Quantitative geneticists use the variance as a measure of scatter in a sample of data; they calculate this statistic by averaging the squared deviations between each measurement and the sample mean. Why don’t they simply measure the scatter by computing the average of the deviations without bothering to square them?