Work

The construction of class and school process variables based on individual questionnaires

Year: 2007
Type: conference-paper
Abstract: In educational effectiveness research, class and school process variables are investigated to determine their association with diverse student outcomes. For example, the school environment and instructional characteristics are studied to establish the association with student achievement. Revealing the class and school process variables that matter for the learning of students could inspire policy makers and school principals to make schooling and teaching more effective. For both practical and financial reasons, student and teacher questionnaires are often the preferred means to capture these process variables. Next, the individual data from these questionnaires are used to construct measures for higher-order process variables at the class level and the school level. Traditionally, this is done through a factor analysis at the individual level, which can be exploratory or confirmatory. This way, scale scores or factor scores are developed, which are subsequently aggregated into the class level or the school level. However, this traditional approach does not take into account the dependency present in the hierarchical structure of educational data. Among others, Muthén argues to explicitly model this hierarchy using multilevel confirmatory factor analysis. This way, a confirmatory factor analysis is performed simultaneously at the individual level and at the group level. Recently, van de Vijver and Poortinga presented an exploratory variant of this multilevel factor analysis, based on the procedure proposed by Muthén. In the current research project, we propose to investigate the surplus value of the multilevel factor analysis methods, as compared to the use of a factor analysis at one level, to construct meaningful, reliable and valid measures for class and school process characteristics. Will the alternative approach of multilevel factor analysis lead to new understandings within the field of educational effectiveness? To this end, the traditional approach of factor analysis at one level will be compared to the alternative procedure of multilevel factor analysis with regard to a number of features: factor structure, meaning and interpretation of the constructed measures for process variables, reliability and criterion validity. The final part of the research project will address simulation studies to explore the conditions associated with differences between the traditional and alternative constructional methods. Several data bases will be used to realize these research goals. Two of them are associated with longitudinal research projects in Flanders, one in primary education (i.e., SiBO: Schoolloopbanen in het BasisOnderwijs [School Careers in Primary Education]) and one in secondary education (i.e., LOSO: Longitudinaal Onderzoek in het Secundair Onderwijs [Longitudinal Research in Secondary Education]). Next, data from a number of pilot studies within the framework of an international research project (i.e., ISTOF: International System for Teacher Observation and Feedback) will be included in the project. Finally, within the framework of the simulation studies, artificial data will be constructed based on our previous research findings. The current research project is relevant to the domain of educational effectiveness in particular as well as to environmental research in general.
Source: Lirias
Authors: D´Haenens, Ellen, Van Damme, Jan, Onghena, Patrick
Institution: KU Leuven
Language: English
Open Access status: green
OpenAlex ID: W6999416564
FWCI:
Cites: 68
Cited by:
Related to: 10