RPG Research Catalog
How Large Should My Sample Be?
One of the toughest questions to answer when designing a study involves the size of the sample needed. And, unfortunately, there’s no magic formula that can tell you exactly what number of respondents you need to be sure that your answers aren’t way off.
Many of the guidelines involve first asking some questions about the sample itself.
Is the study better conducted using a census or a sample? In a B2B study or a study with arelatively small number of respondents, this is actually an important question to ask, because often, extremely small populations can and should see every member included. The advantage of taking a
census is that the data is 100% representative and has zero error due to sampling. For populations of less than 100, a census is the recommended approach.
Is the population more alike, or more different? If the population tends to consist of individuals whose answers very much alike, the sample size won’t need to be as large as a population that consists of individuals who are likely to have some variation. Consumer studies should always assume maximum variability.
Is the study qualitative or quantitative in nature? Qualitative studies rarely require data to be representative of the population, and thus do not require as large of samples. Quantitative studies benefit from larger samples.
Are the results intended to be projected to the population as a whole? If the sample is probability-based (also known as a scientific sample), the necessary sample size can be roughly calculated to account for the potential error. If the sample is non-probability based (also known as a
convenience sample), the larger the sample, the better.
Will the data be used for sophisticated analysis, such as modeling? Many advanced data analysis techniques require a minimum number of responses to work correctly. This should always be accounted for in the sample.
The bottom line is that sample size requires the same care and planning as the research itself. The biggest drawback to having a large sample comes in the cost per completed survey. Larger samples tend to provide more reliable data, but the cost is significantly higher. Many times, a small or mid-sized sample can provide a sufficient amount of information without the heavy cost.
The Research & Planning Group includes sample size estimates and recommendations in its research proposals. These recommendations take many considerations into account and come with the added benefit of decades of experience.