One theoretical model (see Nagler, 2014) suggests that exposure to conflicting health information arouses confusion and backlash, which spill into the rejection of unrelated health recommendations. While this pathway finds support in observational studies, it does not consistently hold when exposure is experimentally manipulated rather than self-reported. This dissertation aims to explore these incongruent effects. In study 1, 200 U.S. adults were randomly assigned to read a research brief about the benefits of e-cigarettes, the risks, neither, or both, using a 2x2 between-subjects design. Next, I assessed confusion, backlash against health recommendations, and intentions to follow unrelated health recommendations. I found significant effects but not as expected—those exposed to the risks-only brief expressed significantly less confusion than others. The studies that follow build from this unexpected result, assessing embedded topic perceptions and moderating traits. Study 2 consists of two iterative pre-tests used to select health topics (2a; n=300) and to develop and pre-test message stimuli (2b; n=507). Study 3—an observational survey—replicates and validates Nagler (2014) with a new set of metrics (n=700). Although pre-specified analyses were replicated, I found a positive direct association between self-reported exposure to conflicting information and intentions to follow established and novel health recommendations. Study 4—a message effects experiment—assesses the effects of both manipulated exposure to conflict and embedded, a priori conflict through a 3x2+1 factorial design (n=700). Those exposed to conflicting messages—vs. neutral messages—reported higher confusion, backlash, and negative affect, and marginally lower novel intentions. Finally, I assessed contingent effects across both studies, asking if moderating traits affected assessed relationships. There were some significant interactions, but they were scattered and did not account for those relationships. Overall, the pattern in the prior literature of incongruence across results was not resolved here. While effects on confusion and backlash consistent with the literature were often found, evidence of contradictory message effects on behavioral intentions was inconsistent and sometimes positive rather than negative as expected.