Interviewer bias

The wider academic term, the interviewer effect, is the distortion of responses to interviewer-administered data collection that arises from differential reactions to interviewers' social style and personality, or to how they present particular questions (Wikipedia).

The hiring meaning is the interviewer's preferences distorting a judgement of a candidate. Employer guidance for that case recommends standardised questions asked of every candidate (GOV.UK, 4 March 2026). Research uses the same remedy, but what gets damaged is the data.

Two neighbouring terms are often confused with it. Social desirability bias sits with the participant, who wants to look good. Interviewer bias is often what triggers it. Confirmation bias sits with the interviewer, who steers towards an answer they already expect, and it is one route to interviewer bias.

Interviewer bias changes answers most when the topic is private and the participant feels socially distant from the interviewer. On socially easy opinion questions the effect can nearly vanish. Two Indian studies show the two ends.

Where interviewer bias is strong and weak

In September 2016, two identical household surveys of mothers ran in two Bihar districts, one with all-male and one with all-female interviewers. Of 156 comparisons across 54 indicators, 58 (37.2%) differed significantly. Male interviewers produced 39 lower estimates and 19 higher ones; 39 of 58 is 67% (Vollmer et al., PLOS ONE, 2021). The gaps clustered on birth preparedness, delivery care, contraception, breastfeeding and reproductive health, while child-health topics showed less gap. A frontline health worker in the study's follow-up explained: "There is no sense of shame if a woman is talking to another woman. But women in the village will definitely feel ashamed while talking to men."

In the Rajasthan study published in 2025, phone numbers for 2,738 men were randomly assigned to male or female interviewers. Answers were near-identical on most gender-attitude questions, and only one question differed noticeably (Hathi et al., Survey Insights, 2025).

Both are structured surveys, not commercial focus groups. The Rajasthan study had only men and eight interviewers. Attitudinal, sensitive, ambiguous, complex and open-ended questions are more prone to interviewer effects (West and Blom, 2017). Qualitative interview questions are mostly open-ended and attitudinal, so the inference is that they are at least as exposed. Plan around topic sensitivity.

Interviewer bias inside an IDI or focus group looks like a question that carries its answer, a reaction that rewards one kind of answer, or a moderator whose voice, accent, dress, age or values change what people will say (Roller, 2019). Many leading questions compromise validity (Nielsen Norman Group, reviewed 15 July 2026).

Settle four things in the brief: moderator match, fixed wording for core questions, neutral probes and a review of recordings. The table below is a recommendation. The High row is an inference from the Bihar gender result, extended to household roles, caste and religion. The Low row is an inference from Rajasthan. The Medium row is a judgement.

Test for the Medium pilot. Pass: the two moderators' transcripts give the same answers on the core questions. Fail: answers differ on two or more core questions, in which case assign by match.

Paste this into your market research brief:

1. Core questions are asked in the guide's exact wording and order in every session.
2. Probes are neutral ("Tell me more", "What do you mean by that?") and never suggest an answer.
3. The moderator reacts the same way to positive and negative answers. Pass: the reviewer finds no difference in tone or praise between positive and negative answers in 10 sampled answers of each.
4. For sensitive topics, the moderator is matched to participants as agreed in the brief. Pass: every sensitive-topic session has the moderator the brief specifies. Fail: any session that does not.
5. The moderator writes a short note after each session on anything that may have shaped answers.
6. Participants consent to recording and review before the session. Pass: consent is on record for every participant before recording starts. See informed consent in research.
7. A second person reviews a sample of recordings against this list.

To test point 7, pick 20 questions at random from each moderator's recordings. Pass: at most one presumes or offers an answer. Fail: two or more, or one moderator clearly above colleagues, in which case retrain or reassign before more fieldwork. The threshold is a recommendation, not a published standard.

Write the guide itself using the in-depth interview method and focus group discussion guide. The format also changes who is in the room: see online vs in-person focus groups.

We recruit participants for in-depth interviews and focus groups, online or in person. Studies may be run by us, by research partners or by other research organisations. We can supply moderators, recruit participants by city and run sessions in languages other than English, so moderator match and participant profile can be settled together in one brief. For group sessions, see how to conduct focus groups in India.

Put the topic, the language and who will moderate in the same brief, so the participant profile and the moderator match are settled together. We offer research buyers a free pilot, sample or trial; ask us what it covers for your study.

Share your research brief with our team by email or WhatsApp through Qualfacto. We aim to reply within one business day (Monday to Friday, 10:00 AM to 6:00 PM India Standard Time).