ABSTRACTIn survey-based tests of causal models, respondents respond to sequences of measurement items designed to measure variables specified in the causal model. Recent research focusing on measurement context effects suggests this method of testing causal models may be vulnerable to measurement context effects that distort the empirical findings. The purpose of this research is to test hypotheses concerning measurement context effects in tests of causal models via an experiment in which the measurement sequence is manipulated. Additionally, respondent knowledge is examined as a possible moderator of the hypothesized context effects. The results provide evidence of a small context effect that supports a hypothesis derived from the Feldman and Lynch (1988) accessibility/diagnosticity theory. Knowledge of the subject of investigation, the act of donating blood, was not a significant moderator of the measurement context effects.