TABLE 2.

Measures of Accuracy for Three Statistical Models Predicting MLD, Using Composite Scores as Predictors, and MLD in Grades 2 and 3 as the Outcome Variable

	Scores Included in Model
	Model 1: KM-R Geometry and Measurement Subtests n = 207	Model 2: KM-R (4 Subtests) and TEMA-2 n = 206	Model 3: KM-R, TEMA-2, DTVP-2, and RAN n = 182*
Sensitivity	77.3%	77.3%	91.7%
Specificity	78.9%	84.2%	78.2%
Positive PV	30.4%	37.0%	22.9%
Negative PV	96.7%	96.9%	99.3%
Correctly classified	78.7%	83.5%	79.1%
ROC	0.860	0.908	0.915
H-L p-value	0.877	0.811	0.972

Note. H-L = Hosmer–Lemeshow goodness-of-fit test, PV = predictive value.

^*Any child missing one of the 12 test scores (four KM-R, TEMA-2, four DTVP-2 subtests, and three RAN scores) was excluded from the analysis.