Click for next page ( 232


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 231
8 Pnonties for Data Collection and Research The pane] has reviewed a broad array of data sets, research studies, and program evaluations for their contribution to the understanding of adoles- cent sexuad and fertility behavior. In previous chapters we summanzed the trends in teenage sexual acidity, pregnancy, abortion, and childbearing as wed as knowledge about the antecedents and consequences of these behav- iors. We have also presented conclusions about the effectiveness of exists Interventions to prevent or delay these behaviors and tO ameliorate their negative consequences. We have found that, over the past several years, researchers have made significant advances in knowledge about teenage sexuality, prep y, and parenting, yet, as we have repeatedly indicated, many questions remain unanswered, and they suggest pnonties for fixture data collection and research. Essential tO frarn~ng an agenda for research is an underlying concept of the applications of increased knowledge. What do concerned policy makers, program administrators, advocates, parents, and adolescents need to know? How would such information make a difference for public and pnvate, collective and individual efforts tO develop solutions tO the prob- lems of early unintended pregnancy and childbearing? The relationship between emp~ncal study, theory building, and policy fo~ulation is ~nter- active and continuously evoking. Advances in one domain lamentably influence new initiatives in others. Implicit in the research questions that have been highlighted in previous chapters is the need to link data cohec- ion, analyses of attitudes and behavior, and the design of ~ntenrentions tO underlying theoretical constructs, for example, theories of adolescent development, theories of social structure, or theories of Sugars ecology. 232

OCR for page 231
232 ADOLESCENT SEXUALITY; PREGNA.~CY AND CHILDBEARING What is the meaning of sexual behavior in the context of psychosocial, cognitive, and physical development? What does it mean in relation to race, faintly structure, and socioeconomic status? What does it mean in different cultural communities and neighborhood environments? Within such frameworks, data needs can be specked, measures can be denved, hypotheses concerning the relationships among relevant variables can be tested, and programmatic approaches can be developed with some logical connection between these often separate activities. The remaining sections of this chapter present the panel's recommenda- tions for Data collection, research on adolescent sexual and fertility behav- ior, program evaluation, and expenmentation. PRIORITIES FOR DATA COLLECTION The panel recommends that data systems that monitor fertility anlfertility- related behaviors be maintained and strengthened. Such data are essential for understanding trends and! correlates of adolescent sexual activity, contraceptive use, pregnancy, abortion, arid childbeanug arid as a basis for policy and program development. Fiscal cutback that affect ongoing data collection programs could seriously damage the quality and availability of these data systems. Data concerning levels and variations in teenage sexual activity, contra- cepi~ve use, pregnancy, abortion, childbearing, and other fertility-related behaviors are the basis for the panel's deliberations. Such data will con- tinue to be essential for fixture research and analysis on these difficult issues. Relevant information is available from several sources, including large-scale surreys, federal and state administration e reporting systems, and service providers. As described in Chapter 2, data Mom each of these sources has particular strengths and weaknesses. Individual data systems vary in their underlying purposes and special emphases as well as their specific characteristics (for example, definitions, sample size, data coDec- tion integrals). For these reasons, and because information on sensitive issues requires validation from more than one source, a multidimensional strategy for data collection is essential. We note here several general issues regarcliT~g the collection of inforTna- tion in large-scare data systems that affect their usefulness In studies of adolescent sama1 and fertility behavior. First, in many cases the definitions of key concepts (for example, pregnancy nsk) are not uniform across data sets, thus making it hit for researchers studying particular phenom- ena or relationships to merge or compare infonnation from different

OCR for page 231
PRIORITIES FOR DATA COLLECTION AND RESEARCH 233 sources. In addition, within individual data sets, standardized information is offers unavailable in sufficient detail to support the desired analyses (for example, data on single years of age an] data on race and eternity). At a minimum, national data sets should include the categories recommended by the Office of Management and Budget (OMB) regarding race (black, white, Asian/Paci~c Islander, Native Amencan' and e~hnic~y tHispmic/ non-Hispan~c) as separate items. More detailed race and ethnics data In large-scale data sources would significantly enhance their value. Second, several large-scale data systems contain ~nfonnation on other behaviors (for example, health, education, and drug and alcohol use) that relate to adolescent behavior or fertility. These ~nforrnation sources constitute on portunities for mon3tonng the concurrence of venous behaviors and for conducting analyses of the relationships between them. Unfortunately, several factors prevent their filll use for these purposes: in some surveys the data are not collected on a regular basis; in others there are serious gaps ~ coverage; In Stat others the data are not collected at the in~vid~ level, or the linkages among relevant topics are not made. Data sets collected primarily to study drug use and alcohol use, for example, would be enhanced by Including even rn~n~mal data on fertility. This would facilitate Interdisciplinary studies of the health consequences of corn . . ton sources of variation in behavior and studies of patterns of substance abuse as they relate to seaman behavior, contraception, and pregnancy. Similarly, data sets focused primanly on education have not eypica~y collected detailed Formation on sexual and ferocity behavior, even though such behavior has been shown to be linked to dropping out of school. The discussion of pnorit~es for data co~ecuOn is organized according to the types of relevant data sources: large-scale surveys, national an] state- level repomng systems, and data collected by service providers. Lar~e-Scaie Surveys Major large-scale surveys that provide cross-sectional information on aspects of adolescent sexual and fertility behavior include general popula- lion surveys, health and ferdlit:y surveys, and youth surreys. Many of these data sources have had long-standing federal support, yet recent fiscal pressures have threatened to compromise the quality of several of the surreys and to dunin~sh their value to fertility researchers. The pane] endorses the protection and maintenance of these data sets and highlights several specific ways in which they usefi~Iness ~ studies of adolescent pregnancy and childbearing might be enhanced.

OCR for page 231
234 ADOLESCENT SEX UAL7Ty PREGNANCY; AND CHILDBEARING GeneralPopulatio?: Surveys General population surveys contain abroad array of descriptive information on ch~actenstics of the U.S. population. Because they pronde lengthy time senes, they permit analyses of popula- tion trends (including fertility) over time. Because of their very large sample sizes, they support analyses of small population subgroups that are difficult to study using other data sources. Two general population sur- veys are especially relevant: the decennial census and the Current Popula . ~ lion Survey. The decennial census provides the largest sample and most complete information on general characteristics of the U.S. population of any available data system. In addition tO identifying patterns of change in household and family composition, racial and ethnic composition, age composition, geographic distnbution, and personal income, it iS in~ralu- able as a benchmark in the development of population estimates at state and local levels. Especially useful to fertility researchers is the ~nfonnation collected on children ever born and marital history, particularly given the available geographic petal] for states, cities, and towns. Data on children ever born we critical tO analyses of fertility among small populations, such as small ethnic groups and recent immigrants. In similar fashion, data on marital history permit researchers to trace patterns of marriage an] family structure among diverse population subgroups. In addition, because of its broad coverage of the population, the decennial census frequently provides the basis for the sampling designs of other data collection acti~rites Unfortunately, however, census data are not detailed in many areas of interest tO researchers studying adolescent sexual and fertility behavior. For example, they do not contain information on sexual activity, contra- ception, pregnancy, and abortion. Although there is significant pressure against expanding the census, the usefulness of these data to fertility researchers could be farther enhanced by inclusion of inflation on rare events, such as adoption. In addition, providing published data on special population groups that do not generally receive attention in other surveys, such as native Americans, would also be helped. This detail, available from the 1970 census, was `deieted from the 1980 tabulation program for reasons of economy, although the detail is available on the tape record. Given the current pressures to further cut the COStS of data collection, it is important to ensure that, at a niinimum, the fertility data included in the 1980 survey will also be Included in 1990. The Current Population Survey (CPS) is the source of monthly esti- mates of employment and unemployment, including extensive detail on

OCR for page 231
PRIORITIES FOR DATA COLLECTION AND RESEARCH 235 population charactenstics. Through the regtliar addition of supplemental questions, the survey also prowdes both annual and one-time information on a broad spectrum of subjects, such as family and personal income, poverty, receipt of noncash transfers, annual work experience, school enrolIrnent, and migration. Among the many supplements that have generally been included annuaTiy on the CPS are questions on fertility and expectations for fixture births. These data permit analyses of fertility as they relate to other social charactenstics, e.g., age, marital status, occupa- tion, and living arrangements. Most important, however, data are not currently collected on births to unmarried women under age 18. Should such data be included in the CPS, it would greatly facilitate analyses of public income transfers and child support to teenage mothers as well as their patterns of labor force participation. Health and Fertility Surveys Ongoing data collection activities related to health and fertility behavior either are or can be made extremely useful for research on adolescents. These surveys include information on sexual activity, contraception, pregnancy, abortion, and childbeanng, and they permit analyses of fertility patterns, infertility, reproductive health, con- traception, and fertility intentions. Three data sets are especially relevant: the National Surrey of Family Growth, the National Natality Survey, and the Health and Nutntion Examinanon Survey. The Nanonal Survey of Family Growth (NSFG) is a basic source of data on the sexual and fertility behavior of U.S. women as well as on pregnancy outcomes, maternal and child health care, and family formation. Because the NSFG represents the con~uai~on of a line of fertility surreys extend- ~ngLack to 1950, is is possible to use these data for analyses of changes over time In patterns of sexual activity, coIltracepnon, pregnancy, an] family formation and composition. In the 1982 surrey and planned for the 1987 survey are data on women ages ~S 14 without regard for mantal status or childbearing history (although parental consent is required for minors who are interviewed). Blacks are overrepresented ~ the sample In order to pronde more reliable data for this subgroup. As a result, these data are especially useful for national estimates of adolescent pregnancy and child- bearing. The NSFG also includes iIlformation on abortion and ad~opi}on, although these data are less reliably reported. The Nai~oIlal Center for Health Staiisucs and other federal agencies supporting the NSFG should be encouraged to explore ways of improving the collection of data on abortion and adoption as a part of this survey. In addition, to the extent

OCR for page 231
236 ADOLESCENT SEXUALITY PREGNANCY AND CHILDBEARING that the survey could be expanded to include community-level data, such as community and local population characteristics (e.g., racial and ethnic composition, unemployment rates, median income), or if it is possible to merge such data with the survey's respondent data, its usefulness would be enhanced. The National Natality Survey (NNS), conducted in 1980, was intended to extend data available from the nation's vita] statistics system: it pro- wdes detailed information on births, late fetal Tosses, and infant mortality In that year. This fohow-back survey sampled one of every 350 birth certificates (oversampling low b~rthweight infants) and collected informa- tion from the mothers, the attending physicians, and the health care facilities. The data set includes valuable information on the health and health care of the mother as wed as other relevant family background charactenstics. Because of its large sample size, it offers special advantages to researchers. Unfortunately, however, only married mothers were asked to complete the mother's questionnaire in 1980, thus creating a significant gap in information concerning birth outcomes to unmarried mothers, a large proportion of whom are adolescents and/or black. The practice of riot contacting umnarried mothers also severely limited information that could be obtained on adoption, since presumably very few married moth- ers relinquish their infants. . . , BAN ~ he Health and Mutation Examination Survey (HANES) is a general source of information on the health status and hearth behavior of the U.S. population. Two Cycles of the survey, each conducted over a three- to four- year penod, were initiated in 1971 aIld 1976. A third cycle is planned to beam In 1988. HANES combines interviews of survey subjects with direct physical examinations, thus enabling researchers to match and compare respondent attitudes and perceptions of health status to objective measures of their physical condition. Successive Cycles of the survey have been targeted to differeIlt segments of the population (e.g., Hispanics) and different sets of health conditions (e.g., sensory defects). HANES contains general information on fereiligr among adult and adolescent women. Questions concerning the sexual and fertility behavior of adult and adoles- cent males, however, were not included in previous cycles. Should such items be ad-deaf ~ the next Cycle, they would significantly expand available knowledge of male attitudes and behavior. Similarly, the child health component of the survey also offers opportunities to improve available data on adolescent health and health-related behavior associated with sexuality md fertility. In particular, increased emphasis on physical and

OCR for page 231
PRIORITIES FOR DATA COLLECTION AND RESEARCH 237 psychological maturity and mental health conditions (e.g., depression, anxiety) would greatly enhance knowledge concerning the relationship between biological, developmental, an] environmental factors influenc- ing adolescent sexual and fertility behavior. Youth Surveys Other ongoing surveys have less direct focus on adoles- cent fertility-related behavior but are valuable sources of data on young adults. These include the National Longitudinal Survey-Youth Cohort and the High School and Beyond Survey, among others. The data on social, educational, occupational, and other aspects of adolescent life make them amenable to analyses directed at the effects and antecedents of early sexual activity and family formation. Many youth surveys include males and thereby facilitate the analysis of male adolescent sexual activity, par- enung, and family foundation in conjunction with education and labor market experiences. The usefulness of these surveys to fertility researchers could be enhanced with greater emphasis on measuring adolescent asp~a- tions and expectations for educational and career attai~nent jointly with measures of marriage and family formation. National arid State-Leve] Reporting Systems Several national and state-level adm~n~stranve reporting systems con- ta~n ~forrnanon on specific `demographic, public health, education, and social service topics. Typically these reporting systems rely on data coIlec- tion by state-level agencies; information Tom the states ,; then comnil~d by designated federal agencies. in, ... . ... r-~~ Or speam Importance to verity researchers is the vital statistics sys- tem, which provides condoning "formation on births, deaths, mar- nages, and divorces. Limited demographic and health wformanon is collected Mom the 50 states, which is summarized and tabulated at the federal level and can be analyzed to highlight health problems among infants and to measure progress made by national health programs. These data serve three major purposes: first, they provide the national monitor- ~ng of year-by-year changes In fertility and provide race and age-specific data for subnat~onal areas. Second, they provide data to abbess research questions of Importance regarding correlates of adolescent fertility behav- for. Third, they are essential in the preparation of population projections. Vital statistics, like census data, offer wide coverage for descnpove studies but limited ir~fonnanon for causal analysis.

OCR for page 231
238 ADOLESCENT SEXUALITY PREGNANCY AND CHILDBEARING The usefulness of ~nta] statistics data to researchers concerned with adolescent sexual and fertility behavior could be improved in several speaDc ways. First, only 41 states and the District of Columbia currently report mother's mantal status on birth certificates. This information is essential to analyses of nonmantal childbeanug as it is related to both outcomes. Because there are so many births to unmarried teenagers, especially black teenagers, information concerning mantal status is signifi- c~t in studying this population group. Second, the possibility of linking both and death records would greatly enhance our understanding of the relationship of factors associated with a pregnancy and its outcome. Adolescents, for example, are at elevated risk of bearing low b~rthweight infants and of having an infant die. A linked birth and death system would enable researchers to relate mother's characteristics (education, mantal status, use of prenatal care, etc.) to birth outcomes. Th=d, better ident~fi- catlon of racial and ethnic groups could also expand research linking special population subgroups to birth outcomes. in this regard, States could be encouraged to modify their birth and death certificates to employ the OMB categories previously discussed. Fourth, the addition of certain new items of infonnai~on could be extremely beneficial for research link- ing health conditions and health care to both outcomes: data on smoking, weight and height prior tO pregnancy, and insurance coverage of both the mother and the newborn. Moreover, although significantly more con~o- versial, inclusion of the mother's social security number would greatly facilitate research that would link successive births to the same woman in order to examine repeat patterns of adverse pregnant y outcome. Many of these suggestions have already been implemented in several states and have proven to be both feasible and valuable. Data From Sentence Providers Infor~aiioI1 concerning the supply, use, and COStS of specific health and fertility-related sentences, such as family planning and abortion services, are collected by sernce proudness aIld compiled by government and private org~nwations concerned with mon~tonng these services. Because these data are typically collected for management purposes, they are frequency inadequate for research purposes. For example, relevant background in- folluaiion on service clients is frequently neglected, and clients' use of services across time is not tracked. For three fertility-related services, data are inadequate and therefore

OCR for page 231
PRIORITIES FOR DATA COLLECTION AND RESEARCH 239 limit researchers' capability to study them in relation to adolescent sexual and fertility behavior: family planning services, adoption, and abortion. Data on family planning services are important to understand how the contraceptive clinic system serves adolescents. Since 1981, the federal government has not provided fiends for monitoring family planning clinic performance, which includes the characteristics of services pro- nders, service components, and patients and the costs of care. The latest available data are for 1983, and these were collected by the Alan Gutt- macher Institute with private support. The federal government should resume monitoring the availability of contraceptive services and their use by all patients, including adolescents. Adoption is a potentially important type of pregnancy resolution about which very little is known. As we discussed in Chapters 2, 4, and 7, this is in part a result of the ~iscontinuanon of the national reporting system in 1975. Currently the only national system that gathers annual information on adoption is the Voluntary Cooperative Information Sys- tem managed by the American Public Welfare Association. This system collects data only on children placed for adoption by public child welfare agencies. Therefore its enumeration of adoptions nationwide is incom- piete and unreliable in assessing trends. While it contains some ~nforma- tion on the characteristics of adopted children and adoptive families, it contains little or no information on the characteristics of the birth parents, the adoption process, and the subsequent fertility of the birth mother. The development of a nationwide adoption information system to account for all types of adoptions (public agency, private agency, independent) on an annual basis would be extremely useful in tracing trends in adoption and the effects of recent federal policies to encourage adoption as an alternative to abortion for unmarred teenagers. Weaknesses in data on abortion hinder analysis of both pregnancy and abortion among adolescents. In particular, information from surveys undercounts the number of women having unintended pregnancies and abortions. Information on abortion is collected by several public and private organizations, and each source has some important shortcom- mgs. State health agencies collect data that are compiled by the Centers for Disease Control: Included is information on the distnbution of abortions by age, race, panty (number of children ever born to a particu- lar woman), and other characteristics, but the incidence of abortion is underestimated because not all providers are covered. The National Center for Health Statistics collects data from selected state vital reg~s

OCR for page 231
240 Al)OLESCENT SEXUALITY PREGNANCy AND CHILDBEARING tration offices: they are valuable for their content but are limited in coverage to a small number of states. The Alan Gut~macher Institute collects data directly from service providers and compiles them: they pronde a more complete enumeration but are limited in background charactenstics. Taken together, these data sources provide a relatively complete Count of abortions and inform citation on relevant background characteristics of women obtaining abortions at the national and state levels. It is unport2ne that these data collection efforts continue with special attention to improving coverage and the timeliness of reports. In surveys of women, such as the National Survey of Family Growth and the National Longitudinal Survey, unintended pregnancies ending in abortion are greatly underreported. Since it is difficult to assess which respondents are most likely not to report such pregnancies, analyses of determinants of pregnancy and abortion are hampered. Efforts should be made to correct for such underreportir~g and to find ways to reduce it in future surreys. PRIORITIES FOR RESEARCH ON ADOLESCENT SEXUAL AND FERTILITY BEHAVIOR The parcel recommends the continued support of a broal-basel research pro- gram on adolescent sexuality ar:diertility to enhance understanding of the causes and consequences of these behaviors and to inform policy and program levelop- ment. Over the past decade, research has added significantly to the knowI- edge of trends, correlates, antecedents, and consequences of adolescent sexual and fertility behavior. These research findings have pronged an essential basis for the panel's deliberations. Numerous studies have ex- am~ned the short- and long-term effects of early pregnancy and child- beanng on young women's health, education, fertility, marital experi- ences, employment, and economic weD-being. Others have explored the effects on the children of teenage mothers and, to a much lesser extent, on young fathers. Still others have examined the somal, developmental, cultural, an] economic antecedents of early sexual activity, contraceptive use, pregnancy, abortion, and childbearing to identity factors affecting sexual pension making. In short, knowledge of adolescent sexuality and fertility has increased substantially, yet many questions remair1 unan- swered. In some cases, the gaps reflect issues that have not been ade- quately studded because of methodological problems; in other eases, new issues have emerged from the accumulation of findings.

OCR for page 231
PRIORITIES FOR DATA COLLECTION AND RESEARCH 241 Throughout this report, we have highlighted a variety of salient questions; many of these have focused on special population subgroups (e.g., small ethnic groups) and rare demographic events (e.g., adop- tion). In framing an agenda for future research, however, it would be usefi~i to fit these questions into a broader framework for understanding factors that influence adolescents' sexual decision making at each stage in the sequence of choices, from the initiation of intercourse to parenthood, and the consequences. As we have emphasized, each choice is complex. Each is affected by the circumstances under which it is made and the range of available options. The histoncal time, the ecological setting, and the developmental characteristics and accumulated life experiences of individual adolescer~ts are all potentially important factors influencing the process of choice and its outcomes. Accordingly, the panel's recom- mendations for future research on adolescent sexual and fertility behav- ior are organized under four domains of antecedents and outcomes: individuals, families, commuruties, and society. In~ivi~uais Much of the emsting research on adolescent sexual decision making and its consequences has focused on teenage girls between the ages of 15 and 19. We know a great deal about the charactensucs of young women in this age group who become sexually active, who experience preg- nancy, and who bear chidden. We know much less about the sexual and fertility behavior of teenagers under age 15 or about adolescent males of ad ages. Similarly, although much of the research has focused on black- white race differences in attitudes and behavior, less attention has been given tO small racial and ethnic groups, such as native Amencans and Hispanic subgroups. Efforts tO preserve and improve large-scale national data sets should be supplemented by efforts to Vitiate more intensive small-scale studies of selected communities and population subgroups. More detailed data are needed on the relation between biological, social, emotional, culturA, and economic factors influencing sexual decision making and itS conse- quences. In particular, ethnographic studies are needed to develop de- tailed profiles of the characteristics, attitudes, and behavior of m~i~d- uals and families living in different circumstances and environments. And more longitudinal studies are needed to examine changes in atti- tudes and behavior over time.

OCR for page 231
250 ADOLESCENT SEXUALITY; PREGNANCY; ANI) CHILDBEARING PRIORITIES FOR PROGRAM EVALUATION RESEARCH The pane] recommends that evaluatio?t to measure the costs, effects, an: effectiveness of service programs be an essential component of intervention strategies. Federal and state-level funding agencies should be urged to set aside adequate supportfor evaluation research, and the research community should be urged to take a more active role in designing and implementing these studies. Although evaluation research methods have become quite sophisti- cated over the past two decades, they are frequently not used in studying the effects of adolescent pregnancy programs. As we discussed in Chap- ters 6 arid 7, there are several reasons for the general scarcity of program evaluation research and the poor quality of findings. These include problems of design: marry programs have failed to clearly define goals and to distinguish direct and indirect outcomes, thus making it ~i~iCUit to specify appropriate hypotheses for study. Measurement problems are also common. Specific types of demographic change (for example, re- ductions in the number of pregnancies) are often difficult to accurately measure because of underreporting of miscarriages and abortions. In addition, a variety of other factors, including local social, demographic, and economic characteristics, can confound program effects or make them difficult to isolate and measure There are also ethical problems: the human dimensions of adolescent pregnancy and childbearing make it difficult to deny a young woman (and perhaps her child) essential sup- ports and services in order to assign her to a nontreatment control group: Finally, several practical problems have also frequently impeded program evaluation research, affecting the quality of the results. Evaluation re- search is expensive and project grants and contracts for seances often fail to include specially earmarked funds for assessment. Sernce providers typically lack the training and skills to design sophisticated evaluations and the time to carry them Out. Moreover, the time frames for measuring program effects and effectiveness are often too short, thus inaccurately characterizing a program's success or failure in meeting specified objec- t~ves. Outcome Venus Process Evaluation Program evaluation can have multiple purposes that are valid but often conflicting. One of these is outcome evaluation: to assess the effectiveness of an intervention strategy in achieving its specified objec

OCR for page 231
PRIORITIES FOR DATA COLLECTION AND RESEARCH 2;1 fives. Often research of this type is undertaken to judge the validity of an approach and to determine whether a program should be continued, modified, or expanded. Typically, outcome evaluation involves estab- lishing "before" and "after" measures of program activity and compar- ing these to a control group that was not exposed to the program. If the treatment remains constant throughout the study period, then the results may in fact represent a valid assessment of the program's success. Another purpose is process evaluation: to monitor how a particular program works and to document program models. Service providers are often less interested in evaluation for its capability to measure outcomes than for its capability to provide information for program improvement. They are interested in knowing how high-risk teenagers get into the service network, how they respond to alternative approaches, and how long they stay in programs. As data are collected, providers frequently see opportunities to modify the program to improve its efficiency and responsiveness to the needs of clients. While such uses of evaluation data are valid, mid-program modifications hinder the usefulness of the find- ings to researchers interested in outcome measurement, since the same treatment is not being analyzed throughout the study penod. Their purposes are different, and both types of evaluation should be pursued, yet it is important for program planners to recognize the circumstances under which one or the other type of evaluation is appro- pnate and is likely to yield useful information. Not every program can or should tee the object of a rigorous outcome evaluation. The problems and expense of colDecting and analyzing data, identifying appropriate control or comparison groups, and following clients over time frequently make good evaluation infeasible. Poor-quality research, which can often be costly ~ personnel and fiscal resources, is of little use in assessing the effects of an intervention or predicting its effectiveness in other settings. Outcome evaluation should be pursued on a limited basis when the availability of resources (i.e., time, financial support, technical assis- tance) an] the circumstances of the program (i.e., the size and composi- tion of the client population, appropriate control or comparison groups, low rates of attrition among treatment and control populations, ~'rli- form and consistent sernces or treatments) will support good research. Process evaluation, on the other hand, should be more broadly pursued. All programs should be encouraged and assisted to collect data on their services and their client populations as a basis for assessing whether they are in fact providing the services they were established to provide, in the

OCR for page 231
252 ADOLESCENT SEXUALITY PREGNANCY AND CHILDBEARING way they were planned, to the target population that was planned, and within the COSt boundaries that were envisioned. Such information can pronde a valuable basis for monitoring progress, identifying problems, and making important management decisions about the operations of the program. it can also pronde the basis for demonstrating to sponsors and others that a program is in fact doing what it was intended to do and justifying its existence and continuation after an initial trial period. Several issues are important in the design and conduct of evaluations. Outcome Measures Evaluation design should be conducted in tandem with the planning an] design of the service program itself. Too often an evaluation compo- nent is superimposed after a program has been in operation for some period of time, thus precluding the possibility of obtaining pretest measures as a baseline. In addition, outcome measures should assess the capability of an intervention to achieve its stated objectives. To the extent that such objectives are not clearly specified in the design of the program, it is difficult to define appropriate outcome measures. art particular, many programs fad! to distinguish direct and indirect outcomes. The reduction of pregnancy, for example, is frequently not an explicit goal of many preventive programs other than family planning services, even if it is an important indirect outcome. Interventions to increase knowledge and charge attitudes, as wed as those intended to enhance life options, are often unable to show long-term reductions in client pregnancy rates. They may, however, significantly affect other outcomes. For example, they may enhance school achievement, which may delay initiation of sexual activity or improve contraceptive use among adolescents. Thus, if evaluation is to be meaningful, both direct and indirect program objec- tives need to be clearly specimen, and the research design must include appropriate measures of the behaviors that the program is aimed at influencing. For example, interventions to encourage delay in sexual initiation must measure changes In sexual behavior among their client populations; programs to improve school performance need to measure not only changes in academic achievement and school retention, but also changes in sexual and fertility behavior. Several other issues concerning outcome measures are also important. First, many evaluations of adolescent pregnancy programs' both preven

OCR for page 231
PRIORITIES FOR DATA COLLECT10~ AND RESEARCH 2~3 rive and ameliorative, rely on self-administered questionnaires to collect data from clients. Often, however, researchers have not been sensitive to the reading levels of their study populations, large proportions of which can be expected to have significant reading difficulties. If seif- a~ministere] questionnaires are used as a means of data collection, they should be acoroDriatelY adjusted to the reading level of the StU8Y DoDUia tlon. 1 1 ~J Cal J ~ Second, few evaluations have used the same measures of knowledge, attitude, and behavior outcomes, thus impeding the comparison of results from study to study and among alternative interventions. Where reliable and valid established measures are applicable, investigators should be encouraged to use them. Where adequate measures do not exist, new instruments should be developed for use by evaluators in a variety of settings, involving adolescents of different genders, ages, races, and other relevant background characteristics. Third, with regard to several specific issues, new measures are needed: (1) developmental age the wide developmental differences between adolescents (boys and gratis) of the same age make chronological age or grade in school inadequate measures of maturity in many cases; (2) "wantedr~ess" and "intendedness"- in the context of pregnancy and childbearing, these terms represent complicated concepts that have not been clearly defined and distinguished; (3) program exposure pretest, interim, and posttest data collection over the time period of an intenren- tion may not adequately take account of uneven exposure by some participants. Clinic and school populations frequently shift, as some clients enter the program after its start, some leave before completion, an] stir] others move in and out more than once during the treatment period. Sampling and Control Groups The representativeness of both study sample and control group signif- icant~y affects the quality and generalizability of evaluation findings. The size and background characteristics of these groups are important. Many of the evaluations we reviewed involved study popuiat~ons that were so small that meanings! conclusions about the applicability of the program's approach to other client groups in other settings were impos- sible. The size of the sample should be large enough to permit general~za- tion. Similarly, the background characteristics of the sample should be

OCR for page 231
254 ADOLESCENT SEXUALITY PREGNANCY; AND CHILDBEARING carefully screened, since various cultural, demographic, religious, and economic factors can influence the variables under study. In addition, too often the program population is chosen as a matter of convenience (i.e., individuals who voluntarily seek services) rather than because it is best suited to answer the relevant questions concerning program effects and effectiveness. Although random samples chosen against clearly specimen Reselection criteria are not always feasible (e.g., too expensive, too few potential study subjects from which to choose), evaluators must take account of the issues of self-selection in study populations that are drawn from clinic samples and voluntary partici- pants in venous school-, church-, and community-based organizations. To the extent that randomization can be incorporated in sample selec- tion, it should be. When it cannot, evaluators should be encouraged to develop innovative research designs and sampling techniques to enhance representativeness. An issue of particular concern in studying program effects on very young adolescents is the need to obtain parental consent for their partici- pation. As previously discussed, getting consent is not always easy or possible, yet eliminating young teenagers whose parents wall not sanc- tion their participation can significantly bias the study findings. Ideally, when measuring the effects of a program on a target popula- tion, researchers should compare those exposed to the treatment with a comparable group who are not exposed. However, identifying and then matching "pure" COIltrOiS to study populations has frequently proved difficult. Because of the broad array of preventive and ameliorative supports and services that has been implemented nationwide, it iS ~i~- cult in many communities to fed individuals and groups who are receiving no services at all as a basis for comparison. To overcome this difficulty, researchers need to explore innovative ways of identifying appropriate comparison groups when pure controls are impossible, for example, by using data from national surveys or from other communi- ties that can be matched along relevant dimensions. Finally, researchers need to be able to minimize attrition among both treatment and control or comparison groups in order to achieve valid results from their studies. When clients leave a program before they have received all the sernces or treatment that was intended or when they are not followed up after their scheduled departure, findings concerning those who stay in and are tracked may be skewed in ways that are difficult to detect. The reasons for a client's leaving prematurely may

OCR for page 231
PRIORITIES FOR DATA COLLECTION CAT) RESEARCH 2~: affect conclusions concerning a program's effectiveness if, for example, only highly motivated clients or those with strong family support sys- tems or those who can travel easily to the service center remain. The same is true for control or comparison groups: attntion can significantly affect findings concerning the outcomes of the program. To avoid these problems, researchers need to have large enough samples so that attrition will not adversely affect the quality of the findings. In addition, they need to develop aggressive and innovative approaches to follow up and track clients after they leave a program. Irzlen~ening Vanables As a part of the research design for a program evaluation, indepen- dent, dependent, and intervening variables and their potential reiation- ship should be clearly specified. A variety of intervening factors can confound the study results unless they are identified and carefi~By con- trolled. For example, as we have previously noted, demographic changes (e.g., reductions in the number of adolescent pregnancies) may reflect a variety of factors other than or in addition to program effects. Changes in the composition of the local population or other changes in the social, economic, or cultural context of the intervention can affect program outcomes. Similarly, the presence, absence, or change in other local health, education, social service, and income programs can influence study results in ways that are sometimes Tout to detect. Researchers need to take account of the possible influence of such intervening van- ables in their research designs. Units of Analysis Studying the erects of specific program components may be as ~mpor- tant, if not more important, than examining the overall outcomes of venous interventions. This is especially true for assessments of compre- hensive programs that integrate and coordinate an array of necessary supports and sernces and may employ several sequential or simultaneous approaches to serving high-nsk teenagers. Many evaluations of compre- hensive programs have failed to separate program components for pur- poses of analysis, and as a result we know very little about which aspects of these programs are most effective and how they individually, or in combination, contribute to overall program outcomes.

OCR for page 231
236 ADOLESCEN T SEXUALITY; PREGNANCy AND CHILDBEARING Cost Data Cost-benefit analyses of alternative programs have been hampered by the lack of available data on cost. Very few" evaluation studies have collected information on the unit costs of service provision (i.e., the COStS per program participant per year) as well as more complex measures or the costs per birth or the potential savings from births that are postponed as a result of the program. Such analyses are needed in order to compare the relative benefits of different types of programs with their costs. As a first step, attention should be given to the development of appropriate measures of "input" an] "output" that can be reasonably applied to different intervention models in different settings. Tme frames The time frames for evaluation are also important. Outcomes may look different at SIX months, one year, two years, and five years Draw- ing conclusions about a program's success or failure on the basis of a short-ter~n follow-up may inaccurately characterize it. For example, to conclude that a program was successful in preventing or delaying subse- quent births on the basis of measurement at one year after a first birth probably overstates its effectiveness, since we know that a high propor- tion of repeat pregnancies and births occur in the second year. I'orlgitu~i- nal research is admittedly expensive and complex, yet follow-ups of selected samples of program participants over time may yield valuable information concerning the duration of program effects for particular subgroups of the target population. Financial and Technical Ass~star~ce Many of the programs we examined lack the necessary capability tO rigorously measure their effects and effectiveness. In addition, retrospec- tive record searches and client foDow-up studies are expensive and time- consurn~ng and cannot typically be conducted by program administra- tors whose primary responsibility is service provision. Research design, data collection, and analysis require ~ level of technical expernse not typically found among project directors and clinical staff. Without a university affiliation or other strong evaluation support both financial and technical most programs are unable to design and carry out the

OCR for page 231
PRIORITIES FOR DATA COLL~CTIO.~ AND RESEARCH 237 kinds of sophisticated evaluations that are necessary to document their processes and outcomes. In this regard, it is important to train young researchers to bridge the gap between their disciplinary domains and the provider community in order to build a strong evaluation ability. Too few scholars of adolescent development are also sophisticated in program evaluation methods. Because academic incentives discourage program research, too few are able and welling to assist service providers in the design and implementa- tion of evaluation components for adolescent pregnancy programs. Fed- eral and local funds should be committed to developing such a capability at the regional and local level. PRIORITIES FOR EXPERIMENTATION The panel recommends thatietleralfunding agencies, privatefoundations, and researchers cooperate in designing, implementing, and evaluating experimental approaches for pregnancy prevention among high-risk adolescents andfor im- provir~g the well-being of teenage parents arid their children. The problems of adolescent pregnancy and childbearing are many and complex. From our comprehensive rewew of research andinterventions,. we have found only limited documentation of successful program models for pregnancy prevention and for the support and care of preg- nant and parenting teenagers and their children. Given the diversity of personal circumstances among adolescents in this country and the uncer- tainty arid controversy surrounding these issues, a multiplicity of ap- proaches wiD continue to be needed. To successfully avoid early unintended pregnancy, young people need both the ability and the motivation to do so. They need information concerning reproduction and contraception, they need effective means of birth control, and they need access to services. They also need to perceive that there are positive benefits associated with postponing family Connation, and that education and job traming can significantly improve their social and economic weD-being. Many specialized and comprehensive programs have been developed to provide education, to deliver contraceptive sernces, and to enhance teenagers' perceptions of life options. We need to continue to mo~tor these existing program development efforts and expand them to include expenmentation with innovative models and Sorrel approaches. In the course of our study, several areas for new program development

OCR for page 231
258 ADOLESCENT SEXUALITY PREGNANCY AND CHILDBEARING surfaced. First, delaying the initiation of sexual intercourse is one means of pregnancy prevention, yet little is known about effective approaches to encourage teenagers to delay sexual initiation. Several kinds of pro- grams are under way that aim to discourage early sexual activity among unmarried teenagers, including family life education programs, asser- tiveness programs, and programs to provide role models for high-nsk youth. These initiatives should continue to be monitored. In addition, new approaches to encourage delay of sexual intercourse should be designed, implemented, and evaluated. Moreover, farther effort is needed to explore ways of providing birth control information to young teenagers, including information on contraception, without making it difficult for those who choose not to initiate intercourse to maintain more traditional attitudes and behavior. Second, the contraceptive pill and, to a somewhat lesser extent, the condom represent effective means of birth control when they are used aDDrocriatelY. Yet research suggests that mane adc~lecc:~ntc finch them ~ ~ ~ ~ - - O D ~ -A ~ methods Cult to use and therefore are frequently inconsistent or ir~effective contraceptors. While additional efforts are needed to encour- age sexually active teenagers to diligently use existing methods, espe- cially the pill and the condom, special attention should be given to developing and testing new contraceptive technologies that are more appropriate to the needs of adolescents. Third, efforts should be devoted to creating positive economic incen- tives for adolescents tO stay in school and avoid pregnancy. For too many high-nsk young people, there are too few disincentives to childbearing. Several interesting proposals have surfaced recently to develop p0Ot programs to reward high-nsk teenagers who complete high school without having a baby by providing annual or lump-sum cash payments. Other possibilities along these lines include the provision of special secondary and postsecondary scholarships or job opportunities. For ex- ample, it appears that many teenage mothers use their AFDC payments to make it possible to finish high school. Experimental programs should be `developed to examine alternative financial assistance mechanisms for secondary education that do not foster welfare dependence. Funds should be made available for carefully designed and evaluated demonstra- tion programs of this type. Similarly, teenage parents need both the ability and the motivation to manage their lives, care for their children, and delay repeat pregnancies. Experimentation with bold approaches to encourage young mothers to

OCR for page 231
PRIORSHIP FOR DATA COLLECTION AND RESEARCH 259 become economically self-sufficient and to encourage young fathers to become financially responsible and actively involved in their children's lives is also warranted. Again, efforts should be devoted to creating positive incentives for these behaviors. Economic rewards are a powerful motivating force and could be an effective tool for encouraging positive life choices among high-nsk teenage parents. Demonstrations to test innovative models should be initiated and carefully evaluated. Finally, parents are responsible for the financial support of they chil- dren, yet many fathers of children born to adolescent mothers, especially teenagers, have not assumed economic responsibility for their families. Further efforts are needed to develop effective means of child support enforcement. In particular, new initiatives are needed to link child support to education and work requirements in the form of registration with the state employment service and participation in job training and job search activities, as well as the provision of work opportunities. For school-age fathers who are still enrolled in school, part-time and summer employment requirements should be tested. For those who are out of school and are not participating in other job training programs, special plot programs to require job training and employment should be devel- oped, implemented, and evaluated as a means of improving the provision of child support.

OCR for page 231