News

Let's face it, social workers don't like statistics. Most of us are apprehensive when we open a textbook on statistics, even when it is aimed at the general reader, not the expert. It is possible to work for decades without learning any more statistical skills than are required to complete a monthly expenses claim.

Yet a basic grasp of statistics is as necessary to social work as the capacity to listen, empathise and offer wise counsel. The inability to understand the principles of sampling, distribution and variance, and, perhaps more importantly, to appraise claims made for statistical significance, can lead us to exaggerate small problems, and ignore big ones. This is especially relevant when claims are being made that children may be at risk.

In a world of competing welfare priorities, numbers matter. Public investment depends on how a social problem is perceived. An unscrupulous guide to getting a child care problem onto the political agenda might contain several core principles. First, choose a topic that the public feels strongly about. Second, conduct a poll of a skewed sample and imply that the findings are representative of the child population as a whole. Third, discuss carefully selected parts of the data, preferably in the form of case studies. Then use ratio statistics (one in six children), gradient figures (up to 20 per cent of children) and conditional conjunctions ("could", "might", "may") to describe the results.

Finally, use an elastic definition that enables as large a population as possible to be included. This approach may bring short-term benefits - and often some gratifying media coverage - but does nothing for the sometimes tedious but essential process of building a solid evidence base for our activities.

Few social workers go through a working week without encountering a situation that a statistical analysis could shed some light on. Is the size of a worker's caseload related to the amount of sick leave he or she takes? Will additional weekly home visits make it more likely that parents will attend sessions at family centres? Does sharing the contents of a case file with users increase their satisfaction?

A basic grasp of some key statistical principles is sufficient to explore these questions. What is probably more important is having sufficient curiosity to want to try in the first place, and a belief that, while we can't do without values, we can't do without numbers either.

Tony Newman is principal officer, research and development at Barnardo's.