- No significant differences in GHQ-12 or SF-12 mental and physical scores across web, social interviewer, and nurse-administered self-completion contexts.
- No statistically significant mode effects on interviewer-recorded doctor diagnoses, including psychiatric conditions; reported odds ratios were non-significant across mode comparisons.
- Findings support embedding self-completion modules within interviewer-administered surveys, while advising caution when designing mixed-mode surveys and interpreting self-reported health data.
JMIR Hum Factors. 2026 Oct 16;13:e89896. doi: 10.2196/89896.
ABSTRACT
BACKGROUND: Probability-based surveys remain essential to epidemiology, but expanding mixed-mode data collection, which combines web self-completion and interviewer-administered surveys, raises concerns that administration mode and interviewer presence may influence health reporting. These concerns warrant further investigation of interviewer-related biases and mode differences in health reporting.
OBJECTIVE: This study assessed whether responses to the 12-item General Health Questionnaire (GHQ-12) and 12-item Short Form Health Survey (SF-12), both self-completed, differed by interviewer presence and role: online without an interviewer, during a social interviewer visit, or during a nurse visit. It also assessed mode effects on doctor-diagnosed conditions recorded by interviewers in computer-assisted personal interviewing (CAPI) and when self-completed online. We further examined possible variation by sex and age.
METHODS: The analysis used randomized mode allocation embedded within wave 12 (2019) of the UK Household Longitudinal Study Innovation Panel. After applying relevant exclusions, the final analytic sample comprised 1111 households (1779 participants) across the 3 randomly assigned survey modes: web-based self-completion, interviewer-administered CAPI, or nurse-administered CAPI. The sample was predominantly White British, with a broadly balanced sex distribution and participants from diverse age, employment, education, and household income groups. Participants did not always complete the survey in their randomly assigned mode. Effects were estimated using a two-stage residual inclusion instrumental variable approach, with randomized mode as an instrument for actual mode, to address confounding arising from nonrandom selection into actual mode.
RESULTS: No statistically significant differences were observed across self-completion contexts for the GHQ-12 or the SF-12 mental and physical component scores. We found no statistically significant evidence of mode effects on doctor-diagnosed conditions. For example, no statistically significant mode effects in psychiatric conditions were identified across the comparisons: nurse relative to social interviewer (odds ratio [OR] 0.72, 95% CI 0.40-1.29; P=.53), web relative to social interviewer (OR 0.45, 95% CI 0.17-1.18; P=.14), and web relative to nurse (OR 0.63, 95% CI 0.27-1.49; P=.59). There was little evidence overall that mode-related differences varied by age or sex.
CONCLUSIONS: Our findings may support embedding self-completion modules within interviewer-administered surveys for the collection of SF-12 and GHQ-12. Meanwhile, potential mode-related differences should be considered when designing mixed-mode surveys and interpreting self-reported health data.
PMID:42886760 | DOI:10.2196/89896
Share Evidence Blueprint
Save to Google Notes
Search Google Scholar
Save as PDF
⭐ My Revision List
