Table 5.
Multinomial logistic regression to predict membership in the 3 latent classes in longitudinal loneliness trajectories over 10 years, 2004-2014. Class 1 (high increasing loneliness), class 2 (intermediate stable loneliness), and class 3 (low stable loneliness) are reference categories.
| Variable | Estimate (SE) | Wald χ2 (df=1) | Significance | Odds ratio (95% CI) | |||||
| Age at Q5a | |||||||||
|
|
Class 1 | 0.08 (0.02) | 15.97 | <.001 | 1.09 (1.04-1.13) | ||||
|
|
Class 2 | 0.03 (0.01) | 6.13 | .01 | 1.03 (1.01-1.06) | ||||
| Sex (1=male, 2=female) | |||||||||
|
|
Class 1 | 0.53 (0.40) | 1.72 | .19 | 1.69 (0.77-3.71) | ||||
|
|
Class 2 | 0.46 (0.23) | 3.87 | .049 | 1.59 (1.00-2.51) | ||||
| ISCEDb | |||||||||
|
|
Class 1 | –0.12 (0.12) | 1.05 | .31 | 0.89 (0.71-1.11) | ||||
|
|
Class 2 | –0.03 (0.07) | 0.18 | .67 | 0.97 (0.85-1.11) | ||||
| SRHc at Q5 | |||||||||
|
|
Class 1 | –0.58 (0.32) | 3.27 | .07 | 0.56 (0.30-1.05) | ||||
|
|
Class 2 | –0.47 (0.21) | 4.93 | .03 | 0.63 (0.42-0.95) | ||||
| Live alone at Q5 | |||||||||
|
|
Class 1 | 1.46 (0.37) | 15.78 | <.001 | 4.30 (2.09-8.84) | ||||
|
|
Class 2 | 0.86 (0.25) | 11.79 | <.001 | 2.36 (1.45-3.86) | ||||
| Perceived social support at Q5 | |||||||||
|
|
Class 1 | –0.33 (0.07) | 23.53 | <.001 | 0.72 (0.62-0.82) | ||||
|
|
Class 2 | –0.22 (0.05) | 17.47 | <.001 | 0.80 (0.72-0.89) | ||||
| Extraversion at Q5 | |||||||||
|
|
Class 1 | –0.06 (0.08) | 0.43 | .51 | 0.95 (0.80-1.12) | ||||
|
|
Class 2 | –0.10 (0.05) | 3.85 | .0498 | 0.90 (0.81-1.00) | ||||
| Daily digital use at Q5 (0=no; 1=yes) | |||||||||
|
|
Class 1 | –1.29 (0.60) | 4.59 | .03 | 0.27 (0.08-0.90) | ||||
|
|
Class 2 | –0.38 (0.27) | 1.91 | .17 | 0.69 (0.40-1.17) | ||||
aQ5: questionnaire wave 5.
bISCED: International Standard Classification of Education.
cSRH: self-rated health.