Table 3. Mixed effect analysis of the association between loneliness, peer support, close friends and suicidal ideation among the in-school adolescents.
Variables | Model O | Model I | Model II |
---|---|---|---|
aOR [95% CI] | aOR [95% CI] | ||
Fixed effect results | |||
Felt lonely | |||
No | 1.00 | 1.00 | |
Yes | 2.49*** [2.25, 2.75] | 1.88*** [1.69, 2.09] | |
Peer support | |||
No | 1.00 | 1.00 | |
Yes | 0.84*** [0.78, 0.92] | 0.89* [0.82, 0.97] | |
Close friends | |||
No | 1.00 | 1.00 | |
Yes | 0.76*** [0.68, 0.85] | 0.77*** [0.68, 0.86] | |
Age group (years) | |||
11–14 | 1.00 | ||
15–19 | 1.08 [0.99, 1.18] | ||
Sex | |||
Female | 1.00 | ||
Male | 0.77*** [0.71, 0.83] | ||
Felt hungry | |||
No | 1.00 | ||
Yes | 1.33*** [1.18, 1.49] | ||
Anxiety | |||
No | 1.00 | ||
Yes | 1.95*** [1.76, 2.17] | ||
Current cigarette smoking | |||
No | 1.00 | ||
Yes | 1.62*** [1.39, 1.89] | ||
Current tobacco use | |||
No | 1.00 | ||
Yes | 1.33** [1.12, 1.58] | ||
Current alcohol use | |||
No | 1.00 | ||
Yes | 1.30*** [1.18, 1.42] | ||
Current marijuana use | |||
No | 1.00 | ||
Yes | 1.24* [1.01, 1.52] | ||
Parent or guardian supervision | |||
No | 1.00 | ||
Yes | 1.04 [0.96, 1.14] | ||
Parent or guardian connectedness | |||
No | 1.00 | ||
Yes | 0.82*** [0.75, 0.90] | ||
Parent or guardian bonding | |||
No | 1.00 | ||
Yes | 0.92 [0.84, 1.01] | ||
Parent or guardian respect for privacy | |||
No | 1.00 | ||
Yes | 0.79*** [0.73, 0.86] | ||
Random effect results | |||
Primary sampling unit variance (95% CI) | 0.020 [0.008, 0.046] | 0.017 [0.007, 0.044] | 0.014 [0.005, 0.040] |
Intra-Class Correlation | 0.006 | 0.005 | 0.004 |
Likelihood ratio test | 13.51 (<0.001) | 10.75 (<0.001) | 7.77 (0.003) |
Wald chi-square | Reference | 374.60 (<0.001) | 877.56 (<0.001) |
Model fitness | |||
Log-likelihood | -8614.7636 | -8441.1852 | -8177.9355 |
Akaike Information Criterion | 17233.53 | 16892.37 | 16389.87 |
Total sample | 19,119 | 19,119 | 19,119 |
Number of clusters | 122 | 122 | 122 |
* p < 0.05,
** p < 0.01,
*** p < 0.001
*Model I consisted of only the key explanatory variables
*Model II was adjusted for the covariates