Interviewer Error Within the Face-to-Face Food Frequency Questionnaire in Large Multisite Epidemiologic Studies

Area of Science:

• Epidemiology
• Nutrition Science
• Survey Methodology

Background:

• Interviewer error is a known issue in surveys.
• Its impact on face-to-face food frequency questionnaires (FFQs) in large epidemiologic studies is under-researched.

Purpose of the Study:

• To investigate the prevalence and types of interviewer error in face-to-face FFQs.
• To identify factors contributing to interviewer error and assess its impact on dietary intake estimates.

Main Methods:

• Utilized audio-recorded dietary data from the China Multi-Ethnic Cohort (2018-2019).
• Identified error-prone interviews using outlier detection.
• Reviewed interviews to categorize and quantify interviewer errors (falsification, coding, reading deviation).

Main Results:

• 7.96% of questions in error-prone interviews contained interviewer errors.
• Most interviewers (98.29%) and respondents (73.71%) had at least one error.
• Errors were more frequent in complex questions (e.g., food quantification, seasonal foods).
• Correcting errors reduced estimated mean and standard deviation of food intakes.

Conclusions:

• Interviewer error significantly impacts face-to-face FFQ data quality.
• Targeted monitoring of high-error interviewers and respondents is needed.
• Simplifying questionnaire design may reduce survey burden and errors.