Table 4. Dwelling infestation by triatomine bugs in 39 municipalities, Cochabamba, Bolivia, 2000–2011^*: linear mixed model results, with year specified as an ordinal, fixed effect (number of parameters k = 15; BIC = 812.32; AICc = 757.11).

	Estimate	SE	CI95
Fixed effects: coefficient estimates			Lower	Upper
Intercept	−1.545	0.922	−3.394	0.304
Year [2003 vs. 2001]	−2.237	0.295	−2.817	−1.657
Year [2004 vs. 2003]	−1.525	0.234	−1.986	−1.065
Year [2005 vs. 2004]	0.324	0.221	−0.110	0.759
Year [2006 vs. 2005]	−0.478	0.182	−0.836	−0.120
Year [2007 vs. 2006]	0.341	0.210	−0.072	0.753
Year [2008 vs. 2007]	−0.095	0.209	−0.506	0.315
Year [2009 vs. 2008]	−0.167	0.183	−0.527	0.193
Year [2010 vs. 2009]	0.183	0.182	−0.175	0.541
Year [2011 vs. 2010]	0.186	0.190	−0.189	0.560
Autocorrelation (time)	0.140	0.047	0.048	0.233
Dry forest	1.401	0.413	0.575	2.227
Human Development Index	−1.778	1.255	−4.330	0.774

*No infestation surveys were conducted in 2002.

The model used N = 325 observations (year-specific municipality-level dwelling infestation rates), weighted by a measure of bug-search effort; the inclusion of the one-year lagged temporal autocorrelation covariate censored observations for year 2000 (see text for details).

BIC, Bayesian information criterion; AICc, second-order Akaike's information criterion.

Fixed-effect coefficient estimates are in bold if different from zero at the 5% level; SE, standard error; CI₉₅, lower and upper limits of the 95% confidence interval.

REML, restricted maximum likelihood; σ², random effect variance estimate; %σ², percentage of total variance that is attributable to differences in average infestation among municipalities (‘Municipality’), also known as intra-class correlation, and to year-to-year variation in infestation rates within municipalities (‘Residual’). Ratio = Municipality/Residual variance estimates.