Ascertainment bias occurs when data for a study are collected such that some members of a population are more likely to be included in the sample than others.
This can result in samples that are not representative of the target population, which makes it hard to generalize the findings from the sample to the population.
Examples of Ascertainment Bias
Here are a couple examples of ascertainment bias in different settings:
1. Prevalance of Diseases
Suppose researchers are interested in estimating how prevalent a disease is in a certain country. To collect data, they ask residents around the country to visit their nearest hospital and get tested for the disease.
Ascertainment bias is likely to occur because residents who are richer and more capable of getting to a hospital/live in an area that has a hospital are more likely to get tested. This means that the disease is likely to appear far more prevalant in rich populations compared to poor ones in this country.
However, this result is misleading because it turns out that richer residents are simply more likely to be included in the sample data.
2. Support of Tax Increases
Suppose a school board is interested in estimating the proportion of households in the school district that would support a tax increase to provide more funding for the school sports teams. To collect data, they go around and survey parents at the school football game on a Friday night.
Ascertainment bias is likely to occur because the parents that are present at the game are likely to have a child who is on the football team, which means they’re far more likely to support a tax increase compared to the typical household in the school district.
This means the proportion of households in the survey that support the tax increase is unlikely to match the proportion of households that support the tax increase in the overall population.
How to Prevent Ascertainment Bias
The easiest way to prevent ascertainment bias is to use a sampling method that gives each member of a population an equal chance of being included in the sample.
Examples of appropriate sampling methods include:
- Simple random sample
- Stratified random sample
- Cluster random sample
- Systematic random sample
In each of these methods, the probability that a given member of the population is included in the sample is equal.
This means each of these methods maximizes the chances that the sample obtained is representative of the target population. Thus, the findings from the sample can be generalized to the overall population with confidence.
The following tutorials provide explanations of other biases that can occur in research: