Reading literacy is a fundamental asset for children. It is important both as a mean to fully interact with peers and society as well as a life-long learning tool. Thus, development of reading literacy at early school ages is a key factor for one’s academic and life success. The IEA’s PIRLS - Progress in International Reading Literacy Study - is an international research effort aimed at evaluating the reading literacy of children at their 4^th year of schooling. In 2016, 50 countries and 11 benchmarking regions participated in the PIRLS.  While participants’ average scores in reading literacy are important to signalize efficient and less efficient educational systems and the international profile of good readers is well described, there is large within-country reading literacy variation. That is, even low performing educational systems have good readers; while good performing systems have also poor readers. The students’ and families’ characteristics, as well as the schools’ and teachers’ variables that contribute to making a good reader, may differ from the international overall profile.  In this paper, I consider a group of top (Russian Federation, Ireland and Sweden), mid (Australia, Germany and Portugal) and low (Malta, Saudi Arabia, and South Africa) contrasting educational systems to determine which key selected student, family, teacher and school variables are commonly associated with the good reader in different PIRLS 2016 achievement reading literacy countries. Mullis et al. (2018) have produced an overall analysis of the key characteristics of the PIRLS 2016 good readers. Their analysis shows that more girls than boys are good readers; good readers have home environments that support learning; attend well resourced, academically oriented and safe schools; have positive attitudes toward reading and low absenteeism. However, it is not clear if these variables have the same impact on reading achievement between different achieving countries, and if so, if their predictive power is homogenous.

Two-level Hierarchical Linear Models (HLM) with student and family level 1 variables and teachers and schools level 2 variables were fitted to students’ reading literacy from the selected top, mid and low PIRLS 2016 performing countries. Appropriate student and school derived sampling weights were used to fit the HLM to the 5 PIRLS reading literacy plausible values considering random intercepts and fixed slopes models with within school ID as the clustering variable. Statistical significance and mid effect sizes (beta ≥ 0.1) were used to identify the key variables that are predictors of reading literacy in each of the selected countries. Multiple beta coefficients comparison was done with student’s t-test to identify countries’ related differences.

While countries’ mean scores show statistically significant differences between students’ reading literacy, top performing countries also have low performing students, while low performing countries also have top performing students. The questions to be answered refer to how reading literacy can be predicted by the student, family, teacher and school variables; and if the variables that are predictors of reading literacy in the top, mid and low performing countries are the same and have similar effects.

Mullis, I. V. S., Martin, M. O., Foy, P., & Hooper, M. (2017). PIRLS 2016 International Results in Reading. Retrieved from Boston College, TIMSS & PIRLS International Study Center website: http://timssandpirls.bc.edu/pirls2016/international-results