The Better-Than-Average Effect

“Eighty-eight percent of US drivers and seventy-seven percent of Swedish drivers placed themselves in the top fifty percent for safety. The math does not allow this.”Ola Svenson, 1981

In 1981 Ola Svenson at Stockholm asked drivers in the US and Sweden to place themselves on a percentile scale against the rest of the driving population. Eighty-eight percent of the US sample put themselves in the top half. By construction, only fifty percent of any sample can be above the median. The founding paper named the effect.

Mark Alicke (1985) ran the trait sweep at Ohio State: 154 personality traits, percentile self-rating. Mean self-rating across all 154 traits came in at the sixty-fifth percentile. Patricia Cross (1977) at Nebraska asked college teachers to rate their own teaching: 94% above average, 68% in the top quartile. Brown (1986) ran the 1976 College Board high-school senior data and found 25% of 829,000 seniors placed themselves in the top one percent for sociability. None below average. Tappin and McKay (2017) found mean self-rated moral character sits at the eighty-fifth percentile, the largest BTAE effect ever recorded.

You are about to take ten traits. Driving safety. Sense of humor. Intelligence. Physical attractiveness. Leadership. Getting along with others. Empathy. Moral character. Future health. Investing. For each, you move a 0-100 slider for where you sit in the general adult population. Zero is the lowest one percent. Fifty is the exact median. One hundred is the top one percent. After you lock in, I show what twenty large samples of self-raters since 1976 put down on the same trait, and we measure the gap.

The math constraint never moves: the true population average is, by definition, fifty. Anything you place above that on average is the bias.