Table 2.
Multilevel logistic regressions for likelihood of estate planning and ACP (N = 10,273).
| Model/Variable | Estate planning |
ACP |
||
|---|---|---|---|---|
| Valid will | Durable power of attorney for health care |
Living will | ||
| Unconditional model | ||||
| Intraclass correlation (ICC) | 0.24 | 0.17 | 0.20 | |
| Random effects | 1.047 | 0.685 | 0.817 | |
| Unadjusted effects model | Odds ratio | Odds ratio | Odds ratio | |
| Intercept | 1.669*** | 0.914*** | 0.879*** | |
| Ever had Alzheimer’s disease | 0.818 | 2.811*** | 1.313* | |
| Adjusted effects model | ||||
| Ever had Alzheimer’s disease | 1.049 | 2.755*** | 1.386 | |
| Demographic characteristics | ||||
| Age | 1.079*** | 1.053*** | 1.047*** | |
| Sex | (ref.=Male) | |||
| Female | 1.242*** | 1.161** | 1.311*** | |
| Race/ethnicity1 | (ref.=Non-Hispanic White) | |||
| Non-Hispanic Black/African American | 0.222*** | 0.524*** | 0.345*** | |
| Hispanic any race | 0.279*** | 0.468*** | 0.370*** | |
| Socioeconomic characteristics | ||||
| Log of total household income | 1.730*** | 1.299*** | 1.379*** | |
| Log of total household wealth | 1.163*** | 1.094*** | 1.112*** | |
| Homeownership | (ref.=non-homeowner) | |||
| Homeowner | 2.115*** | 0.916 | 0.911 | |
| Educational attainment (years) | 1.143*** | 1.120*** | 1.141*** | |
| Employment status | (ref.=Retired) | |||
| Works full-time | 0.640*** | 0.581*** | 0.636*** | |
| Works part-time/partly retired | 0.874 | 0.918 | 0.919 | |
| Other2 | 0.956 | 0.881 | 0.747* | |
| Marital status | (ref.=Other3) | |||
| Married/Partnered | 0.950 | 0.725*** | 0.815** | |
| Parental status | (ref.=Father not alive) | |||
| Father alive | 1.660 | 1.139 | 1.348 | |
| (ref.=Mother not alive) | ||||
| Mother alive | 1.142 | 0.851 | 0.859 | |
| Census region4 | (ref.=South) | |||
| Northeast | 1.174 | 1.087 | 0.874 | |
| Midwest | 1.099 | 1.263*** | 1.104 | |
| West | 0.825* | 1.306*** | 1.043 | |
| Random effects | 0.978 | 0.646 | 0.770 | |
| R2 | 0.066 | 0.057 | 0.058 | |
*
p < .05
**
p < .01
***
p < .001.
Note. Given that multiple hypothesis testing was conducted, the Benjamini-Hochberg correction method was employed to control for the increased Type I error rate.
ref. = reference category.
1
Non-Hispanic Other race (i.e., American Indian, Alaskan Native, Asian, Native Hawaiian, and Pacific Islander) excluded in the adjusted models due to the small number of cases (2.1%, n = 217).
2
Other category included unemployed, disabled, and not in labor force.
3
Other category included separated, divorced, widowed, and never married.
4
Other category of census region excluded in the adjusted models due to the very small number of cases (0.2%, n = 24).