Rolf Turner

Starpath Project, University of Auckland

A dearth of evidence

These days it is very trendy for authorities to demand that initiatives and decisions be "evidence based". This is certainly as true in the education milieu as it is anywhere. The difficulty is that it is, in many cases, virtually impossible to produce "statistically significant" evidence of the success of an initiative. The initiative would have to produce an overwhelming change in order for the improvement to show up as being "significant". This is because of the level of variability with which one has to contend and of the ever-bedevilling fact that precision increases with the square root of the sample size and not with the sample size. I shall give some examples of this problem taken from work done in the context of the Starpath Project.