Improve Surveys: Cognitive Interviewing for Digital Access

According to the PLOS One article published on July 28, 2026, cognitive interviewing exposed widespread misunderstanding of translated and technical survey items, requiring revisions to nearly all tested questions. According to the PLOS One study, iterative rounds of in-language testing improved comprehension for rural Hindi speakers and revealed seven repeatable problem types that undermine digital access measurement.

• 101 cognitive interviews were conducted between 01/04/2023 and 27/04/2023 to test 118 draft questions, according to the PLOS One study published on July 28, 2026.
• According to the PLOS One study, all 44 male respondents and 81% of 57 female respondents reported owning their own phones, while 84% of men and 56% of women reported owning smartphones in the sample.
• According to the PLOS One study, average interview length was 52 minutes and nearly every question required at least one revision across three testing rounds.
• According to the PLOS One study, the team identified seven categories of question problems: terminology, complexity, low resonance of digital concepts, permission/supervision confusion, problematic response options, self-practice bias, and unclear time frames.

According to the PLOS One study, researchers ran three rounds of cognitive interviewing in rural Badaun and Jaunpur districts to test translated survey content and to iterate using daily team debriefs.

According to the PLOS One study, each draft quantitative question was asked verbatim and then followed by qualitative probes such as “Could you tell me in your own words what this question is asking? ” to diagnose comprehension mismatches and inform rewrites.

According to the PLOS One study, the 118 questions were drawn from global instruments and novel items, and the research team used back-translation and multi-person review before field testing to surface translation and cultural issues.

According to the PLOS One study, the research team discovered that formal Hindi translations often failed in practice: one respondent said, “Please speak in Hindi. What does the question mean? ” (Respondent F27), showing that literal translations can be too formal or technical for everyday use.

According to the PLOS One study, iterative testing led to concrete fixes: shorter question stems, examples naming concrete apps (YouTube, WhatsApp, PayTM), three-point or binary response formats, and event-anchored recall (for example, “yesterday” or “in the last seven days”) instead of abstract recalls like “in the past month.”

According to the PLOS One study, survey designers must test translated items with the least educated and most marginalized subgroups rather than assuming urban or more educated respondents will generalize.

According to the PLOS One study, measurement of phone ownership and access in contexts with shared devices requires separate items that capture ownership, primary use, SIM card control, and routine sharing patterns to avoid misclassification.

According to the PLOS One study, Likert scales and hypothetical community-opinion questions produced self-practice bias; simpler binary or three-point formats and direct personal-practice questions produced more reliable responses.

According to the PLOS One study, technology terms like “internet, ” “app, ” “privacy policy, ” and “hack” had low resonance and often required concrete examples or explainer boxes naming local apps and behaviors to become comprehensible.

According to the PLOS One study, cognitive interviewing exposed pervasive comprehension gaps in digital access questions and enabled targeted rewrites that improved measurement among rural Hindi speakers.

According to the PLOS One study, key operational changes that improved questions included using colloquial terminology, concrete app examples, shorter stems, event-anchored recall, and simpler response formats.

If you run multi-language cognitive interviews or need fast, secure synthesis of interview transcripts, Evidano can accelerate transcription, retain original-language terms, and produce thematic and cross-segment analyses to shorten iteration cycles; see our features page to learn more.

Ready to speed your next cognitive interviewing cycle and get reproducible thematic outputs? Try Evidano for free.