Researchers at four U.S. R1 universities describe an "APC Trap", a bind where obligation to publish open access collides with rising article processing charges. This qualitative analysis (n=154 substantive open responses from a 322-respondent survey) published July 1, 2026 documents eight recurring codes and shows 90% of commenters exhibited at least one APC Trap code. Read the original study at PLOS ONE. If you analyze transcripts, survey free text, or stakeholder comments, this post shows how thematic, co-occurrence, and cross-segment analysis turns those 154 responses into reproducible insight.
Key Takeaways
This qualitative study shows that APCs create a widespread "APC Trap" for authors, documented in 154 substantive open responses and published July 1, 2026.
- 90% of substantive commenters (n=139 of 154) exhibited at least one APC Trap code, indicating the phenomenon is widespread in the sample.
- Financial Concern was the most common code (34%), and Diversity/Equity was close behind (32%).
- Data come from a 322-respondent survey conducted Oct 31–Nov 21, 2023 across four U.S. R1 public universities, with 154 open responses analyzed.
Findings snapshot
| Metric | Value | Source / Note |
|---|---|---|
| Publication | July 1, 2026 | PLOS ONE article |
| Survey period | Oct 31–Nov 21, 2023 | Qualtrics instrument (four U.S. R1 institutions) |
| Respondents (completed survey) | 322 | All respondents across 4 institutions |
| Substantive open responses analyzed | 154 (47.8%) | Aggregated per respondent for coding |
| APC Trap codes identified | 8 | Inductive thematic coding with consensus rule (≥4 coders) |
| Respondents with ≥1 APC code | 90% (n=139) | APC Trap widely present among commenters |
| Most common codes | Financial Concern 34%; Diversity/Equity 32% | See Code Book in source |
| Disciplinary & format signals | HSS: +12.5% avg codes; Hybrid articles: +13% avg codes | Hybrid authors more likely to report 'Grants Don’t Cut It' |
What happened, how the qualitative analysis was done
The authors performed an inductive thematic analysis on free-text answers from tailored Qualtrics surveys, consolidating responses, assigning IDs, and coding by multiple researchers.
Eight APC Trap codes emerged after iterative refinement and consensus, with a code applied when four or more coders agreed, and responses were analyzed at the respondent level (n=154 substantive responses).
- Sample: faculty authors at University of Colorado Boulder, UMass Amherst, University of Pittsburgh, University of Tennessee Knoxville (R1 public institutions).
- Key outputs: code assignments per respondent, co-occurrence table, demographic cross-tabs (discipline, institution, OA type).
- Notable method note: automated sentiment tools (NVivo, SentiStrength) produced unreliable results on this dataset, highlighting limits of off-the-shelf sentiment analysis for nuanced academic rhetoric.
Why this matters for researchers, UX teams, and policy analysts
For research managers & librarians
For research managers and librarians, the APC Trap indicates budget strain and altered author choices that require targeted funding and transparent reporting.
Budget signals: 34% of respondents described personal financial strain; many reported grant logistics and timing blocked APC payments.
Decision impact: APCs restrict author choices (journal selection, resubmission behavior) and can shift publication venues away from ideal outlets.
Policy takeaway: Blanket transformative agreements may obscure costs, they do not eliminate the APC Trap; targeted funding and transparent APC reporting remain critical.
For UX / product researchers who run open text studies
For UX and product researchers, academic free-text combines support for open access with outrage about APCs, which complicates automated sentiment analysis.
Complex sentiment: academic responses combine support for OA with outrage about APCs, automated sentiment tools misread this nuance.
Segmented insight matters: HSS authors reported higher per-author APC anxiety; Hybrid vs Gold OA comparisons reveal different pressure points.
Actionable focus: design surveys and follow-ups to surface funding mechanics, not just attitudes.
For policy & equity analysts
For policy and equity analysts, APCs represent a measurable equity risk that can amplify citation and career inequality.
Equity risk: 32% flagged Diversity/Equity concerns; APCs amplify citation and career inequality.
Systemic vs local: Privileged authors express moral quandaries even when able to pay, the problem is structural, not only individual.
Do more, faster with Evidano
Evidano is an AI-powered qualitative data analysis platform that reproduces thematic coding, co-occurrence, and cross-segment tables from survey or transcript corpora
Evidano ingests spreadsheets or transcript text, runs inductive theme discovery, and applies or imports a codebook to reproduce the study’s eight APC Trap codes.
Output: frequency tables by discipline, institution, and OA type and exportable code assignments for audit trails.
Problem: messy open-ended responses across segments → Solution: thematic + cross-segment analysis
Evidano provides thematic and cross-segment analysis to turn messy open-ended responses into structured outputs.
Evidano ingests spreadsheets or transcript text, runs inductive theme discovery, and applies or imports a codebook (reproduce the study’s 8 APC Trap codes).
Output: frequency tables by discipline/institution/OA type and exportable code assignments for audit trails.
Problem: co-occurrence & nuance are hard to surface → Solution: co-occurrence networks & hierarchical codes
Evidano surfaces co-occurrence patterns with matrices and network visualizations so teams can see which codes co-occur most often.
These visuals turn qualitative patterns into stakeholder-ready evidence.
Problem: off-the-shelf sentiment tools mislabel complex responses → Solution: LLMs tuned for qualitative research
Evidano uses LLMs tuned for qualitative coding contexts and applies codebook rules to reduce false positives in nuanced academic text.
This approach avoids the NVivo and SentiStrength failure modes reported in the PLOS study.
Problem: stakeholder buy-in needs quotes and reproducibility → Solution: searchable quotes, audit logs, and visual reports
Evidano provides click-through access from a code or node to the exact respondent quote, with demographics and exportable evidence for committees.
All workspaces are encrypted and data is never used to train third-party models.
Runbook: 7-step workflow to reproduce APC Trap analysis in two weeks
This runbook lists seven concrete steps to reproduce APC Trap outputs from a survey spreadsheet in roughly two weeks.
Step 1: Import your survey spreadsheet (one row per respondent).
Step 2: Normalize demographic columns (institution, discipline, OA type) for cross-tabs.
Step 3: Run automated inductive theme extraction to surface candidate codes; compare with the study’s eight APC Trap codes.
Step 4: Upload a human-validated codebook and run AI-assisted coding; review and adjudicate disagreements.
Step 5: Generate frequency tables, co-occurrence networks, and segment comparisons (HSS vs HM vs NSE, Gold vs Hybrid).
Step 6: Extract representative quotes per code and assemble a one-page stakeholder brief.
Step 7: Share interactive visual reports with stakeholders and export reproducible artifacts for audit.
FAQ: qualitative analysis of APCs
Can I replicate the PLOS study coding scheme?
Yes, you can replicate the PLOS study coding scheme by importing the codebook or using inductive extraction and validating with humans.
A: Yes, import the PLOS codebook or use Evidano’s inductive extraction to recover similar codes, then validate with a small human panel.
Are automated sentiment tools useful here?
No, automated sentiment tools are insufficient by themselves for nuanced academic text, according to the PLOS authors.
A: Not by themselves. The PLOS authors found NVivo and SentiStrength unreliable on nuanced academic text. Combine AI coding with human adjudication.
How do I compare segments (e.g., HSS vs NSE)?
Use cross-segment analysis to compute code frequency, average codes per respondent, and visualize co-occurrence differences between segments.
A: Use Evidano cross-segment analysis to compute code frequency, average number of codes per respondent, and visualize co-occurrence differences between segments.
Is my data secure?
Evidano encrypts workspaces and does not use customer data to train third-party models, according to the platform's stated policies.
A: Evidano encrypts data end-to-end and does not use customer data to train third-party models; see Evidano for details.
Wrapping up, next moves
Treat the APC Trap as a template: map codes, quantify co-occurrence, and segment by discipline and institution to make targeted recommendations.
- Immediate step: download the source dataset and codebook at PLOS ONE and check the linked GitHub referenced by the authors.
- Try a guided demo to import your survey, run inductive coding, and generate code co-occurrence visuals in hours, not weeks: Try Evidano for free.
- Strong CTA: Book a walkthrough to reproduce APC Trap outputs from your corpus and deliver a stakeholder brief within two weeks.