The language of measurement data: mean, sample standard deviation and confidence interval

Why it matters

Cycles using the same method do not always finish in the same time. The mean summarizes these durations with one number but does not describe their spread. Sample standard deviation expresses that spread in units of time. Two tasks can have the same mean but different variation. Report the mean, standard deviation and observation count together for this reason. Before calculating, verify that records cover the same work. A tidy-looking number combining different methods' times may represent neither method correctly. Statistics does not correct measurements with unclear boundaries; it helps assess what defined, comparable measurements tell us.

A sample is only part of all possible cycles under study. Both the mean and spread are estimated from this part. For sample variance, divide the sum of squared deviations by one less than the observation count, n−1. This is because the mean used to calculate deviations was also found from the same data. Since all deviations sum to zero, the last deviation cannot be chosen independently of the others. This helps explain why the divisor is not n. Standard deviation is the square root of variance. If times are in seconds, variance is in s² and standard deviation in s. Tracking units also makes it easier to detect quantities confused in calculation.

A confidence interval for the mean is not a limit on a single cycle's duration. For independent, approximately normally distributed observations, use a t factor when estimating the unknown spread from the sample. This lesson uses t(4; 95%)=2.776 for a 95% confidence level. The first number in parentheses is degrees of freedom; for five observations, 5−1=4. Confidence interval width depends on standard deviation, observation count and the factor. More independent observations can narrow the estimate of the mean when other conditions are similar. They do not automatically reduce physical variation between individual cycles. Increasing measurements and improving the method are different tasks.

95% confidence does not mean that 95% of observed cycles will fall in the calculated interval. If sampling and interval construction were repeated many times under the same conditions, the method is designed to include the true mean in approximately 95% of the intervals. No certainty is promised for a single calculated interval. Dependent records, learning trends or mixed methods weaken this interpretation. Make assumptions visible when reporting the result. This lesson aims to distinguish the mean from an individual cycle and correctly establish inputs for observation count calculations, rather than solve complex distributions. The next lesson builds target precision on this foundation.

Formula