SOME SELECTED FINDINGS OF THE PRINCETON FERTILITY STUDY: 1963 CHARLES F. WESTOFF,1 ROBERT G. POTTER,2 AND PHILIP C. SAGP RESUMEN Deepues de tres anos se hizo una reencuesta de una muestra probabilistica de respondientes que al principio vivian en siete de las areas metropolitanas mas grandes de los Estados Unidos y que habian tenido el segundo hijo seis meses antis de la primera entrevista. Se encontr6 en la encuesta anterior que la efectividad de la contraconcepcion fue comparativamente baja entre el casamiento y elnacimiento del primer hijo, y no mejor6 grandemente entre el primero y segundo hijo. La reencuesta revel6, sin embargo, que despues del nacimiento del segundo hijo la eficiencia de la contraconcepcion mejor6 fuertemente. Los autores llegan a la conclusi6n de que el control de nacimientos se practica primordialmente para espaciarlos en los primeros anos dematrimonio y conuna eficiencia comparetivamente baja. A medida que las parejas alcanzan el tamano deseado de familia la efectividad de su practica anticoncepcional aumenta fuertemente. La encuesta explor6 una iez mas una amplia eerie de hypotesis relativas a factores sociales y psicol6gicos vinculados a la fecundidad. Se encontr6 que la religi6n es un factor importante. Los Cat6licos que han asistido a eecuelos de la iglesiaespecialmente collegios tendian a mostraruna fecundidad mayor quelos otros. Se encontr6 que la variable categoria socio-econ6mica tenia una correlaci6n baja con la fecundidad, y que la correlaci6n era positiva para los Cat6licos y negativa para los Protestantes. La movilidad social no [ue relacionada con la fecundidad. Los resultados completos del estudio se presentan en un monografia El TercerHijo publicada porla Princeton Press. Los autores estanpreparando una tercera entrevista para la muestra referida. Thus far two volumes reporting the results of the so-called Princeton Fertility Study have been published-Family Growth in Metropolitan America 4 and The Third Child.5 The analyses of both volumes have been directed toward two main objectives. The first is to examine the degree to which the number and spacing of births is controlled by fertility preferences-that is, desired family size and spacing preferences-with an accompanying effort to learn more about the means and barriers to such control. Secondly, a broad range of social and psychological factors have been considered for their ability to discriminate, both individually and jointly, selected measures of birth intervals, desired family size, family-planning success, and fertility three to four 1 Office of Population Research, Princeton University. 2 Brown University. 3 University of Pennsylvania. • C. F. Westoff, R. G. Potter, Jr., P. C. Sagi, and E. G. Mishler, Family Growthin Metropolitan America, Princeton, Princeton University Press, 1961. 5 C. F. Westoff, R. G. Potter, Jr., and P. C. Sagi, The Third Child, Princeton, Princeton University Press, 1963. 130 years after the second child. In the present review, results will be presented as a series of propositions grouped around the two major interests just outlined. The original probability sample of 1165 respondents represented couples having a second birth in September of 1956, six months before the first interview, and who were at the time residing in one of the seven largest Standard Metropolitan Areas of the United States. The respondents are further delimited by being native-white and English-speaking with marital histories uncomplicated by divorce, permanent separation, death, or extensive pregnancy wastage. The group reinterviewed three years later (1960) is smaller-90S compared to 1165-mainly because of difficulties encountered in relocating couples and because of refusals. A number of our results may be viewed as replications on a more specialized sample of findings from the 1955 Growth of American Families Study" of Freedman, Whelpton and Campbell including generS Ronald Freedman, P. K. Whelpton and Arthur Campbell, Family Planning, Sterility, and Population Growth, McGraw-Hill, New York, 1959. Some Selected Findings of the Princeton Fertility Study: 1963 131 alizations relating to age of first marriage, consensus on the desirability of two to four children, and the relationship between first use of contraception and pregnancy interval. The birth spacing observed, combined with the young ages at marriage and preferences for two, three, or four children means that most couples attain desired family size while still in their late twenties or early thirties and, as a result, face a "risk period"-perhaps averaging ten or more years to menopause-during which they must practice highly effective birth control if they are to limit their families to desired size. Another finding consistent with their results is the low effectiveness of contraception practiced during the first two birth intervals-a time when nearly all couples are practicing contraception to postpone rather than to prevent pregnancies. Nor did contraception seem to become more effective as one proceeded from the first to the second interval. This lack of improvement is puzzling. Given youthful ages at marriage and preferences for two, three or four children, such ineffective contraception projected over entire reproductive periods implies an average of at least two excess pregnancies per couple. Yet data from other sources plainly belie this projection. In particular, about three-quarters of the "relatively fecund" wives aged 35-39 from the Freedman, Whelpton and Campbell study claimed to have successfully limited the number of their pregnancies. Even though these claims might be somewhat exaggerated, a high enough caliber of family limitation was indicated so that one wondered if effectiveness of contraception did not improve later in marriage when being used to prevent unwanted pregnancies rather than merely to space desired ones. Lacking this explanation, one could only posit wholesale reliance upon sterilization and induced abortion. While the Freedman, Whelpton and Campbell study had shown sterilization to be an important element, the incidence indicated could not suffice as the explanation. Nor does it seem plausible to view induced abortion as the primary means of family limitation in this country. A reanalysis of the contraceptive experience of the first two birth intervals, together with an analysis of contraceptive experience following second birth, has led to altered conclusions about the trend of contraceptive effectiveness with increasing marriage duration. By considering only the first 12 months of experience in any birth interval, and recomputing rates of accidental pregnancy on this basis, it is revealed that the fraction of high-risk couples practicing ineffectual contraception is definitely higher in the first interval following marriage than in the interval followingfirst birth. The incidence of early accidental pregnancy continues to decline when one proceeds to the interval following the second birth. Even more striking, however, is the relationship between contraceptive effectiveness and the number of additional children desired which leads to the following proposition: As couples approach desired family size, the effectiveness with which they practice contraception increases sharplyso much so that this improved contraception would appear to be the main factor in the family limitation being attained in this country. A more decisive confirmation of this proposition will be achieved when the sample is interviewed again after additional experience. As a consequence of this relationship, studies not distinguishing between contraception used for family limitation and contraception used for spacing are likely to underestimate the Ultimate use-effectiveness of a given method. By the same token a comparison of two methods based on experience from a single birth interval will be biased if the users of one method are nearer their family-size objectives than users of the other. A case in point is the rhythm method, favored by Catholics, as opposed to the presumably more effective mechanical-chemical methods favored by Protestants and Jews because 132 DEMOGRAPHY the latter prefer fewer children on the average than do Catholics. The improvement in contraceptive control with increasing marriage duration is clearly not a matter of couples shifting from ineffective to effective methods of contraception in as much as the composition of methods used after the second birth is much the same as that between marriage and first birth. Moreover, the predominantly young ages of the wives preclude sharply declining fecundity as an explanation. Nor has any association been found between reported coital frequency and contraceptive effectiveness. In addition, there does not appear to be any important practice effect, wherein contraceptive skill increases with longer usage, since the relationship between effectiveness of contraception and desired number of additional children remains strong among couples practicing contraception for the same number of intervals. Rather the chief mechanism in the improvement of contraception appears to be an increased regularity of practice. The present study is the first to investigate how much couples know about the location of the fertile period within the monthly cycle and the relevance of this knowledge for contraception. There is considerable variation in the extent and accuracy of women's knowledge of the ovulatory cycle. Only half of the women had the correct information by any liberal interpretation. Knowledge about the timing of ovulation in the monthly cycle, however, does not appear to influence efficiency with contraception, except among those using the rhythm method. Wife's desired family size six months after the second birth has proved to be the best single predictor of fertility over the next three years of all the data collected in the first interview. Several reasons for this are: her desired family size is found to remain rather stable (the correlation between desires expressed at the beginning and at the end of the three-year interval is .70), corroborating a result from a Detroit Area Study;' husband and wife's desires are similar (correlation .66): and contraception becomes more effective once desired family size is attained, so that relatively few of those desiring two children have an unwanted third birth. Contributing to the inverse correlation between length of birth intervals and completed family size, so thoroughly documented by Glass and Grebenik for Great Britain" and more recently by the Census Bureau for the United States," is the inverse correlation between length of birth intervals and desired family size. Familysize preferences are linked to birth spacing chiefly in two ways. Wives desiring larger families tend to plan shorter birth intervals as evidenced by more non-use of contraception for reason of wanting another pregnancy as soon as possible and by earlier interruption of contraception after marriage or a childbirth. At the same time, wives desiring more children tend to practice less efficient contraception and more often forego use of it for reasons unrelated to spacing objectives. The average length of the first two birth intervals has proven the strongest predictor, after desired family size, of the number of pregnancies during the three and one-half year period following the second birth. The predictive power depends on multiple bases, such as the positive correlation of consecutive conception delays, as well as the tendencies of couples desiring large families to seek shorter birth intervals than average and to practice contraception more casually during the initial birth intervals. 7 David Goldberg, Harry Sharp, and Ronald Freedman, "The Stability and Reliability of Expected Family Size Data," Milbank Memorial Fund Quarterly, October, 1959, pp, 369-85. 8 D. V. Glass and E. Grebenik, The Trend and Pattern of Fertility in Great Britain: A Report on the Family Census of 1946, H. M. S. 0., London, 1954, especially Chapter VII of Part I. 9 U.S. Bureau of the Census, Current Population Reports, Series P-20, No. 108, Washington (July 12, 1961). Some Selected Findings of the Princeton Fertility Study: 1963 133 lie women believes that their religion takes a position in favor of large families. Although further research is necessary to isolate the social mechanisms by which a norm of high fertility is sustained, we were able to determine that a Catholic education is one of the social mechanisms supporting such a norm. It seems to operate primarily at the college level, to some extent at the secondary school level, and not at all at the elementary school level. Although a process occurs to select the more religious individuals into Catholic educational institutions, the education itself seems to exert an effect on fertility independent of this selectivity. There is RELIGION One of the main emphases in this study certainly ground for research into the nahas been on the connections between re- ture of the influence exerted since it unligion and fertility from which the follow- doubtedly functions at both formal and informal levels. ing generalizations have emerged: The ethnic origin of Catholics is also a Religious preference, that is preference factor affecting fertility although it is by for the Protestant, Catholic, or Jewish no means independent of religiousness. faith, is the strongest of all major social Catholics of Irish heritage both want and influences on fertility. Catholic couples want the most and Jewish couples the have the most children with those of Italifewest children with Protestants in an in- an origin tending to be at the opposite end of the continuum. termediate position. The influence exerted by religion operSOCIO-ECONOMIC STATUS ates primarily through its effect on the Studies of the fertility of socio-economnumber of children desired and only secic classes of the population have been ondarily through fertility-planning sucabundant in demography and constitute cess. To oversimplify somewhat, Catholics one of the main foundations of the socioloby and large appear to want larger families and they have them. Their compara- gist's interest in the area of fertility. In tively ineffective fertility planning should recent years much research in the United be viewed mainly in terms of child-spac- States has revealed a clear contraction of ing. The assumption, based on findings class difference in fertility. Part of this discussed earlier, is that once their larger contraction may be due to different relafamily ambitions are realized their success tionships between socio-economic status in controlling family size improves. At the and fertility for Protestants and Cathopposite extreme are the Jewish couples olics. who want the fewest number of children Regardless of the way in which socioon the average and whose fertility plan- economic status is measured (including ning is extremely successful. subjective measures), fertility is negativeContrary to our own expectations, we ly associated with the socio-econoInic stafound that only a small minority of Catho- tus of Protestants and positively related 10 P. K. Whelpton and C. V. Kiser, eds., Social to the socio-econoInic status of Catholics, and Psychological Factors Affecting Fertility, Milalthough in neither case is the relationbank Memorial Fund, New York, 5 vols., 1946ship strong. There are several important 58. SOCIAL AND PSYCHOLOGICAL FACTORS Investigation into the social and psychological factors affecting fertility follows a line of inquiry begun with the Indianapolis Study.'? Although some of our earlier interests were directed toward personality and attitudinal levels of explanation, the most successful probes have been in the more conventional sociological areas of differential fertility where our classifications presumably reflect sub-cultural normative systems delineated by religious, educational, occupational, residential and other dimensions. 134 DEMOGRAPHY implications of this finding which we will not elaborate here. Socio-economic status is less strongly related to fertility than is religion-a generalization that is valid even considering the opposite directions of association in the two major religious groups. Another line of research pursued in this study was prompted in large part by Goldberg's re-analysis" of Indianapolis data as well as data from other more recent studies." His observation is that the socio-economic differential in fertility may be largely a reflection of rural-urban differences in the background of the different socio-economic classes. Our analysis confirms this hypothesis in part: In the Protestant population, the negative association of fertility with socioeconomic status diminishes and tends to become positive among couples with several generations of urban living. Among Catholics, however, the positive relationship is maintained at approximately the same magnitude for all classes regardless of rural or urban background. One of the outstanding findings of the Indianapolis Study (of Protestant couples) was that the relationship between socio-economic status and fertility was negative for the total sample but positive for couples who had planned their fertility successfully. This finding was important for what it implied about the future relationship between the two variables as contraception becomes more popular and more effective. Our study confirmed this finding among Protestants as well. Among Catholics, however, socioeconomic status and fertility correlate positively in the total sample as well as among successful planners. 11 David Goldberg, "Another Look at the Indianapolis Fertility Data," Milbank Memorial Fund Quarterly, 38 (1960), pp. 23-36. See also David Goldberg, "The Fertility of Two-Generation Urbanites," Population Studies (March 1959), pp, 214-22. 12Ronald Freedman and Doris P. Slesinger, "Fertility Differentials for the Indigenous NonFarm Population of the United States," Population Studies, 15 (1961), pp. 161-73. The relationship between socio-economic status and the extent of planning fertility successfully also follows an opposite pattern for the two religious groups. For example, among Protestants the relationship between education and fertility-planning success is positive while among Catholics it is negative. The explanation for this apparent paradox is the concentration of Catholics who attended college in the Catholic higher educational system. When the association is examined for Catholics whose education is secular the same pattern is found as for Protestants. This clearly requires an important qualification in traditional generalizations about the relation of education to family planning. SOCIAL MOBILITY Probably no hypothesis received more attention in this study than that connecting fertility with social mobility, a variable that was measured in numerous different forms. There is very little more that can be added to a conclusion of essentially no relationship. There are some indications, as for the socio-economic status dimension itself, that what little association exists follows opposite directions for Protestants and Catholics. ECONOMIC RECESSION Between the first round of interviews in 1957 and the second round conducted in 1960 an economic recession as well as a major steel strike occurred in the United States. We were the recipients quite fortuitously of an experimental opportunity to examine the effects of macroeconomic events on the fertility of individuals directly involved. The steel strike seems to have exerted no effect on fertility, but the recession produced a twofold effect. Two main negative effects were distinguished: a reduction in income and unemployment. The former experience was characterized by lower fertility while the unemployed group actually appeared to have experienced somewhat Some Selected Findings of the Princeton Fertility Study: 1963 135 higher fertility than the bulk of the sample unaffected by the recession. The latter finding was interpreted in terms of planning capabilities underlying both economic and fertility behavior. NEXT STEPS We will skip over other findings, or rather lack of findings with personality variables and numerous measures of social relationships within the family and conclude with a few remarks about the third and final phase of this study. We are now starting to interview the sample for the last time. We are beginning with the oldest couples who have been married the longest time and who expressed desires for no more children and continue the interviewing over a five-year period ending with the youngest women in the sample. The primary objective of this final interview will be to secure data on completed family size with which we will be able to evaluate further the validity of the numerous propositions advanced above. The third interview will also afford us the opportunity to extend our inquiry into the relation of religion and fertility, e.g. by obtaining data on the respondents' communal involvement in religion. We are also asking direct questions on abortion in an attempt to get some estimate of incidence in a good though specialized sample. Incidentally, if our pretest experience is any guide there is no difficulty in asking such questions. And finally, the longitudinal design of our study will allow further methodological inquiries into the reliability of survey data on family planning.