Stephen Gorard

Measurement is fundamental to research-related activities in social science (hence this Handbook). In my own field of education research, perhaps the most discussed element of education lies in test scores. Examination results are measurements, the number of students attaining a particular standard in a test is a measurement; indeed the standard of a test is a measurement. The allocation of places at school, college or university, student:teacher ratios, funding plans, school timetables, staff workloads, adult participation rates, and the stratification of educational outcomes by sex, social class, ethnicity or geography for example, are all based on measurements. Good and careful work has been done in all of these areas (Nuttall 1987). However, the concept of measurement itself remains under-examined, and is often treated in an uncritical way. In saying this I mean more than the usual lament about qualitative:quantitative schism or the supposed reluctance of social scientists to engage with numeric analysis (Gorard et al. 2004a). I mean that even where numeric analysis is being conducted, the emphasis is on collecting, collating, analysing, and reporting the kinds of data generated by measurement, with the process of measurement and the rigor of the measurement instrument being somewhat taken for granted by many commentators. Issues that are traditionally considered by social scientists include levels of measurement, reliability, validity, and the creation of complex indices (as illustrated in some of the chapters contained in this volume). But these matters are too often dealt with primarily as technical matters – such as how to assess reliability or which statistical test to use with which combination of levels of measurement. The process of quantification itself is just assumed.