International studies on achievement are regularly used to study the effectiveness of educational systems. Mostly the average achievement level at a certain moment is considered as an indicator of the quality of an educational system, and the relation between the achievement and a SES-index (also at a certain moment) is considered as an indicator of the equity of the system. Of course, countries differ from each other as for many characteristics. So, it is not surprising that we did not yet learn very much about which variables are really relevant to ‘explain’ the differences as for quality and equity. As demonstrated by Gustafsson (2007) longitudinal studies at the country level can show which factors are linked to change in achievement over time. In this paper we explore the differences between countries as for the change in reading comprehension (as measured in PIRLS, in the fourth grade of primary schools) between 2001 and 2006, and we study whether change in age, SES and class size are relevant to explain these differences. We explore also the country differences as for the equity. Research questions 1. How is the relation between age, SES and class size on the one hand and achievement at the other hand within countries at a certain moment (2001 resp. 2006)? 2. How is the relation between the change in average achievement within countries and the change in average age, SES and class size between 2001 and 2006?

The average age of the pupils participating in PIRLS is not the same for each country. And in some countries the average age is different between 2001 and 2006. We studied the relationship between the average age and the average reading score in PIRLS 2001 and 2006 at the country level by means of correlations. In PIRLS a SES-index is not available. Based on the experiences within PISA we made use of the PCA method to construct an internationally acceptable SES-index in PIRLS 2001 and 2006. The variables that explained too little of the variance in reading achievement and this in most countries were left out for further analyses. Information on class size is available in the PIRLS databases. So, correlations at the country level between (change in) average class size and (change in) achievement are easily obtained.

At this moment we have already the results for age and SES. The correlation at a certain moment between average age and achievement at the country level is rather low. But the change in average achievement is highly correlated with change in average age (r=0, 53). As for SES we have constructed a SES-index. The change in average SES over time within countries is rather low. But there are important differences between countries as for the relation between achievement and SES, and for this relation there are some changes at the country level between 2001 and 2006. As for class size, the results are not yet available at this moment. At the conference we will discuss longitudinal research at the country level as a new and valuable approach to study the effectiveness (and equity) of educational systems.