* lavish buildings and equipment will not raise quality
* curriculum reform will not necessarily raise educational quality
* examinations are a useful way of monitoring school quality
* healthy well-fed children learn better
* amount of learning time affects educational outcomes
* quality depends on good decentralised education management
These points are discussed in more detail in later sections.
1.4 Methodology of school effectiveness
research
Various aspects of research methodology need to be taken into account when evaluating
the evidence to be presented in this summary report, and when applying the evidence in
policy formulation. Firstly, a narrow range of output variables is used in much of the
research, with test and examination results featuring prominently. It could be argued that
these proximate outputs do not necessarily give an accurate prediction of more important
remote outputs, such as whether school leavers become good citizens, or make a positive
contribution to national development. It might also be argued that more attention should
be paid to process variables such as the quality of school management and that an input-
output model is too simplistic (Willms, 1992).
If these arguments are accepted, it might be felt that there is a need for more qualitative
research. Some examples of predominantly qualitative studies are quoted in this report
(e.g. Vulliamy, 1987; Levin & Lockheed, 1991; Urwick & Junaidu, 1991), but the
majority of the cited literature is based on quantitative work.
Not everyone would feel that the scientific paradigm (within which much quantitative
work is located) is always appropriate. There are methodological difficulties, although it
is claimed that the new multilevel models overcome many of them (see Riddell, 1989
and Heyneman, 1989 for a discussion). There are also difficulties of interpretation: for
example, establishing a statistically significant correlation between two variables does
not establish causality. Another point is that some studies (especially transnational
studies) have problems of aggregation.