BEHAVIOR CODING ANALYSIS REPORT:
Evaluating Bilingual Versions of the Non-Response Follow-Up (NRFU) for the 2004 Census Test
Jennifer Hunter, Ashley Landreth
KEY WORDS: pretesting, behavior coding, non-response follow-up, bilingual pretesting, demographics, mCAPI
ABSTRACT
This report documents results of the evaluation of the NRFU data collection using the hand-held computer in the 2004 Census Test conducted in Queens, New York. Audiotapes of 220 NRFU interviews conducted in English and in Spanish were made during the data collection and subsequently behavior coded by bilingual interviewing staff at the Tucson Telephone Center. Training was accomplished by CSMR staff with the assistance of a Spanish-speaking contractor. Results showed that, in general, NRFU interviewers deviated dramatically from the standardized questionnaire. All of the questions in the NRFU instrument exceeded the commonly-accepted error threshold that suggests “ideal” interactions for interviewer and respondent behavior should occur at least 85 percent of the time. Questions administered in Spanish had consistently lower rates of ideal question-asking behavior than questions asked in English, though these differences were not always significant. The trend was also for questions for Person 1 to be asked properly more often than for other people in the household; that is, interviewer correctly administered the question the first time it was asked and made more errors in later administrations. This report also contains a question-by-question evaluation of the questionnaire items, and includes recommendations for revising them.
CITATION:
