Mental Models Daily
Explore
Menu

No. 262 · Evidence · Statistics and Probability

Data Dredging

Search enough data and some pattern will always turn up, by chance alone.

Data dredging means trawling through a large dataset, testing comparison after comparison, until something lines up. With enough tries, coincidences are guaranteed, and they look just as convincing as real effects. That is how a football result once seemed to predict the stock market.

To avoid it, decide your question and your measure before you look. Count how many comparisons were made. Treat surprising patterns as leads, not conclusions, and confirm them on fresh data before you act on them.

In the Almanac

“The Super Bowl indicator”, 1978 · “Biogen’s Aduhelm”, 2019–2024

Both stories in full, three ways to use it and where it misleads.

Get the Almanac

Listen · Watch

Data Dredging

Questions

What is Data Dredging?

Data dredging means trawling through a large dataset, testing comparison after comparison, until something lines up. With enough tries, coincidences are guaranteed, and they look just as convincing as real effects. That is how a football result once seemed to predict the stock market.

How do you use Data Dredging?

To avoid it, decide your question and your measure before you look. Count how many comparisons were made. Treat surprising patterns as leads, not conclusions, and confirm them on fresh data before you act on them.

When does Data Dredging matter?

Data Dredging is one of the 68 mental models for evidence, the situations where you ask: Is this really true? Within that, it belongs to “When research goes wrong”, alongside Anecdotal Evidence, Hasty Generalisation and Non-Response Bias.

What is an example of Data Dredging?

Two stories show it at work: The Super Bowl indicator (1978) and Biogen’s Aduhelm (2019–2024). Both are told in full in The Mental Models Daily Almanac, which has two stories for each of its 365 models.

Which mental models are related to Data Dredging?

Texas Sharpshooter Fallacy, P-Values and Randomness. Data Dredging itself comes from statistics and probability.