HLSC 2P27 Lec. 9 - Sample Size and Power

Too many participants

Waste resources like time and money

Too few participants

Risks insufficient data to answer the study question

Confidence interval (CI)

A statistical range that estimates likely values of a parameter in a source population, based on data from the study population

Narrow CI

Indicates more certainty about the value of a statistic

Wide CI

Suggests less certainty, occurs when sample size is small

Random error

Random difference between study results and true population values (contains extreme values and can lead to incorrect conclusions) (small samples sizes more prone for random error)

Impact of sample size on CI

Larger samples mea closer to true population mean, CI become narrower (greater precision, more statistically significant)

Systematic error (bias)

Systematic flaw in a study design that leads to inaccurate results (unlike random error, bias is consistent and affects results in a specific direction)

Type 1 error

Occurs when study shows a statistically significant result, even though there is no real difference in the source population (a = 0.05) (false positive)

Type 2 error

Occurs when a statistical test fails to detect a significant result, even though a real difference exists in the source population (B) (false negative)

Statistical power

The ability of a statistical test to detect differences when they actually exist in a source population (defined by 1 - B)

Small sample size

Low precision, wide CI, more random error

Large sample size

High precision, narrow CI, less random error