Basic statistical tests - standard error of the mean

Definition

The standard error of the mean quantifies the precision/uncertainty of a sample mean as an estimate of the true population mean - distinct from standard deviation (which describes spread of individual data points), the standard error specifically describes how much the sample mean itself would be expected to vary if the study were repeated with different samples from the same population
Used in constructing confidence intervals around a reported mean, making it foundational to interpreting the precision of any reported average value in clinical research

Standard error is calculated as standard deviation divided by the square root of sample size - this mathematical relationship explains why standard error decreases as sample size increases (larger samples give a more precise estimate of the true population mean), while standard deviation itself (a property of the underlying population's variability) does not systematically change with sample size

Used to construct confidence intervals (typically mean ± approximately 1.96 standard errors for a 95% confidence interval), providing a range within which the true population mean is estimated to lie with a specified level of confidence
A narrower standard error/confidence interval indicates a more precise estimate of the true population mean

Larger studies produce a smaller standard error and narrower confidence interval around the mean, all else being equal, informing sample size planning when precision of the estimate is important
Should not be reported or interpreted as if it were standard deviation, given the different question each answers
Sample size's inverse-square-root relationship with standard error, confidence interval construction from standard error, standard error vs standard deviation distinction (see )

Because standard error equals standard deviation divided by the square root of sample size, quadrupling the sample size only halves the standard error (not quarters it) - this diminishing-returns relationship is a frequently tested, practically relevant point for understanding why very large increases in sample size are needed to meaningfully improve precision beyond a certain point

4 more sections, plus exam facts

Premium unlocks every note across every specialty, and the full exam fact library behind it.