Process Evaluation of HIIT: Qualitative Playbook

Mixed‑methods teams should use process data to explain deviations: PRO‑HIIT’s lower dose delivered was driven by predictable school calendar disruptions, weather, and staffing.

The PRO‑HIIT evaluation shows that teams should capture these reasons in timestamps and structured reason codes to preserve interpretability.

Small qualitative samples can still reveal mechanistic signals, treat small sample findings as exploratory inputs to triangulate with attendance and intensity metrics.

Analytic teams must account for partial measurement, which can introduce bias when intensity is sampled from a subsample only.

Partial measurement (e.g., HR in a subsample) creates potential bias, report sampling frames and integrate uncertainty in interpretation.

Social desirability can affect teacher‑rated session quality and attendance, use unobtrusive measures where possible (for example, automated accelerometer logs).

Teams should record adaptations systematically, noting who changed what, when, and why, because PRO‑HIIT shortened sessions to 6–8 minutes and altered work: rest ratios.

Capture contamination and blinding status: PRO‑HIIT teachers were not blinded, which likely influenced differences in intensity.

Evidano ingests teacher logbooks and HR/RPE spreadsheets so teams can join time‑stamped delivery records with intensity measures and attendance.

Import teacher logbooks and HR/RPE spreadsheets into Evidano to join time‑stamped delivery records with intensity measures and attendance.

Upload focus‑group audio and use Evidano to transcribe with a custom dictionary (for example, 'HIIT', 'SAAFE', local PE terms) and apply automated PII redaction where required.

Evidano supports AI‑assisted thematic coding that combines deductive MRC codes with inductive subthemes to speed qualitative analysis.

Run AI‑assisted thematic coding on transcripts (deductive MRC codes → inductive subthemes). Evidano generates theme frequencies and co‑occurrence networks to show which barriers co‑occur with low dose.

Compare segments (by class, by week, by peer‑coach vs teacher‑led) with cross‑segment analysis to spot where dose delivered and dose received diverge.

Evidano produces exportable visuals, such as word clouds and co‑occurrence networks, for stakeholder briefs and infographics.

Produce exportable visuals (word clouds, co‑occurrence networks, hierarchical code → subcode trees) for stakeholder briefs and infographics.

Use AI chat over your imported documents to write an evidence‑backed implementation brief or to extract verbatim quotes tied to themes for reports.

Evidano encrypts data end‑to‑end and does not train third‑party models on your corpus, which supports handling school data and parental consents.

Codebooks, versions and analytic decisions are saved in the project history for auditability and replication.

PRO‑HIIT was a 12-week school HIIT programme delivered to Year‑7 students in a Chinese secondary school.

The PRO‑HIIT study embedded 6–10 minute HIIT at the start of PE/PA lessons for 12 weeks and reached n = 369 Year‑7 students (12–13 yrs) across eight classes taught by three PE teachers.

The main implementation finding was that fewer sessions were delivered than planned despite high retention.

Planned sessions per class were 60 (5/week × 12 weeks) but delivered class‑level sessions were 26, 25, 24, 31 (avg ≈26.5); retention remained high with T2 = 100% and T3 = 96%.

Intensity was measured with heart rate sampling and sessional RPE, showing moderate-to-high intensity in sampled sessions.

Intensity monitoring included 127 HR measures across 13 sessions with mean HR = 146 bpm (71% HRmax) and peak HR = 175 bpm (85% HRmax); session RPE mean = 5 (range 3–8).

The qualitative component used small focus groups and deductive then inductive coding to identify barriers and facilitators.

Two student focus groups (n=9 students) and a teacher focus group were transcribed in Chinese, translated, and coded in NVivo; main barriers included competing priorities and repetition, facilitators included peer coaches and teacher participation.

Researchers can reproduce PRO‑HIIT style analyses by systematically combining logbooks, HR/RPE data and transcripts and using reproducible AI workflows for coding and cross‑segment analysis.

Follow the seven-step checklist: gather inputs, upload and transcribe, auto‑code with an MRC codebook, run frequency and cross‑segment analyses, visualise co‑occurrence networks, draft an implementation brief, and export reproducible outputs.