Among those who considered health to be an important or very important reason to be physically active, the SQ condition resulted in less underestimation, though at borderline significance.
2
—
—
The sentences
Hierarchical logistic regression analysis using underestimation (scored as 1) versus no misperception (scored as 0 (reference group) as the dependent variable and order, involvement and the six reasons to be active as predictors showed a main effect of involvement, a borderline significant effect of question order and interactions between (1) health as a reason to be active and order and (2) involvement and order (Table 4 ).