Do you use fertility preferences in your research (or want to)? Are you swimming in a ton of measures and don’t know which one(s) to pick? Or do you have one preference measure and want to know if you can compare it to a different preference in another study? Are you concerned that asking three measures in a row might bias your findings? If so, check out our new paper in Population and Development Review, “Do Survey Respondents Distinguish Between Ideal, Desired, and Expected Fertility? An Empirical Test of Survey Design Effects.”
Suppose we asked you three questions. 1. What is the ideal number of children for a family in general; 2. What would be the ideal number of children for you personally; 3. How many children do you expect to have. Would your answers be the same? Or would each question capture something different? Do you think it matters that we asked three questions back-to-back?
Most demographers would tell you these are different concepts. Theories of fertility decision-making are built around the idea that ideals, desires, and expectations capture different pieces of the fertility decision-making process. But some demographers disagree and think that people really don’t have firm well-known preferences and instead just give surveyors a number to answer the question.
But how do ordinary people actually see and report their ideas about family size? And does it matter whether we ask these questions alone or one right after another? This is what motivated our new study.
We compared four common fertility preference measures: general ideal family size, personal ideal family size, wanted fertility, and expected fertility. We randomly varied the sequences in which people saw these questions and then examined whether their answers changed.
Some theories suggest that fertility preferences are highly flexible and constructed in the moment. If that’s true, asking one fertility question should influence how people answer the next one.
Instead, we found something surprising.
First, people consistently distinguished between what is ideal for families in general and what is ideal for themselves. General ideal family size was almost always higher than personal preferences and remained remarkably consistent regardless of question order.
Second, personal ideal and wanted fertility were almost indistinguishable. Respondents treated “what would be ideal for me” and “what I want” as essentially the same thing.
Third, expected fertility was different. Expectations were the one measure that seemed sensitive to being asked after other fertility preference measures. When people answered ideals and desires first, their expectations became lower. Thinking about what they wanted before they were asked what they expected appeared to encourage them to think about the constraints that might prevent them from getting what they expected.
The broader lesson is encouraging for researchers who use fertility preference data. Most fertility preferences appear more consistent than some theories would predict. Asking multiple fertility questions in a survey does not seem to dramatically change how people answer most of them.
But there is an important exception: expectations. If you want to measure what people realistically think will happen, question wording and survey design may matter more than we previously assumed. Expectations asked alone may simply reflect desires. But when respondents are asked expectations after they are asked about desires, expectations may reflect desires filtered through people’s perceptions of financial, relationship, and life-course constraints.
For demographers, the practical takeaway is straightforward. General ideals appear to capture a social norm, distinct from personal preferences. Personal ideals and wanted fertility may be measuring nearly the same thing. Expectations, however, deserve special attention because they seem particularly sensitive to how respondents think about constraints.
Sometimes the most interesting finding is not that people are easily influenced. It’s that they aren’t.
About the Authors
Heather M. Rackin, Louisiana State University
Emily A. Marshall, Franklin & Marshall College
Julia Behrman, Northwestern University