Glossary
Selection bias.
Direct answer
Selection bias is the distortion that arises when the cases studied differ systematically from those the conclusion is meant to cover. It appears during recruitment, during allocation to groups, or through unequal dropout during the study, and it pushes the result in one particular direction.
Selection bias cannot be cured by adding cases: a large but skewed sample merely gives a more precise wrong value. What is always revealing is who does not appear in the study, such as refused, withdrawn or untraceable cases. That absence is passed over in silence far more often than it is explained.
In equine research much data comes from university clinics and from high-value sport horses, a population monitored more closely and managed differently from the average leisure horse. In human medicine the same imbalance arises through volunteers, who are on average healthier and more motivated than the general population.
Related content
- Autumn ACTH in the horse: what a seasonal reference range really tells you
- Why a broken leg ends differently in a horse than in a person
- Sycamore and atypical myopathy: why not every horse on the same pasture falls ill
- Strangles in horses: what is known about silent carriers and the duration of shedding
- Kissing spines on radiographs: what a finding says about pain and what it does not predict
- Crib-biting and weaving: what the evidence says about stereotypies in horses
- A faecal egg count with nothing found: what a negative result does not rule out
- Gastric ulcers in the horse: how well behavioural signs predict the gastroscopy result
The evidence letter
Once a month: new syntheses, checked studies, and what has changed in the state of the evidence.