Illusory Correlation
🇳🇴Illusorisk korrelasjonDefinition
Illusory correlation is the tendency to see relationships between two variables that are actually unrelated – or to exaggerate a weak relationship. We remember the cases that confirm the pattern and forget the ones that contradict it, ending up with a 'rule' that has no basis in data.
Real-world example
'It always rains when I forget my umbrella.' In fact it rains about as often when you remember it – you just don't notice. The brain registers the emotionally loaded combination (forgotten umbrella + rain) and builds a fictional pattern.
Hamilton and Gifford (1976) showed the effect experimentally: Participants read statements from two fictional groups, one smaller and one larger. Negative statements were proportionally equal in both groups, but because negative statements are rarer than positive ones, the combination 'small group + negative behavior' was doubly unusual and therefore more memorable. Participants overestimated how negative the small group was. This is a foundational mechanism behind stereotypes about minorities.
Supplementary perspective
Illusory correlation is one of the mechanisms behind conspiracy thinking, superstition, and folk medicine. Homeopathy, 'lucky socks,' and folk psychology thrive on patients/practitioners remembering the hits and forgetting the misses. Base rates and systematic data (controlled trials, registry data) are the key antidote.
Practical advice
Recognize
- —Notice 'always' claims about relationships you haven't actually counted.
- —Check whether you're only remembering the hits and not the misses.
- —Be extra alert when the pattern supports something you already believe.
Counteract
- —Fill in all four cells of a 2×2 table (X yes/no × Y yes/no), not just the confirming one.
- —Keep a written log to capture both hits and misses over time.
- —Demand base rates and control groups before accepting a 'relationship.'
Ethical use
- —Communicate statistics with absolute numbers and full 2×2 tables, not just the share with both traits.
- —Be cautious with anecdotal 'evidence' in journalism and management.
- —Examine whether your own prejudices about groups are built on unusual, memorable individual cases.