Does your sample size go up or down when you want to not only detect but estimate prevalence in a herd?

• A. Increase
• B. Decrease
• C. Remain the same
• D. Varies with prevalence

Answer: (A)

When you want to not only detect but also estimate prevalence in a herd, you need more samples to get a reliable, precise estimate. Estimating a proportion comes with sampling variability: the standard error of a prevalence estimate is roughly sqrt[p(1−p)/n]. To achieve a given width of a confidence interval (better precision), you must increase the sample size. If you only aim to detect whether the disease is present, you can often use a smaller sample because you’re making a binary yes/no decision about presence (as long as you meet a chosen confidence level for that detection). But for an accurate estimate of how common the disease is, especially if the true prevalence is low, more observations are required to reduce uncertainty. So the general rule is that sample size goes up when you move from mere detection to estimating prevalence.