How Precise Are Our Estimates? Confidence Intervals
In usability testing, like most applied research settings, we almost never have access to the entire user population. Instead we have to rely on taking samples to estimate the unknown population values. If we want to know how long it will take users to complete a task or what percent will complete a task on the first attempt, we need to estimate from a sample. The sample means and sample proportions (called statistics) are estimates of the values we really want—the population parameters.
When we don’t have access to the entire population, even our best estimate from a sample will be close but not exactly right, and the smaller the sample size, the less accurate it will ...