The measurement of public opinion on abortion: The effects of survey design.

A factorial experiment examined the effects of the wording and sequence of survey questions on the measurement of attitudes toward abortion. When a first-trimester pregnancy is specified, 55% of respondents agree that a woman should be able to obtain a legal abortion for any reason, compared with 44% when no pregnancy duration is stated. Specifying first-trimester pregnancies has little effect on the proportion of respondents who agree that abortion should be available for maternal health, fetal defects or rape, but it significantly increases the proportion who agree that a woman should be able to obtain an abortion if she is single, has financial constraints or wants no more children. When gestational lengths from one to six months are presented to respondents in ascending order, agreement that a woman should be able to obtain an abortion for any reason is lower for any given length of gestation than when pregnancy durations are presented in descending order. Forty-eight percent of respondents agree that abo rtion should be legal for any reason when that question is posed after a series of specific reasons; however, 60% do so when it is the first question in the sequence. The difference in agreement with abortion for any reason between Catholics and non-Baptist Protestants, and between Republicans and Democrats, is much smaller when the question is asked first than when it is presented last.

The measurement of public opinion on abortion is both extremely important and fraught with difficulty. The wording and sequencing of questions can affect measured levels of approval (1) and the extent to which various abortion-related attitude items are correlated with one another. (2) This article reports on an experiment that replicates and extends prior work on these measurement issues.

The series of abortion attitude questions asked in the General Social Survey (GSS) is a primary source of data for analyses of trends and differentials in attitudes and in the correlations among specific abortion-related attitudes. (3) The GSS has been conducted annually (with a few exceptions) since 1972 by the National Opinion Research Center at the University of Chicago and therefore makes it possible to monitor changes over time. This article begins with an update of trends from the GSS and then--using the questions from the survey--examines possible design effects on measured levels of approval. The analysis is based on a telephone survey that used a factorial experimental design to evaluate the effect of specifying the stage of pregnancy on agreement that legal abortion should be available, and the effect of a change in where in the question sequence respondents are asked whether abortion should be legal for any reason.

Trends in Abortion Attitudes

The GSS has carefully maintained the same wording and sequence of questions from year to year to avoid introducing context changes into the time series. (*) Respondents are asked whether a woman should be able to obtain a legal abortion for a series of reasons, beginning with "if there is a strong chance of defect in the baby" and ending with "if the woman wants it for any reason."

Figure 1 (page 178), which shows average levels of approval for five-year periods from 1975 through 1994, (+) demonstrates an overall stability in approval levels over time. Two levels of approval are apparent, depending on the reason given for wanting an abortion: About 80% of respondents agree that a woman should be able to obtain an abortion if her health is endangered, if the fetus has a serious defect or if the pregnancy resulted from rape (hereafter referred to as limited reasons). Only 41-46% approve, however, if the woman desires an abortion because she wants no more children, has financial constraints or is unmarried (more inclusive reasons); slightly lower proportions agree that a woman should be able to obtain an abortion for any reason at all.

Second, and more important, the modest decline in agreement from the late 1970s through the late 1980s was followed by a complete recovery in the early 1990s. With the most recent data taken into account, the GSS provides no evidence that opinions on abortion are growing more conservative, as seemed to be the case in the 1980s. (4) On the contrary, the proportion of respondents agreeing that a woman should be able to obtain an abortion for any reason rose from 34% to 43% between 1975-1979 and 1990-1994; most of this change occurred since the late 1980s.