Table 2.
Logistic Multilevel Model Predicting the Probability of Workplace Secondhand Smoke Exposure in Missouri Based on Individual and County-Level Characteristicsa
| Fixed Effects | Odds Ratio (95% Confidence Interval)b |
|---|---|
| Intercept | .10 (.06-.17) |
| Individual characteristics | |
| Age | 0.99 (0.984-0.994) |
| Male sex | 2.65 (2.23-3.17) |
| Race/ethnicity | |
| Non-Hispanic white | 1 [Reference] |
| Non-Hispanic black | 1.08 (0.83-1.40) |
| Hispanic | 1.03 (0.42-2.48) |
| Other race | 0.77 (0.40-1.48) |
| Incomec | 0.88 (0.82-0.94) |
| Education | |
| College graduate | 1 [Reference] |
| Some college | 2.00 (1.32-3.03) |
| High school graduate or General Educational Development certificate | 2.06 (1.35-3.15) |
| <High school graduate | 2.89 (1.64-5.10) |
| Current smoker | 1.91 (1.64-2.23) |
| County characteristicsd | |
| Urban area | 1 [Reference] |
| Large rural area | 1.16 (0.82-1.66) |
| Small rural area | 1.28 (1.02-1.61) |
| Isolated | 1.66 (1.31-2.11) |
| Random effects | |
| For intercept, standard deviation | .21 |
| Model fit | |
| Likelihood ratio | χ2(3) = 18.0; P = .001 |
| Deviance | 42,176 |
| Akaike information criterion | 42,206 |
a
See the Methods section for details of this analysis.
b
Values are expressed as odds ratio (95% confidence interval) unless otherwise indicated.
c
Income was treated as continuous.
d
Rurality was determined by using the Rural-Urban Continuum codes (www.ers.usda.gov/briefing/rurality/ruralurbcon/).