Abstract
In PISA 2009, Finland and Singapore were both ranked high among the participating nations and have caught much attention internationally. However, a secondary analysis of the means for Reading achievement show that the differences are rather small and are attributable to spurious precision. Hence, the two nations should be considered as being on par with each other in achievements and be assigned the same rank. Spurious precision as a problem of interpreting and reporting research findings has caught the attention of researchers in several other disciplines, though not in the field of education, and this needs to be rectified. In spite of the finding of no differences in PISA Reading achievement, principals in Finland and Singapore differ somewhat in school management and involvement in school matters. It is suggested that some intervening variables (e.g., teachers’ quality and instruction) are needed to explain the correlation (or the lack of it) between principals’ management styles and student Reading achievement. It is also suggested that school principals’ management styles might have been influenced by the cultural milieus of the countries and have influenced students’ social-emotional development, which is not measured by PISA.
Notes
1. PISA reports standard error (SE) but not standard deviation (SD). The SDs used in this analysis were calculated with the formula SD=SE∗√N. To evaluate the magnitude of observed differences between Finland and Singapore, the standardised mean difference (SMD) is used as an effect size indicator. The SMD is calculated as SMD = (Finland’s mean – Singapore’s mean)/(Pooled standard deviation).The obtained SMD is evaluated for its magnitude by Cohen’s (Citation1988) criterion:0.0 to 0.2 = trivial or negligible effect0.2 to 0.5 = small effect0.5 to 0.8 = moderate effect0.8 and above = large effect.The independent t-test was not used because the t-value answers the question ‘How likely is the observed difference a chance occurrence?’ or the probability, conventionally referred to as the statistical significance.However, a statistically significant difference (one that is unlikely a chance occurrence) may or may not be practically or educationally important (significant). For a more extensive discussion, see Neill (Citation2008).Moreover, the statistical significance of a t-value is sensitive to the sizes of the groups compared – when group sizes are very large, almost all observed will be statistically significant. In short, this analysis is concerned with magnitude of difference and not probability of difference. However, SMD and t-value answer different statistical questions but are mutually convertible by the formula t = (SMD∗√df)/2, where df = (N1+N2–2).