Urrently smoking tobacco.Statistical methods Descriptive statistics were used to characterize

Author manuscript; available in PMC 2013 June 17.Fillenbaum et al.Pagehealth behaviors, in controlled analyses we examined each task separately wcs.1183 using logistic regression.NIH-PA Author RP54476MedChemExpress RP54476 Manuscript NIH-PA Author Manuscript NIH-PA Author ManuscriptSince the number of people requiring help with certain tasks was small, we first refined each of the three sets of independent variables (demographic characteristics, health conditions, social ties and health behaviors), by running individual chunk tests for each ADL task (e.g., regressing household tasks on health EXEL-2880 molecular weight conditions entered as a chunk). Author manuscript; available in PMC 2013 June 17.Fillenbaum et al.Pagehealth behaviors, in controlled analyses we examined each task separately wcs.1183 using logistic regression.NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author ManuscriptSince the number of people requiring help with certain tasks was small, we first refined each of the three sets of independent variables (demographic characteristics, health conditions, social ties and health behaviors), by running individual chunk tests for each ADL task (e.g., regressing household tasks on health conditions entered as a chunk). All chunks being significant, we selected for inclusion in further analyses the statistically significant variables in each chunk. Three models were then run for each ADL task. Model 1 included only demographic characteristics, Model 2 added social ties srep18714 and health behaviors Model 3 further added health conditions. Multivariate significance tests were carried out using Wald 2 tests. Statistical significance was evaluated using 2-tailed tests, with Pwcs.1183 using logistic regression.NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author ManuscriptSince the number of people requiring help with certain tasks was small, we first refined each of the three sets of independent variables (demographic characteristics, health conditions, social ties and health behaviors), by running individual chunk tests for each ADL task (e.g., regressing household tasks on health conditions entered as a chunk). All chunks being significant, we selected for inclusion in further analyses the statistically significant variables in each chunk. Three models were then run for each ADL task. Model 1 included only demographic characteristics, Model 2 added social ties srep18714 and health behaviors Model 3 further added health conditions. Multivariate significance tests were carried out using Wald 2 tests. Statistical significance was evaluated using 2-tailed tests, with P