Table 2.

(A) Maximum likelihood estimates of structural models with Fechner error terms for each of the four risk preference elicitation methods. Standard errors, clustered on the subject level, are reported in parentheses. (B) Pairwise differences in point estimates of risk preference parameters $\phi$ (lower-triangular matrix) and the standard deviation of noise parameters $\sigma$ (upper-triangular matrix) between the four risk preference elicitation methods

	bret	cem	mpl	scl
Panel A
$\phi$	0.626***	0.838***	0.602***	0.387***
	(0.021)	(0.090)	(0.033)	(0.034)
$\sigma$	0.046***	0.263***	0.977***	0.720***
	(0.002)	(0.048)	(0.066)	(0.057)
$lnL$	− 5,298	− 458	− 600	− 572
No. of Obs.	19,800	1782	1980	990
Clusters	198	198	198	198
Panel B
bret		− 0.217***	− 0.932***	− 0.674***
cem	0.212*		− 0.715***	− 0.457***
mpl	− 0.025	− 0.237**		0.257**
scl	− 0.240***	− 0.452***	− 0.215***

p values are based on pairwise Wald tests. bret, cem, mpl, and scl denote the “bomb” risk elicitation task, the certainty equivalent method, the multiple price list, and the single choice list, respectively. *$p<0.05$, **$p<0.01$, ***$p<0.001$