Pervasive underreporting of abortion in our nation's primary fertility-related surveys compromises the study of abortion, pregnancy and fertility. Relatively little is known about the role of induced abortion in American fertility because it remains a highly sensitive behavior, and social surveys to date have not been able to elicit accurate reporting of women's pregnancy experiences. Researchers have long recognized the underreporting of abortion in U.S. household surveys, but no substantial progress in improving the quality of women's abortion reports has been made in nearly twenty years. To obtain accurate measures of U.S. fertility behaviors and to understand the role of abortion in women's lives, we propose to develop new techniques and improve existing methodologies for measuring abortion reporting, with findings relevant to improvements in the measurement of other sensitive behaviors. This project will focus on two related objectives.
Aim 1 will investigate levels and correlates of abortion underreporting in three widely-used nationally representative surveys. Using a number of National Survey of Family Growth (NSFG) data resources, we will identify individual- and contextual-level correlates of underreporting, exploiting rarely- used geocoded measures of social context. Additionally, we will examine discrepancies in reporting by survey mode for women and men, and the impact of survey-level factors drawn from interviewer and administrative reports. We will compare the patterns of women's underreporting between the NSFG and two other major nationally-representative fertility surveys with varied abortion question wordings and means of administration (National Longitudinal Survey of Youth 1997 and the National Longitudinal Study of Adolescent Health [Add Health]). Finally, we will examine the consistency of retrospective reporting of prior abortions at two points in time for both women and men, utilizing the repeated measurement of full pregnancy histories in Add Health.
Aim 2 will develop, test and evaluate new approaches to improve abortion reporting in future surveys. We will develop strategies to reduce sensitivity of reporting, building on findings from Aim 1, cognitive interviews and focus groups, and analysis of initial experimental tests currently under development. These findings and guidance from an Expert Panel will be used to design and test 10 randomized experimental arms in a national internet survey. Our initial plans for the experimental arms are to test the effect on abortion reporting of different question wordings and the placement of the questions within the survey. These findings will move the fields of fertility behaviors and survey methodology forward by significantly improving the measurement of abortion and other sensitive behaviors in the United States, thereby opening up wider possibilities for research.
Current national surveys of fertility behaviors in the United States are not adequate for studying abortion, or related fertility behaviors, because a majority of respondents do not report abortions they have had. Improving reporting in these surveys would greatly expand opportunities for behavioral and health research, as well as improve the accuracy of research findings needed to inform policy and programmatic arenas, including monitoring the impact of abortion laws and regulations. This project will focus on two related objectives to advance research on abortion and pregnancy: expand our understanding of levels and patterns of abortion underreporting in nationally representative surveys; and develop, test and evaluate new approaches to improve abortion reporting through experimental studies. Findings from the proposed study will create an evidence base for future survey designs for obtaining accurate estimates of abortion and for other sensitive behaviors respondents are reluctant to report.
| Lindberg, Laura; Scott, Rachel H (2018) Effect of ACASI on Reporting of Abortion and Other Pregnancy Outcomes in the US National Survey of Family Growth. Stud Fam Plann 49:259-278 |
