TABLE 4.

Parameter estimates from models predicting disfluency (velocity peak count) when children wrote letters to dictation.

	Straight features	Curved features
Intercept	6.6 [5.4, 7.7]	10 [7.9, 12]
Feature is malformed (vs. correct)	5.2 [1.8, 8.7]**	2.0 [0.12, 3.8]*
Symbol copying fluency	0.97 [−0.24, 2.2]	1.0 [0.37, 1.7]**
Letter copying fluency	1.5 [0.23, 2.7]*	1.8 [1.1, 2.5]***
Phoneme to Grapheme encoding	−1.2 [−2.3, −0.02]*
Phoneme isolation	−0.95 [−2.1, 0.22]
Phoneme blending	1.2 [0.06, 2.3] *

Parameter estimates from a linear mixed-effects model with random by-item and by-subject intercepts. Blank cells indicate that effect was absent in the best-fit model. *p < 0.05; **p < 0.01; ***p < 0.001. All predictor variables were standardised, with the exception of the dummy variable representing malformed features.