The harmonization goal is to estimate personality scores that are not biased by the number of items and the specific inventory used. In the field of IRT, such harmonization is termed ‘test linking’. By fitting IRT models (Lord 1980) to item data, personality scores can be estimated conditional on the observed items and their respective item parameters. This leads to personality scores for individuals that are comparable irrespective of what items were assessed in a particular individual. For example, image an intelligence assessment: If we know that items 1–10 are very easy test items, and items 11–20 are very difficult, we are pretty confident that a person that scores 1 on the items 1–10 is less bright than a person that scores 9 on items 11–20. The exact knowledge of the difficulties of the 20 items allows us to estimate the difference in intelligence.
