The evaluation of social policy began in the USA in the 1950s and 1960s, and reached the UK by the early 1970s (Everitt and Hardiker 1996:42-43) by which time British social policy was focused on Beveridge's 'Welfare State' (Jones 1991:134). The term 'evaluation' has many definitions (Alkin 1990:81-3), but for the purposes of this essay, I will use Patton's broad definition of evaluation as 'any effort to increase human effectiveness through systematic data-based inquiry' (1990:11). Evaluation research differs from traditional research because it doesn't just collect data to increase knowledge, but also to support recommendations for action (Patton 1986:14).
The quantitative approach to evaluating social policy emphasises the neutrality and objectivity of the researcher; conceptualises practice in terms of a clearly defined positivist model; seeks causal relationships between inputs and outputs; and applies different interventions to control and experimental groups so that different outcomes can be measured and compared (Everitt and Hardiker 1996:46).
The qualitative approach rejects the notion of the neutrality and objectivity of the researcher. Instead it seeks to identify participants' understandings of the meanings of processes, situations and effects; treats values as fundamental to participants' understandings; and recognises that a reflexive approach to the practice of evaluation is important for the creation of knowledge (ibid:87). The qualitative approach recognises evaluation as a political activity taking place within a political context, and therefore uses researchers who are independent inasmuch as they have no direct interest in the work being evaluated (Tilley 1999: 92,97). Within the qualitative approach, evaluation is divided into 'formative' evaluation which assesses process, and 'summative' evaluation which assesses outcomes (Reith 1984, in Everitt and Hardiker 1996:88).
The quantitative approach uses quantitative data, e.g. census data, test scores, and surveys requiring the answers to closed-ended or scaled questions (Torres, Preskill and Piontek 1996:97), which are...