Introduction

This chapter explains the importance of evaluating interventions. It distinguishes between examples of random error and systemic bias. Three vignettes illustrate the kinds of everyday problems confronting speech and language therapists going about their daily work: language problems are identified, interventions implemented, and, in many cases, improvements are observed. Showing that an intervention works in an individual case is very hard – especially when dealing with a condition that fluctuates or is self-limiting. When people assess an intervention they aim to sift out any systematic effect of treatment from the background noise. This will be much harder if they have a measure that hops around a lot from one measurement occasion to the next regardless of their intervention – a “noisy” measure. Note that for the purposes of an intervention study, we treat as “noise” or “random error” anything that creates variability in our measurements that is not part of the treatment effect.