Researchers at four U.S. R1 universities report a binding cognitive dissonance around article processing charges (APCs): authors want open access but feel constrained by costs and career pressures. This post translates the PLOS One study findings into a clear, reproducible approach for qualitative analysis of APCs: what to code, what cross-segments matter, and how to speed synthesis with Evidano (Evidano). You will get a compact workflow to reproduce the study’s inductive coding and co‑occurrence checks, and practical steps to produce stakeholder-ready outputs in days rather than weeks.
Key Takeaways
The PLOS One APC Trap study identifies eight interlocking qualitative codes that reveal cost-driven cognitive dissonance among R1 authors, and those codes can be reproduced with a human-in-the-loop inductive coding workflow. Use per-respondent aggregation, inductive codebook generation, multi-coder consensus rules, and co‑occurrence analysis to reproduce the study’s findings; tools such as Evidano speed that pipeline while preserving nuance.
points: ["Dataset: survey collected Oct 31–Nov 21, 2023 (n=322 total; 154 substantive open responses used for qualitative coding).", "Main finding: 90% of substantive responses contained at least one APC Trap code; eight codes were identified via inductive thematic coding.", "Practical pipeline: per-respondent aggregation, multi-coder consensus (3+ reviewers rule), code frequency and co‑occurrence matrices reproduce the study’s outputs within days with AI-assisted tooling."]
Fast take: APC Trap in brief
The APC Trap study identifies eight interlocking codes that produce moral and operational tensions about APCs among R1 authors, reported in PLOS One on July 1, 2026. Read the original paper: PLOS One.
points: ["Dataset: survey collected Oct 31–Nov 21, 2023 (n=322 total; 154 substantive open responses used for qualitative coding).", "Main finding: 90% of substantive responses contained ≥1 APC Trap code; eight codes identified via inductive thematic coding.", "Why it matters: APCs shape publication choices, equity outcomes, and perceived research quality, even at elite US institutions."]
Snapshot: numbers & dates
| Metric | Value | Source | Note |
|---|---|---|---|
| Survey period | Oct 31–Nov 21, 2023 | PLOS One | Recruitment and Qualtrics deployment |
| Respondents (completed substantive portion) | n = 322 | PLOS One | 322 consented & completed substantive portion |
| Substantive open-ended responses analyzed | n = 154 (47.8%) | PLOS One | Responses aggregated per respondent for coding |
| Codes identified | 8 APC Trap codes | PLOS One | Inductive thematic coding with multi-coder consensus |
| Article published | July 1, 2026 | PLOS One | doi:10.1371/journal.pone.0351430 |
What happened: qualitative analysis of APCs and methods
The methods used purposive sampling across four U.S. R1 public universities, six tailored Qualtrics surveys (Gold vs Hybrid OA by three disciplinary groups), and two open-ended prompts producing 154 substantive responses aggregated per respondent.
points: ["Coding approach: inductive thematic coding that produced eight codes and a Code Book, with each response coded yes/no per code.", "Validation: four-or-more coder agreement required for automatic assignment; cases below that threshold were adjudicated by group discussion and consensus.", "Analyses run: frequency counts, co‑occurrence matrix, and demographic cross-tabs (discipline, institution, OA type).", "Automated sentiment tools tested (NVivo Auto Sentiment, SentiStrength) produced unreliable results for this nuanced dataset."]
Key methodological takeaway: human-in-the-loop thematic coding exposed layered moral and operational tensions that keyword and sentiment tools missed, which is precisely the scenario where AI-assisted qualitative platforms accelerate rigor without losing nuance.
Implications for researchers, libraries, and policy teams
For UX & qualitative teams
For UX and qualitative teams: expect multi-code responses and plan for co-occurrence and segment comparisons when analyzing open-ended survey text.
Measure both prevalence (how many responses mention a code) and intensity (how many codes per response), the study found 62% had 1–2 codes and 28% had three or more codes.
For library & policy leads
For library and policy leads: APC policy evaluation needs qualitative evidence, including codebooks that capture equity and career-pressure narratives to justify funding models or transformative agreements.
Cross-tab discipline by funding source matters: in this study, HSS authors had higher rates of Financial Concern and Restriction of Decisions.
For journal editors & publishers
For journal editors and publishers: qualitative feedback exposes reputational and ethical risks, such as perceptions of pay-to-publish and harms to diversity, that raw APC averages do not surface.
Use coded open responses to guide transparent price-and-service reporting.
Do more, faster with Evidano
Import & prepare (minutes)
Evidano is an AI-powered qualitative data analysis platform that preserves respondent IDs and metadata so you can reproduce the same per-respondent aggregation used in the PLOS study.
Bring the Qualtrics CSV or spreadsheet into Evidano; the platform automatically cleans non-substantive answers (for example, “No comment”) and normalizes fields such as institution, discipline, and OA type.
Thematic coding & cross-segment analysis (hours)
Use Evidano to auto-generate an inductive codebook from the corpus and then apply AI-assisted coding across responses, accepting or rejecting suggestions to build consensus coding rules that mirror a multi-coder workflow.
Run co-occurrence networks and cross-segment frequency tables, for example codes by discipline and by OA type, to replicate the study’s co-occurrence and demographic cross-tabs.
Evidano provides hierarchical code and subcode visualizations plus an interactive co‑occurrence graph to surface common code pairings such as Financial Concern plus Restriction of Decisions.
Validate, synthesize & share (hours)
Use AI chat over your coded corpus to draft annotated results, pull illustrative quotes, and export a Code Book, then have human reviewers confirm nuance where sentiment tools struggled.
Export clickable reports and visualizations for stakeholders such as funding committees, library councils, and editorial boards; Evidano encrypts your data and does not use it to train third-party models.
Start here: Evidano for a demo and to test importing a small CSV from your last survey.
Checklist: 7 steps to reproduce this study (practical)
This checklist lists seven practical steps to replicate the PLOS One workflow and produce actionable outputs:
points: ["1) Export Qualtrics responses as CSV and aggregate per respondent.", "2) Import into Evidano and tag metadata (institution, discipline, OA type).", "3) Run inductive codebook generation; review and lock eight to twelve candidate codes.", "4) Apply AI-assisted coding; resolve disagreements with a three-plus reviewer rule or manual adjudication.", "5) Produce code frequency, co‑occurrence matrix, and discipline-by-code cross-tabs.", "6) Pull representative quotes per code and generate an executive summary via AI chat; verify quotes manually.", "7) Deliver a stakeholder report (PDF plus interactive dashboard) with recommended policy actions."]
FAQ: qualitative analysis of APCs
Q: Why did automated sentiment tools fail in the PLOS study?
A: Automated sentiment tools failed because they mis-scored mixed or sarcastic responses and evaluated isolated words rather than coded rhetorical structure. Human-in-the-loop thematic coding captured layered sentiments reliably in this dataset.
Q: Can I compare segments (discipline × OA type) reliably?
A: Yes, you can compare segments reliably if you use per-respondent aggregation and control for uneven subgroup sizes. The PLOS study shows Hybrid authors and HSS respondents had distinct code profiles; cross-tabs and co-occurrence analyses help surface those differences.
Q: How do I preserve nuance while using AI?
A: Preserve nuance by using AI to propose codes and highlight candidate quotes, then validating suggestions with two to four human reviewers. Evidano supports this mixed workflow and preserves an audit trail for reproducibility.
Conclusion: what to do next
The PLOS One APC Trap analysis, published July 1, 2026, shows how open responses reveal ethical and operational strains that numbers alone miss, and a reproducible qualitative pipeline grounds policy moves in people’s experience.
points: "Start small: import one survey into Evidano, run inductive coding, and produce a co‑occurrence map to brief decision-makers within a week.", "Secure trial and demo: [Try Evidano for free to see how Evidano reproduces the APC Trap codebook and cross-segment outputs."]
Ethics note: this post interprets a published research article for research-use workflows; it does not provide clinical or legal advice.