This post refracts a July 1, 2026 PLOS ONE study through the lens of AI-enabled qualitative research. The primary keyword (qualitative analysis of APC Trap) frames what you’ll learn: which themes (n=8) emerge from 154 substantive open comments, what that implies for UX/research teams and policy, and a reproducible 7-step workflow you can run in Evidano (www.evidano.com) to replicate or extend the study. Data were collected Oct 31–Nov 21, 2023 from four U.S. R1 institutions; 322 respondents completed the survey and 154 provided substantive free-text used for coding. If your team analyzes open-ended responses, reports, or policy feedback, this post shows how to go from messy comments to defensible, segment-aware insights fast.
Key Takeaways
Evidano is an AI-powered qualitative data analysis platform that helps teams reproduce this PLOS ONE study’s pipeline and analyze open-ended responses securely. The PLOS ONE study identifies an "APC Trap", a cognitive bind where researchers feel obligated to pay article processing charges (APCs) despite recognizing inequities and unsustainability, based on 154 substantive open responses coded into eight themes. The qualitative analysis shows 90% of substantive responses contained at least one APC Trap code and reveals layered, co-occurring concerns across disciplines and publishing types.
- Published: July 1, 2026, DOI: 10.1371/journal.pone.0351430, source: PLOS ONE.
- Dataset: survey collected Oct 31–Nov 21, 2023; 322 completed surveys and 154 substantive open-ended responses were coded inductively.
- Eight APC Trap codes were identified and applied at the respondent level; 90% of substantive responses (n=139) contained ≥1 APC Trap code, 62% had 1–2 codes, and 28% had ≥3 codes.
Findings snapshot
| Metric | Value | Note / implication |
|---|---|---|
| Completed surveys | 322 | Full quantitative instrument |
| Substantive open responses (qualitative) | 154 | Basis for the APC Trap coding |
| APC Trap codes identified | 8 | Financial Concern; Diversity/Equity; Threshold of Reasonableness; Restriction of Decisions; Grants Don’t Cut It; Academic Obligation; Academic Labor; Quality/Rigor |
| Data collection window | Oct 31–Nov 21, 2023 | Multi-site IRB exempt protocol (IRB Protocol Number: 23–0216) |
| Publication date | July 1, 2026 | PLOS One |
What happened: methods in plain language
The authors used purposive sampling across four U.S. public R1 universities to collect tailored survey responses and combine open-ended prompts into analyzable qualitative data. The authors ran six tailored Qualtrics surveys (by discipline and OA type), aggregated two open-ended prompts per respondent into a single qualitative response (n=154), and applied inductive thematic coding at the respondent level. Inductive thematic coding produced eight APC Trap codes that coders applied yes/no per respondent using consensus (at least four coders) with iterative Code Book refinement, and analyses included code frequencies, co-occurrence matrices, and disciplinary/institutional cross-tabs; automated sentiment tools were tested but found unreliable for these nuanced responses.
Implications: qualitative analysis of APC Trap for researchers & teams
For UX & qualitative research teams
UX and qualitative research teams should code at the respondent level and support multi-label assignment to capture layered, co-occurring themes. Open-ended responses reveal co-occurring themes such as Financial Concern plus Restriction of Decisions, therefore teams should report co-occurrence and segment-level differences (discipline, institution, OA type) to inform targeted interventions.
For library/Open Science leads
Library and Open Science leads should treat APC policies as interacting with perceived inequity and monitor who benefits from institutional funds. Authors with easier APC access were more likely to flag Diversity/Equity concerns, and the study finds Hybrid authors showed more APC Trap codes on average, so monitoring Hybrid versus Gold publishing patterns is important.
For policy makers & funders
Policy makers and funders should consider that subsidies and transformative agreements can hide costs but may not solve inequities, qualitative evidence strengthens the case for alternative collective funding models. Program evaluations should combine grant accounting (who pays) with open comments to reveal moral and operational burdens on researchers.
Do more, faster with Evidano (mapped to this study)
Ingest & prepare
Evidano ingests Qualtrics CSVs and preprocesses free-text for reproducible analysis. Upload surveys and spreadsheets and the free-text responses directly into Evidano for reproducible preprocessing including deduplication, respondent IDs, and institutional tags.
Codebook-first or inductive coding
Evidano supports both codebook-first and inductive methods so teams can mirror the study’s approach. Import an initial Code Book (CSV) or run AI-assisted inductive theme discovery; Evidano produces hierarchical codes and lets you lock or refine them with batch re-coding.
Cross-segment & co-occurrence analysis
Evidano runs cross-segment frequency counts and generates co-occurrence networks to surface code pairings as the paper did. Run cross-segment frequency counts (by discipline, institution, OA type) and generate co-occurrence networks to surface code pairings (for example Financial Concern plus Restriction of Decisions).
Explainability & evidence
Evidano attaches representative quotes to every theme and reports the distribution so teams can quote participants defensibly. Evidano shows both counts and percentages for each theme and links to representative responses for transparency.
Secure, research-first LLMs
Evidano uses research-tuned LLMs and encrypts data so customer data is not used to train third-party models. This model is suitable for sensitive policy or institutional data where confidentiality and control matter.
Checklist: reproduce this qualitative analysis in Evidano (7 steps)
Follow these steps to replicate the study’s core outputs in Evidano.
- 1) Upload Qualtrics CSVs and map respondent IDs plus institution and discipline tags.
- 2) Normalize and merge the two open-response fields into a single text field per respondent.
- 3) Run an inductive theme extraction to surface candidate codes and export suggestions.
- 4) Import or edit a Code Book then run batch coding and resolve disagreements with Evidano’s consensus reviewer tool.
- 5) Generate frequency tables and cross-segment comparisons (Gold vs Hybrid, HSS/HM/NSE).
- 6) Produce a co-occurrence network and export representative quotes per code for transparency.
- 7) Create stakeholder-ready visuals and an executive brief; iterate with AI chat over your documents to draft policy recommendations.
FAQ: qualitative analysis of APC Trap
Can automated sentiment tools handle nuance in these responses?
Automated sentiment tools cannot reliably handle nuance in these responses, the PLOS ONE study found NVivo and SentiStrength produced mixed or unusable outputs because of compound sentiments and sarcasm. Use AI-assisted human-in-the-loop coding to preserve nuance and validate automated outputs with human review.
How do I compare segments reliably?
You should compare segments using respondent-level multi-label coding, then run normalized frequency comparisons and bootstrapped confidence intervals for small groups. Use respondent-level multi-label coding followed by normalized frequencies and bootstrapped CIs for small groups; Evidano automates these cross-segment summaries.
Is my qualitative data safe in AI tools?
Data safety depends on provider policies; the study recommends checking encryption and training-use policies before uploading sensitive data. Evidano encrypts data and does not feed customer data to third-party model training, which is suitable for institutional or policy research.
Conclusion & next steps
The PLOS ONE paper (published July 1, 2026) offers a replicable qualitative frame, the APC Trap, useful for anyone measuring the lived effects of publishing costs. If your team analyzes open comments, policy feedback, or interview transcripts, apply respondent-level multi-label coding, co-occurrence analysis, and segment comparisons to reveal the layered bind the authors describe.
- Ready to pilot? Reproduce the study’s pipeline end-to-end in Evidano and produce shareable visuals and evidence-backed recommendations quickly: Try Evidano for free.
- Ethics note: this post summarizes research findings for methodological and operational use; follow IRB and consent guidance when reusing participant data.