On the use of rotated context questionnaires in conjunction with multilevel item response models

Large-scale Assessments in Education

An IERI – International Educational Research Institute Journal

Table 15 Summary of differences in correlations with context constructs: mathematics

Country	Construct	Rotation design	Corr. with rotation result	SEr	Corr. with original result	SEo
Colombia	ENVPERC	Samecorrjoint	0.13	0.02	0.16	0.03
	GENSCIE	Hilocorrsep	0.07	0.02	0.11	0.02
	INTSCIE	Hilocorrjoint	−0.08	0.02	−0.12	0.02
		Common	−0.08	0.02	−0.12	0.02
	SCIEEFF	Hilocorrjoint	0.17	0.02	0.21	0.02
France	ENVOPT	Hilocorrsep	−0.27	0.02	−0.30	0.02
	INTSCIE	Hilocorrjoint	0.30	0.01	0.33	0.01
	SCINTACT	Hilocorrjoint	−0.15	0.02	−0.19	0.02
		Hilocorrsep	−0.16	0.02	−0.19	0.02
Germany	None
Hong Kong SAR	GENSCIE	Samecorrsep	0.20	0.02	0.17	0.02
Jordan	JOYSCIE	Samecorrsep	0.18	0.02	0.21	0.02
		Hilocorrjoint	0.18	0.02	0.21	0.02
	SCIEEFF	Hilocorrjoint	0.16	0.02	0.19	0.02
Norway	SCINVEST	Hilocorrsep	−0.26	0.02	−0.29	0.01
Poland	JOYSCIE	Samecorrjoint	0.12	0.01	0.15	0.02
	SCIEACT	Common	0.00	0.01	0.03	0.01
	SCSCIE	Samecorrjoint	0.18	0.01	0.22	0.02
Russian	ENVAWARE	Hilocorrsep	0.29	0.02	0.32	0.02
Federation	SCSCIE	Hilocorrjoint	0.09	0.02	0.12	0.02
United States	None

Notes: Only differences between correlation coefficients exceeding 0.03 are reported.
SEr: Standard error of the correlation between the construct and the rotation result.
SEo: Standard error of the correlation between the construct and the original result.
Bolded cell: Correlation coefficient between construct and rotation result is higher than the correlation between construct and original result.