Table 3.
Estimated posterior quantities from fitting the final Bayesian spatial logistic model
| Description of explanatory variable |
Parameter | Mean (SD) | 95% credible interval |
|---|---|---|---|
| Intercept | −3.3833 (0.1356) | (−3.6538, −3.1212) | |
| Individual-level covariates | |||
| Race (White or other) | −0.1544 (0.0932) | (−0.3351, 0.0305) | |
| Sex (male or female) | 0.0852 (0.0659) | (−0.0442, 0.2146) | |
| Housing quartiles | −0.0466 (0.0594) | (−0.1635, 0.0698) | |
| Mobile home community (yes or no) | −0.2448 (0.3210) | (−0.9061, 0.3547) | |
| Age (years) | 0.0117 (0.0098) | (−0.0076, 0.0309) | |
| Interaction: mobile home community, race | 0.8151 (0.3460) | (0.1603, 1.5198) | |
| Interaction: housing index, age | −0.0159 (0.0057) | (−0.0272, −0.0046) | |
| Random effects | |||
| Spatial component | 4.477 (1.397) | (2.351, 7.774) | |
| Family contextual effect | 3.169 (1.015) | (1.663, 5.608) |
Note: The Bayesian spatial logistic model modeled the binary outcome (any adverse event or no adverse event), where the probability was a function of a set of explanatory variables, a spatial component, and a family contextual effect. Adverse events were defined as physical or sexual abuse and accidents or poisonings at home among children in Olmsted County, Minnesota, a mixed rural–urban community, that occurred during the time period April 2004 through March 2009. Estimates of posterior quantities were obtained from the INLA package. The deviance information criterion was 8469.72. Precisions are presented for the random effects. The corresponding mean (95% credible interval) of the estimated posterior distribution for σw and σγ were 0.4892 (0.3590, 0.6502) and 0.5822 (0.4233, 0.7740), respectively.