Table 3.

Factors associated with the WTV during vaccination campaigns.

Logistic regression			LR chi2(10) = 155.38
Number of observations = 304			Log likelihood = −24.606224
LR chi2(10) = 155.38			Pseudo R2 = 0.7595
Variables		Coefficient	Odds Ratio	95% confidence interval		P-value
Information availability/access	Yes*
	No	−4.8	0.008	0.0008	0.0874	<0.001
Information channel	Radio*
	Places of worship	2.7	0.070	0.0043	1.1342	0.061
	Town crier	1.0	0.364	0.0277	4.7699	0.441
	Word-of-mouth	0.7	0.506	0.0513	4.9889	0.560
Benefit of vaccination	Yes*
	No	−2.3	0.102	0.0147	0.7134	0.021
Cattle vaccination frequency	Every vaccination campaign*
	Every year	−4.1	0.016	0.0012	0.2357	0.002
	Many years	−5.7	0.003	0.00005	0.2184	0.007
	Never	−4.3	0.014	0.0006	0.3557	0.010
Vaccination protects others	Disagree*
	Moderately agree	0.8	2.279	0.1125	46.1685	0.591
	Agree	4.9	129.431	1.1750	14257.18**	0.043

A common practice would consist to fuse the reference category with the other levels of the variable that are not significantly different from the reference category. We proceed to these supplemental analyses but this process did not give conclusive results. For the “Information channel” variable, the new reference category resulted from the regrouping provided p-values of 0.868 and 0.771 for the odds ratios.

^*Reference category.

^**Abnormally wide confidence interval can raise with small sample size or when some variables have several categories with small frequencies.