Vital statistics for social work

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.