Fictional Case Study: Investigating a Drop in Reported Attendance

Fictional learning scenario

This is an original RDE teaching case. The organisation, records, observations and outcomes are invented for learning. It is not a client case, testimonial or report of real results.

Context and decision to make

A fictional learning team sees reported attendance fall from 90% to 75% and proposes changing the timetable. The analyst is asked to establish whether learner attendance has actually fallen.

Evidence available in the scenario

The fictional cohort has 20 learners and two scheduled sessions in each comparison week, giving 40 possible learner-session attendances per week. Week A reports 36 attended records out of 40. Week B reports 30 out of 40. The new Week B import includes four records marked JOINED instead of PRESENT and two duplicate rows. A blank attendance status remains unknown, not automatically absent.

Approach and interpretation

The analyst checks the metric definition, source records and import mapping before discussing causes. The published metric uses distinct learner-session pairs marked attended. Four JOINED records have supporting session logs showing attendance and should map to attended. The two duplicate rows repeat pairs already counted. They must not add either attended records or expected learner-session opportunities.

Options and decision

The team could change the timetable immediately, report an unexplained decline, or reconcile the source and explain the uncertainty. They choose reconciliation. An authorised data owner confirms the meaning of JOINED; the analyst documents the mapping change and preserves the original import so the correction can be traced. No new attendance is inferred from a name alone.

Outcome within the fictional scenario

After the fictional reconciliation, Week B has 34 confirmed attended learner-session pairs out of 40, or 85%. The four remapped pairs change the numerator from 30 to 34; duplicates add nothing. The gap from Week A is five percentage points, rather than fifteen. Two of the remaining six pairs have unknown status, so the team reports the 85% confirmed rate with that limitation rather than presenting every non-attended pair as a confirmed absence.

Limitations and next checks

The correction reduces a reporting discrepancy but does not explain the remaining difference. A two-week comparison cannot establish a trend, and missing statuses could still change the result. The team resolves the unknown records and checks more weeks using a stable definition before considering a timetable intervention.

Apply the case

Allow about 20 minutes. Identify the decision, extract the evidence and write down one assumption that needs checking. Separate the observed or stipulated figures from the explanation proposed by the team.

1. Why do duplicate source rows not increase the denominator?
2. What evidence authorises JOINED to map to attended?
3. How would you describe the two unknown pairs in a report?

Check your learning

Investigate definitions and record identity before explaining a trend. A corrected metric is a reporting result, not proof of a behavioural cause.

A useful response names the unit being measured, tests the proposed explanation against the available evidence and proposes a next check that could change the decision. Avoid claiming a general result from the small fictional sample.

Use and attribution

Original educational scenario by Jefferson Otonbara Imgbi. Version 1.0, 8 October 2026. No real customer records, organisation identities or endorsements are used. Personal and internal organisational learning use permitted with author credit retained.

Cover image: AI-generated illustrative scene created specifically for this resource.