This essay is a commentary on a Research Note by James Trussell, Barbara Vaughn and Joseph Stanford entitled "Are All Contraceptive Failures Unintended Pregnancies? Evidence from the 1995 National Survey of Family Growth."

The intended-mistimed-unwanted classification of pregnancies was developed in analyses of fertility surveys conducted in the 1950s and 1960s. Those surveys were restricted to samples of married women. The traditional measures of intendedness were based on a model in which a married couple either selects a family-size target at marriage and then pursues it, or revises it periodically, but at any given time agrees on what the target is (one child, two children, three children, etc.).

This framework works well for most married couples, but it is less plausible for unmarried teenagers, unmarried adults and those who suspect that they are unable to conceive. For these latter groups, a woman's answers may vary from what we would expect from the simple schema above—because of the partner she has at a particular time, whether she intends or wants to marry her current partner, how much she knows about her fecundity or that of her partner, and other factors.

The last three cycles of the NSFG (1982, 1988 and 1995) have collected data from women of all marital statuses, including unmarried teenagers and adults, and from oversamples of minorities. The information collected has been kept consistent throughout NSFG history to make it possible to monitor trends in unintended pregnancy. The growing coverage of NSFG surveys, however, is one of the principal reasons why several new measures of pregnancy wantedness were included in the 1995 survey.1 One of the new measures, the "happiness scale" used by Trussell, Vaughan and Stanford, essentially turns the intended-mistimed-unwanted categories into a continuous variable.

As part of a program of methodological research for Cycle 6 (in 2001), NSFG staff contracted with researchers at the University of Alabama at Birmingham to review the literature, conduct cognitive research and develop recommendations for improved measurement of unintended pregnancy and its explanatory factors.2 A number of the innovations they suggested will be pretested for Cycle 6.

Substantive Additions

We need additional data related to unintended pregnancy to help explain why individuals who claim to be practicing contraception are still becoming pregnant, and why individuals who claim not to want to become pregnant are not using a method. We know who is at highest risk for unintended pregnancy, but we do not know why they are unable or unwilling to prevent or delay pregnancy.

Motivation and desire to avoid pregnancy. Several ways of measuring these concepts have been suggested. For example, items developed by Miller3 include a 10-point scale on which the respondent is asked to rate how hard she tried to prevent pregnancy at the time she conceived a specific pregnancy. To assess the strength of her desire not to conceive, she is asked to rate (on a 10-point scale) how much she wanted to avoid a pregnancy. She is asked similar questions about her partner.

Zabin has used a question about a woman's feelings about having a baby with her partner at that time to get at why a pregnancy might be reported as occurring "too soon."4 For example, the answer to the question, "Right before you became pregnant with the pregnancy that ended in (MO/YR), did you think you might ever want to have a baby with that partner?" may show to what degree feelings about a particular partner play a role in the classification of pregnancies as unintended.

Ambivalence. The Cycle 5 questionnaire contained five paired statements that were used to assess the degree of ambivalence felt by young women about getting pregnant right before each of their recent pregnancies. Analysis of the consistency of responses across pairs within the series, and of the series with the "wantedness" and "happiness to be pregnant" measures, showed that three of the five pairs performed consistently.5 A possible replacement for these items is a series developed by Stevens-Simon,6 in which the respondent is asked which statements most accurately represent her feelings right before she got pregnant. For example, "You felt that having a baby would get in the way of your plans for the future, or would fit into your plans for the future." The response categories are "get in the way," "fit into," "both"and "neither" (if offered). Additional statements about the family being pleased, feeling close to the partner, fitting in with female friends, adding something special to her life, and feeling good about herself would be coded in a similar manner.

Reasons for unintended pregnancy. Questions about reasons why a woman conceived a pregnancy she did not want at the time would add explanatory power to the survey. Klerman and Pulley have developed items to identify such reasons as method failure, improper method use, nonuse of a method due to not expecting to have sex, and nonuse for reasons related to the partner and conditions of intercourse.7


To address concerns about possible recall bias in women's retrospective reports about their pregnancy attitudes, the introduction at the beginning of the "intendedness" series could state that we are interested in knowing about a woman's feelings right before she became pregnant, not her feelings during the pregnancy or after the birth.

In Cycle 5, the second "intendedness" question was worded as follows: "At the time you became pregnant, did you yourself actually want to have (a/another) baby at some time?" Based on cognitive research, Klerman and Pulley recommend testing the following wording to clarify the time periods of interest: "Right before you became pregnant, did you yourself ever want to have (a/another) baby at any time in your life?"

In addition, separating the NSFG "intendedness" questions from the questions on contraceptive use might be useful.8 Having women report contraceptive use in the month of conception right before asking them their intentions about pregnancy could affect the latter responses. For example, if a respondent has just reported that she was using a contraceptive method in the month of conception, she may report that her intended pregnancy was unintended because she wants to appear rational in the eyes of the interviewer.


In Cycle 5, the focus was on a woman's method use during the entire period between the previous pregnancy or first intercourse and the pregnancy of interest. For Cycle 6, we may test a question focusing on the method or methods, if any, used in the month of conception. That is, we may test "Before you became pregnant with your (NTH) pregnancy, which ended in (MO/YR), had you stopped using all methods of birth control?" against "At the time you became pregnant with your (NTH) pregnancy, which ended in (MO/YR), were you as a couple using any methods of contraception?" In addition, the question "Was the reason you stopped using all methods of birth control because you yourself wanted to become pregnant?" might be simplified to "Did you, yourself, want to become pregnant at the time you did?"

The desire to preserve the long-term time series to monitor trends remains strong. Therefore, the impact of question changes on responses will need to be assessed.