FAMILY SIZE INTENTIONS

There have been survey respondents–women or couples–perhaps predominant in some populations, that were unable or unwilling to report family size preferences. However, virtually all contemporary populations, and very likely all future ones, will consist primarily of persons or couples who strategize about family size. Clearly, declining family size preferences constitute a primary cause of fertility transition and will influence post-transition fertility levels. In societies undergoing fertility transition, observed fertility frequently exceeds stated preferences; in post-transition countries of the 1990s the opposite was true. The study of emerging and changing family size preferences and their relation to behavior provides clues to the nature of fertility decisionmaking and to the causes of fertility trends and differentials.

Conceptualization and Measurement

An initial distinction should be made between one's own family size preferences or goals and those deemed appropriate for the average or typical family. The latter concept, usually referred to as "ideal family size," assumes a set of time/place-specific norms or expectations regarding appropriate family size. A common survey item of the 1960s in the United States asked: What number of children do you consider ideal for the typical (American) family? Samples of U.S. women in the 1960s produced modal responses of "two" but substantial proportions chose "three" or "four" children.

The collection and use of data on the ideal family size item has diminished over time. Several reasons account for this waning interest. By the 1980s "two children" had become the predominant ideal family size response, but part of this convergence could be attributed to the widespread public perception that population growth could not continue and that an average of two children per woman was required for population stabilization. Also, the declining prevalence of two-parent families in the period from 1960 to 1990 (along with reduced agreement regarding what constituted "the typical family") exposed the contingent nature of family size norms. If appropriate family size depends on individual and couple characteristics (e.g., marital status, economic status), then useful survey questions need to make explicit the situational context. Finally, some have argued that family size norms would be better represented by questions that asked about a range of acceptable family sizes (e.g., what family size is too small? too large?).

One's own family size preferences are anchored in the life course, represented by upward sloping 45° life lines in Figure 1. Person A was age 15 in 1965 and was age 50 in 2000. Person B was born 25 years laterthan A, reached age 15 in 1990 and 25 in 2000. Let it be assumed that both women were interviewed in January 2000 (represented by a vertical line) and were asked about their past or "retrospective" fertility behavior (solid part of life line) and their future or "prospective" behavior (dashed part of line). Family size preferences are embedded in both retrospective and prospective life line segments.

Prospective Intent

Commonly used prospective questions ask the respondent (for example, respondent B at age 25): Do you intend to have a (or another) child? (And if yes) "How many more children do you intend to have?" These questions raise two fundamental issues. The first is whether fertility intentions–or fertility expectations or fertility desires–should be of paramount, substantive interest. Intentions reflect the respondent's goals (what the respondent plans to do) and, as such, should be strongly linked to subsequent behavior in environments where fertility is controlled. In contrast, expectations invite (sometimes explicitly, but often implicitly) a consideration of impediments that might interfere with one's intentions (such as contraceptive failure or subfecundity) and produce an under-estimate or over-estimate of future fertility. Fertility desires require an even more hypothetical exercise that is linked closely to the concept of "demand for children" (the number of children one would intend if there were no subjective or economic costs to fertility control) and might be seen as a determinant (along with constraints and contingencies) of fertility intentions.

Empirical evidence indicates that many respondents do not detect the differences between these terms or are unable or unwilling to perform the implied conceptual tasks that distinguish them. Further, the demographic literature frequently refers to these questions as family size preferences and ignores the distinctions noted above. Nevertheless, following the admonition that one should ask respondents straightforward questions, intentions seem preferable. Intentions are knowable to the respondent, and thus measurable from respondent reports, and are conceptually important. Intentions are likely to be the stable and dominant component of responses to intended/expected/desired fertility questions. If one wants respondents to consider the additional contingencies implied by expectations or desires, then specific additional questions should be devised to supplement intention questions.

The second issue raised by this pair of questions (i.e., Do you intend a [or another] child? How many more?) is whether prospective intentions are best

FIGURE 1

represented as a fixed target or as a set of sequential decisions. Prior to the 1960s, inquiries concerning fertility intentions were linked closely to a fixed target model (individuals or couples "formulate a desired completed family size and pursued this relative constant target throughout their reproductive life" [Lee, p. 205]). The fixed target model combined with reports of children already born allowed operationalization of intended parity. For example, person B in Figure 1 is 25 years of age at the time of the survey. Her intended parity is the sum of births to date and her reported intended additional births. Mean intended parity for cohorts has frequently been used to anticipate future fertility trends. The accuracy of forecasts based on intended parity depends on the predictive validity of reproductive intentions as explained below.

Fertility researchers have raised serious concerns about the fixed target model. Specifically, strong substantive arguments and substantial empirical work suggest that fertility decisions are better represented as a series of sequential decisions. Children are generally born one at a time, thereby imposing a set of birth intervals. These intervals allow for a set of sequential decisions or at least for reassessments of earlier decisions. Empirical evidence shows that respondents frequently do revise their intentions/behavior based on changed circumstances. Key to this sequential perspective are the claims that: some births may be normative (first and second) and others less so (third births); experience with prior births may affect the decision to have subsequent births; and some relevant factors change in unanticipated ways, and unanticipated factors can affect decisions. The distinction between fixed target and sequential models is important because it challenges whether a numerical intention (for example, two versus three additional children) has any behavioral consequence in abstraction from the current sequential decision (that is, to have another child).

Retrospective Intent

Turning to retrospective experience (i.e., solid lines in Figure 1), demographers have developed a standard procedure and terminology for identifying "wanted" and "unwanted births." Specifically, respondents are asked to recollect their fertility intention at the time of each pregnancy: "At the time you became pregnant did you: want to become pregnant at that time; want to have children in the future, but not now; or not want any additional children?" The first two responses are coded as wanted (although the second is termed a timing failure) and the third category as unwanted. For women who have completed childbearing (e.g., woman A in Figure 1), wanted fertility reflects women's family size preferences and unwanted fertility a component that could have been avoided by effective birth control. The unwanted component of fertility declines as effective contraception and abortion become widely available.

Additional Measures Derived from Birth Intentions

Prior discussion has focused on an individual life line or (its aggregate equivalent) a cohort. However, much fertility estimation is period-based–it answers the question: What is the level of fertility in a given year? The most commonly used/reported measure of period fertility is the total fertility rate, TFR (the number of births a woman would have if she experienced a given set of period age-specific fertility rates). A related concept is the "wanted total fertility rate": What would the TFR be if unwanted births were identified (as described above) and excluded? John Bongaarts in his work of 1990 proposes an alternative measurement based on more limited data. Specifically, using only prospective fertility intentions (asked in a 2000 survey) births in the previous year (1999 to 2000) are divided into wanted and unwanted. The key assumption is that, for each woman, all births to point t on a life line are wanted and all subsequent births are unwanted. Thus, if a woman wants more children in 2000, then a birth last year must have been wanted. In addition, each woman must have one "last wanted birth." Together these observations/assumptions allow calculation of the wanted total fertility rate (or its complement, the unwanted TFR, given by unwanted TFR = TFR wanted TFR).

Fertility intentions also allow measurement of the "unmet need for contraception." If a sexually active woman intends no more children but is not using contraception, then she is at risk of an unwanted pregnancy (she has an unmet need for contraception). Important programmatic efforts aim to make contraception available to those in need. If large proportions of women have an unmet need for contraceptives, then unwanted conceptions will be common. Programmatic success could be measured by declines in unmet need (and fewer unwanted conceptions).

The Predictive Validity of Reproductive Intentions

Suppose that social, economic, and psychological variables are linked to fertility only through fertility intentions. In other words, all relevant factors affect intentions directly and intentions mediate these more distal effects. Indeed, numerous studies show that fertility intentions predict the subsequent behavior of individuals far better than do demographic and social indicators. However, evidence also clearly indicates a more complex process that produces patterned inconsistency between intentions and behavior. Specifically, some groups are better than others at predicting their future behavior; that is, there is an interaction of intent with covariates (e.g., O'Connell and Rogers 1983; Van de Giessen 1992). In addition, some subgroups/periods have higher fertility than others net of intentions, that is, there is a direct effect of group membership/period that bypasses the proximate intention variable (e.g., Thomson 1997; Schoen, Astone et al. 1999). The fact that fertility differences or changes are not always foreshadowed by different or changed intentions challenges the usefulness of intention data for fertility forecasts. In explaining the failure of 1970 intentions data to anticipate the fertility decline between 1970 and 1975, Charles F. Westoff and Norman B. Ryder reasoned that "respondents failed to anticipate the extent to which the times would be unpropitious for childbearing…the same kind of forecasting error that demographers have often made" (p. 449). Thus, intentions and other preference measures can provide clues to future trends and differences, but they should not be expected consistently to perform as reliable indicators of future individual or aggregate behavior.

Variation in the predictive validity of fertility intentions provides clues to the fertility decisionmaking processes. For instance, predictive validity increases if the time frame for the expected behavior is explicit. Consider a pair of twenty-five-year old U.S. women. Both intend to have a first child. But one intends a child soon (within two years) and the other intends to postpone the first birth for at least five years. Research shows that such women behave very differently in the short run (e.g., Rindfuss, Morgan, Swicegood 1988: Chapter 9). Thus, in addition to affirmative and negative responses (to "Do you intend another child?"), a viable strategy–and response–is "wait and see" or "not right now." A question asking if a child is intended in the next few years mimics the operative decision that women or couples make. Cross-national empirical evidence indicates that many women want to postpone the next birth (Lightbourne 1987). Postponement leaves open the possibility that births delayed may be fore-gone. For instance, substantial research indicates that most U.S. childlessness results from a series of decisions to postpone childbearing–rather than from stable childless intentions from a young age.

Despite credible arguments to the contrary, evidence indicates that aggregate preferences of women and men are quite similar (Mason and Taj 1987) and that couples frequently agree on fertility goals (Mason and Smith 2000; Morgan 1985). Nevertheless, consistency between intention and behavior is greater when partners agree (Thomson 1997; Schoen, Astone et al. 1999). Finally, predictive validity increases if one considers the respondents' level of certainty about their stated intention. Some respondents freely admit that they may change their mind or are uncertain whether they will have another child.

See also: Childlessness; Family Planning Programs; Fertility, Below-Replacement; Sex Selection; Unwanted Fertility.

bibliography

Bongaarts, John. 1990. "The Measurement of Wanted Fertility." Population and Development Re-view 16: 487–506.

——. 2001. "Fertility and Reproductive Preferences in Post-Transitional Societies." Global Fertility Transition, eds. Rodolfo. A. Bulatao and John. B. Casterline. New York: Population Council.

Lee, Ronald D. 1980. "Aiming at a Moving Target: Period Fertility and Changing Reproductive Goals." Population Studies 34: 205–226.

Lightbourne, Robert E. 1987. "Reproductive Preferences and Behavior." The World Fertility Survey: An Assessment, ed. J. Cleland and S. Chris. Oxford: Oxford University Press.

Mason, Karen. O., and Herbert L. Smith. 2000. "Husbands' Versus Wives' Fertility Goals and Use of Contraception: The Influence of Gender Context in Five Asian Countries." Demography 37(3): 299–311.

Mason, Karen. O., and Anju M. Taj. 1987. "Differences Between Women's and Men's Reproductive Goals in Developing Countries." Population and Development Review 13: 611–638.

Morgan, S. Philip 1985. "Individual and Couple Intentions for More Children: A Research Note." Demography 22: 125–132.

O'Connell, Martin., and Carolyn C. Rogers. 1983. "Assessing Cohort Birth Expectations Data from the Current Population Survey, 1971–1981." Demography 20: 369–384.

Rindfuss, Ronald R., S. Philip Morgan, et al. 1988. First Births in America. Berkeley, CA: University of California Press.

Ryder, N. B., and C. F. Westoff. 1971. Reproduction in the United States. Princeton, NJ: Princeton University Press.

Schaeffer, N. C., and E. Thomson. 1992. "The Discovery of Grounded Uncertainty: Developing Standardized Questions about Strength of Fertility Motivation." Sociological Methodology 22:37–82.

Schoen, R., N. M. Astone, et al. 1999. "Do Fertility Intentions Affect Fertility Behavior?" Journal of Marriage and the Family 61: 790–799.

Thomson, E. 1997. "Couple Childbearing Desires, Intentions and Births." Demography 34: 343–354.

van de Giessen, H. 1992. Using Birth Expectation Information in National Population Forecasts. National Population Forecasting in Industrialized Countries. N. a. H. C. Keilman. Amsterdam: Swets and Zeitlinger.

van de Walle, E. 1992. "Fertility Transition, Conscious Choice and Numeracy." Demography 29: 487–502.

Westoff, C. F., and N. B. Ryder. 1977. "The Predictive Validity of Reproductive Intentions." Demography 14: 431–453.

S. Philip Morgan