1,257
Views
1
CrossRef citations to date
0
Altmetric
Articles

The blind side: Exploring item variance in PISA 2018 cognitive domains

ORCID Icon & ORCID Icon
Pages 332-360 | Received 10 Dec 2021, Accepted 14 Jun 2022, Published online: 17 Jul 2022

Figures & data

Figure 1. PISA 2018 3-level response data structure.

Figure 1. PISA 2018 3-level response data structure.

Figure 2. PISA 2018 reading literacy variance components (VC) for pupils, schools and items plotted against the item to person variance components ratios (VCR).

Note. The 77 countries are arranged in descending order of variance components ratio. The 8 countries that participated in the paper-based PISA 2018 are marked by an asterisk.
Figure 2. PISA 2018 reading literacy variance components (VC) for pupils, schools and items plotted against the item to person variance components ratios (VCR).

Figure 3. PISA 2018 mathematical literacy variance components (VC) for pupils, schools and items plotted against the item to person variance components ratios (VCR).

Note. The 77 countries are arranged in descending order of variance components ratio. The 8 countries that participated in the paper-based PISA 2018 are marked by an asterisk.
Figure 3. PISA 2018 mathematical literacy variance components (VC) for pupils, schools and items plotted against the item to person variance components ratios (VCR).

Figure 4. PISA 2018 science literacy variance components (VC) for pupils, schools and items plotted against the item to person variance components ratios (VCR).

Note. The 77 countries are arranged in descending order of variance components ratio. The 8 countries that participated in the paper-based PISA 2018 are marked by an asterisk.
Figure 4. PISA 2018 science literacy variance components (VC) for pupils, schools and items plotted against the item to person variance components ratios (VCR).

Figure 5. PISA 2018 pupil variance components by domain plotted against across-countries average pupil variance components.

Note. Countries that participated in the paper-based version of PISA 2018 are denoted by an asterisk. Country-specific variance components are shown with a solid line, and across-countries average variance components with a dashed line.
Figure 5. PISA 2018 pupil variance components by domain plotted against across-countries average pupil variance components.

Figure 6. PISA 2018 school variance components by domain plotted against across-countries average school variance components.

Note. Countries that participated in the paper-based version of PISA 2018 are denoted by an asterisk. Country-specific variance components are shown with a solid line, and across-countries average variance components with a dashed line.
Figure 6. PISA 2018 school variance components by domain plotted against across-countries average school variance components.

Figure 7. PISA 2018 item variance components by domain plotted against across-countries average item variance components.

Note. Countries that participated in the paper-based version of PISA 2018 are denoted by an asterisk. Country-specific variance components are shown with a solid line, and across-countries average variance components with a dashed line.
Figure 7. PISA 2018 item variance components by domain plotted against across-countries average item variance components.

Table A1. Country-wise parameter estimates of the cross-classified mixed effects model for the PISA 2018 reading literacy domain.

Table A2. Country-wise parameter estimates of the cross-classified mixed effects model for the PISA 2018 mathematical literacy domain.

Table A3. Country-wise parameter estimates of the cross-classified mixed effects model for the PISA 2018 science literacy domain.

Table B1. Number of PISA items by domain for computer-based (CBA) and paper-based (PBA) Assessments.

Table C1. Sensitivity analysis resulting country-wise variance components (VC) and variance components ratios (VCR).