Questionnaire Research Report - Series A, Report 3: Results of Three Classroom Experiments with Census of Population and Housing Questionnaires
Donny Rothwell
KEY WORDS: item nonresponse; census forms: FOSDIC data processing; survey format; instruction format; instruction placement
ABSTRACT
Sixteen variants of a standard census-like form, including two alternative new formats, were tested. The forms varied in print color, placement of additional instructions, data processing constraints for age reporting (Film Optic Scanning Device for Input to Computers (FOSDIC) vs. non-FOSDIC), and format. Each respondent (n = 339) completed one variant of the form, as well as an evaluation form. Results showed that: 1) format, spacing and positioning of questions could have a strong effect on item response rates; 2) carryover effects from the inclusion of a difficult or seemingly unreasonable item affects the completion rate of items that follow it; 3) color printing had little effect on response rate or consistency of responses;
4) inclusion of additional instructions had no effect on completion; 5) no difference in responses occurred when a summary question vs. an expanded group of detailed questions was used; and
6) the form which did not call for FOSDIC age reporting was easier to fill out. Recommendations for the form are included.
CITATION:
