Statistical Fallacies

We live in the age of data. Every day - charts, percentages, studies. But numbers can deceive just as well as words. Learning to spot statistical manipulation will give you immunity against most modern propaganda.

• **News**: "Crime rose 50%!" - from 2 to 3 cases in a small neighborhood
• **Advertising**: "9 out of 10 doctors recommend" - from a specially selected group of 10
• **Politics**: Charts with truncated Y-axes to dramatize insignificant changes

Gambler's Fallacy

The roulette wheel has landed on red 10 times in a row. The gambler thinks: "It's *definitely* going to be black now - red has come up so many times!" He bets everything on black... and loses. This is the **gambler's fallacy** - the belief that past random events influence future ones.

**Gambler's fallacy** - the mistaken belief that the probability of a random event depends on previous results. In reality, every coin flip or roulette spin is an *independent* event with an unchanged probability.

**The reverse error - hot hand fallacy**: "He's on a scoring streak, so he's *heated up* and will keep scoring!" If events are independent - the past doesn't predict the future.

Texas Sharpshooter

A Texan shoots at a barn, then draws a target around the bullet holes and brags: "Look, all hits in the bullseye!" This is the **Texas sharpshooter fallacy** - finding a pattern in random data *after* you've seen the data.

**Texas sharpshooter fallacy** - selecting or grouping data *after* it has been collected to create the illusion of a meaningful pattern. The problem: in any random dataset you can find clusters that look non-random.

**Defense:** ask "Was the hypothesis formed BEFORE or AFTER the data?" The scientific method requires: hypothesis first, then testing. Not the other way around.

Cherry-Picking

"Global warming is a myth! Here's a study that refutes it!" - says someone ignoring hundreds of studies confirming warming. This is **cherry-picking** - selecting only the data that supports your position.

**Cherry-picking** - selecting only those data points, examples, or studies that confirm a desired conclusion, ignoring contradictory ones. This is manipulation, even if each individual fact is true.

**Defense:** look for *meta-analyses* and *systematic reviews* - they analyze ALL studies. Ask: "Is this a typical example or an exception?"

Misleading Statistics

"Average salary at the company - $100,000". Sounds good! But if the director earns $900,000 and 9 employees each earn $11,000, the mean is $100k, but the median is $11k. **Misleading statistics** - when the numbers are correct but the interpretation is deceptive.

**Misleading statistics** - using statistically correct data in a way that creates a false impression. Includes: confusing mean/median, manipulating base rates/scales, selecting convenient metrics.

**Defense rules:** 1. Ask: mean or median? 2. Look at absolute numbers, not just percentages 3. Check graph scales 4. Find the base for comparison and context

Key Takeaways

• **Gambler's fallacy**: past random events don't affect future ones. The coin has no memory
• **Texas sharpshooter**: finding patterns after the data creates an illusion. The hypothesis must come before the data
• **Cherry-picking**: selecting only supporting data. Look for the full picture and meta-analyses
• **Misleading statistics**: mean vs median, relative vs absolute risk, graph scales

Related Topics

Statistical fallacies - about critical perception of data:

• Ambiguity Fallacies — Previous category - word manipulations
• Logic in Media — Applying all fallacies to media analysis

Вопросы для размышления

• Which statistics have you taken at face value without checking the context?
• Recall a case when an "impressive percentage" turned out to be insignificant in absolute numbers.
• How would you present negative statistics about your company to make it look positive?

Связанные уроки

• stat-05-hypothesis