AI qualitative analysis museum visitors can convert field notes, coder protocols, and visitor transcripts into reproducible themes, age segments, and targeted exhibit recommendations. Qualitative researchers and museum evaluators face high-volume observational datasets like the one described in PLOS ONE; this post shows how AI-enabled qualitative research methods speed synthesis, surface age-related patterns, and produce actionable suggestions for exhibit design.
Key Takeaways
According to PLOS ONE (Wanča et al., published 21 August 2026), structured observations at the National Museum Prague Children’s Museum produced 720 observation records covering 2, 173 individual visitors between 16 March and 30 May 2024.
- 720 observation records were collected during 16 March–30 May 2024, covering 2, 173 visitors, according to Wanča et al. in PLOS ONE (published 21 August 2026).
- Wanča et al. reported that the exhibition’s intended 6–10 age cohort made up only 23% of child visitors, while 72% were younger children, indicating a clear audience mismatch (PLOS ONE, 21 August 2026).
- Observers recorded an average dwell time of 7 minutes per observation unit, with U1 and U5 averaging 14 minutes and U2 about 2.5 minutes, per Wanča et al. (PLOS ONE, 21 August 2026).
- Quote from Wanča et al.: "The primary target group constituted only 23% of the child visitors, with 72% being younger children, " and the authors summarize the exhibition concept as "learning about the world from a different perspective than provided by the school curriculum."
What happened: Prague Children’s Museum observations and key measures
Answer: Wanča et al. (PLOS ONE, published 21 August 2026) ran a structured observational study at the Children’s Museum in Prague from 16 March to 30 May 2024 and recorded 720 observations covering 2, 173 visitors.
According to Wanča et al. in PLOS ONE (published 21 August 2026), observers used a standardized protocol with 19 trained observers to collect scalar measures (0–3) for interaction dimensions such as creating, communication, focus, and motion, plus open-ended field notes.
According to Wanča et al. (PLOS ONE, 21 August 2026), the dataset shows a marked skew to preschool-aged visitors: the intended 6–10 cohort was only 23% of child visitors while 72% were younger than the target age.
According to Wanča et al. (PLOS ONE, 21 August 2026), engagement scores averaged 2.06 (SD = 0.83) for children, 1.60 (SD = 0.93) for adults, and facilitators averaged 0.27 (SD = 0.63), suggesting facilitators rarely entered interactions except at U5.
Findings snapshot table
| Date / Period | Metric | Value | Implication |
|---|---|---|---|
| 16 Mar–30 May 2024 | Observation records | 720 | Protocol-based observational sample large enough for quantitative ordinal models (Wanča et al., PLOS ONE, 21 Aug 2026) |
| 16 Mar–30 May 2024 | Individual visitors observed | 2, 173 | Enables group-composition analysis and age-segmentation (Wanča et al., PLOS ONE, 21 Aug 2026) |
| Data collection period (published 21 Aug 2026) | Primary target group (age 6–10) | 23% of child visitors | Exhibition design and outreach are misaligned with actual audience (Wanča et al., PLOS ONE, 21 Aug 2026) |
| Data collection period (published 21 Aug 2026) | Younger children proportion | 72% of child visitors | Most visitors may struggle with constructivist, text-light exhibits (Wanča et al., PLOS ONE, 21 Aug 2026) |
| During observations | Average dwell time per unit | 7 minutes (U1/U5 ≈ 14 min; U2 ≈ 2.5 min) | Dwell-time differences identify high-value hubs and weak spots for redesign (Wanča et al., PLOS ONE, 21 Aug 2026) |
Implications for museum researchers and UX teams
Answer: The PLOS ONE study implies researchers should combine age-segmentation with unit-level interaction metrics to prioritize exhibit changes.
- Design alignment: According to Wanča et al. (PLOS ONE, 21 August 2026), when 72% of child visitors are younger than the intended cohort, exhibit complexity and wording must be simplified or age-targeted signage added.
- Measurement: Wanča et al. used ordinal scales and mixed-effects ordinal regression across units; museum evaluators should replicate scalar coding for interaction types to compare units statistically.
- Facilitator strategy: Wanča et al. (PLOS ONE, 21 August 2026) found facilitator engagement is context-dependent (high in U5), suggesting scripts and role definitions can increase pedagogical impact.
- Sampling and slots: Wanča et al. (PLOS ONE, 21 August 2026) observed slot effects (morning slots produced higher interaction), so schedule-based sampling and design testing across time-of-day are essential.
How Evidano helps: map the PLOS ONE workflow to AI-enabled qualitative research
Problem: Large, mixed-format observational data is slow to synthesize
Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents.
Solution: Use Evidano’s ingestion pipeline to import observation protocols, field notes, and CSV coder sheets and run automated thematic extraction to surface high-frequency interaction themes and age-segmentation patterns.
Problem: Manual coding required for scalar measures and unit comparisons
Solution: Evidano supports structured codebooks and hierarchical codes so teams can map the PLOS ONE 0–3 Likert items into reproducible code schemas, then run cross-segment frequency and ordinal analyses to reproduce unit-level contrasts.
Problem: Audio notes and interviews need transcription and PII handling
Solution: Evidano’s speech-to-text pipeline transcribes audio with a custom dictionary and optional PII redaction, mirroring the PLOS ONE workflow where field notes and occasional interviews supplement protocol data.
Problem: Teams need fast hypothesis testing and visualizations
Solution: Evidano produces co-occurrence networks, word clouds, and cross-tab visualizations so researchers can test hypotheses such as age association with creativity or focus (as in Wanča et al., PLOS ONE, 21 Aug 2026) and export charts for stakeholders.
Learn more
For a feature overview relevant to this workflow, see Evidano’s features page.
For teams onboarding observational protocols into AI workflows, Evidano’s platform shortens synthesis time while preserving audit trails and coder provenance.
FAQ: AI qualitative analysis museum visitors
How can AI reproduce the structured observational protocol used by Wanča et al. in PLOS ONE?
Answer: AI can mirror the PLOS ONE protocol by ingesting the scalar coder sheets and training codebook mappings to the 0–3 interaction scales.
Supporting detail: Wanča et al. used standardized Likert items and area diagrams between 16 March and 30 May 2024; feeding those sheets into an AI platform lets you generate code frequency tables and run the same ordinal regressions with exported coder assignments.
Can AI detect the same age-segmentation pattern (23% vs 72%) reported in the PLOS ONE study?
Answer: Yes, AI-enabled thematic and demographic extraction can reproduce age-segmentation once age tags are present or inferred reliably from coder fields.
Supporting detail: Wanča et al. reported that the primary 6–10 cohort was 23% while 72% were younger; an AI pipeline that validates and aggregates age-labels will yield the same percentages and enable subgroup comparisons.
What direct quotes should teams extract and preserve from field notes?
Answer: Preserve succinct, attribution-ready quotes that capture conceptual framing and surprising findings.
Supporting detail: Wanča et al. include usable quotes such as "The primary target group constituted only 23% of the child visitors, with 72% being younger children" and the exhibition’s concept phrased as "learning about the world from a different perspective than provided by the school curriculum, " both of which are extractable and citable.
Is AI analysis ethical for observational museum research?
Answer: AI-assisted analysis is ethical when data are de-identified, consent considerations are respected, and interpretations remain grounded in methods.
Supporting detail: Wanča et al. used semi-overt, non-participant observation and did not collect personal identifiers; researchers must follow equivalent local ethics practices and treat results as inferential rather than diagnostic.
Conclusion & Next Steps
The PLOS ONE study by Wanča et al. (published 21 August 2026) shows how a large, protocol-driven observational dataset (720 records, 2, 173 visitors, observed 16 March–30 May 2024) can reveal audience mismatches and unit-level strengths in a constructivist museum.
For qualitative teams, the immediate next step is to map observation protocols and field notes into a reproducible codebook and run cross-unit and age-segmentation analyses to prioritize design changes.
Evidano accelerates that workflow by ingesting protocols, transcribing audio with speech-to-text, producing thematic and cross-segment reports, and preserving coder provenance.
If you want to test this workflow on your dataset, Try Evidano for free.
Topics
- AI qualitative analysis museum visitors
- qualitative analysis of museum visitors
- museum visitor observation analysis
- AI-enabled qualitative research
Keep reading
- Commentary on NewsAI-assisted qualitative analysis for museum observationsHow AI-enabled qualitative analysis accelerates coding and insight from museum visitor observations. Learn methods and stats from PLOS ONE, and try Evidano.
- Commentary on NewsBoost Exhibit Design: AI Qualitative Analysis for MuseumsUse AI qualitative analysis for museums to convert observational studies into targeted exhibit changes. Learn from PLOS ONE findings and try Evidano to accelerate insight.
- Commentary on NewsImprove Museum Visitor Research: AI Qualitative AnalysisAI qualitative analysis for museums: convert visitor observations into themes, counts, and actionable design changes. Learn from a PLOS ONE study and see how Evidano helps.