School education is such an important process in modern society that education systems and institutions are constantly under review. School effectiveness and the related area of school improvement have been topics for an increasing body of academic research since the 1960s. Research on school effectiveness has suggested that some schools are more successful than others, which provokes questions about what is effectiveness, what are the factors that contribute to effectiveness and how might this information provide the basis for improvement of schools and student outcomes.
School effectiveness research had its origins in the mid-1960s and early 1970s when a prevalent view in the research community, especially with regard to equality of opportunity, was that schools had little influence on children's achievement that was independent of background and social context. Coleman and his colleagues[1] concluded this very lack of an independent effect meant that the inequalities imposed on children by their home, neighbourhood and peer environment tend to become the inequalities which they take into adult life at the end of school. He wrote that for equality of educational opportunity to occur there must be a strong effect in schools that is independent of the child's immediate environment. He believed that this effect was not present in American schools.
These conclusions were of concern to those with responsibilities in education. Studies in school effectiveness originated out of a desire to address the view that schools did not make much difference to young people's life chances. In the late 1970s in the United States, Edmonds[2] and, in the United Kingdom, Rutter[3] responded by embarking on what was to emerge as the first phase of school effectiveness research. The two studies undertaken, independently, by Edmonds and Rutter set out to investigate whether schools in their national contexts showed any effects when account was taken of the differences in their student populations. Their findings, arrived at independently, were similar: schools do make a small but highly significant difference to the life chances of their students
The early existence of independent research projects in two countries asking similar questions and drawing, to a certain extent, on similar methodologies demonstrated the potential for further global investigations. Early studies highlighted the characteristics of schools that appeared to be unusually effective in terms of student performance on standardised achievement tests. There have been important methodological developments that have provided new insights into the effects of schooling and greater confidence that these effects can be positive. Two of these have been:
- the adoption of better statistical methods for analysing the data that are crucial in studies of effectiveness (see box on Multilevel Modelling); and
- the emergence of a view that sees effectiveness measured not just by student achievement at a point in time but by changes in achievement over a longer period of time – that is, by taking a longitudinal approach.
School effectiveness research studies undertaken during the 1980s focused on improving the methodology and replicating the research designs with pupils of different ages and in different settings. The focus during this phase shifted from research into school effectiveness to action research into school improvement.
In Australia there have been a number of research studies in the area of school effectiveness, including a Victorian study[4], that have supported the findings of the international research.
There has been, in the past decade or so, an improvement in the analysis of information gathered from schools by researchers. Usually, this information comes from the students themselves or from the class teachers or school principals. These sources of information are regarded as being at different ‘levels’. Early analyses simply used the school as the unit of analysis and examined relationships between school means. Other studies used the individual student as the unit of analysis but assigned values of school level variables to each student in the school. This method does not take into account the ‘clustering’ effect that measuring similar students can have.
Multilevel Modelling
In recent years there have been methodological and software design advances resulting in statistical techniques known as ‘multilevel’[5] or ‘hierarchical’ linear models[6]. These techniques allow for analysis which examines, simultaneously, differences within schools and differences between schools. This allows for both school level and individual level effects to be investigated and looks at students in schools as part of a multilevel structure. Thus, differences between classes, year groups and schools can be recognised rather than aggregated arbitrarily.
These techniques have been especially relevant to school effectiveness studies which need to take account of school level policies and practices being articulated through classrooms to individual students. They also provide a better allowance for contextual factors which may be related both to school organisation and to student achievement and which must be taken into account in studying any effects of school organisation[7]. An important consequence of these developments is that there is strong evidence of differences between schools in promoting achievement, knowledge about how much variation in achievement can be attributed to school factors and what school factors appear to be associated with higher levels of achievement growth[8]. |
These new methods involving multilevel modelling are now widely applied in educational and social research and have resulted in some rethinking of the conclusions of earlier research. Most importantly, the results of these methods have suggested that previous investigations may have underestimated the effects of schools and overstated the role of student background.