Measuring Identity

Scott Keeter, Anna Brown, and Dana Popky at Pew:

Unlike most Pew Research Center reports, where the emphasis is on original research and the presentation of findings, our goal here is to explain how we do this – that is, how we measure some of the most important core characteristics of the public, which we then use to describe Americans and talk about their opinions and behaviors.

To do so, we first chose what we judged to be the most important personal characteristics and identities for comparing people who take part in our surveys. Then, for each trait, we looked at a range of aspects: why and how it came to be important to survey research; how its measurement has evolved over time; what challenges exist to the accurate measurement of each; and what controversies, if any, remain over its measurement.

These considerations and more shape how we at Pew Research Center measure several important personal characteristics and identities in our surveys of the U.S. public. Here are some things to know about key demographic questions we ask:

• Our main religion question asks respondents to choose from 11 groups that encompass 98% of the U.S. public: eight religious groups and three categories of people who don’t affiliate with a religion. Other, less common faiths are measured by respondents writing in their answer. Our questions have evolved in response to a rise in the share of Americans who do not identify with any religion and to the growing diversity in the country’s population. (Chapter 1)
• Measuring income is challenging because it is both sensitive and sometimes difficult for respondents to estimate. We ask for a person’s “total family income” the previous calendar year from all sources before taxes, in part because that may correspond roughly to what a family computed for filing income taxes. To reduce the burden, we present ranges (e.g., “$30,000 to less than $40,000”) rather than asking for a specific number. (Chapter 2)
• We ask about political party affiliation using a two-part question. People who initially identify as an independent or “something else” (instead of as a Republican or Democrat) and those who refuse to answer receive a follow-up question asking whether they lean more to the Republican Party or the Democratic Party. In many of their attitudes and behaviors, those who only lean to a party greatly resemble those who identify with it. (Chapter 3)
• Our gender question tries to use terminology that is easily understood. It asks, “Do you describe yourself as a man, a woman or in some other way?” Amid national conversation on the subjects, gender and sexual orientation are topics on the cutting edge of survey measurement. (Chapter 4)
• In part because we use U.S. Census Bureau estimates to statistically adjust our data, we ask about race and Hispanic ethnicity separately, just as the census does. People can select all races that apply to them. In the future, the census may combine race and ethnicity into one question. (Chapter 5)
• A person’s age tells us both where they fall in the life span, indicating what social roles and responsibilities they may have, and what era or generation they belong to, which may tell us what events in history had an effect on their political or social thinking. We typically ask people to report just the year of their birth, which is less intrusive than their exact date of birth. (Chapter 6)