Mathematics
Error Variance
Quick fact
Error variance is often called 'noise' in data; in a perfect experiment, it would be zero, but in reality, it's always present and is estimated from the data itself.
Why this is interesting
Why do repeated measurements of the same object never give exactly the same number? That unpredictable scatter is error variance—and it's the key to knowing how much you can trust your data.
Read the full explanation
Understanding Error Variance
Imagine you're trying to find the average height of a group of people. You measure each person several times, but the numbers vary slightly—maybe because the ruler shifts or people stand differently. That variation, which you cannot explain by the person's actual height, is error variance. In statistics, when we build a model (like comparing average heights of men and women), the error variance is the leftover spread after accounting for the group differences. It's the part of the data that the model cannot explain, caused by random fluctuations, measurement errors, or unmeasured factors.
A deeper explanation
Error variance is the variance of the residuals—the differences between observed values and the values predicted by a model. For example, in a simple linear regression, it is the average squared distance of data points from the regression line. This variance is crucial because it tells us how much uncertainty remains in our predictions. In ANOVA, the total variance is split into variance explained by group differences and error variance (within-group variance). The F-test compares these to determine if the group differences are statistically significant. Error variance also directly affects standard errors: larger error variance means less precise estimates, wider confidence intervals, and lower statistical power. Reducing error variance (through careful measurement or controlling extraneous variables) improves the quality of inferences. Understanding error variance prevents overinterpreting noise and helps researchers design better experiments.