Table 2.

Preferences by latent class

Latent class results
Attributes		Class 1: find my cancer (43%)		Class 2: sensitivity/specificity trade-off (41%)		Class 3: wary of screening; particularly false results (16%)
		MLE (SE)	95% CI	MLE (SE)	95% CI	MLE (SE)	95% CI
1. Preferences
Alternative specific constant	Screening A	Reference	–	Reference	–	Reference	–
	Screening B	0.1123 (0.0729)	[− 0.0306; 0.2552]	− 0.0152 (0.0328)	[− 0.0796; 0.0491]	−0.1025 (0.0767)	[− 0.2529; 0.0479]
	No screening	− 3.3865 (0.2871)***	[− 3.9492; − 2.8239]	− 2.8686 (0.1631)***	[− 3.1883; − 2.5489]	0.0071 (0.1591)	[− 0.3047; 0.3190]
Number of cancers tested for	1-Type increase	0.0205 (0.0026)***	[0.0153; 0.0257]	0.0034 (0.0014)*	[0.0007; 0.0061]	0.0002 (0.0022)	[−0.0040; 0.0045]
True positives	1-Case increase	0.0484 (0.0045)***	[0.0397; 0.0571]	0.0076 (0.0014)***	[0.0048; 0.0103]	0.0131 (0.0018)***	[0.0096; 0.0165]
False negatives	1-Case decrease	0.0631 (0.0053)***	[0.0527; 0.0736]	0.0065 (0.0018)***	[0.0030; 0.0099]	0.0363 (0.0027)***	[0.0310; 0.0416]
False positives	10-People decrease	0.0085 (0.0007)***	[0.0070; 0.0100]	0.0046 (0.0004)***	[0.0039; 0.0053]	0.0078 (0.0009)***	[0.0061; 0.0095]
Cancer type unknown	1% Decrease	− 0.0636 (0.0112)***	[− 0.0856; − 0.0416]	0.0163 (0.0057)**	[0.0051; 0.0276]	0.0258 (0.0096)**	[0.0069; 0.0447]
2. Class membership
Constant	–	Reference	–	0.0128 (0.0619)	[− 0.1085; 0.1341]	− 0.9672 (0.0425)***	[− 1.0505; − 0.8839]

Analysis information: respondents = 1700; observations = 13,600; parameters = 23; LL = − 9807.2; BIC = 19833.3; APR = 34.21%

APR adjusted McFadden Pseudo-R2, BIC Bayesian information criteria, CI confidence interval, LL Log-likelihood, MLE maximum likelihood estimate, SE standard error, *p < 0.05; **p < 0.01; ***p < 0.001