Table 1.
Results of EHA on risk of dispersal (n = 29927 person-years for 1544 adult women aged 15 years and over and 31350 for 1651 adult men aged 15 years and over; 375 events for women, and 264 for men)
| Female | Male | |||||
|---|---|---|---|---|---|---|
| Dispersal | Estimate (SE) | Hazard ratio | P | Estimate (SE) | Hazard ratio | P |
| Mother (alive as references category) | ||||||
| Dead | −0.019 (0.158) | 0.982 | 0.906 | 0.332 (0.146) | 1.393 | 0.023 |
| Father (alive as references category) | ||||||
| Dead | 0.142 (0.125) | 1.153 | 0.254 | −0.042 (0.14) | 0.959 | 0.763 |
| Adult brother number | 0.102 (0.04) | 1.107 | 0.01 | 0.094 (0.048) | 1.099 | 0.048 |
| Adult sister number | 0.119 (0.037) | 1.127 | 0.001 | −0.010 (0.047) | 0.99 | 0.823 |
| Birth cohort (<1940 as references category) | ||||||
| 1940 | 0.255 (0.388) | 1.290 | 0.511 | 0.682 (0.366) | 1.977 | 0.062 |
| 1950 | 0.441 (0.329) | 1.554 | 0.180 | 0.41 (0.354) | 1.507 | 0.247 |
| 1960 | 0.698 (0.309) | 2.010 | 0.024 | 0.663 (0.342) | 1.941 | 0.052 |
| 1970 | 0.902 (0.308) | 2.464 | 0.003 | 0.473 (0.35) | 1.605 | 0.177 |
| 1980–1997 | 0.937 (0.325) | 2.553 | 0.004 | −0.626 (0.466) | 0.535 | 0.179 |
| Constant | −14.811 (1.305) | 0 | <0.001 | −15.117 (1.906) | 0 | <0.001 |
We used complementary log–log regression with dispersal (dispersed = 1, stay = 0) as dependent variable. The predictors used in the model are mother dead; father dead as time-varying variables; and adult brother number, adult sister number, and birth cohort as time-invariant variables. Sibling numbers only counted those from the same mother, as it is usually mother’s children who live in the same household; sibling numbers were time-invariant variables, due to timing of leaving home (if ever) not being known for everyone. Significant effects are indicated in bold. SE, standard error.