Table 3:

Multinomial regression predicting the use of focal responses (50%, 0%, and 100%) and other responses.

	Raw responses					Responses rounded to nearest integer
Predictor	50%	0%	100%	Other		50%	0%	100%	Other
Response modes
Visual linear scale vs. open-ended	−.04***	−.05***	−.01***	.10***		−.02***	−.04***	−.01***	.07***
Magnifier scale vs. open-ended	−.07***	−.04***	−.01***	.12***		−.10***	−.06***	−.01***	.17***
Respondent characteristics
High numeracy	−.01**	−.01***	.00	.01***		−.01**	−.01***	.00	.02***
College education	−.01***	.00	.00	.01***		−.02***	.00	.00	.02***
Log of income	.00	.00	.00	.00		.00	.00	.00	.00
Age	>.00***	<.00***	<.00*	.00*		>.00***	<.00***	<.00**	>.00*
Gender (female)	.00	.00	<.00***	.00		.00	−.01*	−.00***	.01
Had diseasea	.01***	<.00*	<.00***	−.01*		.01**	<.00*	<.00***	−.01*
Explanations
No one can know the chances (if follow-up)	.02***	.02***	.00	−.03***		.03***	.02***	.00	−.04***
Other survey conditions
Follow-up condition asking for explanations	.00	.00	.00	.00		.00	.00	.00	.00
Question with follow-up asking for explanations	−.01*	.01**	.00	.00		−.01***	.01**	.00	.00
Dying Condition	−.01***	<.00**	<.00***	.01***		−.01***	<.00*	<.00***	.02***
Numeracy first Condition	.00	.00	.00	.00		<.00*	.00	.00	.00
Condition with questions about getting sick given prevention first	.00	.00	.00	.00		.00	.00	.00	.00

^aRefers to the disease that was mentioned in the question.

^***p<.001

^**p<.01,

^*p<.05

Note to Table 3: Dummy variables for individual questions are not shown. Standard errors varied between .01 and .001. Model statistics: pseudo-R²=.23, Wald χ2(90)=7153.53 (for raw responses), p<.0001; pseudo-R²=.19 Wald χ2(90)=7196.23 (for responses rounded to the nearest integer), p<.0001.