The effect of context on responses to questions has been a prominent focus
in social and political survey research. However, little investigation of c
ontext effects has been done for the measurement of psychological construct
s. A measure of anger experience and expression in development uses vignett
es describing interpersonal situations that provoke varying degrees of an.-
er and require respondents to indicate their affective and/or expressive re
sponse to the situation. In this study, the consequences of pairing the two
questions for each vignette are investigated. Pairing the anger-experience
and likelihood-of-expression questions changes the item's context. Item re
sponse theory analysis similar to that used to detect differential item fun
ctioning was performed. For some of the items, responding to a single or pa
ired question affected the extremity of responses. The findings are discuss
ed in terms of their implications for personality measurement.