Table 4.
Final Stepwise Model Including Individual/Demographic, Psychosocial, and Environment Variables for Walking Leisurely or Fast for Exercise, Minutes per Week
| Predictor Variable | Std Dev of PredVar |
B*SDa or B |
95% CI | P-value |
|---|---|---|---|---|
| Individual/Demographic Variables remaining in model | ||||
| Age (years) | 6.26 | −9.22 | −2.03, −16.41 | .01 |
| Non-Hispanic White (non-White = ref) |
29.89 | 13.35, 46.43 | .0004 | |
| Treatment for osteoarthritis (no = ref) | −19.34 | −1.51, −38.17 | .03 | |
| Census Demographic Variables remaining in model | ||||
| Median age (years) | 5.82 | −10.57 | −2.79, −18.35 | .008 |
| Psychosocial Variables remaining in model | ||||
| Self-efficacy (range: 1–10) | 2.64 | 27.40 | 19.77, 35.03 | < .0001 |
| Social support (range: 0–4) | 0.69 | 25.44 | 18.80, 32.28 | < .0001 |
| Environmental Variables remaining in model | ||||
| Mixed land use: residential, entertainment, retail, office (GIS-determined; range: 0–.88) |
0.25 | −8.01 | −0.04, −15.98 | .047 |
Note. Model adjusted for repeated measures over time, site (Seattle, Baltimore), and subjects’ nesting within census blocks.
a
For a continuous predictor, the regression coefficient is multiplied by its standard deviation. The quantity represents the change in the dependent variable for a 1 SD increase in the predictor. For categorical variables, the regression coefficient is shown. The B*SD effect sizes can be compared within (but not between) models.