The Effects of Questionnaire Design on Reporting of Detailed Hispanic Groups
in Census 2000 Mail Questionnaires
Elizabeth A. Martin
KEY WORDS: decennial census, mail questionnaire, example effects, measurement error
This research note reports results of an experimental replication of a 1990-style mail short form census questionnaire during Census 2000. Panels of households were randomly assigned to receive either 1990-style or 2000-style mail questionnaires in order to evaluate the effects of questionnaire design changes on responses to questions about race and Hispanic origin. The questionnaire changes included dropping examples from both questions. This note considers alternative hypotheses about how examples affect recall and comprehension, and compares experimental panels to assess the effects of examples on detailed Hispanic reporting. There were fewer write-in responses of detailed Hispanic groups and more write-ins of generic Hispanic identities (“Hispanic,” “Latino,” or “Spanish”) in 2000-style questionnaires. This suggests that examples helped respondents understand the specificity of response that was intended by the question; other design differences probably also contributed to the difference in reporting. Some implications of the findings for use of examples in surveys and for comparisons of 1990 and 2000 census data are discussed.
CITATION: Elizabeth Martin. (2002) “The Effects of Questionnaire Design on Reporting of Detailed Hispanic Origin in Census 2000 Mail Questionnaires,” Public Opinion Quarterly 66(4):582-593.