Table 1.
Standardized beta coefficients (standard errors) of logistic regression models predicting self-reported diagnoses (two levels, YES vs. NO) by social behaviors and demographic covariates as well as sensitivity and specificity tests in the online autism sample (n = 575).
| Original model | Sensitivity tests | Specificity tests | |||||
|---|---|---|---|---|---|---|---|
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | ||
| Dependent variable |
SA YES n = 145 NO n = 430 |
SA YES n = 79 NO n = 278 remove those with cognitive delays |
SA YES n = 145 NO n = 430 |
SA YES n = 145 NO n = 430 |
MDD YES n = 371 NO n = 204 |
GAD YES n = 445 NO n = 130 |
|
| Independent variables | Power behavior |
-0.21 (0.10)* p = 0.038 |
-0.32 (0.14)* p = 0.020 |
-0.21 (0.11)* p = 0.044 |
-0.21 (0.11)* p = 0.046 |
-0.13 (0.10) p = 0.190 |
0.03 (0.11) p = 0.776 |
| Affiliation behavior |
-0.07 (0.10) p = 0.455 |
-0.07 (0.13) p = 0.569 |
-0.08 (0.10) p = 0.425 |
-0.06 (0.10) p = 0.582 |
0.01 (0.10) p = 0.959 |
-0.09 (0.11) p = 0.410 |
|
| Sex |
0.58 (0.24)* p = 0.012 |
0.78 (0.37)* p = 0.024 |
0.58 (0.24)* p = 0.013 |
0.19 (0.25) p = 0.448 |
1.21 (0.20)*** p < 0.001 |
1.42 (0.22)*** p < 0.001 |
|
| Age |
-0.02 (0.03) p = 0.551 |
0.02 (0.04) p = 0.588 |
-0.02 (0.03) p = 0.563 |
-0.03 (0.03) p = 0.413 |
0.05 (0.03) p = 0.124 |
0.02 (0.03) p = 0.540 |
|
| Education |
-0.12 (0.06)* p = 0.041 |
-0.20 (0.08)* p = 0.017 |
-0.12 (0.06)* p = 0.042 |
-0.06 (0.06) p = 0.335 |
-0.25 (0.06)*** p < 0.001 |
-0.20 (0.06)** p = 0.001 |
|
| IQ |
-0.03 (0.02) p = 0.076 |
-0.02 (0.03) p = 0.503 |
-0.03 (0.02) p = 0.074 |
-0.04 (0.02)* p = 0.035 |
0.03 (0.02) p = 0.125 |
0.01 (0.02) p = 0.476 |
|
| Cognitive delays |
-0.42 (0.21)* p = 0.043 |
-0.42 (0.21)* p = 0.042 |
-0.43 (0.21)* p = 0.044 |
-0.01 (0.20) p = 0.947 |
-0.18 (0.23) p = 0.439 |
||
| Perceived power |
0.07 (0.49) p = 0.886 |
||||||
| Perceived affiliation |
0.06 (0.31) p = 0.841 |
||||||
| MDD |
0.57 (0.28)* p = 0.038 |
||||||
| GAD |
1.47 (0.43) p = 0.0001 |
||||||
Cells left blank indicate variables that were not included in the corresponding statistical model. For example, compared with the original model, the first sensitivity analysis excluded the variable “cognitive delays.” Both models excluded all variables listed below “cognitive delays.”
***p < 0.001, **p < 0.01, *p < 0.05.