Reading Sub-Competence Correlations: Does the Reading-Specific Factor Represent a Specific Learning or Metacognitive Strategy? Evidence From 74 Countries and Regions in PISA 2009

A study of PISA data from 2009 revealed significant positive and high latent correlations between reading sub-competences, as well as between these sub-competences and mathematical and/or science competences, in nearly all of the 74 countries and federal states (.39 ≤ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"> <mml:msub> <mml:mi>r</mml:mi> <mml:mtext mathvariant="normal">latent</mml:mtext> </mml:msub> </mml:math> ≤ .96; overall mean: <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"> <mml:msub> <mml:mover> <mml:mover> <mml:mi>r</mml:mi> <mml:mo accent="true">¯</mml:mo> </mml:mover> <mml:mo accent="true">¯</mml:mo> </mml:mover> <mml:mtext mathvariant="normal">latent</mml:mtext> </mml:msub> </mml:math> = .83). These high to very strong latent correlations indicate a weak empirical discriminability of the PISA competences. Moreover, as GDP per capita increases across countries and federal states, the latent correlations between these competences and/or subscales tend to become stronger. Consistent with previous studies, these latent correlations could be statistically attributed to a general factor (PISA- g ) and a reading-specific factor (RspecF) within a bi-factor model for each country and federal state. The standardized loadings of the reading competence subscales on PISA- g were stronger than those on RspecF (mean: <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"> <mml:msub> <mml:mover> <mml:mi>λ</mml:mi> <mml:mo accent="true">‾</mml:mo> </mml:mover> <mml:mrow> <mml:mtext mathvariant="normal">PISA-</mml:mtext> <mml:mi>g</mml:mi> </mml:mrow> </mml:msub> </mml:math> = .82 vs. <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"> <mml:msub> <mml:mover> <mml:mi>λ</mml:mi> <mml:mo accent="true">‾</mml:mo> </mml:mover> <mml:mtext mathvariant="normal">RspecF</mml:mtext> </mml:msub> </mml:math> = .47). A close examination of the content of the PISA tasks and their cognitive demands (e.g., information acquisition, reasoning) indicates that PISA- g is more similar to a general cognitive ability. Furthermore, PISA- g demonstrated a consistent correlation pattern with intelligence-related student characteristics (e.g., parental education) in the majority of the countries and federal states examined. Contrary to expectations, the reading-related student characteristics (e.g., learning or metacognitive strategies) did not correlate significantly higher with RspecF than with PISA- g in the majority of cases, if not all. The precise nature of RspecF remains ambiguous, as no student characteristics correlated strongly with it while simultaneously exhibiting non-significant correlations with PISA- g . A comparison showed slightly higher correlations of the reading-related characteristics with the reading competence than with mathematics and science in most cases. Consequently, the discriminant validity of the PISA competences was found to be low. The results suggest that PISA primarily assesses a general cognitive ability.