Skip to main content
. 2014 Feb 12;14:26. doi: 10.1186/1472-6874-14-26

Table 3.

Unweighted multilevel logistic regression of the association between unintended pregnancies and individual-level and municipality-level characteristics among Nicaraguan women

  1. Empty model 2. Model with individual-level variables 3. Model with individual-level and municipality-level variables
Measures of association (AOR, 95% CrI)*
 
 
 
 
 
Individual-level variables
 
 
 
 
 
Controlling behavior by a partner (yes)
---
1.29
(1.13–1.47)
1.28
(1.13–1.44)
Ever sexual abuse by any person (yes)
---
1.32
(1.06–1.62)
1.31
(1.03–1.62)
Ever physical/sexual IPV (yes)
---
1.45
(1.24–1.68)
1.44
(1.24–1.66)
Municipality-level variables
 
 
 
 
 
% of women exposed to controlling behavior by a partner
 
 
 
 
 
Lowest tertile
---
---
---
1.00
 
Middle tertile
---
---
---
1.10
(0.89–1.33)
Highest tertile
---
---
---
1.25
(1.03–1.48)
% of women exposed to sexual violence by any person
 
 
 
 
 
Lowest tertile
---
---
---
1.00
 
Middle tertile
---
---
---
1.05
(0.84–1.28)
Highest tertile
---
---
---
1.00
(0.75–1.33)
% of women exposed to physical/sexual IPV
 
 
 
 
 
Lowest tertile
---
---
---
1.00
 
Middle tertile
---
---
---
1.09
(0.90–1.32)
Highest tertile
---
---
---
1.20
(0.92–1.55)
Measures of clustering
 
 
 
 
 
Area-level variance (SD)
0.081 (0.026)
0.066 (0.025)
0.062 (0.029)
PCV
---
−18.5%
−6%
MOR
1.31
1.27
1.26
VPC 0.024 0.019 0.018

*Adjusted by residency, age, education, socioeconomic status, ever use of modern contraceptive methods, % of women using modern contraceptive methods at the municipality level, % of women illiterate at the municipality level, and % of the population living in extreme poverty at the municipality level. PCV = percentage of change in variance. MOR = median odds ratio. VPC = variance partition coefficient.

Bayesian estimates of the measures of association and clustering are shown. n = 5347.