The chameleon effect hypothesizes that the interpretation of esteem items a
nd the nature of the measurement of the construct are altered by the conten
t of other items in a survey. In each of three studies, responses to esteem
items embedded among items focusing on a specific self-concept domain (aca
demic, artistic, or physical) were more highly correlated to that specific
domain than were esteem items from a broadly based multidimensional self-co
ncept instrument. Confirmatory factor analysis models demonstrated that the
same esteem items embedded in different instruments measured distinct fact
ors. Unlike typical contextual effects showing mean shifts along the same u
nderlying continuum, these results suggest changes in the nature of the con
struct that is being measured so that mean shifts are of dubious relevance
The results have theoretical implications for how individuals form esteem j
udgments and practical limitations for the interpretation of esteem respons
es in correlational and experimental studies.