Predictive validity
How well a score, such as a test or interview result, predicts the thing you actually care about later, usually measured as the correlation between the two.
In plain English
To measure it, you record the scores, wait, and then compare them with the later outcome: job performance, exam results, sales. A correlation of 1 would rank people perfectly; 0 is a coin toss. A method can feel convincing to the person using it and still have low predictive validity, because people rarely see how the candidates they rejected would have done.
Uncle Gary explains
Kevin's boss can tell in five minutes whether a candidate is good. A free-form interview correlates with later job performance at about 0.19; in our simulation it picks the truly best of ten candidates 16% of the time, against 10% for a coin.
Why it matters for your money
Ratings, rankings and 'expert picks' of funds or stocks are predictions too. Ask how well past scores predicted later results, not how confident the person giving them is.
Where we tested it
Related terms
All glossary terms. Educational content, not financial advice.