Consistent with hypotheses, results revealed a significant two - way crossover interaction between health literacy and
message emotionality (factual vs. emotional) on one of the six smoking risk perception items (β = − 0.03, SE = 0.01, p =.01 for item 1).
The
emotionality items and the framing items were summed to create separate composite indices, with higher scores reflecting a greater degree of agreement that
messages were perceived as intended.