. 2016 Jan 11;4(1):cov055. doi: 10.1093/conphys/cov055

Table 3:

Summaries of optimal models for each activity measure

	Parameter	Estimate	SEM	P-value
Y_k = A_day	Optimal model:	M1: A_day = α_kp + FL_j + a_jkp + ϵ_ijkp
	α_kp	−954.1	519.3	0.066
	FL_j	8.7	3.2	0.0083
	σ_akp	132.0	n.a.	n.a.
	σ_jkp	185.2 (73.4–1169.4)	n.a.	n.a.
	ICC_jkp	0.34 (0.013–0.76)	n.a.	n.a.
Y_k = U_max	Optimal model:	M1a: U_max = α_kp + FL_j + FR_j + a_jkp + ϵ_ijkp
	α_kp	−1.53	0.32	<0.001
	FL_j	0.0041	0.00083	<0.001
	FR_j	0.29	0.068	<0.001
	σ_akp	0.029	n.a.	n.a.
	σ_jkp	0.12 (0.03–0.28)	n.a.	n.a.
	ICC_jkp	0.051 (0.011–0.55)	n.a.	n.a.
Y_k = U_inst	Optimal model:	M1a: U_inst = α_kp + FL_j + FR_j + a_jkp + ϵ_ijkp
	α_kp	−2.60	0.47	<0.001
	FL_j	0.00087	0.0011	0.44
	FR_j	0.24	0.10	0.015
	σ_akp	0.052	n.a.	n.a.
	σ_jkp	0.26 (0.19–0.46)	n.a.	n.a.
	ICC_jkp	0.037 (0.013–0.068)	n.a.	n.a.
Y_k = AR_day	Optimal model:	M1: AR_day = α_kp + FL_j + a_jkp + ϵ_ijkp
	α_kp	−582.0	418.5	0.17
	FL_j	5.11	2.59	0.0497
	σ_akp	109.9	n.a.	n.a.
	σ_jkp	111.4 (53.5–998.1)	n.a.	n.a.
	ICC_jkp	0.49 (0.01–0.81)	n.a.	n.a.

Parameter estimates and associated standard errors (where available) are given. Medians are given for ${σ_{j k p}}^{2}$ and ICC_jkp, with the minimum and maximum in parentheses. For all four models, it is assumed that $a_{j k p} \tilde{} N (0, {σ_{a k p}}^{2})$ and $ϵ_{i j k p} \tilde{} N (0, {σ_{j k p}}^{2})$ . P-values are obtained from likelihood ratio tests comparing the optimal model with a nested model excluding the respective parameter.