Abstract:

This paper is directed at the numerate non-statistician. Its aim is to put the subject of weighting into perspective, to bring out the very real dangers and pitfalls of excessive weighting, to show how the effect of weighting can be assessed and to draw your attention in passing to some sampling techniques which can sometimes be used to minimise or totally eliminate the need for weighting surveys. My approach is via case studies rather than through some sophisticated general theory of weighting.