Causes Trump Statistics
People often neglect statistical base rates and instead rely on causal stories when making judgments, leading to systematic errors in probability estimation.
THE IDEA
Causes Trump Statistics
A plausible causal explanation can receive more weight than a relevant base rate. The question is whether the story adds evidence or merely makes one outcome easier to imagine.
Key Example: The Taxi Cab Problem
The classic demonstration involves a hit-and-run cab accident. Witnesses report that the cab was Blue, but the base rate of Blue cabs is only 15% (85% Green). The witness is 80% accurate in identifying colors. When asked for the probability that the cab was actually Blue, most people ignore the base rate and answer 80%, but Bayes' rule gives only 41%.
Helping Experiment (Nisbett & Borgida, 1975)
The cited work concerns learning from an experiment in which students failed to help someone apparently having a seizure. Knowing an aggregate finding and using it to predict an individual’s behavior are different tasks. This contrast matters when judging whether a statistical lesson has changed someone’s expectations.
Bystander Effect (Darley & Latané, 1968)
The bystander experiments examine how the presence of others can reduce an individual’s likelihood of helping. Diffusion of responsibility is a proposed causal explanation for that pattern.
Why It Matters
A coherent story can make a prediction feel justified before its frequency has been considered. Compare the story with the relevant base rate and ask what additional evidence supports departing from it.
When to use it
Evaluating diagnostic tests: When a test is positive for a rare disease, considering the base rate (prevalence) along with the causal story of symptoms.
Investment decisions: Ignoring market base rates due to a compelling narrative about a company's future.
Criminal justice: Jurors relying on a prosecutor's story rather than statistical evidence about recidivism or eyewitness reliability.
Personal relationships: Judging someone's character based on a single vivid anecdote instead of their overall track record.
FROM THEORY TO EVERYDAY LIFE
In practice
Taxi Cab Problem
In a city with 15% Blue cabs and 85% Green cabs, a witness says a cab involved in a hit-and-run was Blue. The witness is 80% accurate in color identification. Most people ignore the base rate and say there's an 80% chance the cab was Blue, but Bayes' rule shows the probability is actually 41%.
Helping Experiment (Nisbett & Borgida)
The cited helping experiment distinguishes being told how often people helped from predicting what a particular person would do. Repeating a study’s result does not by itself show that the result has been incorporated into individual predictions.
Bystander Effect (Darley & Latané)
The well-known finding that people are less likely to help in a group than alone. The causal story of diffusion of responsibility makes this counterintuitive result feel intuitive.
A MOMENT TO REFLECT
Put it to work.
Think of a time you made a decision based on a compelling story rather than statistical evidence. How might the outcome have changed if you had given more weight to the base rates?
In legal settings, eyewitness testimony often overrides base-rate information. How could you use this bias to argue for a more balanced evaluation of evidence?
When evaluating a new medical treatment, what steps can you take to ensure you are not ignoring base rates in favor of a convincing causal mechanism?
Often confused with
References
- Causal Schemas in Judgments Under Uncertainty — Amos Tversky and Daniel Kahneman (1980)
- Attribution and the Psychology of Prediction — Richard E. Nisbett and Eugene Borgida (1975)
- Bystander Intervention in Emergencies: Diffusion of Responsibility — John M. Darley and Bibb Latané (1968)
CONTINUE THE CONVERSATION
Source texts
- Thinking, Fast and Slow (Daniel Kahneman)