Evidano is an AI-powered qualitative data analysis platform that helps teams extract themes from interviews, notes, and survey sheets. This post shows an AI-enabled reproducible workflow for qualitative analysis of STEM outreach using IEEE Spectrum's profile of the Women in Science, Engineering (WiSE) program in India, and explains how to move from transcripts and survey sheets to role-based recommendations in days.
Key Takeaways
This post shows a reproducible 7-step AI workflow to extract themes from STEM outreach data and produce evidence-ready reports in days.
- Evidano is an AI-powered qualitative data analysis platform that ingests interviews, teacher notes, and artifact photos to produce theme frequencies, representative quotes, and cross-segment contrasts.
- WiSE ran five-day sessions at IIT Bombay starting in 2023 with cohorts of about 160–200 rural and tribal girls, supported by ~70 student volunteers and a US $80, 000 grant awarded in 2022 (IEEE Spectrum).
- A 2–7 day workflow (depending on corpus size) yields a theme dashboard, segment comparisons, and slide-ready findings suitable for funders or program teams.
FAQ: qualitative analysis of STEM outreach
What is the WiSE program and who participated?
WiSE ran as a five-day hands-on program at IIT Bombay starting in 2023, selecting cohorts of about 160–200 rural and tribal girls each year, supported by ~70 student volunteers and a US $80, 000 grant awarded in 2022, according to IEEE Spectrum.
Teachers nominated high-performing girls from 68 schools in Maharashtra and Karnataka, and parental follow-ups were part of a four-year engagement commitment after the program.
What does this post recommend for program evaluators?
Program evaluators should prioritize mixed-methods, combining follow-up surveys with thematic coding of participant interviews to measure mechanisms like confidence and role-model impact.
Program evaluators should compare cohorts across years (2023, 2024, 2025) to detect persistence or attenuation of effects.
How long does the proposed workflow take to run?
The proposed 7-step workflow can be executed in 2–7 days depending on corpus size and yields a theme dashboard, segment comparison, and slide-ready findings.
The workflow includes collection, transcription with PII redaction, seed codebook creation, auto-coding with validation, cross-segment analysis, synthesis, and sharing.
Are there published quantitative outcomes for WiSE?
No formal longitudinal quantitative outcome study for WiSE is published in the sources cited here; existing evidence is qualitative participant testimonials reporting increased confidence and changed career aspirations.
Evaluators should therefore treat current evidence as qualitative-first and design follow-up surveys or longitudinal measures to produce quantitative outcomes.
Fast take + source
The fast take: WiSE combined hands-on projects, role-model talks, and parental follow-ups and produced promising qualitative signals around retention and confidence, but no formal quantitative evaluation is published in the cited source.
WiSE ran as a five-day hands-on program at IIT Bombay starting in 2023, selecting cohorts of about 160–200 rural and tribal girls each year, supported by ~70 student volunteers and a US $80, 000 grant awarded in 2022 (source: IEEE Spectrum).
- Why this matters: program teams and evaluators must extract themes (barriers, motivators, persistence signals) from interviews, teacher reports, and field notes to decide what to scale.
- Payoff: this post gives an AI-enabled workflow to extract themes, compare segments (teachers vs students vs parents), and produce visual evidence for stakeholders using Evidano.
Findings snapshot
| Date / Item | Metric | Value | Source | Implication |
|---|---|---|---|---|
| 2022 | Seed funding | US $80, 000 | IEEE Spectrum | Enabled 3-year pilot design |
| 2023–2025 | Program run | Five-day sessions each March; 3 cohorts | IEEE Spectrum | Repeatable seasonal model |
| Per-cohort | Participants | 160–200 girls | IEEE Spectrum | Sufficient scale for thematic saturation across cohorts |
| Outreach | Schools contacted | 68 schools in Maharashtra & Karnataka | IEEE Spectrum | Diverse rural sourcing strategy |
| Staffing | Volunteers | ~70 IIT-B student volunteers | IEEE Spectrum | Peer mentoring resource |
| External event | Pariksha Pe Charcha 2024 | ~4, 000 in-person participants (event) | IEEE Spectrum | Visibility channel for scaling student work |
What happened, methods & signals
This section explains the intervention components, selection method, and qualitative signals observed in the WiSE program.
WiSE combined hands-on Break-Make-Program (BMP) builds (for example, robots, AUVs, glowing bacteria demos) with daily talks by Winspirers and a four-year parent-engagement commitment after the program.
- Selection bias: teachers selected high math/science scorers from 68 schools, so findings reflect motivated students rather than an unfiltered population.
- Intervention components to evaluate include hands-on kits (sustained access), role-model talks, parental accountability meetings, dorm experience, and campus exposure.
- Existing evidence: qualitative participant testimonials report increased confidence and changed career aspirations, but no longitudinal quantitative outcome study is published in the cited source.
So what for researchers and program teams
Program evaluators
Program evaluators should prioritize mixed-methods combining follow-up surveys with thematic coding of participant interviews to measure mechanisms like confidence and role-model impact.
Program evaluators should compare cohorts (2023 vs 2024 vs 2025) to detect whether effects persist or attenuate.
UX / learning designers
UX and learning designers should code BMP artifacts and teacher notes to identify which hands-on activities map to durable learning and diffusion.
UX and learning designers should use co-occurrence analytics to spot links between specific activities and language about confidence or career intent.
Policy & funders
Policy makers and funders should ask for standardized qualitative deliverables such as theme frequency tables, segment comparisons, and a 90-day action memo.
Policy makers and funders should require clear documentation of selection and follow-up protocols to assess equity impacts in rural settings.
Do more, faster with Evidano
Ingest messy inputs
Evidano is an AI-powered qualitative data analysis platform that ingests interview transcripts, teacher notes, survey spreadsheets, and photos of BMP artifacts to start analysis quickly.
Upload interview transcripts, teacher notes, survey spreadsheets, and photos of BMP artifacts into Evidano; use built-in transcription with custom dictionaries for local names and PII redaction to protect participants.
Automated thematic + frequency analysis
Evidano produces theme frequencies, representative quotes, and cross-segment contrasts in minutes once a codebook is seeded or auto-generated.
Run AI-assisted codebook creation or import your existing codes (for example, Confidence, Parental Support, Kit Reuse). Evidano produces theme frequencies, representative quotes, and cross-segment contrasts (students vs parents vs teachers) in minutes.
Cross-segment & visual evidence
Evidano generates co-occurrence networks and hierarchical visualizations to show which activities correlate with outcomes.
Generate co-occurrence networks and hierarchical code to subcode visualizations to show which activities correlate with outcomes for example, robot build leads to 'confidence' plus 'career interest'. These visuals make stakeholder briefings decision-ready.
Secure, reproducible reporting
Evidano provides encrypted storage and reproducible exports so teams can share evidence-ready reports without exposing PII.
All data is encrypted and never used to train third-party models. Export reproducible reports and clickable quote decks for funders or ethics review boards.
Checklist: 7-step workflow to reproduce WiSE-style analysis
This checklist describes a 7-step reproducible workflow that produces a theme dashboard, segment comparison, and slide-ready findings in 2–7 days depending on corpus size.
- 1) Collect: transcripts (student, parent, volunteer), teacher referrals, BMP logs, and short follow-up surveys.
- 2) Clean & transcribe: run Evidano transcription with a custom dictionary for names, terms, local language variants; apply PII redaction.
- 3) Seed codebook: import an initial codebook (Confidence, Retention Intent, Kit Reuse, Barriers) or let Evidano suggest themes.
- 4) Auto-code + validate: auto-code the corpus, then hand-review a 10–20% sample to adjust precision and recall.
- 5) Cross-segment analysis: run frequency tables, chi-squared exploratory checks, and co-occurrence networks to link activities to outcomes.
- 6) Synthesize: generate executive summary, evidence tables, and representative quotes grouped by theme and cohort year.
- 7) Share & act: produce a stakeholder brief and a prioritized 30/60/90 action list for example, scale kit distribution or formalize parent follow-ups.
Ethics & safeguards (short note)
This short note summarizes consent, anonymization, and storage safeguards required when working with minors in qualitative research.
Qualitative work with minors requires consent and careful PII handling. Use anonymization, store parental consents, and treat findings as program-evaluation (non-diagnostic). Evidano supports PII redaction and encrypted storage to help meet these obligations.
Wrapping up & next steps
This conclusion summarizes the practical next steps: run the 7-step workflow on one cohort and produce a 2-page evidence memo for funders.
IEEE’s WiSE offers a compact, repeatable model for rural STEM outreach; the evidence so far is promising but qualitative-first evaluation is essential to turn stories into fundable, scalable practices.
- Next move: run the 7-step workflow on one cohort’s interviews to produce a 2-page evidence memo for funders.
- Try it: import a pilot batch of transcripts into Evidano and generate themes, cross-segment comparisons, and visuals in hours, or Try Evidano for free.