gain power

Often experiments or studies fail due to insufficient statistical power; that is the study is not sensitive enough to detect important real effects. This is particulatly common in human sciences such as HCI because high individual variability leads to inconclusive results due to 'insufficient subjects'. The first line of response is often to attempt to have more participants or trials, but this is not always possible. Careful design can often gain power, typically by manipulating different aspects of the noise-effect-number triangle, for example by having more uniform participants or more extreme tasks.