ResearchR · Survey designMarch 2025 · 5 min read

What self-medication data revealed about student health

Designing the self-medication study end-to-end taught me that determinants rarely live in one variable. Cost, access and prior experience interacted in ways a single regression coefficient couldn't tell — so I let the analysis map behaviour, not just measure it.

Share

My undergraduate research asked a deceptively simple question: how and why do students at USIU-Africa self-medicate? I built the proposal, designed the instrument, collected the responses and ran the analysis in R. Owning the whole pipeline meant I couldn't hide behind any single step.

Determinants don't live alone

I went in expecting a tidy set of predictors. What I found was interaction. Cost shaped behaviour differently depending on proximity to a pharmacy. Prior experience with a drug mattered more for some symptoms than others. A single regression coefficient could report an association, but it couldn't narrate the behaviour — and behaviour was the point.

So I let the analysis map rather than merely measure. Cross-tabulations, stratified models and honest visualisations did more to explain self-medication than any one p-value. The goal shifted from proving an effect to describing a pattern a health-behaviour policy could actually respond to.

Evidence that speaks to policy

The most useful finding wasn't a number — it was a story the numbers supported: students self-medicate when the formal system feels slow, costly or distant. That framing is something a campus health office can act on, and it is the version of the result I care most about.

Share this piece
Research · 5 min read

What self-medication data revealed about student health

Designing the self-medication study end-to-end taught me that determinants rarely live in one variable. Cost, access and prior experience interacted in ways a single regression coefficient couldn't tell — so I let the analysis map behaviour, not just measure it.

MaCaja

Catherine Magawa

Epidemiologist & Biostatistician

Get in touch

Interested in this work? Catherine is open to public health & data roles worldwide.

Email Catherine
Next

Cleaning national survey data without losing the story