When Does the Average Person Not Actually Resemble Anyone in the Group?

An average can accurately describe a population while failing to describe any individual person within it. The problem becomes more pronounced when averages across several characteristics are combined into one statistical profile.

This sounds counterintuitive because additional information usually feels like additional accuracy. If average height tells us something about a population, then adding average weight, arm length, income, age, and other characteristics seems as though it should bring us closer to describing the typical person.

Sometimes the opposite happens. Each additional characteristic can make the statistical portrait more detailed while reducing the number of actual people who resemble it.

Gilbert Daniels and the Search for the Average Man

In 1952, Gilbert S. Daniels analyzed anthropometric measurements from 4,063 members of U.S. Air Force flying personnel.

He selected ten body dimensions useful for clothing and equipment design. Daniels then asked how many individuals could reasonably be considered average across all ten.

He did not require people to match the precise mathematical mean. Instead, he defined an approximately average range around each measurement that included a substantial portion of the population.

For any single characteristic, plenty of people qualified.

The problem appeared when the characteristics were combined.

The Average Population Produced No Average Person

Of the 4,063 men, 1,055 fell within Daniels's approximately average range for stature.

When chest circumference was added, only 302 remained.

Adding sleeve length reduced the number to 143.

Each new measurement eliminated additional people because someone close to the center in one dimension was frequently farther from the center in another.

After ten dimensions, none of the 4,063 men remained.

Every individual average had been calculated from real people.

The resulting average person was not one of them.

Why More Averages Can Make the Profile Less Realistic

Suppose a business knows that its average customer is 45 years old.

That is a straightforward summary of one characteristic.

Now add average income, household size, annual spending, number of purchases, education, commute time, and hours spent online.

The resulting customer profile looks much more informative.

It is certainly a more detailed statistical description of the population.

But it does not follow that the profile describes a more typical individual.

Every additional average imposes another condition an actual customer must satisfy.

The organization can therefore create an increasingly specific person while representing fewer and fewer people.

Population Statistics Are Not Personal Profiles

The distinction comes from understanding what an average actually belongs to.

Average age is a characteristic calculated from a group.

Average income is another characteristic calculated from that group.

Neither statistic promises that the person with average age also earns average income.

Once multiple averages are assembled into one profile, we have moved from describing a population toward constructing a hypothetical individual.

That construction needs to be tested against real people rather than assumed to represent them.

Averages Can Fail in Several Different Ways

The Daniels problem is not the only way averages can mislead.

A highly skewed distribution can produce an average far from what most individuals experience. Income provides a common example because a relatively small number of very high values can pull the arithmetic mean upward.

Distinct populations create another problem. If two groups cluster around very different values, their combined mean can land between them and describe neither group particularly well.

Daniels identified another issue entirely.

The individual means may each be perfectly sensible.

The mistake is assuming they naturally coexist.

Why the Distinction Matters

The difference becomes consequential when an average moves from describing a group to specifying what should be built for a person.

Average height can be useful for understanding a population.

Designing a product around average height, average arm length, average shoulder width, average reach, and average leg length assumes that those characteristics occur together frequently enough to describe actual users.

Daniels's work showed why that assumption can fail.

As more dimensions matter simultaneously, designing for ranges and variability becomes more important.

Designing for Variation

An adjustable seat provides a simple example.

Instead of assuming one fixed dimension will fit the average user, adjustment allows people with different combinations of physical characteristics to use the same design.

The same principle can apply to desks, controls, clothing, safety equipment, software interfaces, and many other products.

The objective shifts from building something for an imaginary central person toward accommodating a meaningful portion of the real distribution.

That distinction can dramatically change design decisions.

Businesses Have Average People Too

Companies regularly create customer personas from demographic and behavioral data.

A customer might be described through average age, household income, spending, family structure, location, purchase frequency, or product usage.

Any one of those measurements may reveal something useful.

The danger comes when the combined persona is treated as though it represents the typical real customer.

A company may then design advertising, pricing, products, or services around a demographic combination that barely exists.

Schools and Employers Face the Same Problem

An “average student” can conceal substantial differences in prior knowledge, learning speed, strengths, weaknesses, language background, and support needs.

An “average employee” can combine experience, working preferences, productivity, responsibilities, and career objectives in ways that describe few actual workers.

The issue is not that population summaries have no value.

It is that individuals arrive as combinations of characteristics.

Those combinations matter.

Ask Whether the Average Actually Exists

Before using a composite average to guide a decision, test it against the population from which it was created.

What exactly is being averaged?

How much variation exists around each number?

Are there distinct groups within the population?

How strongly are the characteristics related?

Most importantly, how many actual people resemble the combined profile?

If the answer is very few, the profile may still be useful as a statistical abstraction.

It should not be mistaken for a person.

Daniels's Zero

The remarkable part of Daniels's research is not that averages were inaccurate.

They weren't.

It is that several accurate averages could be assembled into something that represented none of the 4,063 people from whom the averages were calculated.

That is the distinction worth remembering.

An average can tell us something important about a population without telling us what any particular member of that population looks like.

The more characteristics we combine, the more important it becomes to stop asking only what the average is and start asking:

How many actual people look like the average we just created?