Table 3

Estimated parameters of the alternative-specific conditional logit models predicting perceptions of product harm and motivation to not smoke

	Harm perception (n=693)		Motivation to not smoke (n=774)
	Estimate	SE	Estimate	SE
Within subjects
Brand
Marlboro (ref) versus Pall Mall	−0.223†	0.058	−0.177†	0.053
Marlboro (ref) versus Camel	−0.099	0.057	−0.084	0.054
Marlboro (ref) versus Basic	−0.318†	0.058	−0.208†	0.053
Tar
High (ref) versus low	−0.118*	0.060	−0.069	0.053
Low × numeric information	−0.580†	0.084	−0.276†	0.077
Disease statement
Cancer (ref) versus cardiovascular disease	−0.046	0.058	0.100	0.053
Cancer (ref) versus lung disease	−0.027	0.057	0.074	0.054
Cancer (ref) versus adverse pregnancy outcomes	−0.172†	0.057	−0.207†	0.055
Disease link
No link (ref) versus link	0.041	0.042	0.182†	0.039
Between subjects
Listing of constituents
Listed (ref) versus grouped
1: Choice A (left)	−0.075	0.065	0.029	0.064
2: Choice B (right)	0.036	0.064	0.128*	0.063
3: Both	ref	ref	ref	ref
4: None	0.145	0.217	0.162	0.131
Numeric information
No numbers (ref) versus numeric
1: Choice A (left)	0.289†	0.078	0.110	0.075
2: Choice B (right)	0.332†	0.076	0.080	0.074
3: Both	ref	ref	ref	ref
4: None	−0.032	0.216	0.023	0.131
Smoking status
Smoker (ref) versus susceptible non-smoker
1: Choice A (left)	−0.486†	0.085	−0.340†	0.080
2: Choice B (right)	−0.436†	0.082	−0.234†	0.077
3: Both	ref	ref	ref	ref
4: None	−1.326†	0.376	−0.878†	0.195

The parameter estimate represents the contribution of that attribute level to the final model. Parameter estimates reflect binary coding of attribute levels and control for likelihood of choosing ‘no difference’ and the two ‘no difference’ alternatives (ie, ‘both’ vs ‘none’).
*Significant at the p<0.05 level.
†Significant at the p<0.01 level.