IRI: 28 questions across four kinds of empathy
Empathy gets treated as a single quality, something you either have or lack. Davis argued in 1980 that it is at least four things, and built the IRI to keep them apart: taking another person's perspective, getting absorbed in fiction, feeling concern for someone worse off, and feeling anxious yourself when someone else is in trouble. Twenty-eight items, seven per scale.
Answers run 0 (does not describe me well) to 4 (describes me very well). Each subscale scores 0 to 28, and nine items are reverse-scored. There is no total worth reporting, because the four scales measure different things. A person can be high on concern and high on personal distress at the same time, and those two point in opposite directions when it comes to how well things go.
| Scale | Items | Reverse-scored | What it covers |
|---|---|---|---|
| Perspective taking | 3, 8, 11, 15, 21, 25, 28 | 3, 15 | Spontaneously imagining how things look from someone else's position |
| Fantasy | 1, 5, 7, 12, 16, 23, 26 | 7, 12 | Getting drawn into the feelings of characters in books, films and plays |
| Empathic concern | 2, 4, 9, 14, 18, 20, 22 | 4, 14, 18 | Sympathy and care aimed at another person's misfortune |
| Personal distress | 6, 10, 13, 17, 19, 24, 27 | 13, 19 | Your own anxiety and discomfort in tense emotional situations |
Reliability
Subscale alphas run .68 to .79. Retest reliability over 60 to 75 days runs .61 to .81, and other reports put it between .62 and .71. Those numbers are acceptable for a 7-item scale and no better than that, so a single subscale score carries a fair amount of noise.
Structure
A principal components analysis in 1988 supported Davis's four factors. A later study of 409 students using Schmid-Leiman orthogonalization found the four first-order factors plus two second-order ones, which the authors labelled general empathy and emotional control. So the four scales do group into two broader tendencies if a shorter summary is needed.
What each scale means in practice
Davis reported that the four scales behave differently against outside measures. Perspective taking went with better interpersonal functioning and higher self-esteem, and showed no relationship to intelligence tests. Fantasy went with emotionality and with intelligence, and not with social functioning or self-esteem. Empathic concern tracked emotionality and concern for others. Personal distress went with lower self-esteem and poorer interpersonal functioning.
That last scale is the one people misread. Feeling upset because someone else is upset is not the same skill as understanding them, and it predicts worse outcomes rather than better. A high personal distress score next to a low perspective taking score describes someone who absorbs other people's distress without being able to work with it.
Clinical findings
A meta-analysis of 32 studies compared people with schizophrenia against healthy controls. Empathic concern was lower (g = 0.28), perspective taking lower (g = 0.53) and fantasy lower (g = 0.19), while personal distress ran higher, with a moderate to large effect. Longer illness duration went with lower perspective taking, and gender moderated personal distress.
Adaptations and gaps
Dutch (2007), couples (2010, reworded to ask about a partner) and Colombian Spanish (2017) versions exist, along with a brief B-IRI. No formal Chinese version has been published, so the Chinese items here are a translation of the original list rather than a validated adaptation. Norms are not published either; the author's own page offers means and standard deviations on request. Compare your four subscale scores against each other rather than against an absolute line, since no line exists.
None of the four scales is a diagnosis. A low score on one does not mean you lack empathy, and a high score does not make you kind. If the pattern bothers you, it is workable in therapy. This page is not a diagnosis and does not replace one.