Sample size

Protocols require the study team to justify the number of participants involved in the study. The number of participants may be decided by logistics. For example, if a study is assessing the response of A&E nurses in one hospital to recorded music, there may only be 15 nurses who can take part in the study. Alternatively, it may be that a sample size calculation can be performed before the study commences. Sample sizes should account for dropouts and be realistic within the planned timeframe.

© Sample size calculations are a statistical test used to determine how many people are needed in a study to have a good chance of achieving statistical significance in tests, based on the anticipated size of effect the intervention. If a study is underpowered, it may not achieve significance even though the effect being examined may be occurring (also known as type II error), leaving results ambiguous or even meaning that the intervention is dismissed as not effective.

Sample size calculations can be performed using statistical programs such as G*Power or on paper using graphs called nomograms

For more information on sample sizes, Andy Field provides good introductions in his statistics books.(4,5)

< Prev   CONTENTS   Source   Next >