(go to Outline)
How do we ensure enough precision to make good programmatic decisions after the survey results are analyzed?
Why do we calculate sample size instead of just collecting data on many people?
Remember, if you calculate the sample size for the entire survey sample, you will not have much precision in your analysis of subgroups within the survey sample. For example, of you are doing a survey of three refugee camps and calculate a sample size for the entire survey of 554 households, you will not have much precision when analyzing only the households in one camp. So you will not be able to make estimates for your health indicators or outcomes for only one camp or compare one camp to another.
If a survey has not calculated the sample size beforehand, it may have wasted resources gathering more data than necessary. In addition, because supervision and training are more difficult with more teams and survey workers, not calculating sample size can lead to a greater chance of having bias in the survey results.
We won't go into calculating sample size here. Suffice it to say, many computer programmes do this for you, and many statistics books explain the process.
Here are some websites with explanations of calculating sample sizes: