Why Your Sheet Reports a Median, Not an Average

Every fitting screen in the world offers an average, prominently, in a large font. It is the number golfers repeat afterwards and the number most fitting reports are built around. It is also the wrong summary statistic for what a fitting block actually is, and the reason is not complicated.

What an Average Assumes About Your Shots

An arithmetic mean is the correct summary when the values you are combining are all measurements of the same underlying thing, differing only by small symmetric error. Twelve readings of the same steel rod with the same calipers: average them, and you get closer to the truth than any single reading.

Twelve golf shots are not that. They are a mixture of strikes that did roughly what your swing intends and strikes that did something else entirely, and the something-else strikes are not symmetric. Almost every gross mishit loses distance and almost none gains it. A thin one goes short, a heel strike goes short, a hosel strike goes very short. There is no error mode that adds thirty yards.

The result is a distribution with a long tail hanging off the low end. Averaging that tail into the summary drags the reported number below anything you actually do on a golf course, and worse, it drags it by an amount that depends on how many bad ones happened to land in that particular block.

An Illustration From a Real Block

Two seven-iron shafts, six strikes each, carry distance in yards. The figures below come off a January block with the golfer's permission and the ordering is exactly as it happened.

Read the means and shaft B wins by seven tenths of a yard. Read the medians and shaft A wins by three yards. The entire reversal is caused by one strike: the 141, which was struck low on the face and which the golfer felt distinctly at the time.

The median is the honest answer here. Shaft A is a longer club for this golfer, and it also produced one poor strike in six. Those are two separate findings, both worth knowing, and mashing them into a single mean destroys both.

An average answers the question "what happened on the day". A median answers the question "what does this club do when you hit it", which is the question you are paying us for.

So Where Does the Bad Strike Go?

Onto the same sheet, in its own column, counted rather than blended. This is the part of the method that matters more than the choice of statistic.

Every block on your sheet carries three figures rather than one. The median, which describes what the club does. The interquartile spread, which describes how tightly it does it. And the discard count with reasons, which describes how often it went wrong. Shaft A above reads: median 164.5, spread 2.5, one discard for low face contact.

That is a complete description. If shaft A had produced three discards instead of one, the median would be unchanged and the recommendation would flip entirely, because a club you strike badly a third of the time is not a good club regardless of what it does when you catch it.

Why Not Just Delete the Bad Ones and Average the Rest

Plenty of studios do exactly that, and it is better than nothing. The problem is that it hands the fitter a discretionary lever. Which shots get deleted becomes a judgement made after seeing the numbers, and human beings are extremely good at finding reasons to delete the strikes that spoil the conclusion they were forming.

Our discard rules are written down, applied before anybody looks at the flight figures, and based on strike location and the golfer's own call rather than on the result. A shot is discarded because it came off the hosel or because the golfer said they lost it, not because it was short. Then the median is taken over what remains, which by then is a fairly clean set anyway.

Publishing the discarded rows is what keeps the rule honest. If we deleted a shot you thought was fine, you can see it on your own sheet and ask.

Spread Is the Underrated Column

Golfers read the distance figure and skip the spread, which is backwards for most of the bag. For an iron in particular, a candidate that is three yards shorter and half as scattered is the better club, because approach shots are a distance-control problem rather than a distance problem.

Spread is also where a shaft that is too light for you gets exposed. It rarely shows up as a poorer median. It shows up as a wider spread and a rising discard count, both of which a mean would smooth into invisibility.

Small Samples Deserve Modest Claims

None of this makes twelve strikes into a large sample. It is not one, and we say so during blocks when the numbers get close. Where two candidates finish inside a yard of each other with similar spreads, the correct conclusion is that the block could not separate them, and the decision moves to feel, price, or availability.

Being willing to write "no measurable difference" on a sheet is, in the end, the same discipline as choosing a median. Both are ways of refusing to manufacture a result that the data did not supply.