Same Average, Wildly Different Consistency Statistics

Two batsmen can have exactly the same mean score (53 runs) and the same median score (53 runs) across a season, and yet be genuinely very different players: one's scores range wildly from 0 to 117, while the other's stay tightly between 46 and 60 every single time. Central tendency (mean, median, mode) alone cannot distinguish a wildly inconsistent player from a reliably steady one -- a genuinely different kind of measure, one that captures SPREAD rather than centre, is needed.

Mean deviation measures spread directly: find each value's distance from the mean (ignoring whether that distance is above or below, using absolute value), then average all those distances. For the data set 4, 7, 8, 9, 10, 12, 13, 17 (mean 10), the individual distances are 6, 3, 2, 1, 0, 2, 3, 7, summing to 24, giving a mean deviation of 24 divided by 8, exactly 3. This works, but has a real mathematical limitation: because absolute value strips away the plus-or-minus sign, mean deviation can't be manipulated further using ordinary algebra, which motivates a different approach entirely.

Squaring each value's distance from the mean (rather than taking its absolute value) keeps the algebra fully workable -- but comparing the RAW total of squared distances between two data sets of DIFFERENT sizes is a trap: a set of 31 observations will almost always have a bigger raw total than a set of just 6, purely because it has more terms being added up, not because it's genuinely more spread out. Dividing that raw total by the number of observations (giving the mean of the squared distances, called variance) fixes this completely, making the comparison fair regardless of how many observations each data set actually has.

Standard deviation, the square root of variance, brings the spread measure back into the same units as the original data (variance itself is in SQUARED units, which isn't directly comparable to the data). Two clean, provable rules follow: adding the same constant to every value in a data set leaves the variance completely unchanged (since every value AND the mean shift together by the same amount, so every distance-from-the-mean stays exactly the same), while multiplying every value by a constant k multiplies the variance by k squared (and the standard deviation by k itself, since distances scale by k and get squared).