Generating group-level scores under response accuracy-time conditional dependence

Large-scale Assessments in Education

An IERI – International Educational Research Institute Journal

Table 2 Residual variance estimates: NAEP 2017 data

Timing section	M3	M4	M5	M4 – M3	M5 – M4
T1	0.721 (0.009)	0.616 (0.010)	0.654 (0.009)	0.105	− 0.038
T2	0.719 (0.006)	0.609 (0.009)	0.615 (0.009)	0.110	− 0.006
T3	0.746 (0.009)	0.666 (0.010)	0.674 (0.010)	0.080	− 0.008
T4	0.737 (0.009)	0.572 (0.009)	0.574 (0.009)	0.165	− 0.002
T5	0.736 (0.009)	0.644 (0.009)	0.654 (0.009)	0.092	− 0.010
T6	0.725 (0.009)	0.620 (0.009)	0.630 (0.009)	0.105	− 0.010
T7	0.719 (0.009)	0.588 (0.010)	0.611 (0.009)	0.131	− 0.023
T8	0.733 (0.009)	0.630 (0.009)	0.633 (0.009)	0.103	− 0.003
T9	0.710 (0.009)	0.595 (0.009)	0.600 (0.009)	0.115	− 0.005
T10	0.702 (0.008)	0.618 (0.008)	0.624 (0.008)	0.084	− 0.006
Mean	0.725	0.616	0.627	0.109	− 0.011

M3 is the baseline model, including only contextual variables in the IRT-LRM.
M4 has all the parameters of M3, but also latent speed in the IRT-LRM.
M5 has all the parameters of M4, but also within-item parameters for all RA-RT conditional dependencies.