Averaging a wellness questionnaire destroys the only useful answer

Five items collapsed into one score let four ordinary responses bury the single alarming one.

Sleep, soreness, stress, mood, energy. Five sliders, five down to one, add them up and divide. The composite comes out at 3.8 and the athlete is filed as normal.

Inside that 3.8 could be a player who slept two hours. Fours across the board with a one for sleep averages to the same place as a mild pattern of threes and fours, and only one of those two mornings is worth a phone call. The averaging step was asked to summarise, and what it has done is delete the outlier, which was the only part of the response containing news.

This is a design choice people rarely notice making. It comes from wanting one column to sort by, and sorting is genuinely useful when you have forty players and eleven minutes. But the composite is doing two incompatible jobs. As a rough daily temperature of the squad it is fine. As a flag for the individual who needs attention today it is worse than useless, because it produces a specific false reassurance for exactly the profile you most want to catch: one catastrophic item surrounded by unremarkable ones.

There is a second problem in the arithmetic. The five items do not carry equal information. Soreness moves every day and moves a lot; mood barely shifts for most people. Adding them with equal weight means the noisy item dominates the variance of the composite while the stable item contributes almost nothing, so in practice the score is mostly a soreness score wearing a broader label. Nobody intended that. It falls out of summing things with different spreads.

The fix is to keep the composite for the group chart if you like it, and run a separate rule on the raw items: any single item at the bottom of its range triggers a look, whatever the total says. Two lines of logic. It catches what the mean throws away.

And the response wants to be a conversation rather than an intervention, because a one for sleep might be a newborn at home rather than anything the training plan caused.

Averages exist to turn many numbers into one. Worth remembering what that operation is for, and that the thing it is built to suppress is exactly the thing a monitoring system exists to find.