Table 2.
95% confidence sets of models predicting loneliness from extraversion and neuroticism.
| b 1 | b 2 | b 3 | b 4 | b 5 | K | LL | AICc | w | Adj. R2Model | Adj. R2Full Model | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Cross-sectional | |||||||||||
| Overall loneliness | |||||||||||
| Sample 1 | 0.237 | ||||||||||
| Linear Extraversion and Exponential Neuroticism Model | −0.26 | 0.47 | 0 | 0 | 0.07 | 5 | −608.42 | 1227.02 | 0.42 | 0.233 | |
| Full Model | −0.26 | 0.47 | 0.04 | 0.11 | 0.11 | 7 | −606.59 | 1227.51 | 0.33 | 0.237 | |
| Linear Main Effects Model | −0.25 | 0.49 | 0 | 0 | 0 | 4 | −609.98 | 1228.08 | 0.25 | 0.229 | |
| Sample 2 | 0.347 | ||||||||||
| Linear Main Effects Model | −0.52 | 0.13 | 0 | 0 | 0 | 4 | −271.22 | 550.61 | 0.66 | 0.342 | |
| Full Model | −0.51 | 0.13 | 0.07 | 0.07 | −0.04 | 7 | −268.88 | 552.25 | 0.29 | 0.347 | |
| Emotional loneliness | 0.335 | ||||||||||
| Linear Main Effects Model | −0.64 | 0.30 | 0 | 0 | 0 | 4 | −348.43 | 705.03 | 1 | 0.338 | |
| Social loneliness | 0.165 | ||||||||||
| Linear Main Effect of Extraversion Model | −0.40 | 0 | 0 | 0 | 0 | 3 | −296.95 | 600.00 | 0.84 | 0.161 | |
| Full Model | −0.39 | −0.03 | 0.08 | 0.08 | −0.03 | 7 | −294.44 | 603.36 | 0.16 | 0.165 | |
| Longitudinal | |||||||||||
| Overall loneliness (change) | 0.254 | ||||||||||
| Saturating Effect of Neuroticism Model | 0 | 0.26 | 0 | 0 | −0.04 | 5 | −499.99 | 1010.15 | 0.50 | 0.251 | |
| Linear Main Effect of Neuroticism Model | 0 | 0.23 | 0 | 0 | 0 | 4 | −501.51 | 1011.14 | 0.31 | 0.246 | |
| Full Model | −0.04 | 0.25 | 0.08 | 0.04 | −0.10 | 8 | −497.86 | 1012.14 | 0.19 | 0.254 |
Note: Results are based on NT1 = 346 and NT2 = 283 observations in Sample 1 and NT1 = 237 and NT2 = 129 observations in Sample 2. Following an information-theoretic approach for model comparison, we do not report p-values. Instead, interpretation should be based on the models’ Akaike weights that reflect the relative evidence for all competing models. K = number of estimated parameters; LL = maximized Log-Likelihood; AICc = second-order Akaike information criterion; w = Akaike weight of the model (i.e., likelihood of being the best model in the 95% confidence set); adj. R2 = adjusted R2; b1 to b5 refer to regression coefficients of the full polynomial model Ln = b0 + b1En + b2Nn + b3En2 + b4EnNn + b5Nn2. For reasons of comparability across samples, we transformed all variables of Sample 1 via POMS prior to the analysis. Results for emotional and social loneliness are based on the data of Sample 2 only. Longitudinal analyses additionally controlled for loneliness at T1. In Sample 2, we could not compute longitudinal results because the predictors explained no variance after controlling for loneliness at T1.