Knowledgebase

Determining the size of a sample

There is no "right" size for a sample. The size depends on a number of factors:

• The type of question you are asking. Estimating the population's proportion (2 possible values) for questions like "Did you have a financial audit in your most recently completed fiscal year?" requires a smaller sample than answering a question like "How much did your financial audit cost?"
• Anticipated response rates. If you receive a 50% response rate (unlikely for most surveys of individuals or organizations), then you will need a much smaller sample than if you anticipate a 20% rate.
• The level of confidence and confidence interval that you need in your answer. Do you need to know that the percentage of organizations having financial audits is within 3 percentage points of the sample results, or is a margin of error of +/-10% sufficient? If the former, you may need a sample size of more than 1,000. If the latter, then a sample size of less than 100 is sufficient.
• The size of the population. Once the size of the population exceeds several thousand, it has relatively little influence on the required sample size. For example, a sample of 843 is required for a population of 4,000 if one seeks a confidence interval of +/-3% for a proportion at a 95% level of confidence while the sample size increases to 1056 for a population of 100,000 and 1065 for a population of 500,000.

Attached is an Excel spreadsheet that enables you to calculate the sample size needed for varying levels of confidence and confidence intervals.