Latent variable modeling to develop a robust proxy for sensitive behaviors: application to latrine use behavior and its association with sanitation access in a middle-income country

Table 4 The parameter estimates for the final 2-class model that included 5 indicators: item conditional probability, standard error (SE), and response to the survey question

Indicator	Consistent Latrine Use		Inconsistent Latrine Use
Indicator	Question Response	Item Probability (SE)	Question Response	Item Probability (SE)
During the dry season, I think that most of the men in my village regularly use a latrine.	Yes	0.85 (0.03)	Don’t Know	0.53 (0.11)
During the rainy season, I think all of my neighbors regularly use a latrine.	Yes	1.00 (0.00)	Don’t Know	0.57 (0.11)
During the rainy season, I think that most of the children in my village regularly use a latrine.	Yes	0.96 (0.02)	Yes	0.60 (0.08)
There are too many people in this household for one latrine.	No	0.52 (0.03)	No	0.70 (0.07)
If my household did not have its own latrine, I would use my neighbor’s latrine.	Yes	0.90 (0.02)	Yes	0.71 (0.07)

The overall probability of assignment into the consistent latrine use membership class is 0.78 while assignment into the inconsistent latrine use class was 0.22. The model fit statistics include BIC = 1527; X2 = 395; and relative entropy = 0.86

ISSN: 1471-2458