Abstract
When making affective forecasts, people commit the impact bias. They overestimate the impact an emotional event has on their affective experience. In three studies we show that people also commit the impact bias when making empathic forecasts, affective forecasts for someone else. They overestimate the impact an emotional event has on someone else's affective experience (Study 1), they do so for friends and strangers (Study 2), and they do so when other sources of information are available (Study 3). Empathic forecasting accuracy, the correlation between one person's empathic forecast and another person's actual affective experience, was lower than between-person forecasting correspondence, the correlation between one person's empathic forecast and another person's affective forecast. Empathic forecasts do not capture other people's actual experience very well but are similar to what other people forecast for themselves. This may enhance understanding between people.
Acknowledgements
Monique Pollmann was supported by grant 400–03–102 from the Netherlands Organization for Scientific Research (NWO).
We thank Cary Rusbult, Kaska Kubacka, Jan-Willem van Prooijen & Paul van Lange and the PhD students of the Social Psychology Department at the Vrije Universiteit Amsterdam for helpful comments on an earlier draft of this article.
Notes
1We also conducted an ANOVA including order as a between-subject factor. Because there was no main effect of order of prediction or interaction effects with the other factors this factor was excluded from the analyses.
2To take statistical interdependence into account, we first estimated the variance explained on the dyad level using a hierarchical linear model approach. No reliable effects of the dyad level in explaining variance for the dependent measures emerged. Data in all three studies are hence analysed on the individual level.
3For some participants it was not possible to calculate the across-item correlation because they gave the same response on every item. This is why degrees of freedom vary across analyses.