AI thematic analysis: sand mining governance

Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents. According to PLOS ONE, the study by Asare et al. (published 13 August 2026) used 32 key informant interviews, multiple FGDs, and NVivo 12 to identify 57 thematic codes in two Ghanaian districts. The primary keyword in this post, AI thematic analysis sand mining, describes the process of using AI to speed coding, quantify themes, and surface governance bottlenecks from qualitative transcripts.

Key Takeaways

According to PLOS ONE, illegal sand mining in Ghana persists despite laws, driven by weak enforcement, low penalties, and political patronage.

32 stakeholders were interviewed and two FGDs held between 3 March and 20 May 2021, with follow-up interviews 10–21 January 2022, according to PLOS ONE (Asare et al., 2026).
The study area supplies an estimated 4.55 million m3 of sand annually to nearby markets, according to PLOS ONE (2026).
PLOS ONE (2026) reports that over 80% of sand mining in Ghana is done illegally, and the authors identified 57 themes from the qualitative analysis.
A regulatory quote in PLOS ONE (2026) summarized the penalty problem: "The sanction for sand mining offenses under Section 29... is GHS 200 ($20)." (Key informant, 2021).
A landowner quote in PLOS ONE (2026) captures motive: "As it stands, most landowners believe that while the lands belong to them, regulatory agencies accrue gains from sand mining." (Landowner in Gomoa Buduatta, 2021).

What happened and how it was measured

The study analysed attitudes to law application in Ga South and Gomoa East by interviewing local actors and reviewing legal documents, according to PLOS ONE.

The authors collected primary qualitative data via 32 key informant interviews and two FGDs (7–12 participants each) between 3 March and 20 May 2021, with additional follow-ups 10–21 January 2022, according to PLOS ONE (2026).

NVivo 12 was used for thematic and content analysis and produced 57 themes and a saturation point at 32 interviews, as described in PLOS ONE (2026).

The research combined interview data with a documentary review of the Minerals and Mining Act (Act 703) and Environmental Assessment Regulations (LI 1652), according to PLOS ONE (2026).

Findings Snapshot

Date	Metric	Value	Implication
13 August 2026	Study publication	Asare et al., PLOS ONE	Peer-reviewed qualitative evidence available for policy use
Data collection: 3 Mar–20 May 2021; follow-up 10–21 Jan 2022	Sample	32 interviews; 2 FGDs (7–12 persons)	Saturation achieved; findings reflect local actor perspectives
Reported in PLOS ONE (2026)	Illegal share of sand mining	Over 80%	Regulatory noncompliance is pervasive and systemic
Reported in PLOS ONE (2026)	Sand volume from study areas	4.55 million m3 annually	High economic stakes tied to nearby urban demand
During analysis, PLOS ONE (2026)	Analytic output	57 themes identified	Rich thematic map suitable for cross-segment analysis

Implications for qualitative researchers and policy analysts

Researchers should prioritise reproducible thematic workflows to link interviews to policy recommendations, according to PLOS ONE (2026).

When a study reports 57 themes from 32 interviews, as in PLOS ONE (2026), researchers should use cross-segment frequency analysis to avoid overinterpreting low-frequency themes.
When data collection spans discrete periods (March–May 2021; Jan 2022), as in PLOS ONE (2026), researchers should timestamp transcripts to detect temporal shifts in attitudes.
When enforcement gaps are quantified (for example, a reported over 80% illegal rate in PLOS ONE, 2026), policy analysts should combine qualitative evidence with geospatial monitoring to target scarce regulatory resources.

How Evidano helps (problem → AI-enabled qualitative solution)

Problem: scattered transcripts and long synthesis cycles

Answer: Evidano automates thematic coding and frequency counts so teams can move from raw interviews to a coded evidence matrix in hours, not weeks.

Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents, and the platform supports upload, transcription, and thematic extraction.

Evidano techniques mirror the NVivo approach used in PLOS ONE (2026) but add automated cross-segment frequency analysis and interactive visualizations to validate the 57 themes found in the study.

Problem: inconsistent stakeholder segmentation

Answer: Evidano tags respondents and runs cross-segment comparisons to reveal which themes drive perspectives among landowners, regulators, and contractors.

Evidano can ingest interview transcripts and FGDs and produce segment-level thematic summaries, matching the stakeholder categories reported in PLOS ONE (2026).

For teams that need transcription before analysis, Evidano also offers accurate speech-to-text, which reduces manual transcription time and preserves speaker attribution; see Evidano speech-to-text.

Problem: weak evidence for enforcement priorities

Answer: Evidano converts coded qualitative data into tables, co-occurrence networks, and ranked recommendations to help regulators prioritise interventions.

Evidano links thematic frequency with metadata (date, location, actor type) so analysts can map complaints and extractive hotspots similar to the monitoring gaps highlighted by PLOS ONE (2026).

Learn more about how teams operationalise these features on the Evidano features page.

FAQ: AI thematic analysis sand mining

How many interviews are enough for AI thematic analysis in a sand mining governance study?

Answer: Use saturation as the stopping rule, typically 12–15 interviews for a homogeneous group but more for diverse stakeholders, as discussed by PLOS ONE (Asare et al., 2026) referencing Hennink and Kaiser (2022).

PLOS ONE (2026) collected 32 interviews and reported thematic saturation after coding 57 themes, which supports using AI to speed saturation checks and surface low-frequency but policy-relevant themes.

Can AI replicate the NVivo coding approach used in PLOS ONE (2026)?

Answer: AI can replicate and extend NVivo-style coding by performing initial automated coding, then letting human analysts refine codebooks, as recommended by methodological literature cited in PLOS ONE (2026).

Using AI accelerates codebook iteration and produces frequency tables and co-occurrence analyses that match or exceed manual NVivo workflows while preserving audit trails for reproducibility.

What concrete numbers from the PLOS ONE study should policymakers cite?

Answer: Policymakers should cite the sample and timing (32 interviews; FGDs; data collected 3 March–20 May 2021 with follow-ups 10–21 January 2022) and the reported governance metrics (e.g., over 80% illegal sand mining), all found in PLOS ONE (2026).

Quoting the penalty example from PLOS ONE (2026) may also be persuasive: "The sanction for sand mining offenses... is GHS 200 ($20)." (Key informant, 2021).

Is it ethical to run AI on sensitive interviews about illegal activity?

Answer: Yes, if researchers use encrypted storage, obtain informed consent, and anonymize identifying details; PLOS ONE (2026) followed ethical clearance UCCIRB/CHLS/2020/48 and withheld identifying data.

Evidano stores data encrypted and supports PII redaction to align qualitative AI workflows with ethical research practices.

Conclusion & Next Steps

AI thematic analysis sand mining turns qualitative interviews into actionable governance evidence, as illustrated by Asare et al. in PLOS ONE (13 August 2026).

The PLOS ONE study shows the value of combining document review, 32 interviews, FGDs, and systematic coding to surface enforcement gaps, compensation failures, and coordination problems.

Teams ready to operationalise similar research should use AI-assisted transcription, thematic coding, and cross-segment analysis to shorten synthesis time and increase transparency; see Evidano features for a checklist.

Try a hands-on synthesis: Try Evidano for free

Topics

AI thematic analysis sand mining
AI qualitative analysis
sand mining governance
thematic analysis AI

Keep reading

Browse all articles