We present analyses on the reading engagement of fourth grade students based on data from PIRLS 2006. Reading engagement is comprised of many variables including reading motivation, reading behavior, and self-efficacy in reading (Guthrie & Wigfield, 2000). Engaged readers have positive attitudes toward reading, seek out texts, and make time to read them. Reading engagement is of particular relevance for reading research, as reading engagement is highly correlated with reading achievement. The National Assessment of Educational Progress in the United States (NAEP 1996) shows that more highly engaged readers have higher achievement scores than the less engaged at each of the three ages studied (9, 13 and 17-year-olds). The same national data indicate that engaged readers from low SES families have higher achievement scores than less engaged readers from high SES backgrounds (Campbell, Voelkl, & Donahue, 1997). Analyses on basis of PISA 2000 data confirm these results for the age group of the 15-year olds (Kirsch et al., 2002). In order to prove whether the same holds true for the PIRLS 2006 sample, we conducted a number of analyses. As was done in PISA, we measured reading engagement on the basis of students’ responses to questions covering students’ attitudes towards reading, time spent on reading, as well as diversity and content of reading. In addition, we included the students’ reading self-concept. We conducted confirmatory factor analyses to test whether reading engagement – as a complex construct with many variables – is best measured with one general reading engagement factor or whether the subdomains of the items included (reading motivation, reading self-concept, reading behavior etc.) have an independent explanatory power in addition to one ‘g-factor’. This analysis of dimensionality with confirmatory nested factor modeling (Gustafsson & Balke, 1993) revealed that a model with one general factor alongside orthogonal, more specific factors has a clearly better fit than a general factor model without any other more specific factors. Analyses on basis of this model of reading engagement show that in PIRLS, like in NAEP and PISA, students from low SES backgrounds with high reading engagement achieve better reading scores than students from high SES backgrounds with low reading engagement. - --continues in Methodology---

--- A first tentative conclusion to be drawn from our analyses is that students’ reading engagement is a complex construct with both, general and specific facets at the end of grade 4. With regard to reading achievement, the results of NAEP and PISA can be replicated for the PIRLS 2006 sample: students reading engagement has the potential to “compensate” for low family income and educational background. Therefore, programs to improve the reading abilities of students from socially disadvantaged families might focus on improving their reading engagement.