Multi-Ethnic Comparison of Risk and Protective Factors ... - Springer Link

3 downloads 310 Views 125KB Size Report
San Diego State University, School of Social Work, San Diego, California. Terry Peak ... to predictive differences in school-age pregnancy among ethnic and.
Child and Adolescent Social Work Journal Volume 17, Number 2, April 2000

Multi-Ethnic Comparison of Risk and Protective Factors for Adolescent Pregnancy

E. Helen Berry, Ph.D., Audrey M. Shillington, Ph.D., Terry Peak, M.S.W., Ph.D., and Melinda M. Hohman ABSTRACT: Data from a longitudinal cohort study, the National Longitudinal Survey of Youth, are used to examine the differences in risk and protective factors for adolescent pregnancy among four ethnic groups—non-Hispanic Whites, non-Hispanic Blacks, Hispanics, and American Indians. The objective is the identification of differential predictors for adolescent pregnancy among each ethnic group included in the analyses, as well as better comprehension of the differences among women who experienced a teen pregnancy compared to those who did not. The model for this sample of 5,053 women indicates that higher self-esteem and a higher level of maternal education are protective factors; living in poverty as a young teen, substance use, and adolescent marriage are factors associated with an increased risk for teen pregnancy. Further, the results indicate that unique sets of predictors exist for each ethnic group. Implications of these findings are discussed.

Introduction Adolescent pregnancy has long been a topic of concern. Most research has focused on socioeconomic status (SES) and behavioral variables that correlate with school-age pregnancy. Less attention has been paid E. Helen Berry is affiliated with Utah State University, Department of Sociology, Social Work, and Anthropology, Logan, Utah. Audrey M. Shellington is affiliated with San Diego State University, School of Social Work, San Diego, California. Terry Peak is affiliated with Utah State University, Department of Sociology, Social Work, and Anthropology, Logan, Utah. Melinda M. Hohman is affiliated with San Diego State University School of Social Work, San Diego, California. Address correspondence to E. Helen Berry, Utah State University, Department of Sociology, Social Work, and Anthropology, Logan, UT 84322-0730. 79

 2000 Human Sciences Press, Inc.

80

CHILD AND ADOLESCENT SOCIAL WORK

to predictive differences in school-age pregnancy among ethnic and racial groupings. Differences by socioeconomic status and other background variables are examined relatively often, but comparisons among Whites, Hispanics, African Americans and Native Americans are infrequent; differences for American Indians are even more rare. Given the variability in rates of teen pregnancy between these groups and the continued overall decline in such pregnancies, these differences demand attention. As one example, recent reports (Moore, 1994; Bachu, 1997) show that Hispanic teen fertility increased substantially and now shows a small decline, while African American rates have more or less stabilized. White rates of teen births, after increases in the late 1980s, reached a plateau and began to decline. Conversely, changes in American Indian rates are relatively unknown and unstudied (Snipp, 1992) although the proportion of new mothers under 17 is higher among American Indian/Alaska Natives (8%) than among Hispanics (7.1%), Whites (3.9%) or Asians (2%) (Brown, 1995) and, historically, some 45% of American Indians have a first birth while still in their teens (Snipp, 1996).

Correlates of School-Age Pregnancy When addressing unique differences between racial and ethnic groups, it is important to consider issues correlated with teen pregnancy as a whole—education, marital experience, socioeconomic status, self-esteem, and substance use. Specifically, substantial evidence indicates that school-age mothers also complete less schooling (Hayes, 1987; Hofferth and Moore, 1979; Furstenberg et al., 1987a; Moore and Burt, 1982), and those with the earliest births complete the fewest years of education (Upchurch and McCarthy, 1990) even with socioeconomic status and academic skill controlled (Mott and Marsiglio, 1985). The rate of high school graduation for these mothers, however, has apparently increased since the 1950s (Upchurch and McCarthy, 1990), although this could reflect societal changes that serve to encourage all teens, not just teen mothers, to stay in school (Trussell and Menken, 1978). Teen mothers are more likely to experience divorce (Moore and Waite, 1981) which, in combination with higher rates of non-marital childbearing and lower educational levels, means that such women are more likely to be single heads of families and more likely to be poor. There is, of course, an obvious chicken and egg problem: teenage mothers

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

81

are more often from poorer circumstances and have parents with less education than those who do not become teen mothers. As a result, there is some argument as to whether poverty later in life is a result of the young women’s less advantaged backgrounds or is a result of their experience with early childbearing (Geronimus and Korenman, 1993; Geronimus, 1991; Luker, 1991; Hoffman et al., 1993 a, b; Furstenberg et al., 1989; Furstenberg et al., 1987). Still, with socioeconomic status controlled, those with earlier births seem more likely to be disadvantaged later in life (Ahn, 1994; Nord et al., 1992). Indeed, the single best “predictor” of adolescent childbearing has been lower socioeconomic status (Miller and Moore, 1990; Maxwell and Mott, 1987). Parents with less education themselves are more likely to have children who have teen births and teen pregnancies and achieve less education (Miller and Moore, 1990). Structurally, lower socioeconomic status is also associated with racial and ethnic differences in teen childbearing. African Americans are more likely to bear children at younger ages that Whites; in 1986, births per thousand African Americans ages 15–19 were 104, by 1991 the rate had reached 118 per thousand but declined to 109 in 1995. In contrast, non-Hispanic Whites aged 15–19 had birthrates of 36 per thousand in 1986; in 1991 and 1995 the rates were 43 per thousand. Hispanic rates per thousand females aged 15–19 were 80 in 1986, increased to 107 in 1991 (Moore, 1994) and then decreased slightly to 100 per thousand in 1995 (Bachu, 1997). Rates and trends for American Indians are not as widely reported due to relatively small sample sizes, but in both 1986 and 1991 the rate of live births per thousand was 79. During the same period, however, the percentage of live births to unmarried American Indian women of all ages increased from 46.8 to 55.3 (Bachu, 1997). Since 1990, rural areas have had proportionately higher percentages of teen births (15%) than urban areas (12%) (Lee, 1997). These proportional differences are assumed to be a function of higher teen pregnancy rates as well, although no statistics on pregnancy differences were found. Lichter, Cornwell, and Eggebeen (1993) note that school drop out rates associated with adolescent sexual activity may be higher in rural places although they emphasize that this is more a function of rural poverty than of ruralness per se. Psycho-social factors associated with greater risks of pregnancy among school-age teens included aspects of self-esteem (Oates, 1997), attitude towards peers, economic factors in the community or neighborhood, and tobacco, drug and alcohol use. Oates, specifically, addresses

82

CHILD AND ADOLESCENT SOCIAL WORK

self-esteem as a predictor of early childbearing, itself sometimes considered as a means of increasing self-esteem among less-advantaged youths (see also Corcoran, Franklin, and Bell, 1997). Early tobacco, drug and alcohol use are known to be correlated with school-age childbearing, as well as with lower self-esteem. Other factors known to be correlated with school-age childbearing, though not necessarily predictive of it, include being born into a large family or into a single parent family; and having a mother with relatively less education, and originating in a neighborhood with greater odds of teen pregnancy (Chandy, Harris, and Blum, 1994; Miller and Moore, 1990; Maxwell and Mott, 1987; Michael and Tuma, 1985; Zabin, 1984). Again, little of this research makes comparisons of factors of greater risk between racial or ethnic groups. This paper examines socio-demographic variables to test for differences among four ethnic groups (non-Hispanic Whites, non-Hispanic Blacks, Hispanics, and American Indians). As noted, comparative differences among these groups in terms of ability to predict adolescent pregnancy are not yet well understood. The problem of comparisons between groups is addressed, with special emphasis on understanding predictive differences between the groups. Corollary objectives include: 1) identification of any unique risk factors for adolescent pregnancy among each ethnic group included in the analyses, and 2) better comprehension of the differences among women who experienced a teen pregnancy compared to those who did not.

Methods In order to investigate these questions, we used a nationally representative sample of women of a specific age cohort—the National Longitudinal Survey of Youth (NLSY). The NLSY is a household survey, funded by the U.S. Department of Labor since 1979, that collects information on all individuals in the household between the ages of 14–21 with annual re-interviews. Among national samples, this panel boasts a 90.5% retention rate found at twelve year follow-up (Baker et al., 1993). Investigators used a multistage stratified random sampling technique with a random sample of counties, further broken into block groups, for a total of 75,000 screened dwellings. A total of 12,645 youth, 6,283 of who were female aged 14–21, were interviewed in 1979. The sampling is nationally representative of the age cohort and includes a

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

83

“supplementary” sample that is over-representative of Blacks, Hispanics and economically disadvantaged Whites. The NLSY has a major advantage over census data because it is a sample in which individuals self-identified their ethnicity during the 1970s. As is often noted, between 1970 and 1990 there have been changes in the numbers of persons who self-identify as being of varying ethnic backgrounds. This change is most dramatic for American Indians. As Eschbach, Supple, and Snipp (1998) indicate, between 1970 and 1980, 62% of the American Indian population growth was in excess of natural increase. Between 1980 and 1990, some 35% of the American Indian population growth was due to an increase in self-identification. Further, those who are newly self-identified as American Indian are generally higher in economic and educational status (Eschbach, Supple, and Snipp, 1998). Thus the NLSY is a sample for which the salient ethnicity—Black, Hispanic, White, American Indian—was identified before the changes noted above. The present study utilized a female subsample that, because of small sample size, excluded any respondent who reported her ethnicity as Asian American or “other.” The final sample size was 5,053 women, aged 23–31 in 1988, who completed their teen childbearing years. They were 55% non-Hispanic White, 26% non-Hispanic Black, 14% Hispanic, and 5% American Indian. (Hereafter we refer to the four ethnic groups as White, Black, Hispanic and American Indian.) Teen pregnancy was defined as having reported a pregnancy at 19 years of age or younger (N = 1541). Women who reported a teen pregnancy were compared with all other women (N = 3512). Other independent variables explored include: age at teen pregnancy, educational attainment of the respondent, and teen marriage, each included as control variables; self-esteem as measured in 1980 (early adolescence); associated risky behaviors including teen smoking, drinking, and marijuana use; poverty; and rural residence. Socioeconomic status was measured by whether or not the respondent lived below the poverty line at age 14 and the educational attainment of her parents. In the preliminary analyses we also examined actual dollar income but it was not meaningfully associated with teen pregnancy when controlling for poverty status. Self-esteem is included following Oates’ (1997) article that utilized the ten-item Rosenberg (1986) scale to predict number of children born. Analysis of variance and comparisons of means of unweighted data were used to examine bivariate associations. Bonferroni means tests were used to compare ethnic groups to one another. Logistic regression

84

CHILD AND ADOLESCENT SOCIAL WORK

was utilized to investigate associations between teen pregnancy and independent variables. Logistic regression is designed for models in which the dependent variable is categorical and the independent variables are categorical and/or continuous. The results of the analysis are coefficients that indicate whether a given level of an independent variable increases or decreases the odds of the dependent variable. The Cox-Snell and Nagelkerke R2s are, each, proportionate reduction in error statistics and serve as “pseudo R2”s or multiple correlation coefficients. As with the R2 in multiple linear regression, these are best used to provide comparisons of the goodness of fit of different models relative to total variability within the model. That is, a higher number is associated with less error, or “a better fit,” between the model and the data. For more detailed discussion of logistic regression refer to Hosmer and Lemeshow (1989). The categorical independent variables in each regression model were poverty status in 1978 (poverty = 1); urban or rural residence at age 14 (rural = 1); adolescent marriage (teen marriage = 1); the risky behaviors variables of cigarette smoking, alcohol or marijuana use (teen use of each = 1). Ethnicity variables were dummy coded for being Hispanic, American Indian, White, or Black, with White being the referent group except as noted. Three independent variables—mothers’ educational attainment, fathers’ educational attainment, and self-esteem in 1980— were entered as continuous variables in the regression models.

Results Descriptive Data and Bivariate Analysis Presented here are summary statistics, stratified by teen pregnancy status and ethnic group, and analysis of variance with Bonferroni means tests for differences between ethnic groups within pregnancy categories. An examination of Table 1 reveals that nearly one-third or 1,541 of the 5,053 women in this sample experienced an adolescent pregnancy (≤19 years). With respect to poverty, it is clear that women with a teen pregnancy were overwhelmingly more likely to have been in poverty in 1978 compared to those without a teen pregnancy. The percentage was highest among American Indian women—adolescent mothers reported 47.1% in poverty compared to 25.8% of women without a teen pregnancy. Thus, American Indian women with a teen pregnancy were over two

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

85

and one-half times more likely to be in poverty as young teens compared to those who did not have a teen pregnancy. The same pattern was found for Black and White women, with odds ratios of 2.07 and 1.93 respectively. With regard to pregnancy itself, the overall mean age of first pregnancy was 17.4 for those with a teen pregnancy. Twenty one percent of the total sample married before age 20. Over 50% of women who experienced a teen pregnancy also reported a teen marriage compared to less that 15% of those without a teen birth. For Blacks this is most striking. Only 7.5% of those without a teen pregnancy also had a teen marriage compared to 20.5% who did experience a teen pregnancy. Place of residence at age 14 varied somewhat across ethnic groups. The White and American Indian groups had the highest percentage (26.7 and 22.8, respectively) living in non-Metropolitan areas, while Hispanics (88.2%) and Blacks (82.8%) were more likely to have been living in Metropolitan areas. The socioeconomic variables included in Table 1 are the years of education for the respondent’s mother and father. The overall mean education for mothers and fathers differed substantially between groups; the non-pregnant group had parents with more than 11 years of formal education on average, while the pregnant group had parents with an average of less than 10 years of formal education. Bonferroni means tests, reported as superscripts in the table, indicate that substantial differences appear between ethnic groups with Hispanics having significantly less education for both mothers and fathers compared to the other three groups, with pregnancy status controlled. White women had significantly higher parental educational attainment compared to any other ethnic group. Black and American Indian women reported similar parental education attainments. Overall, parental education was lowest for Hispanic women in both groups. Also included for comparison is the respondent’s educational attainment at the end of her adolescent years. In other words, this is educational attainment of the respondent as measured in 1988 when the respondents are all over age 21. Those without a teen pregnancy attained approximately two more years of education than did those with a pregnancy. An individual’s years of formal education also showed substantial differences. The group who had a pregnancy completed approximately two fewer years of education by the end of adolescence than did those without a pregnancy. However, as with parents’ education, the ANOVA indicates that ethnicity made a difference in educational attainment.

86

CHILD AND ADOLESCENT SOCIAL WORK

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

87

88

CHILD AND ADOLESCENT SOCIAL WORK

The Bonferroni results show that, within the pregnancy categories, the ethnic groups differ from one another—Hispanics and American Indians had similar lower educational attainment, and Whites and Blacks had similar higher levels of educational attainment. An examination of substance use risk factors known to be correlated with teen pregnancy—cigarette, alcohol, and marijuana use—shows that Hispanic women had significantly younger ages of onset for both alcohol and cigarette use when compared to the other groups. This was true both for those with and without a teen pregnancy. Black women were significantly older for their age of onset of marijuana/hashish use compared to the other three groups. Again, this was true independent of pregnancy status. On self-esteem, although the means do not appear to be substantively different, there are statistical differences by ethnicity. Black and White women were similar in their self-esteem in 1980 but had significantly higher self-esteem when compared to Hispanic women. Logistic Analysis Predicting Teen Pregnancy Clearly there are differences in both the risk factors and actual teen pregnancy rates by ethnic group. The ANOVAs shown in Table 1 indicate that there are clear differences in years of completed education, self-esteem, parental education, and ages at onset of tobacco, alcohol, and marijuana use by ethnicity. The proportions of those who are poor, reside in rural areas, or are married, are also quite different from group to group, even within the category of those who experience teen pregnancy compared with those who do not. To discover whether these differences persist when the various risk and background factors are controlled, logistic regression predicting teen pregnancy is utilized. The data are first analyzed using the entire sample of women, with ethnicity included for Blacks, Hispanics, and American Indians. Whites are used as the reference group in this model since this is the group most studied. As noted earlier, the behavioral variables—teen marriage, rural/urban residence at age 14, and poverty status of family of origin—are dichotomized, as are the ethnicity variables. So that the effects of those who did not use alcohol, tobacco, or marijuana do not produce missing values, these variables are also dichotomized. Thus, if the respondent began using any of the three substances as a teen, she is coded as one on each and if she did not begin as a teen, the variables are coded as zero. In the model reporting effects for all women, shown in Table 2, only

Logistic Regression Predicting Teen Prgnancy for All Groups and by Ethnicity: Odds Ratios (95% confidence interval)

Variable

All women

Whites Blacks

Hispanics

Percent Rural Poverty @ 14

n/s 1.69 (1.41–2.00)

n/s 1.54 (1.19–2.00)

n/s 1.74 (1.29–2.36)

n/s n/s

n/s 2.47 (1.05–5.82)

Self-Esteem 1980

.96 (.94–.98) .93 (.89–.96) n/s n/s 1.66 (1.34–2.06)

n/s

.95 (.92–.95) .91 (.85–.98) n/s n/s 1.61 (1.14–2.27)

.93 (.88–.99) n/s

n/s

n/s n/s n/s

n/s n/s n/s

Mother’s education Father’s education Age began alcohol use Age began cigarette use

.91 (.94–.98) n/s n/s 1.66 (1.18–2.33)

American Indians

n/s

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

TABLE 2

89

90

CHILD AND ADOLESCENT SOCIAL WORK

some of the risk factors are statistically significant. Being poor at age 14 increased the odds of a pregnancy by half while higher self-esteem slightly decreased the odds of a pregnancy. Higher mother’s education also somewhat decreased the likelihood of a pregnancy though father’s education was not significant. On the other hand, teen onset of cigarette or marijuana use was associated with a half again increase in the likelihood of a teen pregnancy for cigarette use, and one-third increase for marijuana use. Alcohol use, however, was not predictive of teen pregnancy. Not surprisingly, given the correlation between marriage and childbearing, the experience of teen marriage is associated with eight times (odds = 8.03) the likelihood of a teen pregnancy. Although this is the strongest finding in this part of the analysis, other estimates are that some 26% of white and 73% of black married teen births are conceived prior to marriage (Jones et al., 1994). Thus, it is quite likely to be true with this sample as well—that a majority of the births associated with teen marriage are conceived prior to marriage and that the teen marriages are effects, not causes, of the teen pregnancies. Of greater interest here however, is that although membership in different ethnic groups is associated with different aspects of risk for teen pregnancy, when all women are examined, it is only being Black that increases the likelihood of teen pregnancy. Being Black increased the odds of a teen pregnancy by 3.29 times relative to being any other ethnic group. Identification of Unique Risk Factors We wanted to understand if there were distinct variables that posed risks for teen pregnancy for the different ethnic groups. Despite the finding that only one of these ethnic groups differs substantially from the rest in its likelihood of a teen pregnancy, it is worthwhile to note that each ethnic group examined here are more or less likely to be influenced by the various risk factors. First, Whites are rather less likely than the whole group to be influenced by poverty (1.54). But for Blacks (1.74), and even more so for American Indians (2.47), poverty does increase the odds of a teen pregnancy. For Hispanics, however, this variable is not statistically significant. Self-esteem in early adolescence does not statistically increase the odds of teen pregnancy for Whites and American Indians but higher self-esteem seems to decrease the odds of a teen pregnancy for Blacks (.95) and Hispanics (.93). On the other hand, higher mother’s education is influential in reducing the likelihood of pregnancy for Whites (.93)

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

91

and Blacks (.91) but has no statistical influence on likelihood of pregnancy for Hispanics or American Indians. Teen cigarette use increases the likelihood of a teen pregnancy for Whites and Blacks, although teen marijuana use increases this likelihood only for Whites. Interestingly, none of the three teen substance use variables were statistically associated with teen pregnancy for either Hispanic or American Indian women. Lastly, although teen marriage increases the likelihood for each group to have experienced a pregnancy, it is useful to note that the increase in likelihood varies dramatically from group to group. The association for Hispanics and American Indians was greatest, at 12; though for Whites the likelihood increased by just over eightfold (8.09); and for Blacks, by only 3.61. Father’s education and teen alcohol use have no influence on the total group nor specifically for any of the ethnic groups when other variables are controlled. In other words, while risk factors that have no influence on the whole group also have no influence on any particular group, the risk factors that do increase or decrease the likelihood of a teen pregnancy differ from group to group. The R2s indicate that the independent variables best reduce the likelihood of error in prediction of teen pregnancy for American Indians. Most of this influence is via teen marriage and poverty. However, these same independent variables do the poorest job of prediction of White and Hispanic teen pregnancies. The R2s for all women, for Whites, and for Hispanics, show that 20% or less of the variance in predicting teen pregnancies can be explained by the independent variables. The variables are relatively better predictors of teen pregnancy for Blacks (i.e., Nagelkerke R2 = .36) and American Indians (i.e., Nagelkerke R2 = .45), however, only poverty at age 14 and teen marriage influence both, and teen marriage is substantially less important for Blacks than for American Indians. Thus, the results of the logistic regression models indicate that the risk factors and prediction of an adolescent pregnancy vary among ethnic groups. Further, there appears to be no one, single protective factor that reduces the likelihood of a pregnancy. Teen marriage, though, does appear to be a fairly strong predictor of a pregnancy.

Discussion The goal for this article was to investigate adolescent pregnancy among various ethnic groups and then to contrast each group with the others. The NLSY data provide a unique opportunity to study four different

92

CHILD AND ADOLESCENT SOCIAL WORK

ethnic groups in the same cohort over a span of ten years. Data were collected on the women in early adolescence as well as after they passed age 20. Thus, we can use these data to analyze factors about the women both before and after their risk for adolescent childbearing. For the overall sample, and for each ethnic group separately, except Hispanics, poverty was significantly related to the experience of a teen pregnancy. Another variable strongly associated with teen pregnancy was adolescent marriage. Although, as seen in the descriptive analyses, among those with a teen pregnancy, the average age of first marriage occurred well after the average age of first pregnancy. Although there are other similarities across groups, there are differences in the strength of the associations across ethnic groups. Variables that were not significantly related to teen pregnancy, when controlling for all other associations, include father’s educational attainment, teen onset of alcohol use, and rural residence at age 14. For Whites and Blacks, mother’s education, poverty, teen onset of cigarette use, and teen marriage, statistically alter the odds of having experienced a teen pregnancy. Only for Hispanics and Blacks was self-esteem protective against teen pregnancy. For Hispanics and American Indian women, only two variables were good predictors of pregnancy, yet these models explain the greatest amount of variance of the stratified models. The findings are perhaps most interesting for American Indian women, though they are hardly a homogeneous group. Of particular importance for American Indians seems to be having lived in poverty as a young teen which was most strongly associated with the likelihood of teen pregnancy compared to any other ethnic group studied. Second, American Indian women were most likely to have experienced an adolescent marriage if they also became pregnant as an adolescent. Beyond these two variables, however, the factors that generally combine to predict teen pregnancy for the other groups show no unique statistical impact on American Indians. The independent variables do, however, combine to better predict American Indian teen pregnancies than the variables do for other groups (that is, the R2 are highest). Despite the advantages of data from a national probability sample, there are limitations to the present study that must be kept in mind when interpreting these results. First, this is a secondary data analysis, so there are limitations related to when the data were collected, variable definitions, and variables available for analysis. Another issue is that these data have been collected through personal interviews (with the exception of the 1987 survey that utilized telephone interviews). Whenever data are collected through self-report there may be hesitancy on

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

93

the part of the subjects to fully disclose information regarding sexual or illicit activity and related consequences. Thus, the possibility of under-reporting exists. Policy and Practice Implications Three findings of this study support Schinke’s (1998) contention that adolescent pregnancy and childbearing is a many-faceted problem with no single solution, and that prevention work must take into account racial/cultural as well as regional and class differences. Because of these differences, teens are at varying levels of risk for pregnancy. Prevention efforts are best tailored to the level of risk of the teen. When designing pregnancy prevention programs according to racial/ ethnic/cultural values and contexts, policy must address the problem at the macro, mezzo, and micro levels. Two findings of this study imply the need for intervention at the macro level: poverty and ethnicity with ethnicity assumed to be a surrogate for some differential cultural norms. Prevention programs must continually address poverty, racial discrimination, unfair housing practices, the need for better schools, poor access to transportation, and lack of decent wages for clients. This is perhaps a tall order, and may get lost when working with individual clients. As for cultural norms, prevention programs should not seek to change them. Instead, they should expand the perception of possible alternatives. For example, given the finding of the high rate of teen marriage for American Indians, the “problem” of adolescent pregnancy may be seen as no problem at all. Instead, the real issue may be marriage, not childbearing, prior to completing the adolescent’s schooling. At the mezzo level, given the finding that low level of mother’s education attainment is a risk factor, it may be advantageous to continue the attempts of educators to include parents, specifically mothers, in school-based programs, even in the early grades. If mothers are treated as partners in the educational process, they may be more likely to support ongoing schooling, see other alternatives for their daughters, or to communicate higher educational aspirations to their daughters. Additionally, ensuring that programs are available that will enable teen mothers to complete their schooling may avert some future teen pregnancies. At the micro level, higher self-esteem was an important protective factor for African American and Hispanic girls. Professionals need to continue to include a self-esteem curriculum in pregnancy prevention work, especially content that is specific to varying racial/ethnic/cultural

94

CHILD AND ADOLESCENT SOCIAL WORK

values and contexts. Schinke (1998) suggests including music, legends, storytelling, and characters that appeal to teens and that will instill racial/ethnic pride. While self-esteem was not a predictor variable for White teen pregnancy, early risk-taking behaviors of cigarette and marijuana were. Whether this is related to self-esteem, poverty, or poor educational support cannot be determined; however it does point to the types of intervention that are needed. For White teens the focus of prevention programs may need to be on avoiding risky behaviors, of which unprotected intercourse is but one. The most important implication of the analysis is that successful efforts to reduce adolescent pregnancy must take into account differences in risk and background factors for each group. Any effort to treat teens as a monolithic group will clearly fail since different factors that vary by ethnicity are associated with increased or decreased likelihood of a teen pregnancy. References Ahn, N. (1994). Teenage childbearing and high school completion: Accounting for individual heterogeneity. Family Planning Perspectives 26: 17–21. Bachu, A. (1997). Fertility of American women: June 1995 (Update). Current Population Reports. Table 4. Baker, P. C., Keck, C. K., Mott, F. L., and Quinlan, S. V. (1993). NLSY Child HandbookRevised Edition: A Guide to the 1986–1990 National Longitudinal Survey of Youth Child Data. Columbus, OH: Center for Human Resource Research, October. Brown, M. D. (1995). Strategies to reduce problem conceptions and STDs among young adults. The IHS Primary Care Provider 20(4): 57–68. Chandy, J. M., Harris, L., and Blum, R. W. (1994). Female adolescents of alcohol misusers: sexual behaviors. Journal of Youth and Adolescence 23: 695–709. Corcoran, J., Franklin, C., and Bell, H. (1997). Pregnancy prevention from the teen perspective. Child and Adolescent Social Work 14: 365–382. Cornell, S. and Kalt, J. eds. (1992). What can tribes do? Strategies and institutions in American Indian economic development. Los Angeles: American Indian Studies Center. In, Schaefer, R. (1998). Racial and Ethnic Groups. New York: Longman. Eschbach, K., Supple, K., and Snipp, C. M. (1998). Changes in Racial Identification and the Educational Attainment of American Indians, 1970–1990. Demography 35(1): 35–43. Furstenberg, F. F., Brooks-Gunn, J., and Chase-Lansdale, L. (1989). Teenaged pregnancy and childbearing. American Psychologist 44(2): 313–320. Furstenberg, F. F., Brooks-Gunn, J., and Morgan, S. P. (1987 a). Adolescent Mothers in Later Life: Human Development in Cultural and Historical Contexts. New York: Cambridge University Press. Furstenberg, F. F., Morgan, S. P., Moore, K. A., and Peterson, J. L. (1987b). Race differences in the timing of adolescent intercourse. American Sociological Review 52 (August): 511–518. Geronimus, A. T. (1991). Teenage childbearing and social and reproductive disadvantage:

E. H. BERRY, A. M. SHILLINGTON, T. PEAK, AND M. M. HOHMAN

95

The evolution of complex questions and the demise of simple answers. Family Relations 40(4): 463–471. Geronimus, A. T. and Korenman, S. (1993). The costs of teenage childbearing: evidence and interpretation. Demography 30(2): 281–290. Hayes, C. C., ed. (1987). Risking the Future Vol. 1. Washington, D.C.: National Academy Press. Hofferth, S. L. and Moore, K. A. (1979). Early child bearing and later economic wellbeing. American Sociological Review 44(October): 784–815. Hoffman, S., Foster, E. M., and Furstenberg, F. F. (1993a). Re-evaluating the costs of teenage childbearing. Demography 30(1): 1–13. Hoffman, S., Foster, E. M., and Furstenberg, F. F. (1993b). Re-evaluating the costs of teenage childbearing: Reply to Geronimus and Korenman. Demography 30(2): 291–296. Hosmer, D. W. and Lemeshow, S. (1989). Applied Logistic Regression. New York: Wiley. Jones, J. A., Kahn, J. R., Parnell, A., Rindfuss, R. R., and Swicegood, C. G. (1994). Nonmarital childbearing: Divergent legal and social concerns. Population and Development Review 11: 677–693. Lee, I. (1997). Adolescents. In Goreham, G. A. (ed.), Encyclopedia of Rural America. Santa Barbara, CA: ABC-CLIO. Lichter, D. T., Cornwell, G. T., and Eggebeen, D. J. (1993). Harvesting human capital: Family structure and education among rural youth. Rural Sociology 58: 53–75. Luker, K. (1991). Dubious conceptions: the controversy over teen pregnancy. The American Prospect. Spring: 75–83. Maxwell, N. L. and Mott, F. L. (1987). Trends in the determinants of early childbearing. Population and Environment 9(2): 59–73. Michael, R. T. and Tuma, N. B. (1985). Entry into marriage and parenthood by young men and women: The influence of family background. Demography 22(4): 515–544. Miller, B. C. and Moore, K. A. (1990). Adolescent sexual behavior, pregnancy, and parenting: Research through the 1980s. Journal of Marriage and the Family 52(4): 1025– 1044. Moore, K. (1994). USA Facts at a Glance. January. Washington, D.C.: Child Trends, Inc. Moore, K. A. and Burt, M. R. (1982). Private crisis, public cost: Policy perspectives on teenage childbearing. Washington, D.C.: Urban Institute Press. Moore, K. A. and Waite, L. J. (1981). Marital dissolution, early motherhood and early marriage. Social Forces 60 (September): 20–40. Mott, F. L. and Marsiglio, W. (1985). Early childbearing and completion of high school. Family Planning Perspectives 17: 234–237. Nord, C. W., Moore, K. A., Morrison, D. R., Brown, B., and Myers, D. E. (1992). Consequences of teen-age parenting. Journal of School Health 62(7): 310–318. Oates, G. L. (1997). Self-esteem enhancement through fertility? Socioeconomic prospects, gender and mutual influence. American Sociological Review 62 (December): 965–973. Paisano, E. L. (1991). The American Indian, Eskimo, and Aleut Population. Population Profiles www.census.gov/population/www/pop-profile/amerind.html. Passel, J. S. (1996). The growing American Indian population, 1960–1990: Beyond demography. Pp. 79–102 in Changing Numbers, Changing Needs: American Indian Demography and Public Health, ed. by G. Sandefur, R. Rindfuss, and B. Cohen. Washington, D.C.: National Academy Press. Rosenberg, M. (1986). Conceiving the Self. Reprint, Melbourne, FL: Krieger. SPSSX User’s Guide 3rd Edition (1988). Chicago: IL: SPSS Inc. Schaefer, R. (1998). Racial and Ethnic Groups. New York: Longman. Schinke, S. P. (1998). Preventing teen pregnancy: Translating research knowledge. Journal of Human Behavior in the Social Environment 1: 53–66. Snipp, C. M. (1992). Sociological perspectives on American Indians. Annual Review of Sociology 18: 351–371.

96

CHILD AND ADOLESCENT SOCIAL WORK

Snipp, C. M. (1996). A demographic comeback for American Indians? Population Today, 4–5. Tienda, M. and Jesen, L. (1988). Poverty and minorities: A quarter-century profile of color and socio-economic disadvantage. Pp. 23–61 in Divided Opportunities: Minorities, Poverty and Social Policy, ed. by G. Sandefur and M. Tienda. New York: Plenum Press. Trussell, J. and Menken, J. (1978). Early childbearing and subsequent fertility. Family Planning Perspectives 10: 209–218. Upchurch, D. M. and McCarthy, J. (1990). The timing of a first birth and high school completion. American Sociological Review 55(April): 224–234. Zabin, L. S. (1984). The association between smoking and sexual behavior among teens in U.S. contraceptive clinics. American Journal of Public Health 74(3): 261–263.