Table 2.

Regression model for predicting human imageability judgments from two quantitative measures of context derived from co-occurrence similarity measures, on the test data set consisting of 1848 items.

	Estimate	SE	t-value	p
Intercept	2.26	0.81	2.80	0.005
ARC	2.59	0.82	3.13	0.002
INV-NCOUNT	−0.64	0.06	−9.93	<2e-16

Multiple R²: 0.08; AIC: 5670 F_{(2, 1845)} = 83.29, p < 2.2e-16.