Congratulations, You're in the Top 10% (of a Group Someone Sliced Until You Were)

Every American hospital is, apparently, excellent. So is every school district, most employers, and an alarming share of dental practices. This is not a miracle of American institutions. It's a magic trick, and like all good magic tricks, it depends on you not watching the hands.

Let's watch the hands.

The Slicing Trick: Chop Until Someone Shines

The oldest move in the ranking racket is the arbitrary slice. Take a group, cut it into pieces until one piece looks great, then brag about the piece.

A hospital that can't crack the national top 100 can absolutely be a 'Top 10 Hospital for Minimally Invasive Gallbladder Surgery in the Tri-State Area.' The performance didn't change — the denominator did. This is how US News-style rankings and their countless imitators get mined by marketing departments: not by being excellent, but by finding the category where excellence is easiest.

The tell is always the denominator. When you see a superlative, your first question shouldn't be 'how good?' It should be 'good compared to what, and how many others were in that comparison?'

Regression to the Mean: Why Today's Best Is Often Tomorrow's Average

Here's the deeper problem, and it's pure statistics: extreme results tend to be flukes, and flukes fade.

Pick any group — schools, hospitals, sales teams — and rank them. The top performers will be a mix of genuinely great operations and ordinary ones that had a lucky year. Because luck doesn't repeat, the next year's rankings shuffle, and the top group drifts back toward the middle. This is regression to the mean, and it's been fooling people since statisticians first noticed that very tall parents tend to have slightly shorter children.

The practical consequence: 'best of' lists are snapshots of a moment, and a big chunk of what they capture is noise. A district that jumps from 40th to 8th in test scores probably didn't transform its teaching. It probably had a good draw. Meanwhile, the district that 'fell' to 40th from 8th is often the same school with the same teachers having a bad year.

A Field Guide to Spotting the Racket

You don't need a statistics degree. You need three reflexes:

1. Ask for the base rate. 'Top 10% of workplaces' — says whom, among how many? A list of 40 self-nominated companies is not a ranking; it's a press release. Job postings are the worst offenders: 'awarded top workplace five years running' frequently means the company paid to be on a list that companies pay to be on.

2. Check if the average is doing any work. 'Our students outperform the national average' can mean the district screens out kids who struggle — poverty, disability, English learners — before the comparison ever starts. Averages over unequal groups mostly measure who's in the group.

3. Look for the trend, not the snapshot. One year of data is a coin flip. Three years of consistent performance is a signal. Anyone selling you a ranking based on a single year is selling you noise with a ribbon on it.

Why This Persists (It's Not Just Dishonesty)

To be fair, some of this is structural, not sinister. People genuinely want simple answers — which hospital? which school? — and rankings are a compression algorithm for complicated reality. The problem is that compression discards exactly the information you need: sample sizes, who's being compared, and how much year-to-year wobble is baked in.

The result is a culture where 'above average' has become a participation trophy with better branding. Mathematically, of course, not everyone can be above average. In American marketing, everyone manages it constantly.

So the next time a hospital billboard, a realtor's flyer, or a LinkedIn post announces a top-decile honor, do the thing the design forbids: squint at the fine print, count the denominator, and ask what the ranking would look like next year. Usually the honest answer is 'about average.'

Which, for most things, is exactly where most things are. And that's fine. The trick is refusing to pay a premium for a statistic that was sliced, diced, and printed on a banner.