Four words that change what a statistic means: the edit most readers 2026

Most arguments about numbers are really arguments about sentences. Everyone agrees on the figure. What nobody agrees on is what it is taken to mean, and that gap is usually only a few words wide. Four words that change what a statistic means are rarely dramatic on their own, which is exactly why they slip past readers. See also writing tips.

I started noticing this while proofreading headlines, the same way small slips show up in common grammar mistakes or a misspelled lynchpin. Related context in laptop deals.

Once you spot how a qualifier shifts a claim, you read reports differently. More background in what skills can. Wider detail in typing setup.

The sentence everyone writes

Across 116,382 completed professional singles tennis matches, a player who loses the opening set goes on to win 16.52% of the time.

That sentence is true. It is also the version you will hear in commentary, written as “about one in six”, and as written it cannot be used for anything.

Now add four words to the question behind it. Not “how often does a player come back”, but “come back from what”.

– a first set lost 7-6: they win 21.92% of the time, across 14,265 matches

– lost 6-4: 19.41%, across 25,546

– lost 7-5: 18.89%, across 10,252

– lost 6-3: 17.39%, across 25,970

– lost 6-2: 13.30%, across 19,739

– lost 6-1: 11.06%, across 14,269

– lost 6-0: 7.18%, across 5,833

Top to bottom, a threefold spread hiding inside one number. The two groups in the middle of that list are nearly identical in size, 14,265 and 14,269. One recovers twice as often as the other.

Why the original sentence was the problem

The word doing the damage was “a”. A player. It implies the players in that group are alike, and the data says they are not. They differ by how badly the first set went, and that single dimension carries almost all of the signal.

“One in six” is a true sentence about a group that should never have been treated as one group.

This is the most common failure in writing about numbers, and it has nothing to do with arithmetic. The average is calculated correctly. The sentence around it quietly asserts that the thing being averaged is a single kind of thing.

Three habits that catch it

Name the group out loud. Write “players who lost the first set, at any score, on any surface, at any level” and the sentence starts arguing with itself. If naming the group makes it sound mixed, it is. 

Say what the number cannot tell you. “16.52% across all first-set scores, moving from 7% to 22% depending on that score” is barely longer and is actually usable.

Give the size of each group. 14,265 matches and 14,269 matches can be compared directly. Four hundred against forty thousand cannot, and a reader who sees both counts knows that without being told.

Where context quietly changes the baseline

The same records show there is often no single normal value to report. Across 2,239,440 service games, players hold serve 74.43% on grass. On hard courts it is 69.95%, on clay 65.98%. A hold rate is unreadable without naming the surface.

By competition level it moves again. Men’s professional matches hold 78.62% across 230,715 games and women’s 64.36% across 266,503. Both of those reproduce the rates published elsewhere, from a count tuned to neither, which is how you know the measurement is sound.

Any sentence quoting one of those figures without its condition is making a claim the data does not support.

The habit I built after one embarrassing correction

I once quoted a survey as saying most people preferred a feature, and a colleague asked me to read the exact sentence again. It said most people who had tried it preferred it. Those four words, who had tried it, shrank the claim from everyone to a self-selected group, and my confident summary was suddenly wrong. I keep study routine bookmarked. I point beginners to desk setup.

Now I keep a simple rule: before repeating any statistic, I copy the original sentence and underline every qualifier. Anything like among, of those, reported or so far gets circled, because each one quietly resets the baseline. I also check the surrounding paragraph, since a figure that looks dramatic in a headline often reads as modest in context. It takes thirty seconds and has saved me from more than one correction, and it works whether I am reading a survey, a spec sheet, or a monitor roundup. I covered that in home printing. I tested it, see reading list.

The small discipline

Before publishing a number, read the sentence it sits in and ask which word is doing the generalising. It is usually a short one: a, the, average, typical. Then either defend that word or add the four that make it honest.

The figures here come from the Live Tennis API, a JSON service for developers. It carries live scores, fixtures, players and match events, with completed results and point-by-point history. They are published at blog.livetennisapi.com/studies.json under CC BY 4.0, free to reuse with attribution. Every figure above can be checked.

Method: completed best-of-three professional singles matches, 1 January 2023 to 14 September 2026. Retirements and walkovers excluded. Counted from the point-by-point record. 

Read more: https://grammarreader.com/what-google-ads-actually-costs-law-firms-in-competitive-practice-areas/

https://grammarreader.com/how-continuous-processing-systems-reshape-manufacturing-documentation-standards/

https://grammarreader.com/how-language-precision-shapes-the-classic-mustang-market/