Table 1.

Pairwise-comparisons between active learning models (Bonferroni-corrected P values, test statistics, and effect sizes)

Model A	Model B	p-corr	W	Effect size
LR + Doc2Vec	NB + TF-IDF	< .001	76.5	0.53
NB + TF-IDF	SVM + Doc2Vec	< .001	79	− 0.63
LR + TF-IDF	RF + TF-IDF	< .001	105.5	− 0.48
SVM + Doc2Vec	SVM + TF-IDF	< .001	111	0.65
RF + SBERT	SVM + Doc2Vec	< .001	114	− 0.73
LR + TF-IDF	SVM + Doc2Vec	< .001	118	− 0.68
SVM + Doc2Vec	SVM + SBERT	< .001	76	0.63
LR + SBERT	SVM + Doc2Vec	< .001	126	− 0.68
LR + Doc2Vec	LR + TF-IDF	< .001	129	0.58
RF + TF-IDF	SVM + TF-IDF	< .001	137	0.44
RF + Doc2Vec	SVM + Doc2Vec	< .001	140.5	− 0.23
LR + Doc2Vec	SVM + TF-IDF	< .001	144	0.55
LR + Doc2Vec	RF + SBERT	< .001	145	0.64
LR + Doc2Vec	LR + SBERT	< .001	146.5	0.59
LR + Doc2Vec	SVM + SBERT	< .001	156.5	0.53
NB + TF-IDF	RF + TF-IDF	< .001	157.5	− 0.41
RF + SBERT	RF + TF-IDF	0.001	152	− 0.55
NB + TF-IDF	RF + Doc2Vec	0.002	195.5	− 0.32
RF + Doc2Vec	RF + SBERT	0.002	198	0.40
LR + Doc2Vec	SVM + Doc2Vec	0.038	257	− 0.13
RF + Doc2Vec	SVM + TF-IDF	0.038	256.5	0.33
LR + TF-IDF	RF + Doc2Vec	0.052	264.5	− 0.36
LR + SBERT	RF + TF-IDF	0.066	270.5	− 0.48
LR + Doc2Vec	RF + Doc2Vec	0.074	273	0.11
RF + TF-IDF	SVM + Doc2Vec	0.229	303	− 0.32
RF + TF-IDF	SVM + SBERT	0.292	309.5	0.41
LR + SBERT	RF + Doc2Vec	0.465	324.5	− 0.35
LR + Doc2Vec	RF + TF-IDF	0.528	327.5	0.18
RF + SBERT	SVM + TF-IDF	0.810	323.5	− 0.13
RF + SBERT	SVM + SBERT	1	346.5	− 0.14
RF + Doc2Vec	SVM + SBERT	1	355.5	0.31
LR + TF-IDF	SVM + SBERT	1	514	− 0.06
NB + TF-IDF	SVM + TF-IDF	1	515.5	0.001
NB + TF-IDF	SVM + SBERT	1	543	− 0.01
NB + TF-IDF	RF + SBERT	1	381.5	0.12
LR + TF-IDF	SVM + TF-IDF	1	355.5	− 0.04
LR + TF-IDF	RF + SBERT	1	424.5	0.08
LR + TF-IDF	NB + TF-IDF	1	480.5	− 0.04
LR + SBERT	SVM + TF-IDF	1	492	− 0.03
LR + SBERT	SVM + SBERT	1	527	− 0.05
LR + SBERT	RF + SBERT	1	383	0.10
LR + SBERT	NB + TF-IDF	1	484	− 0.03
LR + SBERT	LR + TF-IDF	1	562	0.01
RF + Doc2Vec	RF + TF-IDF	1	560	0.03
SVM + SBERT	SVM + TF-IDF	1	554	0.02