A look forward and a look back

Birth wantedness reports: A look forward and a look back

Williams, Lindy

Birth Wantedness Reports:

A Look Forward and a Look Back*

ABSTRACT: A number of checks can be done to assess reliability of attitudinal data pertaining to fertility. We ascertain how births that would be considered unintended, based on Time I reports of fertility intentions, are classified by respondents at a second interview after the birth occurred. The 1988 National Survey of Family Growth and a telephone reinterview allow us to identify respondents who initially intended to postpone or stop childbearing, but who then had a birth, and to analyze the reports they gave of the wantedness of the pregnancy leading to the birth. We also examine wantedness responses of women who claimed in 1988 that they intended to conceive within the next few years. Reports are compared across a range of respondent characteristics and circumstances, including changes in marital status since the initial interview. The analysis reveals nontrivial inconsistency between women’s stated birth intentions and their reports about the wantedness of subsequent births. Details across subgroups are examined. Results also provide new information about effects of duration since birth on intention reports.

INTRODUCTION

For years, survey data have been analyzed to provide policymakers with information about pregnancy wantedness,1 one important element of reproductive health.

Unfortunately, however, attitudinal reports are generally collected retrospectively and have long been considered subject to recall error or rationalization (Ryder and Westoff, 1969). There are a number of possible checks that can be done to assess the reliability of attitudinal data, and recently there has been a call for augmenting what has been learned from crosssectional research with prospective measures of wantedness (Bachrach and Newcomer, 1999).

In this paper we ascertain directly how births that would be considered unintended, based on a report of fertility intentions before the birth occurred, are actually classified by respondents after the birth has occurred. Using Cycle IV of the National Survey of Family Growth (NSFG) and a telephone reinterview, we identify respondents who intended to avoid a birth at least for the interval in question. We then focus on women who had a birth during the interval and analyze the reports they gave of the wantedness of the pregnancy leading to the birth. For comparability, we also examine wantedness responses of women who claimed in 1988 that they intended to conceive within the next few years and who then did so. We examine associations of “consistent” and “inconsistent” wantedness reports with a range of socioeconomic and demographic characteristics. We also account for any changes in marital status since the initial interview, changes that could certainly affect attitudes toward fertility.

BACKGROUND

One central reason for studying attitudinal data in social science research has been that they are thought to be predictive of eventual behavior. Early research on this topic often held that attitudes were, nearly by definition, consistent with behavior (Chaffee and Roser, 1986). Both common sense and intuitive appeal suggested that, at the very least, attitudes and behavior should be strongly related (Buckley et al., 1990). In fertility research, as in other areas of sociology, it is generally held that attitudes get translated first into desires and then into intentions. These, in turn, are thought to influence instrumental behaviors, which then determine behavioral outcomes (Miller and Pasta, 1995). A number of analyses have deemed fertility intention data useful in predicting both aggregate (Westoff, 1990) and individual fertility (Hermalin et al., 1979).

Over time, however, evidence concerning the effectiveness of attitudinal data in predicting behavior has been mixed. For example, Chaffee and Roser (1986) have argued that the consistency between knowledge, attitudes, and individual behavior tends at best to be moderate, while Lindzey and Aronson (1985: 251) have argued more forcefully that the low “correlations between attitudes and behaviors have been the scandal of the field for half a century.” In fertility research, current conventional wisdom suggests that high expectations about the predictive ability of fertility intentions are probably unwarranted and may be the result of holding intention data to too high a standard (Schoen et al., 1997).

Furthermore, as we have indicated, questions about the usefulness of attitudinal data are not limited to their role as predictors of future behavior but extend to their role as suppliers of knowledge about past events. Although recent work by Joyce et al. (2000) suggests that retrospectively collected data on pregnancy wantedness may be more reliable than has been thought, the weight of available evidence indicates that such data are subject to recall error and rationalization and are therefore likely to underreport negative events (Ryder and Westoff, 1969; Westoff and Ryder, 1977). Specifically, with duration since pregnancy, women are thought to be increasingly likely to report an unintended pregnancy as intended (Ryder and Westoff, 1969, 1972; Ryder, 1973). In addition, data problems may arise if the respondent has not previously given the matter much thought (Hauser, 1990), if he or she is ambivalent about the topic (Piccinino and Peterson, 1999), or if he or she has some reason (conscious or not) for withholding information (Berscheid, 1985).

One recent analysis of this topic compared aggregate prospective and retrospective assessments of unintended fertility in the United States (Williams et al., 1999). The stopping/avoiding intentions of a nationally representative sample of women were compared with their fertility behavior in the few years following the interview. The study looked separately at women who wanted to postpone versus forego future childbearing, because those who were unsuccessful postponers could be considered comparable to those who had experienced mistimed births or pregnancies in the extant research on pregnancy wantedness (which focuses on retrospective reports); similarly, those who wanted to avoid future births altogether could be considered comparable to those who had experienced unwanted births or pregnancies. That inquiry showed the correspondence between fertility intentions, specifically the intention to avoid a birth at least in the near term, and subsequent behavior, specifically the occurrence of a birth in the short time following the initial interview.

Altogether, 18 percent of women who reported intending to discontinue or delay childbearing had a birth, despite these intentions. Where differences were observed between findings based on that prospective analysis and findings based on retrospective analyses, at least two different sets of conclusions could be drawn. First, certain subgroups may be selectively more apt than other subgroups to fail to report births as unintended when they are asked about their attitudes retrospectively. Alternatively, certain subgroups may be more apt than others to experience situational changes that they had not anticipated that made childbearing more desirable during the interval than it had been as of the initial interview.

This research provides one new window into this process, assessing whether women who had a birth despite intentions to the contrary report that particular birth as intended (inconsistent with their prior intention report) or mistimed/unwanted (consistent with their prior intention report).

ANALYSIS

The analysis examines the extent to which births that would be considered unintended based on prospective assessments are actually reported as such (retrospectively) by the mothers in follow-up interviews. Looking at births that occurred to postponers as well as to those who had wished to avoid all future childbearing as of their interview in 1988, we examine how commonly these births were reported as intended rather than unwanted or mistimed. We can then see whether certain subgroups were more likely than others to underreport their births as unintended as of the 1990 reinterview. For comparison purposes, we also include data on those who said at Time 1 that they intended to become pregnant in the near term.

Specifically, we ascertain to what extent women’s stated fertility intentions, as expressed at Time 1, are consistent with their birth wantedness reports, expressed at Time 2, among those women who experienced a birth in the interval. We control for a range of individual characteristics to determine whether certain subgroups of women report attitudes more consistently than others, net of changes in the context in which attitudes are formed. If the NSFG data on the timing of events over the course of the interval were more detailed, we would control for changes in a wide range of situational factors, since a variety of changes in one’s circumstances no doubt play a role in influencing attitudes (Andrews and Kandel, 1979; Fishbein and Ajzen, 1975). Unfortunately, the data have adequate precision on event timing for marital status only. At the same time, however, marital status is certainly among the most critical changes that could occur to change one’s perception of the desirability of fertility (Reed and McBroom, 1995).

For those who reported wanting no more children at Time 1,(2) we also examine the effect of respondents’ level of certainty about their fertility intentions. It is generally believed that attitudes that are reported as very certain are apt to be more stable over time than less certain ones and less likely to fluctuate with routine changes in daily circumstances (Westoff and Ryder, 1977; Thomson and Brandreth, 1995).

Finally, as we have noted, attitude reporting is thought likely to become less accurate with time; as a result, we also include a measure of how much time has elapsed since conception. The interval covered by this analysis is very short; nonetheless, we assess the extent to which time may have contributed to differential reporting among respondents in our small sample.

DATA AND METHODS

The analysis is based on data from Cycle IV of the NSFG and a subsequent computer-assisted telephone (CATI) reinterview, both of which were conducted by the National Center for Health Statistics. The Cycle IV data were collected between January and August 1988, using a multistage area probability sample of women between the ages of 15 and 44. Interviews were conducted with 8,450 women of all marital statuses, 2,771 of whom were black, 5,354 of whom were white, and 325 of whom reported identification with another racial group. All belonged to the noninstitutionalized population of the United States.

The telephone reinterview took place between July and mid-November 1990. The final sample, including a teen supplement that was drawn to account for the aging of the original sample, included 5,686 women. The final response rate for the telephone reinterview was 67.5 percent; this figure includes, of course, the response rate from the original NSFG interview (82.1 percent).3

The sample for this analysis excludes all those who were pregnant at the time of the original interview (4.1 percent of the sample), because they were asked about their intentions toward births subsequent to the one they were expecting, and were at low risk of experiencing that additional birth within the short period between the two surveys. The sample also excludes those who reported contradictory information during the 1988 interview about whether they were seeking pregnancy at the time (16 cases). It excludes women who were contraceptively sterile or whose husbands were contraceptively sterile (1,306 cases, or 26 percent of the sample), and those who were noncontraceptively sterile or sterile for unknown reasons (a total of 279 cases, or 5.6 percent of the sample). These women were not asked about future birth intentions and were not at risk of an unpredicted pregnancy in the interval. Eight seemingly contradictory cases remain in the sample (four women who wanted no (more) children but whose husbands/ partners did, and four who wanted to postpone but whose husbands/partners did not).

We examine births that occurred in the interval between the 1988 interview and the 1990 reinterview to women who had said in 1988 that they wanted to become pregnant within the next few years (subsample 1), women who said they intended to postpone childbearing at least for the next three years (subsample 2), and women who said they wanted no (more) children (subsample 3). The questions, including the preamble from which our assessment of fertility intentions is derived, are as follows: “Knowing the number of children women have now and the number they expect to have in the future is important in understanding how our population will grow. It is impossible to look into the future and know exactly how things will turn out, but we often have some idea about what we intend to do. Looking to the future do you (and your husbandpartner) intend to have another) baby at some time?” Later in the series respondents were asked, “When do you expect to have your first/next child; that is, in how many years?” From these questions we are able to identify women who intended to have no (more) children (whom we discuss as permanent avoiders), those who intended to have another) child but not within the three years following the 1988 interview (whom we discuss as temporary postponers), and those who did intend to have another) child within the next few years (whom we discuss as pregnancy seekers).4 We examine these groups separately to see how they reported births that occurred to them during the interval.5

THE VARIABLES

As introduced earlier, the terms “wantedness” and “intendedness” and the categories of variables that are derived from these terms (wanted, mistimed, or unwanted; and intended or unintended) are used to describe the mother’s attitude toward each pregnancy at the time of conception.6 Whether or not a pregnancy was classified as “wanted” was determined from a series of questions about the respondent’s use or nonuse of contraception at the time of conception, and about her attitude toward her pregnancy at the time she became pregnant.

“Wanted pregnancies” are classified as such because they occurred at a time when a respondent reported not using contraception or discontinuing its use because she wanted to become pregnant. Pregnancies that were wanted, but occurred sooner than the respondent would have preferred, were classified as “mistimed.” Those that occurred later than the respondent would have preferred are not considered mistimed, since in most cases the delay was not the result of planning or choice. A pregnancy was classified as “unwanted” if the respondent reported that she had not wanted to have another) child at the time of conception or at any point in the future. Unintended pregnancies are those that were either unwanted or mistimed (for further clarification, see Appendix Table A).

Given the short time frame between the two surveys, the total number of births that occurred is small. The total number analyzed in this study is 550. Our descriptive analysis presents data separately for intended, mistimed, and unwanted births. In the logistic regression, however, the data are collapsed into two categories, intended and unintended.7

Age is classified by the age of the respondent as of March 15, 1988. Respondents were classified as black, white, or other, depending on self-reports in the 1988 interview. Parity is classified by the number of live births the woman reported at the time of the 1988 interview.

Women are classified by marital status in 1988 as married, widowed, divorced, separated, or never-married. Those who reported that they were not married but living with their sexual partner are classified as cohabiting. We also include a measure of change in formal marital status between the two interviews. This measure allows us to disaggregate changes occurring after a conception leading to a birth and those occurring before a conception. The variable consists of three categories: (1) no formal change in marital status occurred or the change occurred after the conception leading to the birth in question, (2) the respondent got married or resumed living with her husband before she conceived, or (3) the respondent’s marriage dissolved (she was either widowed, divorced, or separated) before she conceived.

Income in 1988 was measured in accordance with the census definition of poverty status and was calculated by dividing the total family income by the weighted average threshold income of nonfarm residents in households headed by persons under age 65 (U.S. Bureau of the Census, 1987). The income variable is coded as follows: (1) income below the poverty level, (2) income between 100 and 199 percent of poverty, and (3) income over 200 percent of poverty. Education was measured in years of school completed as of the 1988 survey, and is coded as (1) less than a high school education, (2) high school graduate, or (3) one or more years beyond high school.

The working status variable was based on items detailing the respondent’s employment in the week before each survey. Employment status in 1988 was classified as (1) working full-time (35 hours or more), (2) working part-time (1-34 hours), or (3) not working (unemployed, laid off, looking for work, keeping house, in school, or other). Those on temporary leave, for whom active employment status was ambiguous, were omitted.

Certainty of intentions to have no (more) children was derived from a question asked after women reported their future childbearing preferences.8 Unfortunately, the certainty question focused only on the number of additional children the respondent intended to have, and no information was collected on certainty regarding timing. As a result, we have a measure of certainty of intentions only for those who intended to have no (more) children, and we lack an analogous measure of certainty for those who wanted only to postpone, or for those who were intending to conceive in the near term. For postponers, this would have required a survey item ascertaining how certain they were that they would delay having the next child for the specified amount of time. The measure for our subsample of permanent avoiders is coded as (1) very sure they want no more children and (2) not very sure. Time since conception is measured in three categories: 9-19 months, 20-29 months, and 30-39 months.

The respondent’s contraceptive status is also defined as of the 1988 interview and is coded as follows. “Effective methods” are those with failure rates under 5 percent, even with “typical (i.e., imperfect) use.” They include male or female contraceptive sterilization, the pill, and the IUD. “Less effective methods” include the 12 methods that have greater than 10 percent failure rates given “typical use” (Hatcher et al., 1990).9 Women not using contraception and not at risk of an unintended birth are those who were either postpartum at the time of the interview, seeking pregnancy, sterile for other than contraceptive reasons, had not had intercourse since first menstruation, had had intercourse only once, or had not had intercourse in the last three months. The category of women not using contraception and at risk of an unintended birth includes those not reporting any contraceptive use in 1988, and those who did not fall into any of the above “not at risk” categories. These four categories of contraceptive status correspond closely to the ones used by Forrest and Singh (1990) in their analysis of contraceptive use and exposure to the risk of unintended pregnancy.

RESULTS

Table 1 shows weighted percentages of wantedness reports separately for pregnancy seekers, temporary postponers, and permanent avoiders. (Recall that the following situations would be considered consistent: a birth that was reported as intended by a pregnancy seeker; a birth that was reported as mistimed by a temporary postponer; and a birth that was reported as unwanted by a permanent avoider. All other reports can be considered inconsistent with earlier stated intentions.)

Within each subsample, the distribution of births according to wantedness reports did reflect Time 1 stated intentions. The highest percentages of intended births were reported by Time 1 pregnancy seekers, the highest percentages of mistimed births were reported by Time I temporary postponers, and the highest percentages of unwanted births were reported by Time 1 permanent avoiders. The level of consistency varied considerably across intention subgroups, however. The most consistent reporters were the pregnancy seekers, those who had stated at Time 1 that they had intended to become pregnant in the near future. About three-quarters of pregnancy seekers (76 percent) reported the birth in the interval as intended. About half of temporary postponers were consistent (51.3 percent), whereas the permanent avoiders were the least consistent (32.4 percent).

Clearly, a fairly high percentage of responses were inconsistent. Roughly onequarter of Time 1 pregnancy seekers reported their births in the interval as unintended. Nearly half of all temporary postponers reported their births as something other than mistimed; roughly 40 percent reported their births as intended (more favorable), while fewer than 10 percent reported their births as unwanted, a more negative view. And nearly 70 percent of permanent avoiders reported their births as either intended or mistimed.

Levels of consistency also varied by subgroup within intention statuses. For example, older pregnancy seekers were most apt to report consistently between Times 1 and 2 (indicating that their birth had been intended), while close to half of all teens who said at Time 1 that they hoped to become pregnant in the near term, and who then experienced a birth in the interval, reported that birth as mistimed. At the same time, teen postponers were most likely to report a birth in the interval as mistimed, whereas women in the older age groups were more apt to shift to a more favorable response at Time 2. In addition, women with at least some college who were seeking pregnancy in 1988 were most apt to report intended births, but those with the lowest education levels who were avoiding pregnancy in the near term were more apt to report mistimed births.

In general, it is worth noting that a number of these subgroups follow the typical pattern identified in most research on unintended childbearing. Age is one example; younger women are more apt to report mistimed births, While older women are generally more apt to report unwanted births. Also, across intention statuses, women with the lowest household incomes were more likely than those with higher incomes to report their births as unwanted, while women with at least some college were particularly likely to report their births as intended.

Table 2 shows results of logistic regressions for subsample 1, pregnancy seekers. It presents the net effects of the independent variables on the log odds that their births were reported as unintended (mistimed or unwanted) during their 1990 interview. The first model contains the characteristics of the women at the first interview (1988), and the second model mirrors the first but substitutes “change in marital status” for 1988 marital status.

Net of other factors, teens who were pregnancy seekers were more likely than women in their twenties to report their births as unintended, or, in other words, more likely to report inconsistently relative to their initial intentions; older women were somewhat less likely to do so. Neither finding is surprising, as teens’ attitudes toward family formation are typically thought to be in a greater state of flux than is true of older women. One might hypothesize that this apparent shift in attitude could reflect expectations that were unmet (after the 1988 interview) about an existing or potential relationship and its likely trajectory, or a birth that had been thought desirable on its own merits at the time of the initial interview and then proved to be more troublesome than anticipated once it actually occurred. Either explanation seems plausible given the young age of the mothers. To the extent possible we control for change in relationship status (formal marital status) in Model 2, but the addition of the control actually strengthens the effect of the age variable.

Before controlling for change in marital status, zero parity women (1988 status) were less likely than women who had already had one birth at the time of the initial interview to report a birth in the interval as unintended, whereas higher parity women were somewhat more likely to do so. In addition, compared to married women, women of all other marital statuses (as of 1988) were considerably more likely to report a birth in the interval as unintended. Substitution of the marital status change variable causes parity to lose significance. The new marital status variable is not statistically significant itself.

Compared to women who were using effective methods of family planning at the time of the initial interview, those who had not been using a method and who had not been at risk of pregnancy at the time were less likely to report a birth in the interval as unintended. This is true with or without controls for change in marital status. The coefficients for different racial groups are impacted by the choice of marital status variable, however. Once change in marital status was controlled, black respondents who had intended as of 1988 to become pregnant in the near term were somewhat more likely than white women to report their births as unintended. With only the 1988 marital status controlled, the coefficient for African American women is not statistically significant, yet those classified as “other” by the survey were somewhat more likely than white women to report their birth as unintended.

Finally, duration since conception is marginally significant, but not in the direction suggested by the literature. Those for whom the shortest time had elapsed since their birth were less likely than those for whom somewhat more time had passed to report their birth as unintended.

Table 3 shows results of logistic regressions for subsample 2, temporary postponers. Again, it presents the net effects of the independent variables on the log odds that their births were reported as unintended during the 1990 interview. Both models indicate that, compared to 20- to 29-year-olds, temporarily postponing teens were more likely to report their births as unintended (as was true of pregnancy-seeking teens). Model 1 shows that those in the lowest income group, along with those who did not know what their household income was, also were slightly more apt than those in the reference group to report their births as unintended. Both coefficients become insignificant in the second model with the inclusion of the variable measuring change in marital status.

The significant effect of nonuse of contraception among women who were not at immediate risk of pregnancy at Time I also is mediated by substitution of the change in marital status variable for that measuring marital status at the time of the initial interview. Similarly, the fairly strong positive effect of medium duration since conception becomes only marginally significant in the second model. In each model, however, those with the most recent births again were less likely than those for whom more time had elapsed to report the birth as unintended.

While 1988 marital status is not shown to be significant in Model 1, in Model 2 the effect of marital status change significantly affects the log odds of reporting a birth as unintended. As anticipated, those who got married or began cohabiting during the interval were less likely than those whose marital status did not change to report a subsequent birth as unintended.

Table 4 shows results of logistic regressions for subsample 3, permanent avoiders. Again, we present the net effects of the independent variables on the log odds that respondents’ births were reported as unintended during the 1990 interview, a consistent response for this subsample. In this case both models indicate that older women were significantly less likely than women in their twenties to report a birth in the interval as unintended. Women who were never married as of Time I were less likely than married women to report a birth as unintended, black women were more likely than white women to do so, and zero parity women were marginally more likely than women who had already had a birth as of the initial interview to do so. All three of those effects disappear with the inclusion of the change in marital status variable, but the effect of older age remains strongly significant. In addition, duration since conception remains significant and positive in both models; again, those who had experienced a birth most recently remained least likely to report that birth as unintended.

CONCLUSIONS

To engage the ongoing discussion about the usefulness of retrospective attitudinal data, particularly as it pertains to fertility research, this paper is intended to provide one additional piece to the puzzle. Earlier research hypothesized that recall error and rationalization would increase with time, and that attitudes assessed longer after an unintended birth could result in underreporting of unintended pregnancy. Our results suggest the opposite. Women who had had a very recent birth were particularly unlikely, in our study, to report that birth as unintended. Future research should examine this question with a longer interval to assess whether there may be a curvilinear effect of duration since birth on intention status reports.

One previous analysis also compared Time 1 attitudes about pregnancy post– ponement or permanent avoidance with fertility behavior that occurred between Time 1 and Time 2 interviews, and questioned whether certain subgroups (white women, women in their twenties, and more highly educated women in that study) might be more likely than older women, African American women, and less educated women to underreport unwanted births, whereas married women might be relatively more likely than nevermarried women to underreport births as mistimed (Williams et al., 1999). Alternatively, they suggested that these groups might be particularly likely to experience situational change that made childbearing more desirable in the interval than it had been at the time of the first interview.

This research provides one further check on the question of attitude consistency and the nature of pregnancy wantedness reports. Our analysis reveals a nontrivial amount of inconsistency between women’s stated birth intentions and their later reports about the wantedness of subsequent births. Births that would be considered unintended if determined solely on the basis of the woman’s Time 1 intention report were frequently reported as intended by women at the time of their reinterview-41 percent of births in the case of those trying to postpone, and 50 percent of births in the case of those trying to stop childbearing. On the other hand, births that would be considered intended solely on the basis of women’s Time 1 intentions were frequently reported as unintended-24 percent of births in the case of those who intended a birth.

Consistent with some of the speculation from the previous study, our results indicate that black women who wanted no (more) children at Time 1 were somewhat more likely than white women to report their births as unintended. That the effect of race disappeared among permanent avoiders (and became significant among pregnancy seekers) once change in marital status was controlled suggests that changing situational factors are probably important in explaining differences in consistency of attitudinal reporting by race.

Some of our findings run counter to questions posited in the previous analysis, however. First, we found no effect of education on consistency of reporting, net of other variables. Second, never-married temporary postponers did not appear less likely than married postponers to report their births as unintended (although nevermarried permanent avoiders did). In addition, women in their thirties who had claimed they wanted to avoid all future childbearing were particularly unlikely to report their births as unintended net of other factors, including changes in marital status.

Unfortunately, changes in marital status are the only changing circumstances we were able to analyze. Greater precision in the data about the timing of events, such as changes in work experience and the like, would enable us to look in more detail at shifting circumstances that may affect the context in which fertility decisions are made, and thus the apparent “consistency” of attitudes toward the childbearing in general and the timing thereof. Although the samples analyzed in this study are quite small and our ability to measure situational change is limited, this research does update information about the linkages between intentions to avoid childbearing, at least in the near term, subsequent fertility behavior, and eventual attitudinal reports about what, based on the Time I report, would have been defined as an unintended event.

*Support for this research was provided by grant I ROI HD33241-01 from the National Institutes of Health. Work on this project was initiated by Lindy Williams and Kathy London. Dr. London’s untimely death was a great loss to the demographic community in general and to this project in particular. This paper is dedicated to her memory. An earlier version of this paper was presented at the annual meetings of the Population Association of America in New York in March 1999. The authors would like to thank Jim Weed and Jennifer Madans for their useful comments on a previous draft.

1Pregnancy wantedness (or intendedness) refers to a woman’s attitude toward the pregnancy at the time it was conceived. Standard definitions usually distinguish between unwanted pregnancies (those that occurred when a woman wanted no more children), mistimed pregnancies (those that were wanted at some time but occurred too soon), and intended pregnancies (those that were wanted at about the time they occurred). These terms are discussed in detail in the methodology section of the paper and are shown in Table A in the Appendix.

2This is the only subgroup for whom certainty data are available. This issue is discussed in greater detail in the Methods section of the paper.

3The response rate among reinterview respondents was 68.6 percent (7,809 cases were sampled and 5,359 interviews were completed), while that for a teen supplement was 53.4 percent (614 cases were sampled and 327 interviews were completed). Despite efforts to interview women from nontelephone households through letters that encouraged respondents to call a 1-800 number and offered an incentive of $15.00 to complete the interview, and offers of $5.00 incentives to complete a mail-in self-administered questionnaire (SAQ), the combination of unpublished numbers and nontelephone households accounted for the highest percentage (11 percent) of nonresponse. Women who were not located accounted for the next highest percentage (10 percent). The latter was particularly problematic for the teen supplement, however, and that portion of the NSFG sample is not part of the subsample we have selected for this analysis. In addition, because the National Health Interview Survey (a national survey with a response rate of 95 percent) provided the sampling frame for the 1988 NSFG, substantial information is available concerning the characteristics of nonrespondents (both to Cycle IV and to the telephone reinterview). As a result, appropriate weights were developed and applied to both the Cycle IV and reinterview data.

4At the time of the intial interview 1,057 women reported seeking pregnancy, 1,316 hoped to delay pregnancy, and 812 reported wanting no (more) children. Roughly 10 percent of those hoping to postpone a birth, and about 8 percent of those wishing to avoid all future births, reported a birth in the interval. In addition, approximately a third of those who wanted to get pregnant had a birth in the interval.

5We focus on births, rather than on all pregnancies, because of the widespread underreporting of abortion in the NSFG and other national survey data.

6lt should be noted that births that were unwanted at conception do not necessarily result in unwanted children. Mothers who report a pregnancy as unwanted at the time of conception nonetheless may cherish the child that is born as a result of that pregnancy.

7As a result, when we compare results from this analysis with those of previous research, our test is less precise, as our collapsed data combining mistimed and unwanted births do not match exactly those in which the two are disaggregated.

8The question read: “Of course, sometimes things do not work out exactly as we intend them to, or something makes us change our minds. In your case, how sure are you (and your husband/partner) that you will have (SPECIFIED NUMBER [we are interested only in those who reported wanting 01) more bab(y/ies)? Would you say you are very sure or not very sure?”

9These categories reflect methods on the market when this research was conducted; there were no methods with effectiveness between 5 and 10 percent.

REFERENCES

ANDREWS, K. H., and D. B. KANDEL. 1979. Attitudes and behavior: a specification of the contingent consistency hypothesis, Am. Soc. Rev. 44:298-310.

BACHRACH, C. A., and S. NEWCOMER. 1999. Intended pregnancies and unintended pregnancies: distinct categories or opposite ends of a continuum? Fam. Plan. Persp. 31(5):251-252.

BERSCHEID, E. 1985. Interpersonal attraction. In G. Lindzey and E. Aronson (eds.), The Handbook of Social Psychology, pp. 413-484. New York: Random House.

BUCKLEY, M. R., J. A. COTE, and S. M. COMSTOCK. 1990. Measurement errors in the behavioral sciences: the case of personality/attitude research. Educational and Psychological Measurement 50(3):447-474.

CHAFFEE, S. H., and C. ROSER. 1986. Involvement and the consistency of knowledge, attitudes, and behaviors. Communication Research 13(3):373-399.

FISHBEIN, M., and 1. AjZEN. 1975. Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research. Reading, MA: AddisonWesley.

FORREST, J. D., and S. SINGH. 1990. The sexual and reproductive behavior of American women, 1982-1988. Fam. Plan. Persp. 22(5):206-214.

HATCHER, R., F. STEWART, J. TRUSSELL, D. KOWAL, F. GUEST, G. K. STEWART, and W. CATES. 1990. Contraceptive Technology: 1990 1992. New York: Irvington Publishers.

HAUSER, P. M. 1967. Family planning and population programs: a book review article. Demography 4:404.

HERMALIN A. L, R. FREEDMAN, T.-H. SUN, and M.C. CHANG. 1979. Do intentions predict fertility? The experience in Taiwan, 1967-74. Stud. Fam. Plan. 10(3):75-95.

JOYCE, T., R. KAESTNER, and S. KORENMAN. 2000. On the validity of retrospective assessments of pregnancy intention. Unpublished manuscript.

LINDZEY, G., and E. ARONSON. 1985. The Handbook of Social Psychology. New York: Random House.

MILLER, W. B., and D. J. PASTA. 1995. Behavioral intentions: which ones predict fertility behavior in married couples. Journal of Applied Social Psychology 25(6):530-555.

PICCININO, L., and L. S. PETERSON. 1999. Ambivalent attitudes and unintended pregnancy. Advances in Population 3:227-249.

REED, F. W., and W. H. McBROOM. 1995. The impact of marriage on fertility intentions and related values. International Journal of Sociology of the Family 25(l):91-98.

RYDER, N. B. 1973. A critique of the National Fertility Study. Demography 10:495-506.

RYDER, N. B., and CHARLES F. WESTOFF. 1969. Fertility planning status: United States, 1965. Demography 6(4):435-444.

1972. Wanted and unwanted fertility in the United States: 1965 and 1970. In Charles F Westoff and Robert Parke, Jr. (eds.), Demo

graphic and Social Aspects of Population Growth, pp. 467-488. Vol. I of Commission Reports. U.S. Commission on Population Growth and the American Future. Washington, D.C.: U.S. Government Printing Office.

SCHOEN, R., Y. J. KIM, C. A. NATHANSON, J. FIELDS, and N. M. ASTONE. 1997. Why do Americans want children? Popul. Dev. Rev. 23(2):333-358.

THOMSON, E., and Y. BRANDRETH. 1995. Measuring fertility demand. Demography 32(1):81-96.

WESTOFF, C. F. 1990. Reproductive intentions and fertility rates. Fam. Plan. Persp. 16(3):84-96.

WESTOFF, C. F., and N. B. RYDER. 1977. The predictive validity of reproductive intentions. Demography 14(4):431-453.

WILLIAMS, L. B., J. Z. ABMA, and L. PICCININO. 1999. The correspondence between intention to avoid childbearing and subsequent fertility: a prospective analysis. Fam. Plan. Persp. 31(5):220-227.

Lindy Williams and Joyce Abma

Department of Rural Sociology, Warren Hall, Cornell University, Ithaca, NY 14850, e-mail: lbw2@cornell.edu; National Survey of Family Growth, National Center for Health Statistics, 6525 Belcrest Road, Hyattsville, MD 20782, e-mail: jza2@cdc.gov

Copyright Society for the Study of Social Biology Fall 2000

Provided by ProQuest Information and Learning Company. All rights Reserved