors in bringing about the observed trends and differentials in fertility. The Indianapolis Study, 1941, based on house-to-house interviews of 41,498 native white couples, found a large percentage (90 percent) of the couples interviewed using birth control measures.5 A recent study showed that most American families are planned irrespective of religion, income, education, occupation or residence differences.6 While there is much evidence of fertility differ- entials for women grouped on geographic, economic, and social bases, and some evidence that these differences have occurred in a major degree due to individuals con- stituting certain groups having better information about contraception, there has been relatively little attention paid to the conditions affecting the differences in fer- tility. There is a dearth of research on the weight of various factors determining the individuals' or couples' age at marriage, the extent to which family limitation 5C. V. Kiser and P. K. Whelpton, Social and Psy- phplogical Factors.Affegtipg Fertility (New YOrk:.Milbank Memorial Fund), II, Part VI, p. 253. 6R. Freedman, P. K. Whelpton, and A. A. Campbell, Family Planning, Sterility, and Population Growth (New York: McGraw-Hill Book Company, 1959), p. 155. H— I. - b ' Y ' s ‘1 14 is tried, the number of children wanted by couples and the extent to which these factors are subject to social control. The third set of studies constitute the meager literature available on the attempts to measure the in- fluence of psychological, social, and economic factors on the fertility of racial, geographic, and residence groups. Fertility studies based both on census data and private surveys show that differential fertility has arisen mainly due to differences in the use of family limitation practices. Voluntary family limitation is achieved by the conscious efforts of individuals through the use of devices to prevent conception. The acceptance and use of contraceptive devices by a given couple de- pends on the cultural mores of the society of which they are a part, the attitude of their religious persuasions towards family limitation, psychological and social fac- tors, and economic factors which determine the economics of child-bearing and rearing. The effect of psychological factors on fertility is studied in the Indianapolis Survey. The evidence gathered in the survey did not yield a strong relation- ship between psychological variables such as a feeling of personal inadequacy, ego-centered interest in children, . 'R :u va 5; .1 "f A" 15 or felt restriction of present freedom and fertility.7 According to the authors, "Our measures of psychological characteristics were too crude to afford precise differ- entiations, [and] it may be little wonder that the study failed to indicate strong and consistent relations of fer- tility behavior to psychological characteristics among so homogeneous a sample."8 They further stated that the chief lesson to be learned from the study was that fertility is generally more closely related to broad social factors (in- cluding economic) than to psychological factors. They con- cluded that family planning and actual fertility were clear- ly correlated with socio-economic status; but when this var- iable was held constant, the observed relation of fertility behavior to most psychological characteristics considered was generally less pronounced or less regular.9 In the present study, psychological variables were not considered, not due to their unimportance, but essentially due to lack of adequate measures of such variables as would facilitate quantitative analysis. 7Kiser and Whelpton, OE- Cit.. III. Part VI: P- 829- 8C. V. Kiser and P. K. Whelpton, "Summary of Chief Findings and Implications for Future Studies, "Milbank Mem- ppial Fund Quarterly, XXXVI, No. 3 (July, 1958), p. 323. 9Ibid.. pp. 318-19. I he 16 The effect of cultural variables has been long emphasized, both by demographers and sociologists. For example, Landis, in explaining the higher rural birth rate states, Rural life has always emphasized the family, placing family interests above individual inter- ests, and is much more biological in its entire setting than is urban life. . . . Children are taken for granted in the rural family, mating and reproduction are part of the normal scene. . . . There are several logical reasons for a family on the farm where family members are en- gaged in a joint enterprise, have a common so— cial life and live in a community where other families have children and where social functions are family rather than individual affairs, planned to include all age groups rather than one single age group. These general patterns of real culture all he18 to explain the much higher rate of rural areas.1 The influence of religion has also been studied, but the analysis was confined to computing fertility ratios for different religions by different socio-economic groups.11 The use of Census data precludes exact qualifica- tion of cultural concepts. Hewever variables were in- cluded in this study to represent for example the influence of urban culture patterns. The next section discusses the literature on the economic analysis of fertility. 10Paul H. Landis, Population Problems--A Cultural laterpretation (2nd ed., prepared by Paul K. Hall, New Yerk: American Book Company, 1954), p. 234. llFriedman, Whelpton, and Campbell, loc. cit. E 1" .- r, .h 17 Economic Analysis of Fertility The effect of economic variables on fertility has been studied for a considerable period of time. Some of the studies examine trends in economic conditions and trends in fertility for the United States as a whole.12 These studies indicate a positive correlation between fertility and business activity. Recently there has been an increased interest in fertility research, using the theories and econometric methods developed in econ— omic research. Becker uses the conventional theory of consumer demand to analyze the influence of income on fertility. He assumes that children are viewed primar- ily as a conSumer or a productive good. Hewever, in a urban setting, children lose their value as productive 12W} F. Ogburn and D. S. Thomas, "The Influence of Business Cycle on Certain Social Conditions," Journal of the American Statistical Association, XVIII (1922), pp. 324-40. "” D. Kirk and D. L. Nortman, "Business and Babies: The Influence of the Business Cycle on Birth Rates," Egg- gpedings of the American Statistical Association, Social Statistics Section (1958), pp. 151-160. R. A. Easterlin, "The American Baby Boom in its Historical Perspective," American Economic Review, LI, No. 5 (December, 1961), pp. 869-911. Grabill, Kiser, and Whelpton, Op. cit., chaps. 5. 6, 7,pp. 113-203. qAF 4.x. 4 we} ‘ I on (I: as b. (I, u. l8 goods and are what are called in economic theory, superior goods. The number of children "consumed" then, increases as income increases.13 Another development of macro- economic theory, the Relagiye71pcome Hypothesis is used by Freedman to analyze the relationship between economic status and fertility. All the data in the study are from interviews conducted for the "Growth of American Families" study.14 Freedman considers as the income variables not only the husband's actual income, but his income compared with the average earned by men of his age, educational level, and occupation. The ratio between the actual in- come of the husband and the income customary in his socio- economic group is called the relative income. Other econ- omic variables used include the number of years worked since marriage, wife's income, wife's labor force status and wife's future work expectation. The women were divided into two groups, one group married ten or more years and the other married five to nine years. The study shows a stronger relationship when using the relative income, but 13G. S. Becker, "An Economic Analysis of Fertility," ngographic and Economic Change in Developed Countries, Princeton: National Bureau of Economic Research, 1960), PP. 209-31. 14Freedman, Whelpton, and Campbell, op. cit. V‘ s. c. o‘. § 1 p y b 4‘ 19 the two income variables for the husband were the most important. The total variance in family size explained by the six economic variables for the group married ten or more years was only four percent. For the second group, the work variables of the wife were most impor- tant, and the husband's income variables showed no con- sistent relationship to fertility. The total variance explained in this group by the economic variables was eighteen percent. The addition of demographic variables increased the coefficient of multiple determination (R2) from .18 to .24 for the second group and from .04 to .10 for the first group. While the results show a consistent relationship between some of the economic variables and fertility, the proportion of the total variation in fer- tility rates explained is small. This probably can be explained by the purposive nature of the data. The study was restricted to white non-farm "fecund planners." The study showed a concensus among U. S. families to limit the range of both expected and desired family size. The desired norm for the fecund planners was two-four children. Since this is a marked concensus on a limited range of de- sired family size, there is less variance in family size to be explained, and less likelihood that a small group 20 of factors will be able to account for it.15 In another study Adelman, using data gathered by the United Nations, examines the effects of economic var— iables on fertility for a sample of thirty-seven countries whose annual per capita incomes range from $125 (Morocco) to $1900 (United States), with half the incomes below $350.16 Functions were fitted using least squares re- gression methods with age-specific birth rates as the dependent variables, and the percentage of the labor force employed outside agriculture, an index of educa- tion, population density, and the level of real national income per capita as the independent variables. The re- sults of the study supported the hypothesis that, ceteris paribus, age-specific birth rates tend to vary directly with per capita income in the long run, that urbanization leads to a reduction of births in the long run, that the level of education is negatively correlated with fertility, and that over-population generates its own antidote by 15D. S. Freedman, "Relation of Economic Status to Fertility," American Economic Review, LIII, No. 3 (June 1963). pp. 414-26. 161. Adelman, "An Econometric Analysis of Popula— tion Growth," American Economic Review, LIII, No. 3 (June, 1963), pp. 314-39 C. v. .s. .a I Il'lILIl 21 leading to fewer births. One criticism of the statistical procedure relates to the possible high inter-correlation among the independent variables, especially between the percent of the labor force employed outside agriculture and income, which would affect the significance of the regression coefficients through increased standard errors. The study was a pioneering effort in that it has used statistical methods to show that economic variables account for a significant amount of the fertility differ- entials among countries. Easterlin's study of the post-war baby boom incor- porated an analysis of rural-farm fertility. Although the main emphasis of the study was on the non-farm sector, it yielded important implications for rural-farm fertility. Easterlin correlates birth rates with the Kuznets cycles of economic change. His findings provide an historical ex- planation of rural farm fertility trends. The accelerated decline of rural fertility in the Twenties and early Thir- ties are attributed to a set-back in the growth of farm in- come. The subsequent baby boom in the rural areas would be explained by the resurgence in farm income growth in the —* 17R. A. Easterlin, op. cit., pp. 889-90 Av- v. Ar I... r [1) . 22 late Thirties and Forties. The decline in the rate of growth in fertility in the 1950's is explained by the tapering off of the boom and the substantial drop in farm income growth. The increased significance of non-farm sources of total in- come and the progressive rise in the proportion of the farm's population not engaged in agriculture have tended to reduce the effect of farm income decreases on farm fertility. Other studies have attempted to examine the effect of urban con- centrations on farm fertility. These have been examined in Chapter III. In summary, although a considerable amount of liter- ature is available on fertility analysis, a major proportion is devoted to the measurement of trends and differentials and to explanations as to their occurrence. Studies anal- yzing the influence of economic variables on fertility are of recent origin. There is only one instance of the appli- cation of economic models to the study of fertility. The study of the influence of economic conditions on rural-farm fertility was incidental to the main emphasis of Easterlin's study. There are no major studies which analyze relation- ships between fertility rates and economic variables within a county, among the various political units by urban-rural residence or by color. Me a. A‘ «ad :a In. 23 A substantial amount of research is still to be carried out on the influence of economic variables on fer- tility by residence groups and color within the country. Econometric analysis of rural-farm fertility could be fruitful in understanding how differences in economic variables lead to differences in fertility among rural- farm residents in the various geographic units. It is hoped that it will also contribute to the meager litera- ture in this area. v. ‘1 I'. y" ‘\ CHAPTER III THE CONCEPTUAL FRAMEWORK: A DISCUSSION OF THE HYPOTHESES A chapter outlines the conceptual framework within which the study of inter-regional rural-farm fertility dif— ferentials was conducted. It discusses the variable used to measure fertility. The independent variables are intro- duced, their rationale discussed, and their expected rela- tionships with the dependent variable examined. An Index of Fertility Several measures are used as indices of fertility in the literature. These include crude birth rates, speci- fic birth rates, gross reproductive rates, net reproductive rates, the general fertility rates, the nuptial, or marital, fertility and the age-specific fertility rate. The fertility rates are measures used to refine and supplement the birth rates. The limitations and advantages of these various measures are discussed in the literature.1 lLandis, op. cit., pp. 157-8. Peterson, op. cit., Appendix, pp. 622-28. Grabill, Kiser, and Whelpton, op. cit, pp. 13, 128, Appendix A, pp. 400-04. 24 a" (T, h. ‘I« 25 In this study, the marital fertility rate was used as the dependent variable. This ratio refers to the number of children ever born per 1,000 ever married females age 15-44. The census data on fertility for 1960 were based on answers to a single question on the number of babies ever born, which was accompanied by a specific caution to exclude stillbirths.2 The enumerator was instructed to leave out adopted children, stepchildren, or other children not born to the women; on the other hand, the count was to include children born to the women before the present mar— riage, children no longer living, and children away from home, as well as children borne by the women who were still in the home. As it is possible in any census enumeration, over- counting or under-counting of children ever born could occur. Over-counting is possible through the inclusion, intentional or unintentional, of an adopted child, a stepchild, or a fe- tal death. An undercount may occur from the reporting of only births from the current marriage, or exclusion of ille- gitimate children (who should be included). However, it is believed very likely that many of the unwed mothers living L; 2P. C. (1), 1C, op. cit., p. xxv. - 4 I“. 1 ‘V7 a... u n,. . ...- (I) o . Vo \. 1). 26 with an illegitimate child reported themselves as having been:married, and therefore were among the women who were expected to report the number of children ever born, and that many of the mothers who married after the birth of an illegitimate child counted that child. Errors in the data are more probable for the rural women than for the urban women, and for the non-white women, especially in the South, than for white women. The above statements are speculative. The enumerator is not expected to probe in- tensively to obtain a precise answer to all parts of the records, especially when additional, often personal, ques- tions are required to be sure. However, it is probable that the reports on children ever born do not contain any systematic bias, and that the measure offers a meaningful index of rural farm fertility. The Conceptual Framework The study was cross-sectional in nature, in that a number of geographical units were investigated at the same point in time. The data were observations on economic and demographic factors which vary from community to community. The hypotheses postulated that fertility rates of rural-farm Women vary from community to community in accordance with inter-community variations in social and economic factors. 27 The statistical analysis of relationships among variables follows the specification of a model. A model is, loosely, a set of statements that are consistent with the advanced economic and statistical assumptions made by the investigator. The hypotheses to be tested are usually drawn from the model. In this study, the specification of such a model was based on assumptions about the nature of the factors, and about the relationships of the factors which were expected to explain intercommunity differentials in fertility rates of rural-farm women. The hypotheses to be tested can be grouped loosely with respect to these assumptions: (1) Some of the factors which account for different fertility levels of women in rural-farm communities vary from community to community according to the location of the community with respect to other communities, and with respect to the size of other communities. Three alterna- tive measures of urban influence were tested by estimation of three equations which had the fertility rate as the de- pendent variable, but each of which included only one of the above measures. (2) Fertility rates may also differ among communi- ties due to variations in certain factors among communities 1.1 q I..-‘ ,.- ”:7 the. op! bl.- \C] pl' #4 28 and their population. These factors include income, educa- tion, employment, occupational distribution, and demographic variables. Variations in the above factors among communities lead to differential fertility rates. Variables measuring these factors were included in the estimated equations. (3) Fertility rates may differ among regions due to differences in the effect of these factors in different re- gions. This applies both to the affect of urban influence and to the socio-economic variables. The equations were es- timated individually for each census division, and region, of the United States, and for the conterminous United States. In the following sections, the independent variables and their expected relationships with the fertility of rural- farm women are discussed. In addition, the expected relation- ships among the independent variables are noted and discussed. The Discussion of the Hypotheses Urban Influence Urban influence is one of the factors which is ex- pected to explain fertility differentials among rural com- munities. Urban influence stems from the presence of urban centers near the communities. The growth of large urban centers has been the result of the rapid industrialization of the economy. Large urban centers with industrial bases ur“ .— n“: 51- “ a; 9-. 5v. p. u. a. A)! 29 have grown up throughout the united States. A map showing the distribution of population in the United States indi- cates that around each concentration of population marked by a city, the density of population decreases as the dis- tance from the city increases. At all distances, the hin- terlands of large metropolitan communities are more urban in comparison than the hinterlands of small metrOpolitan communities. Further, the characteristics of the popula- tion are expected to be different in communities which are nearer to the central city than those which are farther away from the city. It is hypothesized that differentials in fertility rates of rural-farm communities are affected by the location of the community with regard to urban cen- ters and the size of the centers. There is a considerable body of evidence which in- dicates that the rural population which lives within ready access of large population centers differs in its charac- teristics from the rural population located at more remote distance from such centers. One of the earliest studies was conducted by Brunner and Kolb.3 They made comparisons of concentric zones, extending out of cities, on a large —_ 3 E. deS. Brunner and J. H. Kolb, Rural Social Trends (New York: McGraw-Hill, Incorporated, 1933), pp. 111-43. a... Soul .I. .a. All. 9‘?) 30 number of factors. Taking the city as the center and the adjoining counties as tier one, the next counties as tier two, and so forth, they found measurable differences in the characteristics of the rural population between the tiers. The results, insofar as they pertain to fertility, show that urban characterisitics such as lower birth rates and fewer children were present in the city and tier one coun- ties, and that such characteristics diminished as the dis- tance of the county from tier one increased. Other studies have generally indicated that both the size of the urban center and the distance away from the center are important factors in producing inter-community differentials in rural- farm fertility.4 At this point it is relevant to discuss the ration- ale underlying the hypothesis that fertility levels of rural- 4For example, see W. S. Thompson and N. E. Jackson, "Fertility in Rural Areas in Relation to their Distance from the Cities, 1930," Rural Sociology, V, No. 2 (June, 1940). However, in this study, no strong relationships were found between fertility differentials and gradations of urban in- fluence. D. J. Bogue, The Structure of the Metropolitan Com- nmnity (Ann Arbor: University of Michigan, 1949). The rele- vance of Bogue’s study is limited only to an analysis of rural population density. 0. D. Duncan and A. J. Reiss, Jr., Social Character- lptics of Urban and Rural Communities, A VOlume in the Cen- §_§ Monograph Series (New York: John Wiley and Sons, 1956), p. 157. 31 farm communities are inversely related to the distance of the community from urban centers and to the size of the ur- ban center. A study of the fertility of the rural farm sec- tor cannot be carried out independently without reference to the influence of urban centers. Such a task was pos- sible, for example, in 1860, when the United States was pre- dominantly rural and sixty percent of the pOpulation lived on farms. Communications between the farm and city were minimal, due to lack of transport facilities and mass media. Today the isolation of the farm with the concomitant unique social characteristics has all but disappeared. Approxi- mately seventy percent of the population of the United States were urban residents in 1960, while the rural-farm population was 7.5 percent. Seventy-six percent of the urban residents lived in urbanized areas.5 Urban areas exert an influence over the rural areas which transcends the census boundaries between the two sec- tors. These influences, as they relate to fertility, are mainly cultural in nature, and are associated with urbanism. urbanism is the rise of new culture patterns, new attitudes, and values which are characteristic of urban populations. M . 5P. C. (1), 1C, op. cit., p. ix. An urbanized area 18 a city of 50,000 population or more, along with a densely Populated urban fringe surrounding it. 32 There is an observed tendency of families to be smaller in highly urbanized areas. It is often argued that there are distinct culture patterns in the cities which encourage low fertility.6 Modern city life provides a particularly favor- able environment for the development of attitudes motivating family limitation. Many reasons are adduced for such atti- tudes. Urban culture patterns are based on competition, individualism, and a struggle for status. City family life is less cohesive as family members participate in other in- stitutions and have a broader contact outside the family. This can be contrasted with the rural family which is still, to a considerable extent, a social and economic unit.7 In the city, children do not have the same economic status as on a farm. There is no scope for employment of children. On the other hand, they have to be provided with an expen— sive technical or professional education to enable them to compete in the highly selective urban labor markets. Medi- cal care, living expenses, maternity expenses, are high in 6United Nations, "The Determinants and Consequences of Population Trends" (New York: United Nations, Department of Social Affairs, Population Division, 1953), p. 78. W} S. Thompson, Population Problems (New York: Mc- Graw-Hill, l943),pp. 207-10. 7Landis, loc. cit. 33 the cities. It costs less physically, as well as economi- cally to rear children where there is plenty of housing space and adequate playgrounds, as on a farm. The above factors tend to limit the size of the urban family. The role of women is also different in cities. The equal status given to city women in many aspects of public life and employment has reduced the traditional role of women as homemakers and mothers. Both city and rural women have a variety of interests, such as in clubs and associa- tions. However, the main interest of such clubs, etc.,of rural women is centered around the home and family. Further, status aspirations are more prominent in large cities. Where the consumption habits of a family are determined, as it has been often proposed, by the actions of its peers in the social scale, children and consumption goods compete for the available resources. ”Keeping up with the Joneses" is more difficult if there is a large family re- quiring support. Social mobility is also a factor tending towards fewer children in cities. The desire to improve one's status is associated with the phenomenon of Dumont's social capillarity. Just as a column of liquid must be thin to rise under the force of capillarity, so also must a family ‘n bet I‘. In 9,. v. I 1 '1 I4- h. I 34 be small in order to rise in the social scale. Social mo- bility is more feasible with one or two children than with a large number. The above facets of urban life lead to attitudes motivating small families and fewer children. The effect of urban culture patterns on rural—farm fertility arises from the diffusion of these patterns from urban centers. In the United States, by virtue of the urban complexion of the population, urban culture is the dominant and dy- namic culture. Urban and rural culture patterns tend to intermingle with one another. But the relative dominance of urban culture constantly influence and change rural cul- ture patterns. Hence the attitudes of rural-farm communi- ties towards fertility will be influenced by those of the urban communities. The extent of the influence is deter- mined by the degree of exposure to urban culture patterns, which could occur either by personal contact or by mass communications media. It follows from the above discussion that the dis- tance of a rural community from an urban area will be directly related to the amount of exposure of the rural people to urban culture. In the case of personal contact, the amount of exposure to urban influence is clearly some function of distance. In the case of exposure via mass Ax. ,a, .. ow .\- V1 ~§v wll rl‘-u'+.I-L, 35 communications Inedia, it is very likely that the closer a rural community is to an urban area, the more probable it is that the newspaper would be a city paper rather than a local one from a rural town, and that the radio or tele- vision station will be a city rather than a local one, and so forth. In the area of mass communications, it is hypothe- sized that the size of the population of the urban area is likely to be important. Large cities offer a greater di- versity of cultural systems in their mass communications than do small cities because of the diverse nature of the population in them. Further, large cities not only have numerically a large number of television and radio stations, but also, their mass communications media have a wide geo- graphic coverage. Hence, it is hypothesized that urban culture pat- terns are favorable to low fertility and that those patterns diffuse over from the urban centers, influencing rural value systems. The extent of the exposure of rural culture to urban culture is both a function of distance of the rural community from the urban area and of the size of the city. Fertility rates in rural-farm communities near to urban cen- ters are expected to be lower than those in rural-farm com- munities near to small urban centers. a». .n. we» a... VA ox. fr. 4“ 36 To transform these hypotheses into operational vari- ables, it was necessary not only to emphasize the proximity of the rural-farm community to the center, but also the size of the population of the center. One of the most relevant concepts of the urban community, expressed in terms of cen- sus data, is the SMSA.8 It includes the inhabitants of the county or counties surrounding a major city, and their pat- terns of work and communications with the city. It provides a functional and a geographic and demographic entity. One of the three alternative measures used in this study to rep— resent urban influence on rural—farm fertility was called the distance variable. Each county was assigned a number which was a linear function of its distance from the near- est SMSA. The hypotheses underlying the variable are the following: a) the influence of an SMSA on the fertility rates in nearby counties is a linear function of the distance of the county from the SMSA; b) rural-farm fertility levels do not vary among counties in which cities of 50,000 or more are located because of varying population size of the city; C) the effect of a large SMSA on fertility levels in a com— munity "x" miles distant are the same as the effects of a 81n general, an SMSA is a county in which a city of 50,000 or more is located. See P. C. (1), 1C, op. cit., p. x, for a detailed discussion of the subject. pf.- .V 37 small SMSA on fertility levels in a community "x" miles distant. The other two measures alter the hypotheses ex- pressed by the distance variable. They represent the hy- potheses that, a) the influence of any SMSA on the fertil- ity levels of nearby counties is a joint linear function of the distance between the community and the SMSA, and of the population size of the SMSA. Implied here is that the influence of Chicago is greater and extends further than the influence of Denver; b) the effects of the pre- sence of‘a city, 50,000 population or more, in a county, is a linear function of the population size of a city up to a population size of two million. It is hypothesized that cities of two million or more have similar influences on the fertility levels in the county in which they are lo- cated, and on outlying counties the same distance away. The two measures other than the distance variable differ with respect to the maximum area over which they hypothesize a city of a given size extends its influence. The size-distancel variable expresses the hypothesis that a city of two or more million population extends its in- fluence up to a maximum of 250 miles. Cities larger or smaller than one million are hypothesized to extend their influence in proportion according to their population size. h ‘.,| ril’hfl... 38 The size-distance2 variable expresses the hypotheses that a city of two million or more extends its influence up to a maximum of 250 miles, whereas a city of one million extends its influence up to a maximum of 150 miles. Again, cities greater or smaller than one million extend their influence in proportion according to their population size. The pro- cedures for assigning values to counties allowed interven- ing cities to add or cancel out the effect of any particular city on a particular county. In summary, differentials in the fertility levels of rural-farm women in communities at the center and at the periphery of an industrial—urban matrix can be explained by differentials in the economic demographic and cultural in- fluences of the matrix upon the community. The economic and demographic influences include differentials in educa- tion, rate of female labor force participation, and occu- pational status; the differentials in cultural influence arise from the differential diffusion, determined by the location of the community to the center, of urban culture patterns and the concomitant attitudes toward family size and acceptance of contraception. It is also hypothesized that the cultural influence was a function of city size as well as of distance from the city. 39 Age Distribution of Married Females The fertility rates specific to the age group of married females reveal that the number of children ever born increases as the age of the group increases. For the rural— farm sector of the conterminous United States in 1960, the number of children ever born per 1000 ever married women age 15-24 was 1,436; for the age-specific group 25-34, it was 2,745; and for the group 35-44, it was 3,125.9 The number of children born to a woman is in part a function of time° As the age of the women increases, the number of children born increases. Intercommunity fer— tility rates will be affected by proportions of married wo- men in the different age groups. The previous discussion, therefore, suggests the following hypothesis. Fertility rates of rural-farm fami- lies are inversely related to the percent of married females who are in the younger age-groups. The variables used to represent this hypothesis in the statistical analysis are the percentage of ever married females age 15-44 who are 15-24, and the percentage who are age 25—34. ‘ 9P. c., (1), 1c, 02. cit., Table 82, p. 213. 40 Educational Level The level of education of the husband and wife has both a direct and an indirect effect on fertility rates. Previous studies have examined fertility differentials and educational level, and have found an inverse relationship using the wife's education, or the husband's education, or the couple's education to measure the educational attain- ment.10 The lowest fertility is shown for the cases where both the husband and wife have received a college education. The low fertility of college or university graduates has been particularly stressed. The direct effect of education is the observed relationship between the level of education and the use of family limitation practices. Education pro- motes rational attitudes towards family planning. Indirect- ly education affects some of the factors tending towards low fertility, such as income, female employment and occupation. The higher the educational level of the county, the greater will be the proportion of the labor force employed in the higher-income jobs. A high educational attainment by females enlarges their scope of gainful employment. It follows from the preceding discussion that educa- tional attainment of the rural pepulation can be expected to ¥ lOGrabill, Kiser, and Whelpton, op. cit., pp. 198-217. 41 affect fertility levels. It was hypothesized that there was an inverse relationship between fertility and education level. The variable used to measure educational attainment was the median years of school completed by males and fe- males 25 years and over. The educational level of married females and males would have been more appropriate, but such data were not available. Hewever, no bias should re- sult unless there is a significant relationship between educational levels and marriage rates in an age group. Occupation Another factor that is hypothesized to affect rural- farm fertility is the occupational status of the farm popu- lation. The pattern of differential fertility by occupations indicates an inverse relationship. Typically, farm laborers and farmers and farm managers are in the high fertility group. The service workers and craftsmen are in the middle group and the clerical, managerial, and professional person— nel have the lowest fertility.11 In agriculture, unlike other occupations, children can be gainfully employed after a certain age is reached. Children are assets in tobacco _* llGrabill, Kiser, and Whelpton, op. cit., Table 54, pp. 131-20 v. o I O b .oo A w. . . .I j . ntl »E,. , I j i I l . .nw. Ibh'? 42 farms and other enterprises using hand labor. A great deal of tractor work is done by teen—agers on farms. In economic terms, as the organization of the farm is usually based around the family, the distribution of the product is not determined by marginal productivity considerations. Hence total value product and not net profit is maximized. Therefore, an addi— tional farm hand increases total product and children consti- tute economic assets on farms. In the case of farm laborers and farm foremen, they constitute both in the national occupational structure and within the rural-farm sector, the lowest stratum in terms of socio-economic status. Hence in accordance with the hy- pothesized relationship between socio-economic status and fertility rates, the proportion of farm laborers and farm foremen will be positively related to fertility. The variables used in the equation to represent occu- pation groups in the county were a) the percent of the employed male rural-farm work force who are farmers and farm managers, and b) the percent of the employed rural-farm work force who are farm laborers and farm foremen. The other occupational groups were not included. The two groups included formed 68 percent of the employed rural-farm work force of the United F. h -\u 43 States in 1960.12 It is hypothesized that the proportion of the popu- lation engaged in agricultural pursuits tends to raise the fertility rates in a county. Further, it has been observed that the fertility rates of farm laborers tend to be gener- ally higher than those for the farmers, and farm managers, so that, not only would the proportion of the total popu- lation engaged in agricultural pursuits tend to increase the fertility rate in the county, but the proportion of the agricultural population who are farm laborers and farm foremen tend to accentuate this rise in fertility. It was postulated that a positive relationship exists between these variables and fertility. Female Labor Force Participation Rate and Female Personal Income WOmen in the labor force and female income were two other variables hypothesized to affect fertility rates. In a sense, these two variables refleceed the alternatives to having children. The variable used in the statistical an— alysis to represent female employment was the percent of females age 14 and over who were employed. Both married —— 12P. C., (1), 1C, op. cit., Table 87, p. 216 . 1'! IO“. sub W" ()- rt’ ’<: 9‘.» v 44 and unmarried women were included in this measure. It has been observed that fertility rates for married women were considerably lower for those in the labor force than for those not in the labor force. But the relationship may be less clear. According to the Bureau of Census: women in the labor force at a given point of time generally have fewer children than the women not in the labor force. . . . In the last decade or so, however, a life pattern seems to be developing among many married women in which they work until the arrival of the first baby, temporarily with- draw from the labor force while their children are young, and then return to the labor force after their children are old enough to require little care. Even in terms of children ever born, young women who happen to be in the labor force are less fertile than those who happen not to be in the labor force; but it is possible that the two groups will not differ very much in the size of completed families. The above pattern is of recent origin, but it is clear that working wives, at a given point of time, are less fertile than non-working wives. In the case of un- married women in the labor force, it is hypothesized that they constitute those women who have postponed marriage, or have chosen a professional career which is incompatible with the making of a home. It was hypothesized that there was an inverse relationship between fertility and female labor force participation rate. E 13U. 8. Bureau of Census, Census of Population, 1950, IV. Special Reports, Part V, Chap. 6, "Fertility," p. 11. ”'fi': yup)... V‘I" 'l on“ N v évcn ov.‘, ii a 35.. u . ’r‘ 'n UV HE ? -.c:a. TP PA 8”»; n: i K). V . \‘a- 'u-vfl . b ”C 'n» h “H.‘ ‘- '0“! .‘VJ s “A any 45 Female personal income was expected to affect fer- tility rates in a county. It reflected the opportunity in- comes which women would have foregone if they had not been employed. In a sense, it represents loosely, for the work- ing wives, the opportunity cost of bearing and rearing chil- dren. The measure of female income used was the median female personal income for county, and it was hypothesized to be inversely related to fertility. Total Family Income Family income is a major factor affecting fertility. Income has been used as a measure of socio-economic status and studies relating family income to fertility have often shown an inverse relationship. Within socio-economic classes, the same studies have shown a positive relationship, for the upper income brackets. In this study, the variable used to represent family income was the median family income for the county, and it was hypothesized that it was negatively re— lated to fertility. If the distribution of median family incomes in counties is skewed to the right, the expected relationship may not hold. The rationale for the above hypothesis can be ex- plained in terms of the transitional phase of differential v..‘ -~~t ° PI! M.“ 4 . . ”C., ~ g l i 46 14 Family limitation practices found acceptance fertility. and use first in the upper-socio-economic groups, and later spread downwards to the middle groups and finally to the lower socio-economic groups. Hence, when family income is used as an index of socio-economic status, fertility rates can be expected to be inversely related to family incomes. A relationship between family income and female income could be expected, especially for the working wives. Female in- come increases family income and is a factor conducive to the inverse relationship between family income and fertility. Family income, employment of the wife, and fertility are also related. Couples with low family income are more likely to be those in which the wife does not work, and has rela- tively more children; while couples with higher incomes are likely to be heavily weighted with those that include work- ing wives. Color The equations with the children ever born as the de— pendent variable were estimated for non-white females using the same independent variables for the three divisions in the South and the southern region. The rationale behind a 14See text, Chap. II, p. 11. sq.» pa u..i.r..~‘ a .3551 “h; \ but: . ‘h A“ j-Vn. R a. Vb Sm," "VJ; v.1 Sr 71 (D «(j 3“ ‘u‘. ‘., I I) 47 separate set of equations for the non-white females in the South was that they formed a distinct group in this region. The distribution of the non-white rural-farm population among the regions in the United States in 1960 was as fol- lows: .5 percent in the Northeast region, 2 perqent in the North Central region, 93 percent in the southern re- gion, and 4.5 percent in the western region. The results of the non-white analysis can be ex- pected to be different from those of the white analyses. About 98 percent of the rural non-white population of the South is negro, and the difference in the analysis arise from the distinct nature of the problems facing them. There is no cultural trait that affects the fertility rates of negro women independently of the variables discussed above. But the particular social organizations of the southern re- tion affects some of the variables. Discriminatory practices among the employers would restrict the scope of female employment, or limit employ- ment to low-paying jobs. Even educated non—whites may be faced with discrimination in employment. The most important difference arises in the influence of the urban centers. The influence of urban centers may be less among non—white communities. There has been an Observed . .F‘ f v.1 - .‘ vo‘. {Lea java .G.3 5.x, Lea/.4 . Main "l we: 0"“ Nu.” ‘ y“!- ‘§ A a“, ‘\\u ¥\‘ ’u- ‘v .V‘ ‘ n 48 tendency of non-white migration to be concentrated towards the northern cities rather than to the large southern cities. Further, the selective nature of non-white migration is limi- ted to the younger age-groups, which will be reflected in a large proportion of the women remaining in the rural popula- tion, being in the higher age-specific groups. In accordance with the hypotheses expressed for the white women, a larger proportion of women in the higher age—groups is directly re- lated to the number of children ever born. It is also probable that urban culture may receive a different response from the non-white communities in the South region. The newspapers and other mass communications media in the cities are dominated by white culture. In the context of the racial problems in the South, the influence of urban culture patterns may be different in the non-white analyses to the extent that the non—whites reject the pat- terns of urban culture disseminated by the mass—media con— trolled by white interests. Regional Differences Equations were estimated for the white fertility rate for each census division, each region, and for the conterminous United States. The effects of the variables hypothesized to explain fertility differentials were Ila-n WU \ Auvy 49 expected to be different in the various regions. These dif— ferent effects arise due to differences in the regions them- selves. For example, in the southern region there is a pre- dominance of small family farms and the type of farm enter- prise is different from that of the North Central region. In the South, the type offarming is labor intensive, being mostly tobacco or cotton farms. It is expected that children would be more useful on a tobacco farm in the South than on .a wheat farm in the North Central region. Further the degree of industrialization and urbanization differs among the vari- ous divisions and regions. The Northeast has been urban since the nineteenth century while the South has relatively x little urbanization. The Northeast and North Central regions have been settled for a considerable period of time. The western region is still undergoing the process of settlement although the frontier disappeared in the last century. The differences in the composition of regions and divisions affect fertility rates. In the preceding sections, the dependent variables and the independent variables were introduced. Some of the expected relationships among the independent variables were discussed. The rationale for the use of a separate set of equations for the non-white population in the South region I).Pl.p' ‘.l'(,§. 50 was presented, and, finally, expected regional differences in the effects of the independent variables were discussed. In the following chapter, the equations are presented for- mally, the variables specified, and the statistical hypo- theses stated. CHAPTER.IV THE STATISTICAL FRAMEWORK: THE SOURCES OF DATA AND THE METHODS OF ANALYSIS The Sources of Data In the Eighteenth Decennial Census of Population, taken as of April 1, 1960, statistics on social and economic characteristics of persons were enumerated on the basis of a 25 percent sample of the population.1 The data included statistics on children ever born, years of school completed, employment, occupation, and income by residence and color. For the purposes of the PC series of Census publications, the data were enumerated with the county as the unit of observation. The census schedules were microfilmed and placed on magnetic tape. The data used in this study were from a copy of the PC tape obtained by Michigan State Uni- versity for use in preparing a Population Monograph. The statistical analyses were carried out by the Armour Research ‘ ) 1See p. c. (1), 1c, op. cit., pp. xli, for sampling procedures, accuracy of data, and methods of estimating pop- ulation characteristics. 51 52 Corporation of the Illinois Institute of Technology in Chi- cago on a Remington-Rand Univac 1105 Computer. The Equations Estimated This study undertook a cross-sectional investigation of the effect of economic and other variables on average fer: tility rates in the rural-farm parts of counties. In a cross- sectional analysis, a number of geographic divisions are in- vestigated at a point of time. Although the use of cross: sectional data has certain advantages over time-series, es- pecially from the point of View of availability of comparable data, it raises a number of statistical problems. It assumes that historical and environmental conditions, not included in the study, have a constant affect on the population of each of the geographic units and that these conditions do not affect the quantifiable variables which are expected to ex- plain fertility differentials. If these assumptions are not held, biases in the estimated statistics will result. Large standard errors, low correlation coefficients, nonesignifi- cance of the estimated regression coefficients, and bias in the regression coefficients will be observed. In this study, the basic unit studied was the county but it is not unreal; istic to assume that culturally, the populations of the various counties, in a division or a region of the United States form 'l‘." a .6"; 51‘: ni‘h‘ rr (3 p: (n 5.) Us; . ml h‘l‘ re; .F s“ r) H r) 53 a more homogeneous entity than, for example, a cross-section of nations across the world. Least squares regression analysis was used to esti- mate two sets of equations. In the first set the dependent variable was the number of children born to white females, and in the second set it was the number of children born to non-white females. The above sets will be referred to as the "white fertility" equations and the "non—white fer- tility” equations. The "white fertility” equations were fitted for each census division in each region, for each region, and for the conterminous United States as a whole. In all, there were forty-two ”white fertility" equations. The "non-white fertility" equations were fitted for the three divisions in the South and for the southern region. Twelve ”non-white fertility" equations were estimated. The Regression Models: "White Fertility" Equations In this set of equations, three equations are in- cluded. They are identical with the exception of one vari— able. In Chapter III, the three measures of industrial- urban influence were discussed briefly. Equation (1) in- cludes the distance variable but excludes the size—distancel and size-distance2 variables. Equation (2) includes the ya ‘I an: ~Hu\ 0!)- 54 size-distancel variable and omits the distance variable and size—distance2 variable. Equation (3) includes the size- distance2 variable and omits the distance and size-distancel variables. ”White fertility" equation (1) Yi = a + c4xi4 ...+ C12X112 + Ui where i = 1,2,. . ., N j = 4,5,. . ., 12 and Yi is the ith observed value of the dependent variable xij is the ith value of the jth independent variable 2 U- is the ith random disturbance term 1 a is the constant term Cj is the unbiased coefficient of the jth independent variable Variable specification The dependent variable, Yi, . . ., the number of children 21t is assumed that the U‘s in all equations are in- dependent and follow some probability distribution, with zero mean and finite variance. See R. J. Foote, Analytical Tools for Studying De- mand and Price Structures (Washington: United States Depart- ment of Agriculture Handbook No. 146, 1958), p. 58. ”fir I .. any *6: A - "n. A? v‘. a‘: 5 . ’1' 1 (I) 55 ever born per 1000 ever married white women age 15-44 in the rural-farm part of a country in 1960 was used as the dependent variable. This variable, taken to be an index of rura1;farm fertility, was discussed in detail in Chap- ter III. The independent variables, X5 X4 measured the percent of the employed white male work force who were farmers and farm managers in county in 1960. The information referred to the job held during the reference week. Therefore, the use of the measure assumed a minimal occupational mobility. X5 measured the percent of the employed white male work force who were farm laborers and farm foremen. The comments made for variable X4 are applicable to this vari- able, also. X6.measured the white female labor force participa- tion rate. WOmen in the labor force were measured by the percent of white rural-farm female labor force in the cen- sus period. X7 measured the median white female income for the county. The data were for median female income in 1959. X8 measured the median years of schooling completed '11 56 by rural-farm white males and white females, age 25 and over, in a county. X9 measured the median income of white rural-farm families in each county. The data measures median family incomes in 1959. X10 and X11 measured the age distribution of the rural-farm white female population. X10 measured the per- cent of ever married white females age 15-44 who were 15- 24, and X11 measured the percent of white rural-farm fe— males age 15-44 who were age 25—34. Distance Variable X12 One of the variables that measured urban influence was X12. It measured the distance of the county from the nearest SMSA. The value zero was assigned to all counties in which cities of 50,000 or more population were located in 1960. All counties which were located within 50 miles of an SMSA were assigned the value one. The value two was assigned to all counties located 50 to 100 miles from an SMSA. Counties which were located from 100 to 150 miles fron an SMSA were assigned the value three. The value four was assigned to a county located from 150 to 200 miles from an SMSA. The values five and six were assigned to all AA”? ybu- a .W‘JI ..Va "hr” in: 57 counties located from 200 to 250 and from 250 to 300 miles from an SMSA, respectively. No county in the conterminous united States was located more than 300 miles from an SMSA. In determining the distance from the central city of the SMSA to the most distant boundary of the county was used.3 "White fertility" equation (2). Y1 = a + c4xi4. . . . cllxill + c13x13 + Ui where 1 = l, 2, . . N 3:4,5,....11,13 and Yi is the ith observed value of the independent variable Xij is the ith value of the jth independent variable Ui is the ith random disturbance term a is the constant term C- is the unbiased coefficient of the jth indepen- dent variable. 3The above section and the sections dealing with X13 and X14 are summarized from Bryant, W; K., "An Analysis of the Inter-Community Income Differentials in Agriculture in the united States," Unpublished Ph.D. Thesis, Michigan State University, 1963, Chap. IV, pp. 80-84. «M a DIP‘ uuvu an , Ill [.rtliiilh'.’l'l ..7 Dual , i It...’ I In .. PM. 58 Variable Specification The variables in this equation were identical to those of the "white fertility" equation (1), with the ex- ception that distance variable X12 is omitted and size- distancel X13 is included. The size-distancel variable takes into account not only the location of the city from an SMSA, but also,the size of the SMSA. SMSA counties (counties in which cities of 50,000 or more population were located) were given a value of one for every 100,000 population. SMSA counties with population between 50,000 and 100,000 were given a value of .5. No SMSA county was given a value of greater than 20. This restriction expressed the assumption that SMSA's of two million or more have similar influences on the fer— tility levels of rural—farm women in the counties in which they are located. Counties within 50 miles of the central city of the SMSA were assigned a value two less than that assigned to the SMSA county. Counties between 50 to 100 miles of an SMSA were assigned a value two less than the value assigned to counties within 50 miles of the SMSA. This Procedure was followed until the value zero was reached. 1’1 l m T" a l at {vv v. fin‘ Vin \Hu ..— s .n .. My. Ad t ‘ ..I,xl)EEPrI .. k'.»l.l., 5' . l\)).s' l,|-: . ll- 59 An implication of this scheme is that no SMSA of two mil- lion or more population is assumed to influence the level of fertility rates in a community which is more than 500 miles distant. An SMSA of one million was assigned a value of ten. Such SMSA's then could influence counties at a distance up to a maximum of 250 miles. SMSA's larger or smaller than one million could influence counties in proportion to their population size. In the case where one county could be assigned two values, one value from one SMSA and another value from a different SMSA, the value assigned was the greater of the two. In a number of cases, when one SMSA was in the range of another SMSA, the value of the SMSA county plus the value derived from the influencing SMSA were assigned subject to the restraint that the value assigned could not be greater than the value assigned to the influencing SMSA. By this procedure, each county in the United States was assigned a number ranging from zero to twenty. "White fertility" equations (3) Y1 = a + C4Xi4 . . + Cllxill + C14Xil4+ Ui K), 3n t? (‘9‘ a an OAU ''''' 60 where and Yi is the ith observed value of the dependent variable X13 is the ith value of the jth independent variable Ui is the ith random disturbance term a is the constant term Cj is the increased coefficient of the jph inde- pendent variable. Variable Specification The variables in this equation were identical with those in the "white fertility" equation (1) with the excep- tion that the distance variable, X11, was omitted and the size-distance2X14 variable was included. The size-distance2 variable expressed a similar hypothesis to that of the size- distancel variable in that they took into account both the distance of the county from the SMSA and the size of the SMSA. They differed in that the sizeedistancez variable expressed the hypothesis that urban centers extended their influence shorter distances than was hypothesized by the AN» u!» A‘ ~ . \l Van-l F \ 4" tan: ~\h nun .5 LP. tha 61 size-distancel variable. The same values were assigned to SMSA counties by the size-distance2 variable as were assigned by the size- distancel variable. The rules for assigning values to non- SMEA counties were similar to those used for the size-dis- tancel variable with the following exceptions. The value assigned to a county X and (X+50) miles from an SMSA accor— ding to the size-distance2 variable was four less than the value assigned to counties between (X-SO) and X miles from the SMSA. It expressed the hypothesis that no SMSA in- fluenced fertility levels in a community which was more than 300 miles distant. As with the size-distancel variable, the size-distance2 variable assigned values from zero to 20 to each county in the United States. "NOn-white fertility” equations "Non-white fertility equations (1), (2), and (3) used identical variables to those in the "white fertility" equations (1), (2), and (3), with the exception that X4 — x11 referred to the non-white population rather than to the white population. Variables X12, x13, and X14 were identical in both sets of equations. These equations were estimated for each division in the South and for the southern region as a whole. ..AI-' .I ‘ y...»- .A& .14 Ly : o n... h... 839 Wu” ...C. a .“H! 5v. ‘Hh ‘ua. X'X ’3. kn: On, (1‘ .rfi 62 Constant Terms It should be noted that the estimated equations do not contain variables with respect to a) the percent of (whites or non-white) employed rural male work force who are in occupation categories other than that of farmers and farm managers, (white and non-white) age 15-44 who are 35-44. These variables were omitted to gain a determinant solution. For example, the three female age variables, if all had been included, would have been linearly dependent, resulting in X'X matrices which would have been singular. The constant terms are expected to contain functions of the effects of these omitted variables, but no interpretation of the con— stant terms with regard to the effects can be made without knowledge of assumptions about functions.4 The Beta Coefficients In presenting the equations estimated, beta coeffi— cients were used in addition to the usual partial regression coefficients.5 The regression of Y, the dependent variable, 4See Bryant, op. cit., pp. 85-6, for discussion of the interpretation of constant terms. 5For the formulae for estimating beta coefficients and a discussion of their value see M. Ezekiel and K. A. Fox, Methods of Correlation and Regression Analysis (third edition, New YOrk: John Wiley and Sons, Incorporated, 1959). cm X “:6 “..e kite: RNA Cs ass 63 on X1 and X2, the independent variabks, can be presented in the form of the following equation: Y = blxl + b2X2 The beta coefficient is calculated by the following formula: S S . x1 . _ x1 Sy sy l H | where bi = the estimated beta coefficient of X1 bé = the estimated beta coefficient of X2 bl = the estimated regression coefficient of X1 b2 = the estimated regression coefficient of X2 8x1: the standard deviation of X1 3x2: the standard deviation of X2 Sy = the standard deviation of Y The estimated equation in terms of beta coefficients would be as follows: Y" = biXi + béXé p. 148, and p. 196. Ezekiel and Fox use the term beta coe— fficient which is used in this study. G. W; Snedecor, Statistical Methods, (fifth edition) Ames: The Iowa State College Press, 1956, p. 416. Snedecor uses the term standard partial regression coefficient. nun. c,, u‘\ 3U ~\v ~ 64 where Y" = the predicted value of the dependent variable X1- X1 X' = the standard deviate of X and x' = S x1 Thus, if the Standard deviate of X1 changes by 1 (in a posi- tive or negative direction) and if the standard deviate of X2 remains constant, then the predicted Y, (Y") deviates -—-—, from the estimated mean of Y, (Y) by the amount of bi (in a positive or negative direction). Beta coefficients are pure numbers which take into account the variation in the independent variable relative to the dependent variable. Partial regression coefficents cannot be directly compared as they vary with the units in which they are measured. But the beta coefficient makes them comparable by expressing them in terms of their own standard deviation. The beta coefficients are summary measures of the importance of the effect of each indepen- dent variable on the dependent variable. The sign of the beta coefficient indicates the direction of the effect. Beta coefficients were estimated for all independent vari- ables in all equations and are presented in the results. §lmp1e Correlation Analyses Simple correlation coefficients were calculated‘ 65 between each of the independent variables in each of the equa- tions. These showed the degree of intercorrelation among the independent variables and wherever pertinent, the implications will be discussed in conjunction with the results. Simple correlation coefficients were also calculated between each of the independent variables and dependent variables in each of the equations. Statistical Hypotheses In Chapter III some of the relationships between the independent variables and the dependent variable were discussed. In this section, these relationships are ex- pressed in relation to the equations presented earlier. Table 4.1 shows the expected signs of the estimated regres- sion coefficients for the independent variables, in both the "white fertility" and "non-white fertility" equations. The same signs were expected for both sets of equations. The expected signs of the regression coefficients represent tentative hypotheses as to the direction of the influence of the independent variables on fertility rates. One-tailed "t" tests were used to ascertain whether the estimated re- gression coefficients were significantly different from zero. The one-tailed "t" test is statistically stronger and more definitive than the two-tailed tests. However, “x; . n b U f "(r .\ l (l l) [‘1.‘ .li'fslll..|||.v.itc ..,. 66 TABLE 4.1 Expected Results of the Analysis of the Factors Influencing the White and Non4White Fertility Rates of Rural Farm Females in a County EXpected Signs of Estimated Regres- Independent Variables sion Coefficients Equation Equation Equation 1 2 3 Distance from nearest SMSA (x12) + Size-distancel (X13) - Size-distance2 (X14) Percent of male employed work force who are: farmers and farm managers (X4) + + + farm laborers and farm foremen (X5) + + + Percent of females, age 14 and over, in county, who are employed (X6) - - Median female personal income (X7) - Median years of school com- pleted by males & females age 25 & over (X8) - ' ' Median family income (9) ‘ Percent of ever married fe- males, age 15—44, who are: age 15-24 (X10) - — _ a e 25-34 X — — _ 9 ( 11) 4s ‘, ‘I 2: 67 it is expected that regional deviations in the expected signs will occur. Under these circumstances, the use of one-tailed "t" tests in place of the conventional two-tailed tests raise certain problems. The following is a hypothetical example of the problems created by the use of one—tailed "t" tests. Assume that it is hypothesized that the sign of a particular regression coefficient is positive and that the computed "t" value is —55.2620. The two-tailed test will reject a null hypothesis that the regression coefficient is zero. The one-tailed positive "t" test will accept the null hypothe- sis that the regression coefficient is not positive. Hew- ever it does not indicate whether the regression coefficient is different from zero. The actual "t" value is very much less than zero, and a one-tailed negative "t" test will re- Ject a null hypothesis that the regression coefficient is positive or zero. In the tables presenting the results, regression coefficients which are significant but have signs Opposite to those expected are noted separately, with the symbols (aa). The hypothesis, tested in this case, is that the regression coefficient is significantly different from zero in the direction indicated by its sign. Another statistical procedure used was the multiple Eon Eu”; 4 w": i 531; In be 68 6 This test comparison or contrast tests offered by Scheffe. identifies the significant differences among regression co- efficients of an independent variable. This test was per- formed for each independent variable in the "white-fertility" equations, among divisions and among regions and in the "non- white fertility" equations among the three divisions in the southern region. This chapter presented a discussion of the statisti- cal framework, the statistics used, and the tests performed. In the following chapter, the results of the analyses will be presented, interpreted, and discussed. 6H. Scheffe, The Analysis of Variance (New York: John Wiley and Sons, Inc., 1959), pp. 66-72, and K. A. Brownlee, Statistical Theory and Methodology in §g$ence and Engineering (New YOrk: John Wiley and Sons, Inc., 1959). pp. 252-4. SEE a: '11)! .. .t' 6' .I-Io. CHAPTER V THE RESULTS OF THE ANALYSIS Introduction In this chapter, the results obtained from the analy- ses of fertility rates are presented and interpreted by geo- graphical regions. The "white fertility" equations are dis- cussed for each division in a region, followed by a discus- sion of the equations estimated for the region. The "non- white fertility" equations, which were estimated for the three divisions in the southern region and for the southern region are discussed next. Finally, the results of the "white fertility" equations estimated for the conterminous United States are discussed. Three equations were estimated for each geographic . area. Each equation included a variable representing an alternative hypothesis about the nature and extent of the effect of urban influence on the fertility rates of rural farm communities near urban centers. The three measures of urban influence were the distance variable, the size- distancel variable and the size—distancez variable. The 69 . ("TI «u. y «a .0: t? hu. 70 other variables were identical in all three equations esti- mated for a geographic area. In interpreting the results, one equation was se- lected for each geographic area. The selected equation was expected to explain variations in fertility rates bet- ter than the other two equations; statistical devices have been developed which indicate a basis for choice among equations. Two such devices, the standard error of esti— mate, and the multiple correlation coefficient (R), are 1 The standard error of estimate indi- used in this study. cates the closeness with which estimated values from an equation agree with the original values. The multiple cor— relation coefficient measures the combined importance of the independent variables as a means of explaining varia- tions in the dependent variable. In this study, the vari— ables in the three equations estimated for any geographic area were similar with the exception of the variable ex— plaining urban influence. Hence, the equation with the highest multiple correlation coefficient would explain the most variation in fertility rates and also would indicate which of the three proximity variables was the better mea- sure of urban influence in a geographic area. In a majority lEzekiel and Fox, op. cit., pp. 118-92. 5a: ta: ir 71 of equations, those with the highest multiple correlation coefficient (R) had also the lowest standard error of esti- mate. One equation was chosen, on the above bases, for de- tailed discussion and interpretation. Each table in this chapter summarizes the results of the three equations for a geographic area. They present the multiple correlation coefficient, the standard error of estimate, the estimated beta coefficients of the indepen- dent variables, and the results of the "t" tests of the partial regression coefficients. In the Appendix, the de— tailed results including computed "t" values and partial regression coefficients for an equation are given in one table. Each table in this chapter, then, is a summary of three tables included in the Appendix. The Northeast The Northeast region of the United States as defined in the census is comprised of two divisions, the New England division and the middle Atlantic division. The New England division constitutes six states: Maine, New Hampshire, Ver- mont, Massachusetts, Connecticut, and Rhode Island. The middle Atlantic division contains three states: New YOrk, New Jersey, and Pennsylvania. r) 72 The New England Division A summary of the results of the three equations esti- mated for the New England division is presented in Table 5.1. Tables 1, 2, and 3 of the Appendix present estimated regres- sion coefficients, and the computed "t" values for this divi- sion. The coefficient of multiple determination (R2) in equation (1) was .4583; in equation (2) it was .4546, and for equation (3) it was .4571. The simple regression coeffi- cient between white fertility rates and distance (X12) was .4548; between fertility rates and size-distancel (X13) it was -.4510, and between fertility rates and size-distance2 (X14) it was -.5085. None of the estimated partial regres- sion coefficients of these three variables were significantly different from zero at the .05 level of significance. Equa- tion (1) was selected for detailed discussion, as it had the highest coefficient of multiple determination and also the lowest standard error of estimate. The median female personal income (X7) was the most important relative to other variables as measured by the beta coefficients in equation (1). Next was the median family in- come (Xg). The regression coefficient of median female in- come (X7) and the regression coefficient of median family NJ vL — At V) PE 73 TABLE 5.1 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960 New England Division Equation Equation Equation 1 2 3 Multiple Correlation Coefficient .6770 .6743 .6761 Standard Error of Estimate 284.1576 285.0873 284.4690 Beta Coefficients1 Independent Variables (relative importance) Distance from nearest SMSA (X12) -.l421 Size-distancel (X13) -.O3l9 Size-distance2 (X14) -.0115 Percent of male employed work force who are: farmers and farm managers (X4) .1379 .1272 .1215 farm laborers and farm foremen (X5) .2027 .1927 .1805 Percent of females, age 14 & Over in county, who are employed(X6) .3076 .3032 .3184 Median female personal income for county (X7) Median years of school completed by males & females age 25 and over (x8) —.0248 .0143 .0306 Median family income (X9) —.4861** —.3817** —.3267 Percent of ever married females. age 15—44, who are: age 15-24(X10) -.1546 -.1886 -.2032 -.5152** -.4661** —.4609** age 25-34(X11) -.1392 —.l368 —.l418 1See Appendix, Tables 1, 2, and 3 for complete results **Significant at .05 level 74 income were significant at the .05 level of significance. The partial regression coefficients of all other variables in equation (1) were not significantly different from zero at the .05 level of confidence. Of the variables that were not significant, female employment (X5), and distance vari- ables (x12) had signs contrary to those expected. There was a large degree of intercorrelation among independent variables in the New England division. This intercorrelation may have resulted in the estimates of the partial regression coefficients having larger variances, and thus tending toward reducing their statistical signifi- cance. It could also affect the regression coefficients sufficiently to change the signs. Female employment rate (X5) was highly correlated with female personal income (X7) (r6.7=.8319). The percent of farmers and farm managers (X4) was correlated with the percent of farm laborers and farm foremen (X5) and with the median years of school completed (X8) (r4.5= .6299; r4.8= -.6858). Both median female personal income (X7) and median family income (X9) were correlated with distance (X12) (r7 12= -.6245; r9_12= -.6858). Although the simple correlation coefficient between female employment (X6) and white fertility rates was -.2524, 75 the sign of the partial regression coefficient of this vari- able (X6) was positive, and was not significantly different from zero. This is not unexpected in view of the high inter- correlation between female employment (X6) and female personal income (X7). It should also be noted that while simple cor— 2w— relation between the dependent variable and a particular in- dependent variable measures a direct relationship, the partial ~regression coefficients are net, having taken into account the influence of other independent variables included in the study. The interpretation of the results was made difficult because only two of the nine variables in equation (1) had partial regression coefficients that were significantly dif- ferent from zero. Nevertheless, variations in the variables included in equation (1) explained about 45 percent of vari- ation in white fertility rates in the New England division. Female personal income (X7) and median family income (X9) were the most important variables in expressing differential fertility in this division. The results of the "t“ tests of the regression coefficients of these two variables showed that they were significant. Hence, the results of the anal- yses bear out the postulated relationship between income and fertility. They show a strong inverse relationship between fertility and income variables. The simple correlation 76 between fertility rates and female income (X7) was -.4l67, and between fertility rates and family income (X9) was .4432. Apparently income was a key factor in depressing fertility rates in this division. The relationship between distance (X12) and fertility was difficult to explain. The distance variable was one of the alternative hypotheses about the extent of urban influence. Not only was the regression coefficient non-significant, but its sign was the reverse of that expected. The regression coefficients of the percent of farmers and farm managers (X4) and the percent of farmers and farm foremen (X6) had signs which were expected, but were non- significant. The high intercorrelation between these two variables may have resulted in large standard errors, thus resulting in their non—significance. The regression coeffi- cients of the percent of married females age 15-44 who are 15-24 (X10) and who are 25-34 (X11) also had signs which were consistent with the hypotheses, but were non—significant. The regression coefficients of the variables in equa- tions (2) and (3) for the New England division had the same signs as those in equation (1). In equation (2), the size- distancel variable (X13) and the size-distance2 variable in equation (3) had negative signs which were consistent with va re of ‘VBIP'Il-lzl , FIE». I: 77 the hypotheses. In summary, in the New England divisions, the results of the equation showed a strong inverse rela- tionship between white fertility rates and the income vari- ables. No clear relationship was shown between the other variables and fertility rates. The Middle Atlantic Division Table 5.2 contains a summary of the results of the equations for the middle-Atlantic division. The detailed results of the equations are shown in Tables 4, 5, and 6 of the Appendix. I Of the three equations estimated for the middle Atlantic division, equation (3) had the highest coefficient of multiple determination (R2 = .2818), and the lowest stan- dard error of estimate. The size-distance2 variable seemed to be the most representative of the three variables mea- suring urban influence. The simple correlation coefficient between white fertility rates and distance (X12) was .3495; between fertility rates and size-distancel it was .3942. and between fertility rates and size-distancez it was -.4318. The regression coefficients of the three variables were all significantly different from zero at the .05 level of con- fidence. Mn. 5. Pe Ill-MI E'E'L.ihlr [LID ["5 b ‘4‘ 78 TABLE 5.2 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960 Middle Atlantic Division Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Distance from nearest SMSA (X12) Size distancel (X13) Size distance2 (X14) Percent of male employed work force who are: farmers and farm managers (X4) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (x6) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) Percent of ever married females, age 15-44, who are: age 15-24 (X age 25-34 (X 10) 11) k Equation Equation Equation 1 2 .4263 .4793 3 .5309 312.1851 302.8961 292.4568 Beta Coefficients1 (relative importance) .3586** -.4766** .0600 .1469 .2624** .2506** .0919 .1433 -.0623 -.2449 -.0526 —.0722 -.0216 -.0017 —.1142 —.l335 —.1280 —.l364 -.6019** __2009aa .2446** .1488 -.2687** -.1261 .0820 -.1438 —.1438 1See Appendix, Tables 4, 5 and 6 for complete results **Significant at .05 level aaSignificant at .05 level but the relationship was Opposite to that hypothesized. 79 The cultural influence of urban areas as measured by the size-distance2 variable was the most important vari— able relative to the others in explaining variations in fer— tility rates. Median female income (x7) was the next impor- tant, followed by the percent of farm laborers and farm foremen (X5). The regression coefficients of X7 (female income) and of X5 (percent of farm laborers and farm fore- men) were significant at the .05 level of confidence. The regression coefficient of the percent of farmers and farm managers was significantly different from zero at the .05 level of confidence, but the relationship was the opposite of that hypothesized. None of the other variables in equa— tion (3) had regression coefficients which.were signifi- cantly different from zero. There was less intercorrelation among the indepen- dent variables. The highest intercorrelation was again be— tween female employment (x6) and median female income (X7) (r6.7 = .8293). This intercorrelation could account for the unexpected sign of the partial regression coefficient of the female employment variable (X6) and its non-signifi— cance. The regression coefficients of all variables in equation (3) had signs which were consistent with the 80 hypothesized relationships between them and fertility, with the exception of those for the percent of farmers and farm managers (X4), female employment (X6), and median family in- come (Xé). The regression coefficients for median family income (xg) and that for female employment (X6) were not significantly different from zero. The regression coeffi- cient for farmers and farm managers was significant at the .05 level, but in. the opposite direction. The results substantiated the postulated hypothesis that the closeness of a rural community to an urban area and the size of the urban area tend to lower the fertility rate in the community. This is shown by the results of all three equations. In all three equations, the beta coefficients for the variables representing urban influence were the highest and the regression coefficients of these variables were consistent with the expected signs and were signifi- cant. In all three equations, the sign of the regression coefficient for the percent of farm laborers and farm fore- men was consistent with that expected and the regression coefficients themselves were significant at the .05 level of confidence. AU 11" IIIIIP n 81 In the middle-Atlantic Division, urban influence was the most important factor in explaining variations in fertility rates, followed by female income and the percent of the work force who are farm laborers and farm foremen. The Northeast Region The results of the equations estimated for the Northeast region are presented in Tables 7, 8, and 9 of the Appendix. They are summarized in Table 5.3. The variables in equation (3) explain more of the variance in fertility rates of rural-farm women in the NOrth east region than either of the other two equations (Ri = 2410, R; = 2828, kg = 3066). Equation (3) had the highest coeffi- cient of multiple determination and also the lowest standard error of estimate. The size—distance2 variable appeared to measure the influence of cities more closely than either distance (X11) or size-distance1 (X12). Thus, differences in distance from cities as well as differences in the sizes of the cities, were important factors affecting differen- tials in fertility rates. As measured by the absolute size of the beta coeffi- cients, the size-distance2 (X14) variable was the most impor- tant factor which affected fertility rates in equation (3). 82 TABLE 5.3 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960 Northeast Region Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Distance from nearest SMSA (X12) Size-distancel (X13) Size-distance2 (X14) Percent of male employed work force who are:, farmers and farm managers (Xh) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (XS) Median family income (X9) Percent of ever married females, age 15-44, who are: age 15-24 (X10) age 25-34 (X11) g Equation Equation Equation 1 2 3 .4910 .5318 .5538' 304.7671 296.2525 291.2765 Beta Coefficients1 (relative importance) .0305** -.3692** -.4476** .0248 -.0590 -.1057 .2524** .2573** .2336** .1053 .1430 .1723 -.1518 -.2743** -.2898** -.0104 -.0474 -.0511 .2014** -.1427** -.1325 -.1589** —.1874** -.1894** -.0953 -.1128 —.1239 1See Appendix, Tables 7, 8 and 9 for complete results ** Significant at .05 level 6:. l L, [I [Illui‘l‘ 83 Next was female income (X7), followed by percent of farm laborers and farm foremen (X5), percent of married women age 15—44 who are 15—24 (Xio), and family income (Xé), in that order. The partial regression coefficients of the other variables in equation (3) were not significantly different from zero. These variables also had signs which were expected. Of the variables that were non—significant, the percent of farmers and farm managers (X4) and female employment (X6) had signs contrary to those hypothesized. Intercorrelation among the independent variables was low, with the exception of that between female employ- ment (X6) and female income (Xfi) (r6.7 = .8282). This may be a factor affecting the sign of X6 (female employment). The percent of farmers and farm managers (X4) was corre- lated with the percent of farm laborers and farm foremen (X5) (r4.5 = .4122); education level (X8) was correlated with female income (X7) and with family income (X9) (r8.7 = .4246, r8_9 = -.4182). The influence of urban areas had the strongest in- fluence on the fertility rates of rural—farm communities in the Northeast region. Irrespective of the measure used, the results, in all three equations, show that urban influence was important. The regression coefficients of all three 84 measures were significantly different from zero at the .05 level of confidence and had signs that were consistent with those hypothesized. Clearly, the presence of large cities and the proximity of rural-farm communities to these cities affected the rural cultural patterns regarding family size. The influence of large cities was not unexpected in the Nertheast region, as there are large concentrations of SMSA's in this region. Some of the SMSA's are New York (10.7 million), Philadelphia (4.3 million), Pittsburgh (2.4 million), Boston (2.6 million), Newark (1.7 million), and Paterson-Clifton-Passaic (1.2 million). Of the 215 SMSA's in the united States, 48 are located in the North- east region.2 The estimated regression coefficients for female income (X7) and family incomes (X9) were negative and highly significant. These relationships bear out the argument in Chapter III that incomes are inversely related to fertility rates. The positive and significant regression coefficient for the percent of farm laborers and farm foremen (X5) were 2United States Bureau of the Budget, Standard Met- ropolitan Statistical Areas, (Washington: U. S. Government Printint Office, 1961). 85 in accordance with expectations. The negative regression coefficients of the age distribution variables of married women (X10 and‘Xll) were also expected. The low but nega- tive regression coefficients for education supported the expected inverse relationship between fertility rates and educational attainment. The positive regression coeffi- cient for X6 (female employment) can be rationalized in terms of the intercorrelation between female employment and female income. The negative regression coefficient for the percent of farmers and farm managers was contrary to that hypothesized. In summary, in the NOrtheast region, which is an area of high urbanization, it was not surprising that in- fluence of cities, as measured both by the distance of com- munities from large cities and by the size of the city, was the most important factor affecting variations in fer- tility rates among rural—farm communities. The effects of the income variables were also important. The other vari- ables had a relatively minor effect. The North Central Region The North Central region is comprised of two divi- sions, the East NOrth Central division and the West North 86 Central division. The East North Central division contains five states; Michigan, Indiana, Ohio, Illinois, and Wiscon- sin. The west North Central division contains Iowa, Minne— sota, Missouri, North Dakota, South Dakota, Nebraska, and Kansas. East North Central Division Table 5.4 summarizes the results of the three equa- tions for the division. Tables 10, 11, and 12 in the Appen- dix contain the detailed results. The coefficient of multiple determination for equa- tion (1) was .3601; for equation (2) it was .3596, and it was .3746 for equation (3) yielded the highest coefficient of multiple determination and also had the lowest standard error of estimate. Hence, it was chosen for detailed dis- cussion. Median years of school completed by males and fe- males age 25 and over in a county (x8), was the most im- portant variable affecting intercommunity differentials in fertility rates of white rural-farm women, in all three equations. In equation (3) the size-distance2 variable (X14) was next in relative importance, followed by percent of married females age 15-24 (X10), the percent of farm la- borers and farm foremen (X5), median female income (X7) and P! M Ix‘ll PI, .[| 1 .1 ll 87 TABLE 5.4 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. East North Central Division Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Distance from nearest SMSA (X12) Size-distance1 (X13) 14) Percent of male employed work force who are: Size-distance2 (X farmers and farm managers (X4) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (Xk) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) Percent of ever married females, age 15-44 who are: age 15—24 (X10) age 25-34 (X11) ¥ 1See Appendix, Tables 10, 11, and ** Significant at .05 level aaSignificant at .05 level but the to that hypothesized. .6001 .5997 Equation Equgtion Equation 1 3 .6121 309.8051 309.9162 306.2579 Beta Coefficientl (relative importance) .2567** -.2682aa .2007** .2438aa -.1796** -.3042** -.0472 .0267 .2347** .2495aa .1942** .2049aa .1692** .3161** .0990** .2393 -.2648** -.2288aa .1812** .2033aa -.1720** -.3034** -.0965** -.2277** .2942 12 for complete results relationship was opposite 88 median family income (X9), in that order. Educational attainment (X8) had the strongest in- fluence on variations in fertility rates in the East North Central division. The simple correlation between X8 (edue cation) and fertility rates was the highest relative to that between fertility rates and any other variable in equation (3) (ry.8 = -.4261). Education was also posi- tively correlated with family income (X9) (r8.9 = .4252). The regression coefficient of X8 (education) was negative, and it was significant at the .05 level of confidence. The higher the educational attainment of the people in the county, the lower was the fertility rate of the county. The influence of urban areas was important in ex- plaining variation in fertility rates among rural-farm com- munities. The regression coefficients of all three mea— sures of urban influence used (X12, X13, and X14) had signs consistent with the hypotheses and they were significantly different from zero at the .05 level of confidence. The negative and significant regression coefficient of Xlo(per- cent of married women age 15-24) substantiated the hypothe- sis that a larger proportion of women in the younger age group tended to lower fertility rates. Hewever, the posi- tive and non—significant regression coefficient of X11 89 (percent of married women age 25-34) was unexpected, es- pecially in view of the low intercorrelation between X10 and other variables. The percent of farm laborers and farm foremen (X5) was also important. The results substantiated the argument that a relatively larger proportion of farm la- borers and farm foremen in the rural work force tended to increase the fertility rates in rural—farm communities. Hewever, the regression coefficient for X4 (the percent of farmers and farm managers) had a negative sign and it was significant. Both female income (X7) and family income (X9) influenced variations in white fertility rates in this division. The regression coefficients of both variables were negative. The regression coefficients of X7 (female income) and that of X9 (family income) were significant at the .05 level of confidence. These results bear out the hypothesized inverse relationships between the income variables and fertility rates. The positive and significant regression coeffi- cient of female employment (X6) was contrary to the ob- served phenomenon that, at a given point of time, women in the labor force have lower fertility rates than those 90 outside the labor force. However, the high intercorrela- tion between female income (X9) and female employment (X9) (r6.7 = .6998) may be a cause for the unexpected results. In general, the results of the analysis for the East North Central division substantiated the hypotheses regarding the effect of economic and cultural variables on variations in fertility rates. Except for three variables, female employment, percent of farmers and farm managers, and percent of married women age 25-34, the rest of the variables included in the study showed consistent relation— ships with fertility, as hypothesized. West North Central Division Table 5.5 presents a summary of the results of the analysis for this division. Tables l3, l4, and 15 in the Appendix show the detailed results. Of the three equations, estimated for this division, equation (2) accounted for most of the variation in fertility rates (RE = .3218, R3 = .3324, R3 = .2992). It also had the lowest standard error of estimate of the three equations. The median years of school completed by males and females age 25 and over, (X8), was the most important variable rela- tive to other variables in explaining variation in fertility rates among communities. The next most important variable 91 TABLE 5.5 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. West NOrth Central Division Equition Equation Equgtion Multiple Correlation Coefficient .5673 .5766 .5470 Standard Error of Estimate 307.0689 304.6467 312.1330 Beta Coefficient1 Independent Variables (relative importance) Distance from nearest SMSA (X12) .1766** Size-distancel (X13) -.26l4** -.0548 Size-distance2 (X14) Percent of male employed work force who are: farmers and farm managers (X9) .0694** -.0268 .0821** farm laborers and farm foremen (X5) .3135** .3042** .3450** Percent of females, age 14 and over, in county, who are em- ployed (x6) .0838aa .0626 .0805aa Median female personal income for county (X7) -.1658** —.1495** -.1772** Median years of school completed by males and females age 25 and over (X9) Median family income (xg) ‘ —.0125 -.0493 -.0285 Percent of ever married females, age 15—44 who are: age 15-24 (X10) -.3111** -.2939** -.2807** age 25-34 (X11) .0245 .0087 .0329 -.3357** -.3868** —.3512** 1See Appendix, Tables 13, 14 and 15 for complete results **Significant at .05 level aaSignificant at .05 level but the relationship was opposite to that hypothesized. 92 was the percent of farm laborers and farm foremen (X5) fol- lowed by the percent of married females age 15-24 (X10), the size-distancel variable (X13), and median female in- come (X7). Educational attainment was the most important vari- able. The regression coefficient of X9 (educational attain- ment) was negative and significant. Education may be a fac- tor which determines to a great extent the acceptance of urban cultural values in a community. The regression coef- ficient for the percent of farm laborers and farm foremen (X5) was positive and significant. The regression coeffi- cient of X10 (percent of married females age 15-24) was negative and significant. These results are consistent with the hypotheses. The regression coefficient of the size-distancel variable (X13) was negative and significant at the .05 level of confidence. It indicated that the proximity of rural communities to urban areas, as well as the size of the population of the urban area, were important factors in lowering fertility rates, but that the urban city ex- tended its influence over a larger area than would be shown by the size-distance2 variable. The regression coefficience for X12 (distance) and X13 (size—distancez) had signs which 93 were expected, but only that of X12 was significant. Median female income (X9) accounted for some of the variation in fertility rates in this division. Higher median family incomes had a negative effect on the fertility rates in a county relative to that with a lower median. The regression coefficient of X9 (family income) was negative, but was non-significant. The percent of females age 14 and over who were employed (X9) was highly correlated with female income (X7) and with education (X9) (r = .6995. 6.7 = .4954). This r6.8 may account for the unexpected sign of the regression coeffi- cient of X6 (female employment) and its non-significance. In equations (1) and (3) the regression coefficient for X6 (female employment) was again positive but was signifi- cant at the .05 level of confidence. X 4 (percent of farmers and farm managers) was cor- related with X (percent of farm laborers and farm foremen) 5 (r4.5 = .6507). The regression coefficient of X9 was nega— tive and non—significant. However, in equations (1) and (3), the regression coefficient for X4 (percent of farmers and farm managers) was consistent with the positive rela- tionship hypothesized between farmers and farm managers and fertility rates, and was also significantly different poi 94 from zero at the .05 level of confidence. It was difficult to rationalize these conflicting results in the three equa- tions. The percent of married females age 15-24 (X10) was also an important variable affecting white fertility rates. The regression coefficient of X10 was negative and signifi- cant. waever, the regression coefficient of married fe- males age 25-34 (X11) was difficult to explain especially in view of the low intercorrelation among the variables. In summary, the variables included in equation (2) substantiated the hypotheses formulated regarding their re- lationships with fertility rates. Except in the case of married females age 25—34 (X farm managers and farm 11) I foremen (X4) and female employment (X6), the other variables had the expected relationships. Of these variables, only median family income (X9) was non-significant. The North Central Region Table 5.6 presents the summary of the results of the three equations for the North Central region. Tables 16, 17, and 18, in the Appendix contain the detailed results by equation. Equation (1) accounted for 30.37 percent; equa- tion (2) for 31.92 percent, and equation (3) for 29.76 percent 95 (TABLE 5.6 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. North Central Region 4 J Equation Equation Equation 1 2 3 Multiple Correlation Coefficient .5511 .5650 .5456 Standard error of Estimate 314.5378 311.0129 315.9007 Beta Coefficient1 Ipdependent Variables (relative importance) Distance from nearest SMSA (X12) .2280** Size-distance1 (X13) -.3214** - _ ** Size distance2 (X14) .2170 Percent of male employed work force who are: farmers and farm managers (X4) -.0878aa -.1653aa -.1085aa farm laborers and farm foremen (X5) .2356** .2493** .2625** Percent of females, age 14 and over, in county, who are em- ployed (x6) .1162aa .1006aa .1093aa Median female personal income for county (X7) —.l704** -.1552** -.l756** Median years of school completed by males and females age 25 and over (X9) -.3203** -.3614** -.3438**' Median family income (X9) .0463 , .0838aa .5938aa Percent of ever married females, age 15-44 who are: age 15-24 (x10) -.2779** -.2615** -.2518** age 25-34 (x11) .0424 .0318 .5074aa ‘ 1See Appendix, Tables 16, 17, and 18 for complete results ** Significant at .05 level aaSignificant at .05 level but the relationship was opposite to that hypothesized. 96 of the variance in fertility rates among rural-farm counties in the NOrth Central region. The size-distancel variable more closely measured the extent of urban influence. Hew- ever, there was not much difference among the statistics computed for the three measures of urban influence. The simple correlation between fertility rates and the distance variable was .2252; between fertility rate and size-distancel, it was -.1780, and between fertility rates and size-distance2 it was -.1550. The regression coefficients of all three variables were significant and had signs which were expected. The median years of school completed by males and females age 25 and over (X9) was the most important relative to others in equation (1). This importance of the influence of education on fertility rates was shown in the results of the analysis for the two divisions in this region. The next most important was size-distancel (X13), followed in declining importance by the percent of married females age 15-24 (X10), the percent of farm laborers and farm foremen (X5), the per- cent of farmers and managers (X9), median female income (X7), female employment (X6) and family income (X9). All these variables significantly affected fertility rates among rural- farm communities. 97 The regression coefficient of X8 (educational attain- ment) was negative and significant. This bears out the argu- ment that a higher education level in a county is likely to be related to the complex relationships that determine fer- tility rates. The negative and significant regression coeffi- cient of the size-distancel indicates the importance of both closeness of rural communities to a city and the size of the city. The regression coefficients of X 0 (married females 1 age 15-24), X5 (percent of farm laborers and farm foremen), and X7 (female income) were significant at the .05 level of confidence and showed relationships which were expected. In equations (1) and (2), too, these variables showed signifi- cant relationships with fertility rates. The percent of farmers and farm managers (X9) was significantly related to fertility rates but in a direction contrary to that hypothesized. It may be that the high de- gree of intercorrelation between X9 (farmers and farm mana- gers) and other variables obscured the effects of this vari- able on fertility rates. The percent of farmers and farm managers (X9) was correlated with median family income (X9), with married females age 15-24 (X10) and with the size-dis- tancel variable, (14.9 = .6287; r4.10 = .4045; r4.” = .6658) . 98 In equations (1) and (3), which showed similar negative re- lationships of X9 and fertility rates, the correlation be- tween X4 (percent of farmers and farm managers) and distance was .4641 and between X9 and X14 (size-distancez) it was -.5754. Female employment (X6) showed a significant positive relationship to white fertility rates. Here again there was a high correlation between female employment (X9) and female income (X7) (r6.7 = .6264). The regression coefficient of family income (X9) was positive and significant. Although it was correlated with size-distancel (r9 13 = .5870), it was difficult to rationalize the relationship between fer- tility rates and family income. The relationship of X11 (married females age 25—34) to fertility rates was positive, but the regression coefficient of X11 was non-significant. In summary, the results show that in the divisions of the North Central region as well as in the region itself, education was the most important factor affecting variations in fertility rates. The influence of the proximity to and size of urban centers was also important. The percent of married females age 15-24 and female income had equal impor- tance in each of the divisions and in the region. Farm la- borers and farm foremen also affected fertility rates and 99 the results of this variable were as expected in the divi- sions and in the region. Female employment was positively related to fertility rates in the East North Central divi- sion and for the region as a whole. This was contrary to the expected relationship. Finally, the proportion of married females who were age 25—34 had no significant in- fluence in either division or in the region. The Southern Region The Scnlthern region is comprised of three divi- sions. Eight states plus the District of Columbia consti- tute the South Atlantic division. These states are Dela- ware, Maryland, Virginia, West Virginia, North Carolina, South Carolina, Georgia, and Florida. The East South Cen- tral division contains four states; Kentucky, Tennessee, Alabama, and Mississippi, and the West South Central divi- sion contains Arkansas, Louisiana, Oklahoma and Texas. The South Atlantic Division A summary of the results for this division is pre- sented in Table 5.7. Tables 19, 20, and 21 of the Appendix contain the detailed results of the analyses. The percent of variance in white fertility rates explained by equation (1) was 16.08; by equation (2) it was 100 TABLE 5.7 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. South Atlantic Division Equition Equation Equgtion Multiple Correlation Coefficient ' .4011 .4490 .4243 Standard Error of Estimate 339.9806 331.6208 336.0785 Beta Coefficients1 Independent Variables (relative importance) Distance from nearest SMSA (X12) .0810** Size-distancel (X13) -.3095** i — -. ** S ze distance2 (X14) 2217 Percent of male employed work force who are: farmers and farm managers (X9) -.2176aa -.2177aa -.2252aa farm laborers and farm foremen (X5) .1273** .1224** .1452** Percent of females, age 14 and over, in county, who are em- ployed (X6) -.0884 —.1063 -.1027 Median female personal income for county (X7) -.0095 -.0384 -.0335 Median years of school completed by males and females age 25 and over (X8) -.2796** —.2680** -.2482** Median family income (x9) -.0922** .1121aa .0037 Percent of ever married females, age 15-44 who are: age 15-24 (x10) —.0897** -.1050** -.1050** age 25-34 (x11) .0708 .0716 .0634 1See Appendix, Tables 19, 20, and 21 for complete results **Signi£icant it .05 level aaSignificant at .05 level but the relationship was opposite to that hypothesized. 101 20.16, and by equation (3) it was 18.00. Equation (2) accounted for more of the variance than any of the other equations, suggesting that the size-distance1 variable more closely measured the influence of urban areas. urban in- fluence was the most important variable affecting white fertility rates. The median years of school completed (X8) was next most important,followed by the farmers and farm managers (X9), farm laborers and farm foremen (X5), family income (X9), and married females age 15-24 (X10), in that order. The regression coefficients of median female in- come (X7), married females age 25-34 (X11), and female em- ployment (X5) were not significantly different from zero. The influence of large cities as measured by the size-distancel variable had the strongest effect on varia- tions in fertility rates of white women in rural-farm coun— ties in the South Atlantic division. This is also indicated by the results of equations (1) and (3). The distance vari- able (X12) in equation (1) and the size—distance2 variable in equation (3) were significantly related to fertility rates. Hewever, the relative importance of these variables in the equations clearly indicates that distance from a city had a relatively weaker effect than the combined effect of distance and city size. 102 Educational attainment (X9) had a positive and sig- nificant effect on the variations in white fertility rates. The negative effect of farmers and farm managers (X9) was not expected. The regression coefficient of X9 was negative as well as significant. Here again, the relationships shown were clearly opposite to those hypothesized. Both farm la- borers and farm foremen (X5) and family income were impor- tant variables. Farm laborers and farm foremen (X5) were positively related to fertility rates. The positive rela- tionship in the case of family income was inconsistent with the hypothesis. Family income (X9) was correlated with size- distancel (X13) (r9.l3 = .7062). The regression coefficients of both these variables were significantly different from zero at the .05 level of confidence. The negative effects of the relative prevalence of married women in the age group 15—24 (X10) was consistent with the hypotheses. The positive regression coefficient for x11 (married women 25—34) was non-significant. Female employment (X6) and median female income (X7) were nega- tively related to fertility rates. These relationships were expected, but the regression coefficients of these two variables were non—significant. There was a high de- gree of intercorrelation between these two variables 103 (r6.7 = .7519), which may have resulted in large standard errors of the regression coefficients and thus in their non-significance. Although the percent of variance in fertility rates explained by the variables in this division was small, the results provide evidence for the hypothesized relationship between these variables and white fertility rates. The di- rection of influence of these variables was as expected, except in the case of family income, married females age 25-34, and percent of farmers and farm managers. Of the variables that showed relationships opposite to those ex- pected, married females age 25-34 was non-significant. The East South Central Division Table 5.8 contains the summary of the results of the three equations estimated for this division. Tables 22, 23, and 24 of the Appendix present the detailed results for each equation. Equation (3) accounted for most of the variance in white fertility rates in the East South Central division (R: = .3354; R: = .3388; R§ = .3540). The size-distance2 variable was a more representative measure of urban influence than either distance or size-distancel variables in this 104 TABLE 5.8 Some of the Factors Influencing the Number of Children Ever Born to Married‘White Females in 1960. East South Central Division Multiple Correlation Coefficient Standard Error of Estimate .5792 2 .5821 Equation Equation Equation 1 3 .595 321.8234 321.0057 317.28 9 Independent Variables Distance from nearest SMSA (X12) Size-distancel (X13) Size—distance2 (X14) Percent of male employed work force who are: farmers and farm managers (X9) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) Percent of ever married females, age 15-44 who are: a e 15- 4 9 2 (X10) age 25-34 (X11) Beta Coefficients I (relative importance) .0851** -.3474aa .2042** -.4913** .0758 -.0810 —.0815** -.1886** .0070 -.1158** -'.3250aa -.2050** -.5059** .0913 -.1137** -.0325 -.1962** -.0109 -.1673** -.3364aa .2252** -.4962** .0911 -.0784 -.0334 -.2084** -.0093 1See Appendix, Tables 22, 23, and 24 for complete results **Significant at .05 level aaSignificant at .05 level but t9e relationship was opposite to that hypothesized. 105 division. As shown by the beta coefficients, the percent of females age 14 and over who are employed (X9) was the most important variable relative to other variables in equa- tion (3). The next most important were the percent of farm- ers and farm managers (X9) and the percent of farm laborers and farm foremen (X5), followed by married females age 15-24 (X10) and the size-distance2 variable. Female employment had a predominant influence on fertility rates in this division. The regression coeffi- cient of X9 (female employment) was negative and signifi- cantly different from zero at the .05 level of confidence. It bears out the hypothesis that women in the labor force have a lower fertility than those outside the labor force. The direction of the influence of farmers and farm managers (34) was negative and significant, while that of X5 (farm laborers and farm foremen) was positive and significant. The effect of the predominance of married women in the younger age groups was negative as shown by the signs of X10 (married women age 15—24) and X11 (married females age 25-34) but only the regression coefficient of X10 was significant. Urban influence was also an important factor affect- ing fertility rates. The results of all three equations 106 tended to confirm this hypothesis. The regression coeffi- cients of all three variables measuring urban influence were significant and had signs which‘were expected. Hewever, size-distance2 more closely measured the effects of urban influence. It confirmed the hypothesis that both size of the city and the proximity of rural communities to the city were important factors affecting fertility rates. Female income (X7) education (X8) and family income (X9) showed no significant influence on fertility rates. waever, the direction of the influence of education and family income was consistent with expectations. Female in- come (X7) was correlated with female employment (X6) (r6.7 = .6445). In general, the variables explained a larger per- centage of variation in fertility rates of white females in the East South Central division than in the South Atlan— tic division. The expected relationships between fertility rates and the variables were confirmed except in the case of female income and farmers and farm managers. Hewever, only female employment, farm laborers and farm foremen and married females age 15-24 showed significant relationships with fertility rates. 107 The west South Central Division A summary of the results for this division is pre- sented in Table 5.9. Tables 25, 26, and 27 in the Appendix contain the complete results of the analyses. Equation (2) yielded the highest coefficient of multiple determination relative to the other two highest coefficient of multiple determination relative to the other two equations estimated for this division (Rf = .3361; R3 = .3432; 12% = .3326), and also had the lowest standard error of estimate. Therefore equation (2) is discussed fully in this section. Years of school completed by males and females age 25 and over (X9) was the most important variable affecting fertility rates of white females in rural-farm communities in the west South Central division. In declining order of importance were fam- ily income (X9), farm laborers and farm foremen (X5), farmers and farm managers (X9), and size—distance1 (X13). The regres- sion coefficients of the other variables were not significantly different from zero. Intercorrelation among the independent variables was very low. Education had the greatest influence on white fer- tility rates. The direction of the influence was negative, tending to confirm the hypothesis that a higher educational level in a county depressed fertility rates. Median family 108 TABLE 5.9 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. west South Central Division Equation Equation Equation 1 2 3 Multiple Correlation Coefficient .5798 .5859 .5768 Standard Error of Estimate 323.3908 321.6679 324.2382 Beta Coefficientsl Independent Variables (relative importance) Distance from nearest SMSA (X12) -.0686 Size-distance1 (X13) -.1240** Size-distance2 (X19) -.0154 Percent of male employed work force, who are: farmers and farm managers (X9) -.0752 -.1264aa -.1029aa farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em— ployed (X6) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) Percent of ever married females age 15-24, who are: age 15-24 (X10) age 25-34 (X11) .2697** .2151** .2542** -.0518 -.0333 -.0337 .9423 -.0631 —.0713 .4082** -.3677** .2805** -.2974** -.3651** -.3157** -.0436 -.0393 -.0386 -.0381 -.0346 -.O348 1See Appendix, Tables 25, 26, and 27 for complete results ** Significant at .05 level aaSignificant at .05 level but the relationship was opposite to that hypothesized. 109 income (X9) was inversely related to fertility rates. This was consistent with the hypotheses. The relative prevalence of white rural-farm.males who were farm laborers and farm foremen (X5) was positively related to fertility rates in a community. Hewever, the relative prevalence of farmers and farm managers was nega- tively related to fertility rates. The latter was incon- sistent with the hypotheses. The influence of large urban centers was negative and significant. Fertility rates of white women in rural- farm communities near large urban centers were lower than those of women in communities further removed from large cities. However, in equation (1) the regression coeffi- cient of size-distance2 was negative but non-significant. The size-distancel variable in equation (2) was the least important relative to the other significant variables. In all three equations, the most important vari- ables were education and median family income. The direc- tion of their influence was negative as expected, and the regression coefficients were significant. The direction of influence of the other variables including those that were non-significant were consistent with expectations, with the exception of farmers and farm managers. 110 In summary, education and family income were of overwhelming importance-in explaining fertility differen- tials in ghe west South Central division. A point to note is the consistency (with those expected) of the signs of all the $ariables, with the exception of farmers and farm managers. The Southern Region The results of the three equations for this region are summarized in Table 5.10. The complete results are pre- sented in Tables 28, 29, and 30 of the Appendix. The coef- ficients of multiple determination for the three equations were almost identical (R‘la = .2354; R3 = .2521; R§ = .2451). Hewever, equation (2) is discussed in detail as it has the highest coefficient and also the lowest standard error of estimate. Median years of school completed by males and fe- males (Xg) was the most important variable in explaining fertility differentials of white females among rural-farm communities in the southern region. The next most impor- tant was X9 (farmers and farm managers), followed in de- clining importance by female employment (X6), size—distancel (X13), farm laborers and farm foremen (X5), married females 111 TABLE 5.10 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Southern Region 'Multiple Correlation Coefficient Standard Error of Estimate Equation Equation Equation 1 .4852 2 .5021 3 .4951 338.5959 334.8846 336.4448 Independent Variables Distance from nearest SMSA (X12) Size-distance1 (X13) Si - ze distance2 (X14) Percent of male employed work force who are: farmers and farm managers (X9) farm laborers and farm fore- men (X5) Percent of females, age 14 and over, in county, who are emJ' ployed (X9) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X9) .Median family income (X9) Percent of ever married females. age 15-44, who are: age 25-34 (X11) —__‘ 1See Appendix, Tables 28. 29, and **Significant at .05 level Beta coefficients1 (relative importance) .0445 -.2451aa .1458** —.2130** .0326 -.2695** -.1599** -.0786** .0376 -.1640** -.2436aa .1315** -.2320** .0225 -.2898** —.0734** -.0774** .0338 -.1338** .1522** -.2215** .0197 -.2570** -.0862** -.0862** .0324 30 for complete results a aSignificant at .05 level but the relationship was opposite to that hypothesized. 112 age 15-24 (X10) and family income (X9). Median female in- come (X7) and married females age 25-34 (X11) were non-sig- nificant. The negative direction of the influence of educa- tion‘was expected. A county with a high median of years of school completed would have a lower fertility rate than one with a lower median. This high importance of education was found for two of the three divisions of the southern region. The relative prevalence of farmers and farm mana- gers had a negative influence on fertility rates. This was not expected. This negative relationship was also clearly shown in the three divisions of this region. It may be that a relative prevalence of farmers and farm managers in a county tends to depress fertility rates. Female employment (X6) was also an important factor affecting fertility rates. The regression coefficient of X9 was negative as expected. urban influence as measured by the size-distancel variable was inversely related to fer- tility rates. This relationship was expected. Although the size-distancel variable occupied a lower rank in im- portance, it emphasized the combined effects of both dis- tance and size of the cities in fertility rates. 113 The coefficient of farm laborers and farm managers (X5) was positive as expected. The coefficients of married females age 15-24 (x10) and family income (X§) were negative. These relationships were consistent with the hypotheses. A high intercorrelation was present only between female employa ment (x6) and female income (x7) (r63, = .7010). The results of the analyses for the southern region show that in spite of the low percentage of variance explained by the variables, the relationship between these variables and fertility rates were consistent with those hypothesized. Of the nine variables, included in the analyses, seven were sig- nificantly related. Of these, the direction of influence of only one, (farmers and farm managers), was contrary to that expected. The Western Region The Mbuntain division and the Pacific division com- prise the two divisions in the western region. The Mountain division is constituted by eight states: Montana, Idaho, wy- oming, Colorado, New Mexico, Arizona, Utah, and Nevada. The Pacific division contains Oregon, Washington, California, Hawaii, and Alaska. Alaska and Hawaii were omitted in this study. 114 The Mountain Division Table 5.11 presents a summary of the results of this division. Table 31, 32, and 33 of the Appendix con- tain the complete results of the analyses for equations. The percent of variation in fertility rate explained by the variables was 26.51 in equation (1), 26.52 in equation (2) and 25.97 in equation (3). Although the percent of Variation explained was not very much different from those of other divisions, the interpretation of the results was different. Equation (2) is discussed in detail in this sec- tion, but the results of equation (1) and equation (3) were very similar. The important variables were married females age 15-24 (x farmers and farm managers (X4), and median 10)' family income (Mg), in that order. The rest of the variables had regression coefficients which were not significantly dif- ferent from zero. The same variables were significant in equations (1) and (3) with the exception that in equation (1), female employment (XS) was also significantly different from zero at the .05 level of confidence. Married females age 15-24 (X10) showed a negative relationship with fertility rates as expected. Married fe- males age 25-34 (x11) was also inversely related to fertility, 115 TABLE 5.11 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Mountain Division Multiple Correlation Coefficient Standard Error of Estimate Equation Equation Equation 1 2 3 .5149 .5150 .5097 402.3271 402.2858 403.7895 Independent Variables Distance from nearest SMSA (X12) Size—distance1 (X13) Size-distance2 (X14) Percent of male employed work force who are: farmers and farm managers (xz)' farm laborers and farm ,foremen (X5) Percent of females, age 14 and OVEr, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school completed by males and females age 25 [and over (x3) Mbdian family income (X9) Percent of ever married females, age 15-44, who are: age 15-24 (X10) age 25-34 (x11) Beta Coefficientsl (relative importance) -.1051 -.1035 -.227?aa -.2781aa -.2744aa -.0104 —.0325 -.0328 —.1466** -.1340 -.1382 -.1236 -.1049 -.1092 .0840 .0788 .0322 .1466aa .1321aa .1344aa -.3985** -.3744** -.3838** -.0014 -.OO25 -.0023 1See Appendix, Tables 31, 32, and 33 for complete results **Significant at .05 level aaSignificant at .05 level but the relationship was opposite to that hypothesized. 116 while family income (X9)was directly related to fertility. The direction of the influence of farmers and farm managers and familyincome was contrary to expectations. Urban in- fluence showed no significant relationship in any of the three equations. The results of the analysis for this division does not bear out the hypothesized relationships between the variables and fertility rates. Of the three variables that had significant influences on fertility, only married fe— 'males age 15-24 was related in the expected direction. The positive relationship between family income and fertility may be explained by the predominance of the adherents of the Mormon Church, especially in Idaho and Utah. Although the expected relationship between income and fertility may be negative, religious influence on fertility may be strong enough to create a positive relationship. Further, there is an element of pioneering in the people of this division which is consistent with high fertility. The Pacific Division Table 5.12 is a summary of the results of the three equations for this division. Tables 34, 35, and 36 in the Appendix contain the detailed results. 117 TABLE 5.12 Some of the Factors Influencing the NUmber of Children Ever Born to Married White Females in 1960. Pacific Division Multiple Correlation Coefficient Standard Error of Estimate Equation Equation Equation 1 .5413 2 3 .5535 ' .5569 224.2901 222.1728 221.5622 Independent Variables P Distance from nearest SMSA (X12) Size-distancel (X13) Size-distance2 (X14) Percent of male employed work force who are: farmers and farm managers (X4) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) ‘Median female personal income for county (X7) Median years of school completed by males and females age 25 and over.(X8) Median family income (X9) Percent of ever marriedkfemales, age 15-44, who are: age 15-24 (X10) age 25-34 (X11) Beta coefficients1 (relative importance) .0443 -.0004 .0644 .1236 -.1117 -.3815** -.3840** -.2611** -.O474 -.1850 -.1823** -.0472 -.O312 .0796 .0718 .1088 .1242 -.0847 -.0809 —.3701** -.3721** -.2862** -.3025** -.2565** -.2658** -.O446 -.O421 1See Appendix, Tables 34, 35, and 3”: Significant at .05 level 36 for complete results 118 Equation (3) accounted for more variation in fer- tility rates than either of the other two equations (RE = .2930; 12% = .3063: 123 = .3101). In order of relative im— portance, the education variable (X8) had the strongest in- fluence on fertility rates, followed by family income (X9), married females age 15—24 (X10) and size-distance2 (X14). The rest of the variables did not show any significant re- lationship with fertility rates. Education was inversely related to fertility rates. Counties with males and females age 14 and over who had a high median of years of school completed had lower fertil- ity rates than those with a lower median. Family income was also inversely related to fertility rates. These re- lationships were also consistent with expectations. Both age groups of married females were inversely related to fertility, but only the group age 15-24 had a significant relationship. The influence of urban areas as measured by the size-distance2 variable was inversely related to fer— tility. These relationships, too, were consistent with expectations. Among the variables that had no significant influence, only farmers and farm managers (X4) and female employment (X6) had signs contrary to those expected. Fi- nally, although five of the variables were non—significant, 119 the variations in the significant variables were strong enough to account for a third of the variation in fertility rates. The absolute values of the beta coefficients for the variables that were significant were much larger rela- tive to the others. Education, family income, and married females age 15-24 had equal importance in their effects on fertility rates. The western Region Table 5.13 is a summary of the results of the three equations. More complete results are presented in Tables 37, 38, and 39 of the Appendix. Equation (2) accounted for more variance than did any of the other equations (RE = .2162: R: = .2350; R§ = .2237). Intercorrelation among the inde- pendent variables in this region was very low. The percent of married females age 15-24 (X10) was the most important variable in equation (2), followed in declining order of importance by size-distancel (X13), me- dian family income (X9), farmers and farm managers (X4), and median female income (X7). The regression coefficients of the rest of the variables were not significantly differ- ent from zero. 120 TABLE 5.13 Some of the Factors Influencing the Number of Children Ever Born to Married White Females in 1960. Western Region Multiple Correlation Coefficient Standard Error of Estimate Equation Equation Equation 1 2 ‘ .4650 .4848 3 .4730 375.0500 370.5576 373.2605 Independent Variables Distance from nearest SMSA (X12) Size-distancel (X13) Size-distance2 (X14) Percent of male employed work force, who are: farmers and farm managers (X4) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county,who are em— ployed (X6) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) Percent of ever married females, age 15-44, who are: age 15-24 (X age 25—34 (X 10) 11) k 1See Appendix, Tables 37, 38, and **Significant at .05 level aaSignificant at .05 level but the to that hypothesized. Beta Coefficientsl (relative importance) -.0394 —.0932 -.1532§a .0378 .0359 —.0964 -.0987 -.1705** -.1247** -h0054 —.0181 -.3254** -.l941** -.2829** —.2851** -.0179 -.0197 -.1748 -.1299aa .0329 -.O963 -.1370** -.0129 -.2666** -.2851** -.0181 39 for complete results relationship was opposite 121 The relative prevalence of married women in the age group 15-24 was inversely related to fertility rates of white females in this region. This relationship was consistent with expectations. The size-distancel variables showed the expected negative relationship with fertility rates. For this region, of the three variables measuring urban influence, only the size-distancel variable showed any significant rela- tionship. Median family income and female income were inversely related to fertility rates. A county with a higher median family income or female income tended to have a lower fertil- ity rate than one with low medians. This inverse relationship between fertility rates and the income variables was consis- tent with expectations. Farmers and farm managers negatively affected the fertility rates in a county. This relationship was not expected. In general, the results of the analyses for the Western region provide evidence in support of the hypothe— sized relationships between fertility rates and the variables included in the study. The direction of the influence of each variable was consistent with the hypothesized relation- ship except in the case of farmers and farm managers. The unexpected relationships of farmers and farm managers which 122 has occurred throughout the analyses will be discussed in a later section. Finally, the regional equations for this region explained a lower percent of variance in fertility rates than did those for the individual divisions of this region. The "Non-White Fertility Rates" Analysis Equations were estimated for the non-white females for the divisions in the southern region and for the southern region as a whole. Twelve equations were estimated for the non-white fertility rates. Using the same criterion used for the ”white fertility" equations, one equation was selec- ted for discussion from the three equations estimated for each division and for the region. The South Atlantic Division Table 5.14 presents a summary of the three equations estimated for this division. Tables 40, 41, and 42 of the Appendix contain more complete results. Equation (2) was selected for detailed discussion, as it had the highest coefficient of multiple determination (Rf = .2732; kg = .2740; 8% = .2711). The percent of married females age 15-24 (X10) was the most important variable in eXplaining variations in fertility rates of non—white women 123 TABLE 5.14 Some of the Factors Influencing the Number of Children Ever Born to Married Non-White Females in 1960 South Atlantic Division Equation Equation Equation 1 2 3 Multiple Correlation Coefficient * .5218 .5235 .5207 Standard Error of Estimate 545.5685 553.8758 555.0050 Beta Coefficients1 Independent Variables (relative importance) Distance from nearest SMSA (X12) -.0528 Size-distance X -.1296 1 ( 13) Size-d sta e X .063 1 nc 2 ( 14) 2 Percent of male employed work force, who are: farmers and farm managers (Xi) .0334 .0436 .0332 farm laborers and farm foremen (X5) .0956 .1034 .0887 Percent of females, age 14 and over, in ounty, who are em- ployed (X6) .1345 .1301 .1379 Median femalé personal income -.l356 -.1217 -.1296 for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) -.2998(* -.2845** -.3008** -.2085** -.0920 -.2498** Percent of ever married females. age 15-44, whoJare: age 15-24 (X10) age 25-34 (X11) -.4158** -.4115** -.4177** -.2355** -.2479** -.2394** 1 Tables 40, 41, and 42 for complete results See Appendix, ** Significant at .05 level 124 in communities in this division. The next most important was education (X8), followed by married females age 25-34 (x11) . The relative prevalence of married women in the younger age groups (15-24 and 25-34) tended to lower the fertility rates of non-white females. This relationship was significant and was consistent with expectations. Family income was also an important factor which tended to influence fertility rates. The results show a clear inverse relationship between family income and fertility rates. The rest of the variables did not show significant results. Hewever, with the exception of female employment, the rest of the variables affected fertility rates in di- rections as were expected. In equations (1) and (3), me- dian family income had a significant influence on fertility rates. None of the measures of urban influence showed any significant relationship with fertility rates. There was a high degree of intercorrelation among the variables in this division. Farmers and farm managers (Rh) correlated with farm laborers and farm foremen (X5) and with female income (X7) (r4.5 = —.5212; r4.7 = -.427l). Farm laborers and farm foremen (X5) was correlated with fe— male income (X7) and education (X18)(r5 7 = .4329; r = -.5752). 5.8 125 Female employment (X6) was correlated with female income (X7) (r6.7= .6644). The two age group variables were cor- related with each other (r10.11 = -.4198). The size-dis- tancel (X13) was correlated with education and family in- come (r13 8 = .4650; r13.9 = .8415). These intercorrela- tions may have affected the significance of the regression coefficients of these variables. The East South Central Division Table 5.15 summarizes the results of the three equations for this division and Tables 43, 44, and 45 of the Appendix contain the complete results for the equations. Equation (2) explained most of the variance in non-white fertility rates than did the other two equations (RE = .2896; 8% = .3216; 8% = .2934). In order of declining importance, married females age 15-24 (X10) was the most important, followed by median family income (X9), size-distance1 (X13) and married females age 25-34 (X11). The other variables did not show signifi- cant relationships with fertility rates. As in the South Atlantic division, the age-distri- bution of married females affected fertility rates signifi- cantly. The relative prevalence of married females in the 126 TABLE 5.15 Some of the Factors Influencing the Number of Children Ever Born to Married an-White Females in 1960 East South Central Division Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Distance from nearest SMSA (X12) Size-distance1 (X13) Size-distance2 (X14) Percent of male employed work force, who are: farmers and farm managers (X4) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) Percent of ever married females, age 15-44, who are: a - 4 ge 15 2 (X10) a e -34 9 25 (X11) Equation Equation Equation 1 2 3 .5382 '.5671 .5417 561.3958 548.5985 559.8786 Beta Coefficients1 (relative importance) .9358 -.2301** -.O662 ~.0565 .0198 -.0707 .0297 .1267 .0153 .0015 —.Ol35 -.OO72 -.l776 -.1275 -.1570 -.2065** -.1261 -.2077** -.3540** -.2713** -.3568** .3417** -.2965** -.3361** .2444** .1981** -.2360** 1See Appendix, Tables 43, 44, **Significant at .05 level and 45 for complete results 127 younger age-groups tended to have a depressing influence on fertility rates. These relationships were consistent with the hypotheses. Median family income was strongly related to fer- tility rates. The results indicated that a county with a high median family income tended to have a lower fertility rate than did those with lower median family income. In all three equations, median family income showed a signi- ficant relationship with fertility rates. Urban influence, too, tended to depress fertility rates. The effect of urban influence was measured by the size—distancel variable, which took into account the dis- tance of a rural community from the city as well as the size of the city itself. The results supported the hypothe- sis that cities extended their influence longer distances than was hypothesized by the size—distance2 variable. Both the distance variable in equation (1) and size—distance (2) variable in equation (2), although the direction of their influence was consistent with the hypotheses, were non-sig- nificant. Intercorrelation among the independent variables may have resulted in the regression coefficients of some of the variables being non-significant. Farmers and farm 128 managers (X4) was correlated with farm laborers and farm foremen (XS), median female income (X7) and education (X9) = .4144). Farm laborers = .7603; = -.5404; r ”4.5 1"4.7 4.8 and farm foremen was correlated with female income and edu- cation (r5.7 = .6878; r5.8 = .4850) and female employment -(X6) was correlated with female income (r6.7 = .6351). The positive correlation between education and farmers and farm managers may be rationalized on the grounds that, given the low median years of schooling of non-whites, farming con- stitutes a higher occupational group among the occupations available to non—whites. The West South Central Division A summary of the results for this division is pre- sented in Table 5.16 and the complete results are presented in Tables 46, 47, and 48 of the Appendix. The results of the analyses of non-white fertility rates for the West South Central division present a confusing picture. On the one hand, the coefficient of multiple corre- lation for any of the three equations was the highest rela- tive to that for the same equation in any other division or region. On the other hand, only one variable (family income) had a significant effect on fertility rates. 129 TABLE 5.16 Some of the Factors Influencing the Number of Children Ever Born to Married Non-White Females in 1960 West South Central Division Multiple Correlation Coefficient Standard Error of Estimate Equation Equation Equation 1 .6655 2 3 .6834 .6793 579.2482 566.5324 569.5510 lgfiependent Variables Distance from nearest SMSA(X12) Size-distancel (X13) Size-distance2 (X14) Percent of male employed work force, who are: farmers and farm managers (X4) farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school completed by males and females age 25 and over (X8) Median family income (X9) Percent of ever married females, age 15-44, who are: age 15-24 (X10) a e 25-34 X 9 ( 11) Beta Coefficients 1 (relative importance) -.0769 .1347 .1049 -.1128 .0852 .0008 .1943 .1915 .1307 .1468 .1081 .0815 -.0448 .0731 .0793 .0864 -.0578 .0129 -.6081** —.6676** -.7070** —.0989 .0394 —.1446 —.1337 .008q .0304 1See Appendix, Tables 46, 47, and 48 for complete results '1: 1: . Significant at .05 level 130 The percentage of variance explained by equation (1) was 44.28; by equation (2) was 46.70; and by equation (3) it was 46.14. There was no difference in the results of the three equations. In all three equations, only fam- ily income (X9) was significant. It had the highest beta coefficient, the absolute value of which was very much larger than the others. Simple correlation between fer- tility rates and family income was high (ry 9 = .6401). The regression coefficients of the variables measuring urban influence were non—significant and had signs con- trary to those expected. The regression coefficients of the other variables except that of female income and mar- ried females age 25-34 had signs consistent with expecta- tions. Intercorrelation, though low, was fairly widespread among the independent variables. Farmers and farm managers (X4) was correlated with farm laborers and farm foremen (X5), female employment (X6) and female income (X7) (r4 5 = .4742; r4 6 = -.44l4; = -.668l). Female employment was corre- r4.7 lated with female income and family income (r6.7 = .5586; r6.9 = .5240). In summary, although the variables in the study accounted for a high proportion of the variation in fertility 131 rates in this division, as compared to other divisions, only family income was significantly related to fertility rates. Family income tended to be the most important and the only variable which had a significant influence on variations in fertility rates. The Southern Region Table 5.17 summarizes the results of the three equations for the southern region. The complete results are presented in Tables 49, 50, and 51 of the Appendix. The percent of variance explained by the three equations were almost identical (R3 = .2442; R3 = .2458; R3 = .2487). Equation (3) was chosen for detailed discus- sion. The regression coefficients of four variables in equation (3) were significantly different from zero. They were, in order of declining importance, median family in- come (X9), married females age 15-24 (X10), education (X8), and married females age 25-34 (X11). The regression coefficients of the other variables were non-significant. Family income was inversely related to fertility rates. The results showed that a county with a higher median family income tended to have a lower fertility rate than one with a lower median family income. A high median of years 132 TABLE 5.17 Some of the Factors Influencing the Number of Children Ever Born to Married Non-White Females in 1960 Southern Region Equation Equation Equation 1 2 3 Multiple Correlation Coefficient .4942 .4958 .4987' Standard Error of Estimate 590.0039’ 89.4096 588.2745 " Independent Variables Distance from nearest SMSA (X12) —.0317 Sizef-distance1 (X13) Size-distance2 (X14) Percent of male employed work force, who are: farmers and farm managers (X4) -.0547 farm laborers and farm foremen (X5) -.Ol74 Percent of females, age 14 and over, in county, who are em- ployed (X6) -.0639 Median female personal income for county (X7) -.0592 Median years of school completed by males and females, age 25 and over (X9) -.2188** Median family income (X9) -.3574** Percent of ever married females, age 15-44, who are: age 15-24 (X10) -.3045** age 25-34 (X11) —.138l** -.0753 -.0513 .0168 .0729 .0440 .2016** .2973** .2925** -.1365** Beta Coefficientsl (relative importance) .0995 -.0470 .1872 .0545 .0579 -.2282** -.4175** -.3165** -.1402** 1 ** Significant at .05 level See Appendix, Tables 49, 50, and 51 for complete results 133 of school completed by males and females also tended to depress fertility rates in a county. These relationships were consistent with those hypothesized. The age distribution of married females affected the fertility rates inversely. This relationship was con- sistent with the hypothesis that the relative prevalence of married females in the younger age groups tended to de- press the fertility rates in a county. The regression coefficients of distance, size-dis- tancel, and size-distance2 variables were all non-significant. These results are consistent with the statements made earlier that the effect of urban cultural influence on non—white attitudes towards fertility will be different to the extent that the non-white population react differently to the in- fluence of largely white-dominated mass media. There was correlation between the farmers and farm managers (X4) and farm laborers and farm foremen (X5) and female income (X7) (r4_5 = -.5789; r4.7 = -.5336), between female employment (X6) and female income (X7) (r6 7 = .5796). Size-distance2 and family income were also correlated (r = .6590). 14.9 Comparison of the white and non—white analyses in the South indicates differences in the importance of factors 134 affecting fertility rates. For the white population, edu- cation was the most important factor followed in declining importance by occupation, urban influence, female employ- ment, female income and the proportion of females age 15-44, who were 15-24. The proportion of females 15-44 who were 25-34 did not have any significant influence on white fer- tility rates in the South. In contrast, occupations, urban influence, female employment and female income did not have any significant influence on non-white fertility rates in the South. Education, family income and the age distribu- tion of married females were the important factors affecting fertility rates. In summary, the equations for the region explained a lower percentage of variance than those for the divisions in the southern region. In the equations chosen for discus- sion for the divisions as well as those for the regions, X4 (farmers and farm managers), X5 (farm laborers and farm foremen), X6 (female employment), and X7 (female income) showed no significant relationship with fertility rates. The effects of urban influence on fertility rates were im— portant only in the East South Central division. 135 The Conterminous United States Table 5.18 is a summary of the results of the equa- tions for the conterminous United States. Tables 52, 53, and 54 of the Appendix contain the complete results per equation. The equations were estimated for the white pop- ulation only. The percent of variance in fertility rates explained by the equations was low. (11% = .1510; 8% = .1656; R§ = .1596). Hewever, all the variables in equation (2), which.was selec- ted for discussion, were significantly related to fertility rates. The regression coefficients of all the variables were significantly different from zero at the .05 level of confi- dence. Intercorrelation among the independent variables was low with the exception of a relatively high intercorrelation between female employment (X6) and female income (X7) (r6.7= .6759). Education (X8) had the most important influence on fertility rates of white rural-farm females in the conter- minous United States, closely followed by urban influence (X13). Median family income (X9) was next most important, and ranked in order of declining importance were married females age 15-24 (X10), farm laborers and farm foremen (X5), female employment (X6), median female personal income 136 TABLE 5.18 Some of the Factors Influencing the Number of Children Ever ' Born to Married White Females in 1960. Conterminous United States Equation Equation Equation l 2 3 Multiple Correlation Coefficient .3886 .3996 .4070 Standard Error of Estimate 358.7741 355.3657 356.6103 Beta Coefficients1 Independent variables (relative importance) Distance from nearest SMSA (X12) .1912** Size-distancel'(Xl3) -.2540** .. ' _. it * Size distance2 (X14) .2359 Percent of male employed work force, who are: farmers and farm managers (X4) -.0721 -.0511 -.0497aa farm laborers and farm foremen (X5) .1476** .l362** .1543** Percent of females, age 14 and over, in county, who are em— ployed (X6) .0744** -.0925** -.O794** Median female personal income for county (X7) .0804** —.0709** -.0866** Median years of school completed by males and females age 25 and over (X8) .2257** -.2676** -.236l** Median family income (x9) .1058aa .2292aa .2210aa Percent of ever married females, age 15-44, who are: age 15-24 (X 10) .l876** -.1903** -.1913** age 25-34 (x11) .0608aa .0599aa .0633aa 1See Appendix, Tables 52, 53, and 54 for complete results * *Significant at .05 level aa Significant at .05 level but the relationship was opposite to that hypothesized. 137 (X7), married females age 25-34 (X11), and farmers and farm managers (X4). Education was negatively related to fertility rates. The prevalence of males and females with a high median of years of school completed tended to depress fertility rates in a county. This relationship was consistent with expecta- tions. Urban influence as measured by the size-distancel variable was inversely related to fertility rates. The combined effect of both distance of a county from the city, and the size of the city was important in measuring urban influence. The regression coefficients of distance in equa- tion (1) and size-distance2 in equation (3) were both signi- ficant at the .05 level of confidence, which indicated the overall importance of urban cultural influence on fertility rates. Median family income was positively related to fertility rates. This relationship was unexpected. It could be that the equations estimated for the conterminous United States passed through the negative planes of the equations estimated for the regions. The relationship be— tween family income and fertility rates was negative in the regions and divisions. 138 The negative relationship between married females age 15-24 and fertility rates was consistent with expecta- tions. The relative prevalence of farmers and farm managers was positively related to fertility rates. Both female in— come and female employment were inversely related to fertil- ity rates. These relationships were consistent with the hy- pothesis that women in the labor force and female income represented alternative opportunities. A high rate of fe- male labor force participation and female income mean that women forego bearing and rearing children in order to work. The positive relationship between married females age 25-34 and fertility rates was unexpected. Similarly, the negative relationship between fertility rates and far- mers and farm managers was contrary to expectations. In summary, although the percent of variation in white fertility rates explained by the variables was small, the relationships between the variables and fertility rates were significant. The direction of the influence of all the variables, with the exception of family income, married females age 25-34, and farmers and farm managers, were con- sistent with those hypothesized. In the next section, the relevance of the results to the hypotheses is discussed. 139 Consistent deviation of the direction of the influence of the variables is also discussed and explained. Summary Table 5.19 is a summary of the results of the total analysis. The table presents the rank and the sign of each variable by the selected equation for an area. Only the var— iables whose regression coefficients were significant at the .05 level of confidence are included. Broadly, certain uni- form patterns of the influence of the variables included in the study on fertility rates can be indicated. For the whole analysis, no equation showed any significant relationship be- tween fertility rates and the distance variable. However, fertility rates in most geographic areas were affected by urban influence as measured by either the size-distancel or size—distance2 variables. The results indicated that urban influence on fertility rates was a joint function of both the distance of a rural community from a city and the size of the city from which the influence spreads. Median family income, married females age 15-24, and education affected fertility rates in a majority of the areas discussed. Female employment and female income, and farm laborers, and farm foremen affected fertility rates in 140 TABLE 5.19 mmmmry of the Results of the Analysis of Fertility Rates, by Division, Region, and for the Conterminous United States ' in Rural-Farm Counties, 1960. X X X X X Area 4 5 6 7 8 r* s** r s r s r s r s White Analysis . . . . . . Conterminous United States 9 - 5 (+) 6 (-) 7 (-) 1 (-) New England (+) (+) + 1 (-) (-) huddle Atlantic 4 - 3 (+) + 2 (-) (-) Nertheast - 3 (+) + 2 (-) (-) East North Central 3 - 6 (+) 5 +- 7 (-) 1 (-) west North Central - 2 (+) + 5 (-) l (-) North Central 5 - 4 (+) 7 + 6 (-) 1 (-) South Atlantic 3 - 4 (+) (-) (-) 2 (-1 East South Central 2 — 3 (+) l (-) (-) west South Central 4 - 3 (+) (-) (-) (-) South 2 - 5 (+) 3 (—) -+ 1 (—) Mountain 2 - - (-) (_) + Pacific - (+) + (—) l (-) West 4 - (+) (-) 5 (-) (-) N0n4White Analysis.. South Atlantic (+) (+) + (-) 2 (-) East South Central (+) (+) (-) (-) (-) West South Central (+) (+) (-) + (+) South — - (-) (~) 3 (-) *Rank of variable as indicated by the beta coefficient. Only those variables whose regression coefficients were significant at the .05 level were ranked. **Sign of the regression coefficient. Parentheses indi- cate consistency of sign with expectations. X4=Farmers & farm managers; X5=Farm laborers & farm fore- men; X6=Female employment; x7=Fema1e income; X8=Educa- tion; X9=Family income; X10=Married females, age 15-24; X11=Married females, age 25-34; X12=Distance; X13=Size— distancel; X14=Size—distance2. 141 TABLE 5 . l9--Continued — _ . ) ) M... (( ( ...x l_. + X ) \II) \.I ‘1) \ul ) )1] _ _ _ _ _ . _ . . . + 3 ( (ll\ ( (( ( ( (I.\ 1 X 2 4.2 .1 :34. 2 3 \II . 2 I.\ 1... K. \II)) )\.I \II)) )) ) + __ _ + + + .+ __ + . _ . _ . + . l ((I.\ (II\ ((( (:1 II\ X1 8 34 4 . ) \II)\.I ))\-I )\II)) )1!) )\.l)\.l .____.._.__.__ ...— 0 I\ ((( ((II\ ((111.1 ((I..\ (((( 1.. x 4. 4 433 64 6 131 11.. 2 ))\I \.I))\.I )\|I) )))) +\._I+\mI_—_____.__ ...— ll\ l.\ ((( (((l.\ ((I.\ (((II\ 9 x 3 2 5 8 8 5 27 323 211... 142 a lesser number of areas. Married females age 25-34 showed any significant influence on fertility rates only for the non-white analysis and for the analysis of the conterminous United States. The direction of influence of these variables was consistent with that hypothesized with a few exceptions. A major inconsistency, however, was the unexpected relationship between farmers and farm managers and fertility rates. In all the areas, where the regression coefficient of farmers and farm managers was significant, the results indicated a negative relationship between fertility rates and farmers and farm managers. Earlier, it was hypothesized that the relative prevalence of farmers and farm managers in a county was directly related to fertility rates. The unexpected signs of the farmers and farm managers variable may be explained by the following arguments. While the prevalence of the population engaged in agricultural pur- suits is positively correlated with fertility rates, there is also the observed inverse relationship between fertility rates and socio—economic status. Fertility rates are low for the professional and managerial occupational groups and high for those in the farming and unskilled occupation groups. In the rural-farm sector of the united States, the farmers and farm managers and farm laborers and farm foremen constituted 143 68 percent of the white male employed work force and 74 percent of the non-white male employed work force in 1960. For the regions of the united States, the percent of the work force in these two occupational categories for both white and non-white ranged from 61 in the Northeast regions to 73 in the Nerth Central region. These figures do not include unpaid family workers who would have increased the proportion. The percent of professional, technical, and kindred workers in the white rural-farm labor force for the United States was 2.9; for the non-white, it was 2.7. Managers, officials and proprietors, excluding farms, con- stituted 2.9 percent for the white labor force and 3.8 for the non-white labor force. Thus it may be contended that as the upper occupation groups constitute a negligible proportion of the labor force, the rest of the occupations form a scale with farmers and farm managers at the top. Within this limited occupational scale, the inverse rela- tionship between fertility rates and occupation holds, with the farmers and farm managers forming a select group with respect to fertility. Within a rural-farm county, given the high proportion of the work force in agriculture, farmers and farm managers are in a higher social scale than those in other occupations in agriculture. The above 144 observations provide a rational basis for the negative effects of the prevalence of farmers and farm managers upon the fertility rates of rural-farm women. Finally, it was difficult to determine which of the equations; the divisional, the regional, or the na- tional equations, represented the best relationships be- tween fertility rates and independent variables. In gen- eral, the divisional equations explained a higher percen- tage of the variation in fertility rates than the equation for the relevant region. However, the divisional equations had a fewer number of variables that were significant than did the regional equations. The equation for the conter- minous United States explained the least amount of variance relative to that for any division or region. But the vari- ables in the equations for the conterminous United States were all significant. The significance of the regression coefficients may have been effected by the relative larger number of observations in the regional analyses than in the divisional analyses. waever, it is difficult to state categorically that a divisional or regional equation was the most representative of the hypothesized relationships. In the next chapter, regional and divisional variations in the effects of the variables are discussed. CHAPTER VI COMPARISON OF THE RESULTS AMONG DIVISIONS AND REGIONS Introduction In this chapter, the results of the multiple com- parison or contrast tests are presented and discussed. This test identifies significant differences among the regression coefficients of an independent variable. The regression coefficients of each independent variable were compared among all divisions and all regions for the "white fertility“ equations and among the divisions of the southern regions for the "non-white fertility” equations. The test identified differences in the effects of all the variables included in the study with the exception of distance and size-distancez variables. The results of the multiple com- parison tests among the divisions are presented in Table 6.1. Comparison Among the Divisions: "White Fertility" Equations In this section, the regression coefficients of each independent variable are compared among the nine divisions 145 146 TABLE 6.1 Summary of the Results of the Multiple Comparison Tests Among Divisions of the Conterminous united States Divisions Compared Independent Variables x4 X5) X5 X7 X8 X9 X10 x11 X13 "White Fertility" ' Equations New England-Mid Atlantic 0 0 O O O O O 0 0 ” -East North Central 0 O O 0 O O O O 0 “ —West North Central 0 O O O O 0 O 0 0 ” —South Atlantic 0 O O 0 O 0 O 0 0 " -East South Central 0 O O 0 0 O 0 O 0 ” -West South Central 0 O 0 O O 0 O O 0 " —Mountain 0 O O O 0 O 0 0 0 " -Pacific 0 O 0 O 0 O 0 O 0 Mid Atlantic- East North Central 0 O O O 0 0 O 0 O " —West North Central 0 O O O O O O O O ” -South Atlantic 0 0 0 0 0 O 0 O 0 ” -East South Central 0 O O O O O O O O ” —West South Central 0 0 0 0 0 0 0 O O " -Mountain 0 O O 0 0 O O 0 0 " —Pacific 0 O 0 0 0 0 O 0 0 East North Central— West North Central 0 0 O O 0 O O 0 0 ” -South Atlantic 0 0 1 O O 0 0 O O " —East South Central 0 O 1 0 0 O 0 0 0 ” —West South Central 0 O 0 0 O 0 1 0 O " —Mountain 0 O l O 1 O O 0 0 " -Pacific 0 O O O 0 0 O 0 0 West North Central- 1 South Atlantic 0 O *0 O 0 O l O O " -Mountain 0 0 O O l O O O 0 “ -Pacific 0 0 0 0 0 O O 0 O " East South Central 0 O 1 O 0 0 0 0 0 " -West South Central 0 0 ‘0 0 0 l 1 O 0 147 TABLE 6.l--Continued Divisions Compared Independent Variables x4 x5 X6 X7 X8 X9 X10 X11 x13 South‘Atlantic- East South Central 0 0 l 0 0 0 O 0 0 " -West South Central “ 0 O 0 0 0 l 0 0 0 " -Mountain 0 0 0 0 l 0 l 0 0 " —Pacific 0 O O 0 O 0 O 0 0 East South Central— West South Central 0 0 O O l O 0 0 " -Mountain 0 O 0 O 0 O O 0 " -Pacific 0 O O 0 O O 0 O 0 West South Central— Mountain 0 O O O l l l O O " -Pacific 0 0 O 0 0 0 0 O 0 Mountain-Pacific O 0 0 0 0 0 0 0 0 "Non-White Fertility” Equations South Atlantic— East South Central 0 0 O 0 0 l O O 0 ” —West South Central 0 O 0 0 1 l l l 1 East South Central— West South Central 0 0 0 0 0 l 0 l 1 1 denotes that there were significant differences in the regreSsion coefficients of the independent variable for the two divisions compared. 0 denotes that there were no significant differences be- tween the two. X4=farmers & farm managers; X5: farm laborers & farm fore— men; X6=fema1e employment; X7=female income; X8=education; Xg=family income; X10=married females age 15-24; Xll=mar- ried females age 24-35; X13=size-distance1. 148 for which white fertility rates were analyzed. The effects of female employment on white fertility rates were different in the East North Central division and the East South Central division. Its effects in the East North Central division were different from those in the South Atlantic, East South Central, and Mountain divisions. It also had a different effect on the fertility rates in the East South Central division than it did in the West North Central, South Atlantic and West South Central divi— sions. Reference to Table 5.19 in Chapter V will make these differences in the effects of the variables more clear. Fe- male employment had a significant and positive effect on fertility rates in the East North Central division. In the East South Central division, the regression coefficient of female employment was negative and it ranked first, in rela- tive importance, of the five significant variables in that division. It had no significant influence on fertility rates in the West North Central division, the South Atlantic division, and the Mountain division. The effect of median years of school completed by males and females was different in the Mountain division. Comparisons between the divisions indicated that education 149 had a different effect on white fertility rates in the Moun— tain division than it did in the East North Central, West NOrth Central, South Atlantic, and‘West South Central divi- sions. In the Mountain division, the effect of education, though non-significant, was positive. In the East North Central, West North Central, South Atlantic, and West South Central divisions, education had a strong negative influence on fertility rates. It ranked first in relative importance in the East North Central, West North Central, and West South Central divisions, while in the South Atlantic division, it ranked second. Apparently, the weight of the Mormon popu- lation was felt in the results for the Mountain division. Although this division had a high level of education, the influence of the Mormon faith, which encourages large fami- lies, counteracted the negative influence of education on fertility rates. Hence high levels of education and high rates of fertility were characteristics of this division. Family income had a different effect on fertility rates in the West South Central division than it did in the West North Central, the South Atlantic, the East South Cen- tral,*and the Mountain divisions. In the West South Central division, family income was significant and negative. It ranked second in relative importance among the five 150 significant variables in that division. In the South At- lantic division it ranked fifth in importance and was the least important in the Mountain division. Family income had no significant influence on fertility rates in the West North Central and East South Central divisions. The relative prevalence of married females age 15-24 had different effects in the South Atlantic and West South Central divisions. Married females age 15-24 had a different effect on the fertility rates in the South Atlan- tic division than it did in the West North Central and Mountain divisions. In the South Atlantic division, mar- ried females age 15-24 was the least important variable, while it had no significant influence in the West South Central division. It ranked first in relative importance in the East North Central and West North Central divisions. Farmers and farm managers, farm laborers and farm foremen, median female income, married females age 25—34, and size-distancel had similar effects on fertility rates in all divisions. The multiple comparison tests indicated the divisions among which the effects of the other variables were different. These differences in the effects were con— fined to the East North Central, West North Central, South Atlantic, East South Central, West South Central, and the 151 Mountain divisions. "Non—White Fertility” Equations Among the three divisions in the Southern region, the comparisons indicated that the effects of farmers and farm managers, farm laborers and farm foremen, female em- ployment, and female income on non-white fertility rates were similar. Education had a different effect on non-white fer- tility rates in the South Atlantic than it did in the West South Central division. Although the regression coeffi- cient of education was not significant in either division, it was positive in the South Atlantic division and negative in the West South Central division. The effect of family income on fertility rates in the South Atlantic division were different from those in the East South Central and West South Central divisions, and its effects in the East South Central division were different from those in the West South Central division. Family income did not have a significant influence on fer- tility rates in the South Atlantic division. It ranked second of the four significant variables in the East South Central division, and was the only important variable in the West South Central division. It should be noted that 152 the equation estimated for non-white fertility rates for this division explained the highest percent of variance in fertility rates relative to that estimated for any other division, region, or the nation for the white fertility rates or for the other two divisions in the southern region and the southern region for the non-white fertility rates. However, variations in family income accounted for most of the variance in fertility rates in the West South Central division explained by the equation. Married females age 15-24 had a different effect on fertility rates in the South Atlantic division than it did in the west South Central division. This variable ranked first in relative importance among the variables affecting fertility rates in the South Atlantic division, while it had no significant influence in the West South Central divi- sion. The effects of married females age 25-34 and size- distancel were different in the West South Central division compared to the South Atlantic and East South Central divi— sions. In the West South Central division the regression coefficient of married females age 25-34 was non-significant and positive, while it was significant and negative in the South Atlantic and East South Central divisions. Similarly, the regression coefficient of size-distancel was not 153 significant and positive in the west South Central division and negative in the other two divisions. The results of the multiple comparison tests indi- cated that the effects of all the variables with the excep- tion of family income were similar in the South Atlantic and East South Central divisions. The effects of most of the variables on non-white fertility rates in the West South Central division were different from those in the South At- lantic and East South Central divisions. Comparisons Among Regions The comparison of the regression coefficients of each independent variable was done for the estimated white fertil- ity equations among the regions. Table 6.2 presents the re- sults of the comparison among regions. The results of the multiple comparison tests among the regions indicated that female employment affected fertil- ity rates differently in the North Central region than it did in the Southern and Western regions. In the North Central region female employment had a significant and positive in- fluence on fertility rates. It had a signigicant and nega- tive influence in the Southern region and in the Western region the relationship, though not significant, was also negative. 154 TABLE 6.2 Summary of the Results of the Multiple Comparison Tests Among Regions of the Conterminous united States Independent Variables Regions Compared x4 x5 X6 x7 X8 X9 x10 X11 x13 NOrtheast-North Central 0 0 O 0 l l 0 0 0 " -South 0 0 0 O 3 0 0 O 0 " -West 0 O 0 0 O 0 O 0 North Central-South O“ 0 l l 0 1 1 0 0 " -West 0 0 l 0 l l 0 0 0 South-West 0 0 0 0 l 0 l q 0 1 denotes that there were significant differences in the regression coefficients of the independent variables, for the two divisions compared. 0 denotes that there were no significant differences be- tween the two. X4=farmers & farm managers; X5=farm laborers and farm fore- men; X6=fema1e employment; X7=female income; =education; =family income; Xl =married females age 15-24;.X11=females age 25-34; X13=size- istancel. Female income affected fertility rates differently in the North Central and the southern regions. The rela- tionship between fertility rates and female income was sig- nificant and negative in the Nerth Central region, while in the southern region it was positive, though not significant. The effects of education on fertility rates was different in the North Central region as compared to the same regions. The influence of education was negative and significant in the North Central and southern regions. In these two regions, 155 it ranked first in relative importance among the variables affecting fertility rates. In the Northeast and Western regions education had no significant influence on fertility rates. Family income affected fertility rates differently in the North Central region than it did in the other three regions. In all the regions it had a significant and nega- tive influence on fertility rates, but it was more important relative to other variables in the Northeast, southern, and western regions than in the North Central region, where it was the least important. Married females age 15-24 affected fertility rates differently in the southern region than it did in the North Central region, where it was the least im- portant. Married females age 15-24 affected fertility rates differently in the southern region than it did in the North Central and western regions. Again, in these three regions, married females age 15-24 had a significant and negative in- fluence on fertility rates. However, the relative importance of this variable relative to the others differed among the equations for the regions. It ranked first of the five sig- nificant variables in the western region and third of the. eight significant variables in the North Central region. In the southern region, it ranked sixth of the seven significant 156 variables in relative importance. The effects of farmers and farm managers, farm laborers and farm foremen, married females age 25-34, and size—distancel were similar in all the regions. In summary, comparisons of the regression coeffi- cients of the independent variables indicated that there were significant differences in the effects of female em- ployment, education, family income, and married females age 15-24, especially between the divisions in the North Central region, the divisions in the southern region, and in the Mountain division. For the non-white analyses, there were significant differences in the effects of education, married females age 25—34, and size—distancel variables be— tween the East South Central and West South Central divi- sions. The income variables, age distribution of married females and size—distancel affected non-white fertility rates differently in the South Atlantic division than they did in the West South Central division. Except for the effects of family income, the effects of the other vari- ables on fertility rates in the South Atlantic and the East South Central divisions were similar. Among the regions, the effects of some of the variables were different be- tween North Central region and the southern and western 157 regions. The greatest differences were in the effects of female employment, female income, family income, and mar- ried females, age 15-24, between the North Central and the southern regions. There was no significant difference in the effects of these variables between the northern and western regions. These findings provided evidence in sup- port of the hypothesis stated in Chapter III. It was hy- pothesized that fertility rates varied from community to community, in part due to differences in the divisions and regions in which the communities are located, CHAPTER'VII SUMMARY AND CONCLUSIONS This chapter is divided into two sections. In the first section, a general summary of the important conclu- sions, derived from the study, is presented along with the implications of the conclusions for public policy. In the second section, suggestions are made which may be considered in subsequent analyses of this nature and areas for future research are noted. Summary of Findings and Their Implications for Policy The importance of some of the factors affecting differences in fertility rates among rural-farm communi— ties was studied for the white population as well as for the non-white population. The analysis of "white fertil- ity" rates was conducted at the divisional, regional, and national levels. "Non-white fertility" rates were analyzed only for the three divisions in the South and for the south- ern region. The importance of some of the factors was in- dicated by the analyses. 158 159 The importance of the influence of some of the fac- tors and the lack of any significant effect of others on the fertility rates of rural-farm communities in a geographic area was striking. Also striking was the consistency in the importance of some of the factors in influencing fertility rates among the divisions and among the regions. High me- dians of years of school completed by males and females age 25 or over, high median family incomes and the relative pre- valence of married females age 15—24 tended to depress white fertility rates. The relative prevalence of farm laborers and farm foremen tended to increase fertility rates in rural— farm communities. Those communities which were located near large urban centers tended to have lower fertility rates than those located farther away. Education was the most important variable which affec— ted fertility rates in seven of the fourteen geographic areas in which white fertility rates were analyzed. In all these units, education was inversely related to fertility rates. Median family income was significantly related in eleven of the fourteen areas and urban influence, as measured either by the size-distancel or size-distance2 variables affected fertility rates significantly in twelve of the fourteen areas, the only exceptions being, the New England and Mountain divisions. 160 Another important conclusion derived from the anal- yses was that the same factors also were important in their effects on non:white fertility rates. High family income, education and urban influence tended to lower non-white fer- tility rates. A relative prevalence of married females age 15-24 tended to raise fertility rates. The only exception was the significant influence of married females age 25-34 on non-white fertility rates. The study, therefore, clearly indicated the importance of education, income, and urban in? fluence as factors tending to depress the fertility rates of both white and non-white females. These findings raise important implications for pub- lic policy. Agriculture in the United States today suffers from the twin problems of surplus from products production and surplus people. Policies designed to cure one will nec- essarily have a bearing on the second problem. The high fertility rates of the farm sector in the previous century helped the rural areas to provide a steady supply of labor to the cities. Agriculture was the dominant sector in the economy. The rapid industrialization of the economy in the twentieth century has reduced the importance of agriculture. Technological improvements have increased the productivity of inputs in agriculture greatly, but, due 161 to the inelastic nature qf the demand for farm products, the supply of farm products exceeds the demand for them at price levels acceptable to farmers. The resulting low prices and low incomes have to be bolstered with massive price sup- ports. Along with these problems, is also the problem of poverty of certain agricultural areas. This poverty is not as much due to market conditions as to the climatic and top- ographical nature of the specific area itself. 'The indus— trial cities still absorb much of the excess population of the rural areas, but industrial choice of labor has become highly selective, requiring skill and education. The above is a much simplified View of the farm problem, but the dis- cussion of it is common in the literature and needs no elab- oration here. The relevance of the economic problem in agriculture in this study stems from its importance to the fertility rates of the rural sector. The replacement index for the United States is 2.2 children per woman. Rural—farm women average 3.33 children, compared with 2.88 children for the rural-non-farm women, and 2.26 children for the urban women.l lPopulation Bulletin, (Population Reference Bureau, Inc., XIX, No. 3, May, 1963), p. 54. ‘~ 4162 Thus, the high fertility rates of rural-farm.women not only abet the rapid population growth of the nation, but also complicates the pressing need for the farm people to move out of agriculture. High fertility rates among rural-farm women are not undesirable, per se, but they do aggravate the low income problems in agriculture. It has long been observed that the rural-farm sector is characterized by low incomes and high fertility rates. However, it is expected that pplicies designed to eradicate the problem of low incomes will have an effect on rural-farm fertility. Policies which are designed to increase income to a given minimum will not have any effect on fertility rates. Most programs of this nature tend to allocate the subsidies according to family size. Accord- ingly, a large family would receive a larger amount than a small family, and it may appeal for families to have more children. This reaction is observed among city families who are on relief. Policies which are designed to eradicate low incomes through long-term plans would affect fertility rates. These policies may be education, movement of industry into rural areas, or maintenance of full employment in the non-farm economy to accelerate the outflow of the farm population. 163 It was noted earlier that education was an important factor depressing fertility rates of both whites and non-whites. Education not only enlarges the scope of employment, both in the rural community and in the city, but it also influ- ences the response of people to the cultural influence of cities. The higher the educational level attained by the people in a community, the more likely they are to read newspapers, etc., of cities, and the more amenable they will be to changing their values regarding family limita- tion. Education, by enlarging the scope of opportunities available, will reduce those who seek employment as farm .1aborers and farm foremen, who were shown in the study to have a positive influence on fertility rates. Further, a farm laborer with a high education will be more susceptible to urban influences than one with less or no education. The enlarged Scope of employment in higher paying jobs through education will accentuate the impact of education on fertil— ity rates through higher family incomes. Urban influence was also an important factor in de- pressing fertility. Policies designed to attract industry to hitherto rural areas and the growth of small cities into larger cities will have an effect on fertility rates, The spread of urban culture would be abetted by the movement of 164 industry and growth of cities in rural areas. Policies de- signed to solve the problems of low incomes in agriculture also stress education and movement of industry into rural areas. It should be noted therefore, that such policies will have a dual effect of not only solving the income problems in the rural areas, but also of reducing fertility rates. Hence, in the long run, such public policies will help to solve both the economic and demographic problems of the farm population. L Evaluation of the Study Least squares regression techniques were used to study the relationships between fertility rates of rural- farm women and sociological and economic variables. In accordance with the hypothesis that there were regional differences in the effects of these variables on fertility rates, equations were estimated for each division, region, and for the conterminous United States. The results of the analyses were presented and discussed by each division and region. In this section, improvements in subsequent analyses and suggestions for future research in fertility analyses are noted. Although the percent of variance in fertility rates explained by the variables was low, most of the variables 165 were significantly related to fertility rates. However, modification might be made in the specification of some of the variables in order to improve the results. Gener; ally, intercorrelation among the independent variables was low, with the exception of that between female employment and female income. In all the equation, there was a high degree of intercorrelation between these two variables, which may have resulted in large standard errors of the regression coefficients and thus introduced an element of bias into the results. This could account for the few geo- graphic areas where these variables were both significantly related to fertility rates. Omission of either one of these variables may yield better results. The percent of variance explained by the variables was substantially more in the equations for a division than for the equation for the relevant region. The equation for the New England division explained 45.83 percent of the var- iance. Similarly, the equations for the majority of the divisions in a region explained a higher percent of the variance than that for the relevant region. The coefficient of multiple determination for the conterminous United States was the lowest in absolute value relative to that for any division or region. Cultural systems have a greater influence 166 on the attitudes of a couple regarding the number of chile dren they have, as compared to economic conditions. It is quite obvious that the population in a small geographic area such as a state or division is more homogenous than the population of a region or the United States. The per- cent of variance in fertility rates explained by a set of predominantly economic variables will be less in a cultur- ally heterogenous community than in one which is culturally homogenous. Hence, the higher percent of variance explained by the analysis for the divisions than for the regions or for the nation, in the study. A useful area of research would be the analysis of fertility rates in culturally ho- mogenous geographic areas, where economic variables would account for much of the variability in fertility rates. A further step in the above process would be to analyze the influence of religion. In this study, the influence of religion was not studied. The United States Census does not enumerate data regarding religion. However, it is acknowledged that religious beliefs have an influence on the attitude of couples toward family limitation practices, etc. Some faiths do not allow the use of artificial devices to prevent conception, while others are tolerant of such practices. The use of a variable to represent the percent 167 of the population who.belong to a particular faith would enable the study of the effects of religious beliefs on fertility rates. The influence of religion could be an- anlyzed in a study where the data pertain to a small geo- graphic area. Finally, fertility rates were hypothesized to be linearly related to each of the independent variables. The equations estimated indicated variatiqns in fertility rates as linear functions of the independent variables. However the functional relationships between fertility rates and some of the variables may be other than linear. It has been observed that high fertility rates are found among the lowest and highest income brackets. Similarly, the effects of education and urban influence on fertility rates may not be linear. The use of other functional forms, such as logarithmic, quadratic, or polynomial, may reflect the true relationship between these variables and fertility. Initial graphic analyses may show some indication of the nature of the relationships. If the true relationships are other than those hypothesized, a change in the func- tional form of the equations would yield a higher coeffi- cient of multiple determination. 168 The above suggestions provide areas for further re; search. Hewever, much of the suggested research rests on the availability of adequate data. The present study was made possible by the availability of census data in a form capable of analyses by a computer. The availability of cen- sus data in such a form is a necessary prerequisite not only for the analysis of fertility rates but also of other social and economic variables. The provision of data from future censuses on tapes ready for electronic computer processing will greatly facilitate the exploitation of the research potential in census data hitherto relatively untouched. LITERATURE CITED Adelman, I. "An Econometric Analysis of Population Growth," American Economic Review, vol. LIII, No. 3, June, 1963: PP. 314-339. Becker, G. S. "An Economic Analysis of Fertility,” Demo- graphic and Economic Change in Developed Countries. National Bureau of Economic Research, Princeton. Bogue, D. J. The Structure of the Metropolitan Community. University of Michigan, Ann Arbor, 1949. Brownlee, K. A. Statistical Theory and Methodology in Sci— ence and Engineering. New York: John Wiley and Sons, Inc., 1959. Brunner, E. de 8., and Kolb, J. H. Rural Social Trends. New York: McGraw-Hill Book Company, Inc., 1933. Bryant, W} K. An Analysis of Inter-Community Income Differ- entials in Agriculture in the united States. unpub- lished Ph. D. Dissertation. Department of Agri. Ec- onomics, Michigan State University, 1963. Coale, A. J., and Hoover, E. M. Population Growth and Econ- omic Development of Low-Income Countries; A Case Study of India. Princeton: Princeton University Press, 1958. Duncan, 0. J., and Reiss, A. J., Jr. Social Characteristics of Urban and Rural Families. A VOlume in the Census Monograph Series. New Yerk: John Wiley and Sons, Inc., 1956. Easterlin, R. A. "The American Baby Boom in Historical Per- spective," American Economic Review, vol. LI, No. 5, December, 1961, pp. 869-911. 169 170 Ezekiel, M., and Fox, K. A. Methods of Correlation and Re- gressidn Analysis. 3rd ed., New York: John Wiley and Sons, Inc., 1959. Foote, R. J. Analytical Tools for Studying Demand and Price Structures. USDA Agriculture Handbook 146, Washing- ton, 1958. Freedman, D. S. "Relation of Economic Status to Fertility," American Economic Review, V01. LIII, No. 3, June, Freedman, R. Family Planning Sterility and Pdpulation Growth. New York: McGraw-Hfill Book Company, Inc., 1959. Grabill, W; H. et. al. The Fertility of American women, A volume in the Census Monograph Series. New Y0rk: John Wiley and Sons, Inc., 1958. Kirk, D., and Nortman, D. L. "Business and Babies: The In— fluence of the Business Cycle on Birth Rates," Pro- ceedings of the American Statistical Association, Social Statistics Section, 1958, pp. 151-160. Kiser, C. V. "Differential Fertility in the United States," Demographic and Economic Change in Developed Coun- tries, Princeton: Princeton University Press, 1960, pp. 77-112. Kiser, C. V., and Whelpton, P. K. Social and Peychological Factors Affecting_Fertility. New York: Midbank Mem- orial Fund. Vol. II, Part VI and V01. IV, Part VIII. Kiser, C. V., and Whelpton, P. K. "Summary of Chief Findings and Implications for Future Studies," Milbank Memor- ial Fundyguarterly, V01. XXXVI, No. III, July, 1958. Landis, Paul H. Population Problems - A Cultural Integpre- tation. 2nd ed. prepared by Paul K. Hall, New York: American Book Company. 1954. Leibenstein, H. Economic Backwardness and Economic Growth. New Y0rk: John Wiley and Sons, Inc., 1957. 171 0gburn, W. F., and Thomas, D. s. "The Influence of the BuSw iness Cycle on Certain Social Conditions," Journal of the American Statistical Association, Vol. XVIII, 1922, PP. 324-40. Peterson, W3, Population. New York: Macmillan Company, 1961. Population Bulletin. (Population Reference Bureau, Inc.) Vol. XIX, No. 3, May, 1963. Scheffe, H. The Analysis of Variance. New York: John Wiley and Sons, Inc., 1959. Snedecor, G. W; Statistical Methods, 5th ed. Ames: Iowa State College Press, 1956. Thomspon, W} S., and Jackson, W. E. "Fertility in Rural Areas in Relation to their Distance from the Cities," Rural Sociology, Vol. V., No. 2, June, 1940. Thompson, W. 3., Population Problems. New York: McGraw-Hill Book Company, 1942. Thompson, W; S., and Whelpton, P. K. Population Trends in the United States. New York: McGraw Hill Book Com- pany, Inc., 1933. United Nations, The Determinants and Consequences of Popu- Vlation Trends. New Y0rk: United Nations, Department of Social Affairs, Population Division, 1953. U. S. Bureau of Census. U. S. Census of Population, 1960, United States Summary__Genera1 Social and Economic Characteristics, P. C. (1) 1C. U. S. Bureau of Census. U. S. Census of Population 1950. IV Special Reports Part V. Chepter VI "Fertility". U. S. Bureau of the Budget, Standard Metropolitan Statisti- cal Areas. Washington: U. S. Government Printing Office, 1961. Whelpton, P. K. "Geographic and Economic Differentials in Fertility," Annals of the American Academy of Pol iti cal and Social Science, V01. 118, November, 1936. 172 Whelpton, P. K. Needed Population Research. Lancaster: Science Press Printing Company, 1938. woofter, T. J. "Trends in Rural and urban Fertility Rates," Rural Sociology, Vol. 13, No. 1, March, 1959. APPENDIX The results of the analysis of factors influencing the number of observed children ever born to ever married females age 15-44 per 1,000 ever married females age 15-44 per rural farm county in 1960. 173 174 TABLE 1 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 New England Division ‘ _—‘ A - Multiple Correlation Coefficient .6770 Standard Error of Estimate 284.1574 Partial. Beta Computed Independent Variables Regression Coeffi- "t" Coefficients cients values Constant term 4481.3371 .6090 Distance from nearest SMSA (X12) -33.0490 -.1421 -.5868 Percent of male employed work force who are farmers and farm managers (X4) .6747 .1379 .8350 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.3578 .2411 1.4186 Percent of females, age 14 and over, in county, who are employed (X6) 2.1612 .3076 1.2657 Median female personal income for county (X7) —.1164 -.5152 -2.0192** Median years of school com- pleted by males and females age 25 and older (X8) . -.8746 -.0248 -.1506 Median family income (xg) —.0515 -.4861 -2.5467** Percent of ever married fe- males, age 15—44, who are age 15-24 (X10) -1.l715 -.1546 -1.1871 Percent of ever married fe— males, age 15-44, who are age 25—34 (X11) **Significant at .05 level —.6482 -.1392 —1.l451 175 TABLE 2 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 New England Division Multiple Correlation Coefficient .6743 Standard Error of Estimate 285.0873 Partial Beta Computed Independent Variables Regression Coeffi- "t" Coefficiaits cients values Constant term 3922.5037 3.5320** Size-distancel(X13) -1.4601 -.0319 -.1719 Percent of male employed work force who are farmers and farm managers (Xh) .6223 .1272 .7663 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.2905 .2291 1.3337 Percent of females, age 14 and over, in county who are em— ployed (X6) 2.1307 .3032 1.2441 Median female personal income for county (X ) —.1054 —.4667 -l.8920** 7 Median years of school com- pleted by males and females age 25 and older (X8) .5506 .0156 .1027 Median family income (X9) -.0404 -.3817 -2.1898** Percent of ever married fanales, age 15-44, who are age 15-44 —1.4286 -.1886 —1.4551 (X10) Percent of ever married fe- males, age 15-44, who are -.6372 -.1368 -l.1225 age 25-34 (X 11) **Significant at .05 level 176 TABLE 3 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 New England Division Multiple Correlation Coefficient .6761 Standard Error of Estimate 284.4690 Partial Beta Computed Independent Variables Regression Coeffi- "t" Coefficients cients values Constant term 3682.0618 2.3451** Size-distance2 (X14) -5.8597 -.0157 -.4884 Percent of male employed work force who are farmers and farm managers (X4). .5947 .1215 .7335 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.2089 .2146 1.2342 Percent of females, age 14 and over, in county who are employed (X6) 2.2372 .3184 1.2992 Median female personal income for county (X7) -.1041 —.4609 -l.8781** Median years of school com- pleted by males and females age 25 and older (X8) 1.0771 .0306 .1969 Median family income (X9) -.O346 -.3265 -l.6349 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) —1.5394 -.2032 -1.5609 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) —.6602 -.l418 —l.16l6 **Significant at .05 level 177 TABLE 4 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 Middle Atlantic Division Multiple Correlation Coefficient .4263 Standard Error of Estimate 312.1851 Partial Beta Computed Independent Variables Regression Coeffi- "t" Coefficients cients values Constant term 3509.9077 .7012 Distance from nearest SMSA 134.6246 .3586 3.7856** (X12) Percent of male employed work force who are farmers and farm managers (X4) —.2755 -.0606 -.6372 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.8840 .2624 2.4600** Percent of females, age 14 & over, in county, who are employed (X6) .7360 .0919 .5671 Median female personal income for county (X7) -.0137 -.0623 —.3938 Median years of school com- pleted by males and females , age 25 and older (X8) -1.9907 -.0526 -.5961 Median family income (X9) —.0131 --0215 “-2202 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -l.0802 -.1142 -1.1040 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) -.9100 -.1280 -l.3632 **Significant at .05 level 178 TABLE 5 The Results of'the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 Middle Atlantic Division = Multiple Correlation Coefficient .4793 Standard Error of Estimate 302.8961 Partial Beta Computed Independent Variables Regression Coeffi- "t" Coefficients cients values Constant term 3766.0854 5.1534** Size-distancel (X13) -23.8587 -.4766 -4.7444** Percent of male employed work force who are farmers and farm managers (Xi) -.6680 -.1469 -1.5264 Percent of male employed work force who are farm laborers and farm foremen (X5) -1.7997 .2506 -2.4200** Percent of females, age 14 and over, in county who are employed (X6) 1.1473 .1433 .9060 Median female personal income for county (X7) -.0541 -.2449 -1.5210 Median years of school com- pleted by males and females age 25 and older (X8) —2.7303 -.0722 -.8401 Median family income (X9) -.0010 —.0017 -.0179 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -l.2630 -.l335 -1.3271 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) -.9698 -.1364 -1.4972 **Significant at .05 level 179 TABLE 6 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 Middle Atlantic Division Multiple Correlation Coefficient .5309 Standard Error of Estimate 292.4568 Partial Beta Computed Independent Variables Regression Coeffi- "t" Coefficients cients values Constant term 1829.6769 l.9143** Size-distancez (X14) -36.2973 -.6019 -5.7138** Percent of male employed work force who are farmers a and farm managers (xa) -.9137 -.2009 -2.1205a Percent of male employed work force who are farm laborers and farm foremen (X5) 1.7562 .2446 2.4456** Percent of females, age 14 and over, in county who are employed (X6) 1.1917 .1488 .9842 Median female personal income for county (X7) -.0594 —.2687 -l.7373** Median years of school com- pleted by males and females age 25 and older (X8) -4.7676 -.1261 —1.4969 Median family income (X9) .0496 .0820 .8587 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -1.3606 -.l438 -1.4802 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) -1.0103 -.1421 -1.6160 **Significant at .05 level aaSignificant at .05 level but to that hypothesized the relationship was opposite 180 TABLE 7 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 Northeast Region Multiple Correlation Coefficient .4910 Standard Error of Estimate 304.7671 Partial Beta Computed Independent Variables Regression Coeffi- “t" Coefficients cients values Constant term 3135.6082 .8262 Distance from nearest ‘SMSA (x12) 90.9336 .3050 3.2865** Percent of male employed work force who are farmers and farm managers (X4) .1094 .0248 .3097 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.6551 .2524 3.1142** Percent of females, age 14 and over, in county who are employed (X6) .7781 .1053 .7794 Median female personal income for county (X7) -.0295 -.1518 —l.0767 Median years of school com- pleted by males and females age 25 and older (X8) -.3401 -.0104 -.1291 Median family income (X9) -.0122 -.1325 -1.3193 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -1.3801 —.1589 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) -.5679 -2.0719** -.0953 -1.3076 **Significant at .05 level 181 TABLE 8 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 Northeast Region Multiple Correlation Coefficient .5318 Standard Error of Estimate 296.2525 Partial Beta Computed Independent Variables Regression Coeffi- "t" Coefficients cients values Constant term 4121.5466 7.5225** Size-distance1 (X13) -18.0111 -.3692 -4.6513** Percent of male employed work force who are farmers and farm managers (X4) -.2594 -.0590 -.7181 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.6875 .2573 3.2847** Percent of females, age 14 and over, in county, who are employed (X6) 1.0565 .1430 1.0867 Median female personal in- come for county (X7) -.0534 -.2743 -l.9817** Median years of school com- pleted by males and females age 25 and older (X8) -l.5400 -.O474 -.6071 . Median family income (X9) ":0185 "2014 "2‘3909** Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -1.6277 —.1874 -2.4969** Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) -.6719 -.1128 —l.5874 **Significant at .05 level 182 TABLE 9 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 Northeast Region Multiple Correlation Coefficient .5538 Standard Error of Estimate 291.2765 Partial Beta Con Independent Variables Regression Coeffi- ' Coefficients cients vs Constant term 3922.5534 5. Size-distance2(Xl4) —25.3799 -.4476 -5. Percent of male employed work force who are farmers and farm managers (X4) -.4650 -.1057 -1. Percent of male employed work force who are farm laborers and farm foremen (X5) 1.5321 .2336 3. Percent of females, age 14 and over, in county who are employed (X6) 1.2730 .1723 1. Median female personal income for county (X7) -.0564 -.2898 —2. Median years of school com- pleted by males and females age 25 and older (X8) -1.6610 -.0511 - Median family income (X9) -.0131 --1427 -1- Percent of ever married fe— males, age 15-44, who are 4 2 _ -l.6452 -.189 - . 15 24 (X10) Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) —.738l v.1239 -l. ** Significant at .05 level 183 TABLE 10 The Results of the Analysis of Factors Influencin The Number of Children Ever Born To Married White Females in 1960 East North Central Division Multiple Correlation Coefficient .6001 Standard Error of Estimate 309.8051 Partial Beta Independent Variables Regression Coeffi- Coefficients cients Constant term 4693.2001 Distance from nearest SMSA (X12) 108.8512 .2567 Percent of male employed work force who are farmers and farm managers (X4) -.9423 -.2682 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.9071 .2007 Percent of females, age 14 and over, in county who are em- ployed (X6) 2.2161 .2438 Median female personal income for county (X7) —.0694 -.1796 Median years of school com- pleted by males and females age 25 and older (X8) —10.2505 -.3042 Median family income (X9) -.0246 -.0472 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -2.4722 -.2154 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) .2266 .0267 **Significant at .05 level aaSignificant at .05 level but the relationship was to that hypothesized 184 TABLE 11 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 East North Central Division M Multiple Correlation Coefficient .5997 Standard Error of Estimate 309.9162 Partial Beta Co Independent Variables Regression Coeffi- Coefficients cients v Constant term 6375.6173 l3 Size-distance1 (X13) -l9.1639 -.2347 -5 Percent of male employed work force who are farmers and farm managers (X4) -.8768 -.2495 -5 Percent of male employed work force who are farm laborers and farm fore- men (Xs) 1.8456 .1942 4 Percent of females, age 14 and over, in county who are employed (X6) 1.8620 .2049 3 Median female personal income for county (X7) -.O653 -.1692 -2 Median years of school com— pleted by males and females age 25 and older (X8) -lO.6515 -.3l6l -6 Median family income (X9) -.0516 -.0990 -2 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -2.5829 —.2251 Percent of ever married fe- males, age 15-44, who are age 25—34 (X11) .2026 .2393 **Significant at .05 level aaSignificant at .05 level but the relationship was op] to that hypothesized 185 TABLE 12 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 East North Central Division _A._ Multiple Correlation Coefficient .6121 Standard Error of Estimate 306.2579 Partial Beta C< Independent Variables Regression Coeffi- Coefficients cients ; Constant term 6135.2305 ll Size-distance2 (X14) -21.0501 -.2648 -( Percent of male employed work force who are farmers and farm managers (X4) -.8038 —.2288 -5 Percent of male employed work force who are farm laborers and farm fore- men (XS) 1.7217 .1812 2 Percent of females, age 14 and over, in county who are employed (X6) 1.8483 .2033 2 Median female personal in- ' come for county (X7) -.0664 --l720 '4 Median years of school com— pleted by males and females age 25 and older (X4) -10.2240 -.3034 -E Median family income (X4) -.0503 --0965 ’3 Percent of ever married fe- males, age 15-44, who are 2277 I - -2.6129 —. —. age 15 24 (X10) Percent of ever married fe- males, age 15-44, who are age 25-34 (x11) .2491 .2942 **. . . Signiflcant at .05 level 01 aaSignificant at .05 level but the relationship was to that hypothesized 186 TABLE 13 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 west North Central Division Multiple Correlation Coefficient .5673 Standard Error of Estimate 307.0689 0 Partial Beta Independent Variables Regression Coeffi— Coefficients cients Constant term 3698.5026 Distance from nearest 45.4846 .1786 SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) .2218 .0694 Percent of male employed work force who are farm laborers and farm foremen (X5) 2.1731 .3135 Percent of females, age 14 and over, in county, who are employed (X6) .6179 .0838 Median female personal income for county (X7) -.0658 -.l658 — Median years of school com- pleted by males and females age 25 and older (X8) -10.1053 -.3357 - Median family income (X4) -.0050 -.0125 Percent of ever married fe— males, age 15—44, who are age 15-24 LXlO) -2.8224 -.3111 - Percent of ever married fe— males age 15—44, who are - .1673 .0245 25 34 (X11) **Significant at .05 level aaSignificant at .05 level but the relationship was 0 to that hypothesized 187 TABLE 14 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 west North Central Division Multiple Correlation Coefficient .5766 Standard Error of Estimate 304.6467 Partial Beta C Independent Variables Regression Coeffi- Coefficients cients Constant term 4818.9654 L Size-distancel (X13) -l7.4058 -.2614 - Percent of male employed work force who are farmers and farm managers (X4) -.0857 -.0268 Percent of male employed work force who are farm laborers and farm foremen (X5) 2.1084 .3042 Percent of females, age 14 and over, in county, who are em- ployed (X4) .4616 .0626 Median female personal income for county (X7) -.0594 -.l495 - Median years of school com- pleted by males and females age 25 and older (X8) -11.6407 -.3868 +1( Median family income (X9) -.Ol97 -.0493 - Percent of ever married fe- males, age 15-44, who are age 15—24 (X10) -2.6662 -.2939 -. Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) .0594 .0087 **Significant at .05 level 188 TABLE 15 The Results of the Analysis of Factors Influenci The Number of Children Ever Born To Married White Females in 1960 'West North Central Division Multiple Correlation Coefficient .5470 Standard Error of Estimate 312.1350 Partial Beta Independent Variables Regressions Coeffi- Coefficients cients Constant term 3972.1069 Size-distance2 (X14) -6.0640 -.0548 Percent of male employed work force who are farmers and farm managers (X4) .2623 .0821 Percent of male employed work force who are farm laborers and farm foremen (X5) 2.3916 .3450 Percent of females, age 14 and over, in county who are em- ployed (X6) .5934 .0805 'Median female personal income for county (X7) -.0704 -.1772 Median years of school com- pleted by males and females age 25 and older (X4) —lO.5692 -.3512 Median family income (X9) -.0114 -.0285 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -2.5459 —.2807 Percent of ever married fee 'males, age 15-44, who are age 25-34 (X11) .2245 .0329 **Significant at .05 level aaSignificant at .05 level but the relationship was to that hypothesized 189 TABLE 16 The Results of the Analysis of Factors Influenci The Number of Children Ever Born To Married White Females in 1960 North Central Region Multiple Correlation Coefficient .5511 Standard Error of Estimate 314.5378 Partial Beta Independent Variables Regression Coeffi- Coefficients cients Constant term 3602.2467 Distance from nearest SMSA (X12) 61.6676 .2280 Percent of male employed work force who are farmers and farm managers (X4) —.2273 -.0878 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.8063 .2356 Percent of females, age 14 and over, in county who are em- ployed (X5) .9171 .1162 Median female personal income for county (X7) -.0662 -.1704 Median years of school com- pleted by males and females age 25 and older (X8) -l0.0826 -.3203 Median family income (X9) .0034 .0463 Percent of ever married fer males, age 15-44, who are ‘ age 15-24 (X10) -2.7406 -.2779 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) .3039 .0424 **Significant at .05 level aaSignificant at .05 level but the relationship was to that hypothesized 190‘ TABLE 17 The Results of the Analysis of Factors Influencing The Number of Children Ever Born To Married White Females in 1960 North Central Region Multiple Correlation Coefficient .5650 Standard Error of Estimate 311.0129 Partial Beta C01 Independent Variables Regression Coeffie Coefficients cients v; Constant term 4083.5405 16 Size-distancel (X13) -18.4918 -.3214 -8 Percent of male employed work force who are farmers and farm managers (X4) -.4286 —.1653 -4 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.9118 .2493 8. Percent of females, age 14 and over, in county who are employed (X6) .7939 .1006 2. Median female personal income for county (X7) -.0603 -.1552 -4. Median years of school com- pleted by males and females age 25 and older (X4) -11.3734 -.3614 +13 Median family income (X9) .0061 .0838 2 Percent of ever married fe— males, age 15-44, who are age 15-24 (X10) -2.5796 -.2615 -9 Percent of ever married fe- males, age 15—44, who are age 25-34 (X11) .2279 .0318 l **Significant at .05 level aaSignificant at .05 level but the relationship was op] to that hypothesized 191 TABLE 18 The Results of the Analysis of Factors Influenci The Number of Children Ever Born To Married White Females in 1960 North Central Region w Multiple Correlation Coefficient .5456 Standard Error of Estimate 315.9007 Partial Beta Independent Variables Regression Coeffi- Coefficients cients Constant term 3810.3413 Size-distance2‘(X14) -l6.7763 -.2170 Percent of male employed work force who are farmers and farm managers (X4) -.2809 -.1085 Percent of male employed work force who are farm laborers and farm foremen (X5) 2.0127 .2625 Percent of females, age 14 and over, in county who are employed (X6) .8628 .1093 Median female personal income for county (X7) -.0682 -.1756 Median years of school com- pleted by males and females age 25 and older (X8) -10.8202 -.3438 Median family income (X9) .0043 .5938 Percent of ever married fer males, age 15-44, who are age 15-24 (X10) -2.4838 -.2518 Percent of ever married fe- males, age 15-44, who are 25-34 (X11) .3636 .5074 **Significant at .05 level aaSignificant at .05 level but the relationship was to that hypothesized 192 TABLE 19 The Results of the Analysis of Factors Influenci The Number of Children Ever Born To Married White Females in 1960 South Atlantic Division Multiple Correlation Coefficient .4011 Standard Error of Estimate 339.9806 Partial Beta Independent Variables Regression Coeffi- Coefficients cients Constant term 4139.7692 Distance from nearest SMSA (X12) 39.1581 .0810 Percent of male employed work force who are farmers and farm managers (X4) -.7026 -.2176 Percent of male employed work force who are farm laborers and farm foremen (X5) .8137 .1273 Percent of females, age 14 and over, in county who are em- ployed (X6) -.4587 —.0884 Median female personal income for county (X7) -.0022 -.0095 Median years of school com- pleted by males and females age 25 and older (X8) -9.7731 -.2796 Median family income (X9) --.0102 --0922 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -.6761 -.0897 Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) .4116 .0708 **Significant at .05 level aaSignificant at .05 level but the relationship was to that hypothesized 1937 TABLE 20 The Results of the Analysis of Factors Influencin; The Number of Children Ever Born To Married White Females in 1960 South Atlantic DiviSion Multiple Correlation Coefficient .4490 Standard Error of Estimate 331.6208 Partial Beta 1‘ Independent Variables Regression Coeffi—I Coefficients cients Constant term 3633.4022 Size-distance1 (Xi3) —18.8976 -.3095 - Percent of male employed ‘work force who are farmers and farm managers (X4) -.7027 -.2177 - Percent of male employed work force who are farm la- borers and farm foremen (X5) .7825 .1224 Percent of females, age 14 and over, in county who are employed (X4) ‘ -.5515 -.1063 - Median female personal income for county (X7) -.0088 -.0384 Median years of school com- pleted by males and females age 25 and older (X8) -9.3691 -.2680 - Median family income (X4) .0124 .1121 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) —.7916 -.1050 - Percent of ever married fe- males, age 15-44, who are age25-34 (X11) 4 .4165 .0716 **Significant at .05 level aaSignificatn at .05 level but the relationship was c to that hypothesized 194' TABLE 21 The Results of the Analysis of Factors Inf: The Number of Children Ever Born To Married White Females in 1960 South Atlantic Division Multiple Correlation Coefficient J Standard Error of Estimate 336.( ; _ Partial ] Independent Variables Regression Cc *Coefficients cj Constant term 3748.6768 Size-distance2 (X44) —16.7l95 -. Percent of male employed work force who are farmers and farm managers (X4) -.7270 -. Percent of male employed work force who are farm laborers and farm foremen (X5) .9280 Percent of females, age 14 and over, in county who are em- ployed (X6) -.5331 —. Median female personal income for county (X7) -.0077 -. Median years of school com- pleted by males and females age 25 and older (X8) -8-6771 '- Median family income (X9) .0041 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -.7914 -. Percent of ever married fe- males age 15-44, who are age 25-34 (X11) .3689 **Significant at .05 level aaSignificant at .05 level but the relationshj to that hypothesized 195. TABLE 22 The Results of the Analysis of Factors In The Number of Children Ever Born To Married White Females in 1960 East South Central Division Multiple Correlation Coefficient Standard Error of Estimate 321 Partial Independent Variables Regression Coefficients Constant term 5143.3374 Distance from nearest SMSA (X12) 48.0974 Percent of male employed work force who are farmers and farm managers (X4) -1.1112 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.3997 Percent of females, age 14 and over, in county who are em- ployed (X6) -2.7369 Median female personal income for county (X7) .0313 Median years of school com— pleted by males and females age 25 and older (X8) -3.3285 Median family income (X9) -.0323 Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) -l-5493 Percent of ever married fe- males, age 15-44, who are .0517 age 25-34 (x11) **Significant at .05 level aaSignificant at .05 level but the relations to that hypothesized 196 TABLE 23 The Results of the Analysis of Factors Ir The Number of Children Ever Born To Married White Females in 1960 East South Central Division Multiple Correlation Coefficient Standard Error of Estimate 321 Partial Independent Variables Regression Coefficients Constant term 4995.9436 Size-distance1 (X13) -12.5435 Percent of male employed work force who are farmers and farm managers (X4) -l.0416 Percent of male employed work force who are farm laborers and farm foremen (X5) 1.4048 Percent of females, age 14 and over, in county, who are employed (X6) -2.8181 Median female personal income for county (X7) .0378 Median years of school com- pleted by males and females age 25 and older (X6) -4.951l Median family income (X9) -.0129 Percent of ever married fe— males, age 15-44, who are age 15-24 (x10) -l.6121 Percent of ever married fe— males, age 15-44, who are 25-34 (X11) -.0799 **Significant at .05 level aaSignificant at .05 level but the relations to that hypothesized 197. TABLE 24 The Results of the Analysis of Factors The Number of Children Ever B01 To Married White Females in 196 East South Central Division Multiple Correlation Coefficient Standard Error of Estimate 3 Partial Independent Variables Regressior Coefficient Constant term 4828.2269 Size-distance2 (X14) -36.1566 Percent of male employed work force who are farmers and farm managers (X4) -l.078l Percent of male employed work force who are farm laborers and farm foremen (X5) 1.5432 Percent of females, age 14 and over, in county, who are em- ployed (X6) -2.7645 Median female personal income for county (X7) .0376 Median years of school com- pleted by males and females age 25 and older (X8) -3.4146 Median family income (X9) -.0132 Percent of ever married fe- males, age 15-44, who are e 15-24 X -l.7ll6 a9 ( 10) Percent of ever married fe— males, age 15-44, who are 25-34 (X11) —.0682 **Significant at .05 level aaSignificant at .05 level but the relatic to that hypothesized. 198 TABLE 25 The Results of the Analysis of The Number of Children To Married White Female West South Central E ==—_——__—————-—————————_____ Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Re Coe Constant term Distance from nearest SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) **Significant at .05 level 754 199 TABLE 26 The Results of the Analysis < 'The Number of Childr< To Married White Fem; west South Central W Multiple Correlation Coefficien1 Standard Error of Estimate Independent Variables Constant term Size-distance1 (X13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are employed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) Percent of ever married fe- males, age 15-44, who are 25-34 (x11) L * Significant at .05 level it aaSignificant at .05 level but t to that hypothesized ' 200 TABLE 27 The Results of the Analysis C The Number of Childrc To Married White Fema west South Central M Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Constant term Size-distance2 (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— males, age 15-44, who are age 15-24 (X10) Percent of ever married fe- males, age 15—44, who are **Significant at .05 level aaSignificant at .05 level but to that hypothesized \I t 201 TABLE 28 The Results of the Analysis of The Number of Children To Married White Female Southern Regic Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Re Coe Constant term 458 Distance from nearest SMSA (X12) 2 Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (XS) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X6) Percent of ever married fe— males, age 15-44, who are age 15-24 (X10) Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) **Significant at .05 level aaSignificant at .05 level but the to that hypothesized 202 TABLE 29 The Results of the Analysis The Number of Childr To Married White Fem Southern Re W Multiple Correlation Coefficien' Standard Error of Estimate Independent Variables Constant term 4 Si e-di t X z s ancel ( 13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are a e 15—24 X g ( 10) Percent of ever married fe— males, age 15-44, who are age 25-34 (X11) **Significant at .05 level aaSignificant at .05 level but 1 to that hypothesized 203 TABLE 30 The Results of the Analysis of The Number of Children To Married White Female Southern Regic Multiple Correlation Coefficient Standard Error of Estimate Independent Variables RE Coe Constant term Size distancez (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em— ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— males, age 15—44, who are age 15-24 (X40) Percent of ever married fe— males, age 15—44, who are age 25-34 (X11) 444 **Significant at .05 level aaSignificant at .05 level but to that hypothesized the 204 Tkmu§3l The Results of the Analysis c The Number of Childre To Married White Fema Mountain Divi Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Constant term Distance from nearest SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em— ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— males, age 15-44, who are age 15—24 (X10) Percent of ever married fe- males, age 15—44, Who are age 25-34 (X11) **Significant at .05 level aaSignificant at .05 level to that hypothesized 205 TABLE 32 The Results of the Analysis The Number of Child: To Married White Fen Mountain Dix Multiple Correlation Coefficier Standard Error of Estimate Independent Variables Constant term Size-distancel (X13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are employed (X6) Median female personal income for county (X7) Median years of school come pleted by males and females age 25 and older (X8) Median family income (X9) PerCent of ever married fe— males, age 15-44, Who are age 15-24 (X10) Percent of ever married fe— males, age 15-44, who are **Significant at .05 level aaSignificant at .05 level to that hypothesized 206 TABLE 33 The Results of the Analysis The Number of Child: To Married White Fen Mountain Div Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size-distance2 (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em- ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15-24 (x10) Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) **Significant at .05 level aaSignificant at .05 level to that hypothesized 207 TABLE 34 The Results of the Analysis The Number of Childr To Married White Fem PaCifiC Divi Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Distance from nearest SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of sChool com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— males, age 15-44, Who are age 15—24 (X10) Percent of ever married fe- males, age 15—44, who are age 25-34 (X11) **Significant at .05 level 208 TABLE 35 The Results of the Analysis The Number of Childr To Married White Fem Pacific Divi .Multiple Correlation Coefficien Standard Error of Estimate Independent Variables ‘__ Constant term Size-distancel (X13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em- ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— males, age 15—44, Who are age 15-24 (X10) Percent of ever married fe— males, age 15—44, who are 25—34 (X11) —.l **Significant at .05 level 209 Insul36 The Results of the Analysisl The Number of Childr) To Married White Fem, Pacific Divi. Multiple Correlation Coefficien' Standard Error of Estimate Independent Variables Constant term Size-distance2 (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, Who are eme ployed (X6) Median female personal income for county (X7) Median years of school com— pleted'by males and females age 25 and older (X8) .Median family income (X9) Percent of ever married fe- tmales, age 15-44,*who are - 4 age 15 2 (X10) Percent of ever married fe— males, age 15—44, who are age 25—34 (X11) __I **Significant at .05 level 210 TAEU337 The Results of the Analysis The Number of Childl To Married White Fen western Reg ,Multiple Correlation Coefficier Standard Error of Estimate Independent Variables Constant term Distance from nearest A x SMS (12) Percent of male employed work force who are farmers and farm.managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em— ployed (X4) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) .Median family income (X9) Percent of ever married fe— males, age 15-44, who are a 15- 4 X. 9e 2 ( 10) Percent of ever married fe- males, age 15-44, who are age 25-34 (X14) **Significant at .05 level 211 TABLE 38 The Results of the Analysis The Number of Childr To.Married‘White Fem Western-Reg fi- ___-.-__-———--- #h.._____ ___,.——_____h_______ _ Multiple Correlation Coefficien Standard Error of Estimate —L Independent Variables Constant term < Size-distancel (X13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and.farm.foremen (X5) Percent of females, age 14 and over, in county Who are em- ployed (X5) Median female personal income for county (X7) Median years of school com- pleted by males and females 'age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15—44, who are age 15- 24 (X10) Percent of ever married fe- males, age 15—44, Who are age 25-34 (X11) **Significant at .05 level aaSignificant at .05 level to that hypothesized 212 TABLE 39 The Results of the Analysis The Number of Childr To Married White Fem ‘Western Reg Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size-distance2 (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm.foremen (X5) Percent of females, age 14 and_ over, in county who are em— ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15—24 (X10) Percent of ever married fe— males, age 15-44, who are age 25-34 (X11) .ZI **Significant at .05 level aaSignificant at .05 level to that hypothesized 213 TAEUB40 The Results of the Analysis The Number of Childr To Married NonAWhite F South.Atlantic Multiple Correlation Coefficien Standard Error of Estimate Independent Variables constant term Distance from nearest SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are employed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) Percent of ever married fe— males, age 15-44, who are age 25—34 (X11) **Significant at .05 level 214 TABLE 41 The Results of the Analysis The Number of Childr To Married Non4White E South Atlantic Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size—distancel (X13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are emr ployed (X6) Median female personal income for county (X7) Median years of school com- pleted'by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— ‘males, age 15-44, who are a e 15—24 X, 9 ( 10) Percent of ever married fe— males, age 15—44, who are age 25-34 (X11) **Significant at .05 level 215 TABLE‘42 The Results of the Analysis The Number of Childr To Married Non-White F South Atlantic m Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size-distance2 (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em- ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— males, age 15-44, who are Percent of ever married fe- ‘males, age 15-44, who are age 25-34 (X11) **Significant at .05 level 216 TABLE1¢ The Results of the Analysis The Number of Child: To Married N0n4White'] East South Centra Multiple Correlation Coefficie1 Standard Error of Estimate Independent Variables Constant term Distance from.nearest SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em- ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) .Median family income (X9) Percent of ever married fe— males, age 15—44, who are 15— 4 x age 2 ( 10) Percent of ever married fe— :males, age 15—44, Who are age 25-34 (X11) *9: Significant at .05 level 217 TABLE4¢4 The Results of the Analysis The Number of Childr To Married Non-White F East South Centra Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size-distance (X ) l 13 Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em- ployed (X6) Median female personal income for county (X7) Median years of school com: pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44,‘Who are a e 15-24 X 9 ( 10) Percent of ever married fe— males, age 15—44, who are age 25-34 (X11) **Significant at .05 level 218 TABLE 45 The Results of the Analysis The Number of Childr To Married N0n-White F East South Centra Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size-distance2(Xl4) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em— ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) .Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) Percent of ever married fe- -ma1es, age 15-44, who are age 25—34 (X11) **Significant at .05 level 219 TABLE 4E The Results of the Analysis The Number of Child: To Married Non—White I west South Centre Multiple Correlation Coefficier Standard Error of Estimate Independent Variables Constant term Distance from nearest SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em— ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15—44, who are age 15-24 (X10) Percent of ever married fe- males, age 15-44, who are 25—34 (X11) **Significant at .05 level 220 TABLE 47 The Results of the Analysis The Number of Childr To Married Non-White F ‘West South Centra Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size-distancel (X13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county whp are em- ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X6) Percent of ever married fe— males, age 15—44, who are age 15-24 (X10) Percent of ever married fe— males, age 15-44, who are 25-34 (X11) **Significant at .05 level 221 TABLE 48 The Results of the Analysis c The Number of Childre To Married Non-White Fe 'West South Central Multiple Correlation Coefficient Standard Error of Estimate Independent Variables Constant term : Size-distance2 (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em— ployed (x6) 1 Median female‘personal income for c0unty~(X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15—24 (X10) Percent of ever married fe— males, age 15-44, who are 25-34 (X11) L **Significant at .05 level 222 TABLE 49 The Results of the Analysis The Number of Childr To Married Non-White F Southern Re —I Multiple Correlation Coefficien Standard Error of Estimate Independent Variable Constant term Distance from nearest SMSA (X12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em— ployed (X6) Median female personal income for county (X7) Median years of school com~ pleted by males and females age 25 and older (X8) .Median family income (X9) Percent of ever married fe— males, age 15-44, who are age 15—24 (X10) Percent of ever married fe— males, age\15—44, who are age 25-34 (X11) **Significant at .05 level 223 TABLE 50 'The Results of the Analysis C The Number of Childre To Married Non-White Fe Southern Reg —_I Multiple Correlation_Coefficient Standard error of Estimate Independent Variables Constant term Si e-distance X z 1 ( 13) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em— ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) Percent of ever married fe— males, age 15-44, who are a -34 X 9e 25 ( ll) **Significant at .05 level 224 TABLE 53 The Results of the Analysis The Number of Child] To Married N0n-White I Southern Re Multiple Correlation Coefficier Standard Error of Estimate Independent Variables Constant term Size—distance (X 2 14) ‘ Percent of male employed work , force who are farmers and farm managers (X4) ! Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em— ployed (X6) Median female personal income for county (X7) pleted by males and females Median years of school com— ‘ 1 age 25 and older (X8) 1 | Median family income (X9) Percent of ever married fe— males, age 15—44, who are age 15-24 (X10) l l Percent of ever married fe— 1 males, age 15-44, who are l age 25-34 (X11) *1': Significant at .05 level 225 TABLE 5 The Results of the Analysis The Number of Child To.Married White Fe Conterminous Uni‘ Multiple Correlation Coefficie Standard error of Estimate Independent Variables Constant term Distance from nearest A X 8M3 ( 12) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em— ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X8) Median family income (xg) Percent of ever married fe- males, age 15—44, who are age 15-24 (X10) Percent of ever married fe- males, age 15-44, who are 5-3 2 4 (X11) **Significant at .05 level aaSignificant at .05 level but to that hypothesized 226 TABLE 53 The Results of the Analysis The Number of Child: To Married White Fen Conterminous Unit Multiple Correlation Coefficier Standard Error of Estimate Independent Variables Constant term Size-distance2 (X13) Percent of male employed work force who are farmers and farm managers (X ) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county who are em- ployed (X6) Median female personal income for county (X7) Median years of school com— pleted by males and females age 25 and older (X8) Median family income (X9) Percent of ever married fe— males, age 15—44, who are — 4 X age 15 2 ( 10) Percent of ever married fe- males, age 15-44, who are -34 X age 25 ( ll) ** Significant at .05 level aaSignificant at .05 level but to that hypothesized 227 TABLE 54 The Results of the Analysis The Number of Childr To Married White Fem Conterminous Unit .Multiple Correlation Coefficien Standard Error of Estimate Independent Variables Constant term Size-distancel (X14) Percent of male employed work force who are farmers and farm managers (X4) Percent of male employed work force who are farm laborers and farm foremen (X5) Percent of females, age 14 and over, in county, who are em— ployed (X6) Median female personal income for county (X7) Median years of school com- pleted by males and females age 25 and older (X4) Median family income (X9) Percent of ever married fe- males, age 15-44, who are age 15-24 (X10) Percent of ever married fe- males, age 15-44, who are age 25-34 (X11) **Significant at .05 level aaSignificant at .05 level but to that hypothesized We I 5‘ 1’.’ 3:17 .EflilLY munuywmm 1111 03 3 9 2 4| (MINI