Ability or trait: two tests, two objects

Before knowing whether someone is gifted, one must decide whether one is talking about what they can do at best or what they usually do. Differential psychology has long distinguished maximal performance, assessed by tasks with correct answers, from typical performance, reported by the person.

This distinction determines everything else: the method, the mode of validation, and the kind of conclusion possible. Konstantinos Petrides and Adrian Furnham applied it to our subject by distinguishing an ability emotional intelligence from a trait emotional intelligence.

The reference instrument, published in 2002 by John Mayer, Peter Salovey and David Caruso, comprises around one hundred and forty items across eight tasks: assessing the emotion present on a face or in a landscape, saying which mood would favour a given activity, predicting how a state will evolve, choosing the most effective conduct.

The central difficulty remains: deciding what a good answer is. Two methods coexist, the majority response of a large sample and the judgement of an expert panel. They agree, which the authors present as an argument, and the objection remains that a majority is not a truth.

The questionnaire developed by Petrides comprises, in its long version, some fifteen facets grouped into four factors: well-being, self-control, emotionality, sociability. It openly measures a perceived emotional efficacy, that is, a self-assessment.

That frankness changes how the results should be read. The analyses situate this measure, moreover, within the space defined by the five major personality traits, which its critics take for redundancy and its promoters for an assumed theoretical choice.

If the two families measured the same thing, their scores would converge. The correlations reported are weak, generally below 0.30, sometimes close to zero depending on the sample.

The practical consequence is direct: taking two emotional intelligence tests and obtaining contradictory results indicates no incoherence in the person. It indicates that the two instruments bore on different objects.

An ability task resists manipulation better, since one must know the right answer in order to give it. A self-report questionnaire adjusts easily: studies that ask participants to present themselves in their best light obtain markedly higher scores.

Hence a simple and often ignored rule: self-report suits a self-knowledge exercise, where one has no interest in flattering oneself, and suits a hiring decision badly, where one does. The problem is not the quality of the instrument, it is the situation in which it is taken.