As part of the 2004 Census Test, the Coverage Research Follow-Up (CRFU) interview was conducted. In order to pretest and refine the CRFU instrument a behavior-coding project was undertaken by staff from the Statistical Research Division (SRD) in conjunction with the Decennial Statistical Studies Division (DSSD). The behavior coding method was used to evaluate the CRFU to explore the extent to which interviewers delivered the survey questions as worded and whether any response issues were present.

Field and telephone staff collected over 200 audio-taped interviews, with respondents’ permission, in the summer and fall of 2004—209 of which were used for this analysis. The identical paper-and-pencil instrument (PAPI) was used for both personal visit and telephone survey modes. Telephone interviewers from the Hagerstown Telephone Center applied a set of behavioral codes to interviewer/respondent interactions for each question.

The following section describes several facets of the results: general observations; logistic regression by survey mode; and particular questions deemed either high or medium priority for question redesign due to signs of administrative and response difficulties.