Sample Size

Sample Size:

Important component of planning stage – there are methods available to estimate sample sizes before conducting a study, so that there is a good chance of obtaining a statistically significant result. Study needs to be large enough to ensure that the results are generalisable, but not so large that the research resources are wasted.

Sample size is a judgement issue based on what is considered a clinically important difference between groups.

If too small:
• type 1 or type 2 errors
• power inadequate to show how significant and clinically important difference is
• unethical
If too large:
• small differences found that are not clinically important
• waste of resources
• unethical

Sample Size Calculations:
Several formula are available for sample size calculations, but information is needed on four variables are needed:
• Size of response to control intervention
• Expected size of response to experimental intervention
• Significance level (arbitrary determination; usually set at 0.05 by convention; risk of committing a type 1 error)
• Power (probability of rejecting null hypothesis; probability of a type 2 error is beta, so power is defined as 1-beta; by convention it is set at 0.80.

Comments are closed.