Lost in the crowd? Using eye-tracking to investigate the effect of complexity on attribute non-attendance in discrete choice experiments

Spinks, Jean; Mortimer, Duncan

doi:10.1186/s12911-016-0251-1

BMC Medical Informatics and Decision Making

Table 4 Summary of main results from eye-tracking regression of attribute non-attendance (ANA)

From: Lost in the crowd? Using eye-tracking to investigate the effect of complexity on attribute non-attendance in discrete choice experiments

	(1) Number ANA, fe			(2) Number ANA, re			(3) Number ANA (conv), re			(4) Number ANA (CM), re			(5) Consistency^a
	b	se	p	b	se	p	b	se	p	b	se	p	b	se	p
complexity	0.578^b	0.210	0.006	0.578^c	0.230	0.012	0.749^b	0.247	0.002	0.863^b	0.309	0.005
complexity2	−0.036^d	0.019	0.057	−0.036^d	0.020	0.081	−0.050^c	0.024	0.037	−0.061^c	0.027	0.027
joint	−0.182^d	0.097	0.062	−0.182^d	0.097	0.060	−0.046	0.132	0.728	−0.013	0.094	0.887
late appointment				0.481^d	0.275	0.080	0.438	0.308	0.155	0.680^d	0.412	0.098	0.019	0.011	0.112
forward order				0.005	0.140	0.973	0.154	0.210	0.462	0.029	0.228	0.897	−0.006	0.007	0.412
female				0.104	0.144	0.471	0.017	0.187	0.927	0.234	0.188	0.214	0.003	0.007	0.655
age				0.001	0.037	0.982	−0.022	0.046	0.630	−0.015	0.053	0.775	−0.004^c	0.002	0.038
age2				0.000	0.000	0.577	0.001	0.000	0.134	0.000	0.001	0.482	0.000^c	0.000	0.017
<uni education				0.218	0.200	0.275	0.431^d	0.223	0.053	0.010	0.255	0.970	−0.002	0.010	0.817
student				−0.019	0.156	0.905	0.024	0.187	0.896	−0.125	0.237	0.598	−0.009	0.009	0.346
vit (self-selected)				−0.250	0.176	0.154	−0.183	0.239	0.444	−0.328	0.277	0.236	−0.011	0.008	0.190
vit (prescribed)				0.116	0.171	0.496	−0.189	0.192	0.325	0.456^c	0.218	0.036	0.014	0.008	0.104
other CAM				0.027	0.145	0.853	0.088	0.154	0.568	−0.022	0.188	0.906	0.007	0.006	0.312
Constant	−1.348^c	0.532	0.012	−1.737^c	0.765	0.023	−1.975^d	1.055	0.061	−1.970^d	1.083	0.069	0.087^c	0.041	0.046
Observations^e	255			255			255			255			32
R ²	.210			.276			.356			.245			.612

Abbreviations: ANA attribute non-attendance OR attributes not attended (to), complexity2 complexity squared, age2 age squared, uni university, conv conventional medicine, CM complementary medicine, vit (self-selected) taken a vitamin, mineral or herbal supplement not prescribed by a medical doctor in the past 12 months; vit (prescribed) taken a vitamin, mineral or herbal supplement prescribed by a medical doctor in the past 12 months; other CAM used other complementary and alternative medicine products or therapies (here it includes Western herbal medicine; Chinese medicine; CAM practitioners, or indigenous or traditional folk therapies)
^aAs measured by the mean(s_ij-S)² where s is the proportion of attributes attended to in choice set j by individual i and S_i = mean (s) for individual i [whereby a higher value indicates less consistency and more deviation in terms of attribute non-attendance]
^b, ^c,^d indicates significance at the 1, 5 and 10 % levels respectively
^eObservations are based on data from 32 participants, however, eye-tracking data is absent for question 8 for one participant (124)
We test for the appropriateness of using a random effects model using a robust Hausman test using a Wald test and cluster-robust standard errors (Wooldrige, 2002) after excluding participant 124 for whom there is missing eye-tracking data for question 8 (the scalar theta cannot be calculated for an unbalanced panel). The null hypothesis assumes that individual effects are random and both fixed and random effect estimators are consistent. The test does not reject the null (p = 0.652). We also perform an over-identification test with the null-hypothesis (participant 124 included) that the group means are uncorrelated with the idiosyncratic error term. The test does not reject the null (p = 0.911). From this we conclude that the random effects estimator results are appropriate

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com