Glossary
Confounding factor.
Direct answer
A confounding factor is a characteristic related both to the cause under study and to the outcome observed, able to create an apparent link or hide a real one. Comparing turnout with health means allowing for age, use and housing, otherwise those characteristics are what gets measured.
A confounder can only be allowed for if it was recorded; unknown confounders remain a fundamental problem in observational studies. That is precisely the strength of random allocation: it also spreads unknown characteristics evenly on average across the groups. Statistical adjustment after the fact replaces that protection only in part.
In equine research, breed, age, discipline, fitness, shoeing and ground conditions are typical confounders, and they rarely occur singly. In people they are more often smoking, income and education. The trap is identical in both settings: whoever starts out healthier tends to choose the measure that will subsequently look effective.
The evidence letter
Once a month: new syntheses, checked studies, and what has changed in the state of the evidence.