a one-unit increase in support for taking one’s medicine
predicted 22% higher odds of improvement in SRH over
the data collection period (p = .037). Model 4 demonstrates
that a one-increment increase of support for getting enough
physical activity predicted 26% higher odds of health
improvement (p = .011). Finally, as shown in Model 6, a
one-unit increase in support for going to health care providers predicted 22% higher odds of health improvement
within 7 years (p = .043). A one-unit increase in support
for keeping one’s weight under control was associated with
17% higher odds of health improvement; however, this was
only marginally significant (at p = .08). The other measures—reported support from friends/family for following a
meal plan, taking care of one’s feet, and testing one’s blood
sugar—did not significantly predict change in SRH over
time in multivariable models. We repeated analyses imputing missing data on all covariates using multiple imputations. As shown in Table 3, these results are fairly stable
across models, a few notable differences in statistical significance across models (e.g., support for taking medicine
is only a significant predictor in the fully adjusted model).