Table 3 Tukey post-hoc test results for sensitivity and specificity metrics across prediction source. [RC6 (PC): Restructured Clinical 6 (Positive Control); F&V: Fenigstein & Vanable; R + N: RBS + NEGE (Combined); R: RBS; N: NEGE; R + N(R)/RNr: RBS + NEGE (Reduced); RO (NC): Random Observations (Negative Control)].
From: Using machine-learning strategies to solve psychometric problems
Metric | Sensitivity | Specificity | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Scale | RC6 | F&V | R + N | R | N | RNr | RO(NC) | RC6 | F&V | R + N | R | N | RNr | RO(NC) |
RC6 | – | p < .001 | p < .001 | p < .001 | p < .001 | p < .001 | p < .001 | – | p = .005 | p = .003 | p < .001 | p < .001 | p < .001 | p < .001 |
FV | p < .001 | – | n.s | p < .001 | n.s | p < .001 | p < .001 | p = .005 | – | n.s | n.s | n.s | p < .001 | p < .001 |
R + N | p < .001 | n.s | – | p = .02 | n.s | p = .08 | p < .001 | p = .003 | n.s | – | n.s | n.s | p < .001 | p < .001 |
RBS | p < .001 | p < .001 | p = .02 | – | n.s | n.s | p < .001 | p < .001 | n.s | n.s | – | n.s | n.s | p < .001 |
NEGE | p < .001 | n.s | n.s | n.s | – | n.s | p < .001 | p < .001 | n.s | n.s | n.s | – | n.s | p < .001 |
R + N(R) | p < .001 | p < .001 | p = .08 | n.s | n.s | – | p < .001 | p < .001 | p < .001 | p < .001 | n.s | n.s | – | p < .001 |
RO(NC) | p < .001 | p < .001 | p < .001 | p < .001 | p < .001 | p < .001 | – | p < .001 | p < .001 | p < .001 | p < .001 | p < .001 | p < .001 | – |
