Almost nobody is average on every dimension at once, so designing to the mean fits no one

What it is

The design-to-the-average fallacy is this: a person may be close to the population average on any single body dimension, but almost no real individual is close to average on several independent dimensions at once. A cockpit, seat, or console that only fits someone when their height, sitting height, arm length, and leg length all match simultaneously — if designed around an "average person" assembled by averaging each dimension separately — corresponds to essentially no real human being. The result fits no one, not "fits most people except the edges."

Why it happens

This conclusion isn't a guess — it comes from a measurement that became one of the founding stories of human factors engineering. In the early 1950s, the US Air Force found that cockpits and equipment designed around 1926 anthropometric data and an "average pilot" baseline were producing frequent fit failures. It commissioned measurements of ten body dimensions on over four thousand pilots and checked, dimension by dimension, whether each pilot fell within ±30% of the average — an "average range" — on each. The result: even requiring simultaneous fit on just three or four dimensions left very few pilots qualifying, and requiring all ten dimensions to fall within their average ranges at once left zero pilots. Not one of the more than four thousand men was "average" on every dimension.

The mathematics behind this is the same mechanism discussed elsewhere in this group for compounding coverage: if roughly 30% of people fall within the "average range" on a given dimension, and the dimensions are largely independent, then the share falling within the average range on all ten dimensions simultaneously is approximately 30% raised to the tenth power — a number far below 1%. The more dimensions involved, the more the probability of an "average person" existing collapses exponentially toward zero. This isn't bad data or poor sampling — it's a mathematical property of multivariate distributions: the average is a statistical property of the distribution, not a portrait of any specific individual, and once enough dimensions are involved, an individual matching every one of them essentially does not exist in reality.

Studying it

Verifying this is extremely direct: take real measurements of the target population across the relevant dimensions, check each individual against each dimension's own "average range," and then track how the share still qualifying on every dimension drops as the number of dimensions checked simultaneously increases. This calculation can be re-run for any product's target population using the specific set of dimensions that product actually cares about — it doesn't depend on whether a historical finding still applies to today's population.

Where it stops holding

This fallacy targets fixed, non-adjustable designs that require several independent dimensions to be satisfied by the same individual at once. It does not invalidate single-dimension percentile selection — choosing the 5th or 95th percentile on one dimension and leaving adequate margin in the corresponding structure remains valid, because it doesn't assume an "average person" exists; it simply draws a coverage line on one dimension. The fallacy specifically targets combining averages across multiple dimensions into a fictional "standard person" and then locking a fixed design to that composite.

Applying it

• For any design requiring several independent body dimensions (seat height, fore-aft position, backrest angle, pedal distance) to fit the same person simultaneously, don't lock each dimension to its average value — give every dimension an independent adjustment mechanism so people of different builds can each dial in a combination that fits them, instead of everyone sharing one fixed combination computed from averages.
• The acceptance criterion cannot be "does an average-sized person fit" — by the mechanism above, a person matching every average dimension essentially doesn't exist, so that acceptance criterion is wrong from the start.
• Verification: recruit real participants from the target population spanning the full percentile range on each relevant dimension (not people close to average), have each of them use the actual adjustment mechanism to find a configuration that fits them, and confirm the adjustment range itself covers the two extremes of body variation in the population — not merely that "most people" can manage.

Related

• Same group: A11.06.1 Percentile selection and design coverage · A11.06.2 Hand length, finger width, and grip dimensions · A11.06.4 Static dimensions vs. dynamic working dimensions
• Adjacent: A8.25 Reach envelope and range of motion
• Search terms: average man fallacy · Gilbert Daniels · cockpit anthropometry · adjustability