The use of epidemiological tools in conflict-affected populations: open-access educational resources for policy-makers

A survey selected only 83 young children to measure the prevalence of acute protein-energy malnutrition. It found that the prevalence was 13% and the 95% confidence interval was 5% - 21%. A few days later, as part of a nutrition screening, every child in the same age range in the population was weighed and measured. It turned out that only 7% of all the children in the population had acute malnutrition.

What might account for this difference between 13% from the survey and 7% from the universal weighing and measuring?

	a)	Presence of bias
	b)	Large sampling error
Correct. A large sampling error might account for the difference. The confidence interval of 5% to 21% is very wide indicating a large sampling error. As a result, the difference between the point estimate and the true population value may be large solely because of sampling error. But a large sampling error does not preclude the presence of bias. Both may result in a survey point estimate being quite different from the true population value. Possibly correct. However, a large sampling error might also account for the difference. The confidence interval of 5% to 21% is very wide, indicating a large sampling error. As a result, the difference between the point estimate and the true population value may be big only because of sampling error. But a large sampling error does not preclude the presence of bias. Both may result in a survey point estimate being quite different from the true population value. Your answer has been saved.