mcdonald-choosingresponse-2004.pdf (598.03 kB)
Choosing response scale labels and length: guidance for researchers and clients
journal contributionposted on 2004-11-01, 00:00 authored by P Darbyshire, H McDonald
There has been much debate over the optimal format for scales, particularly in regard to two key issues - the labelling of points and the overall length of response scales. This paper reviews the evidence regarding the advantages of different scale types and lengths, and provides guidance as to what scale types suit different research objectives. Using a direct comparison of 400 responses on 5-point and 11 -point scales to the same question, by the same people, we examine some of the important differences previously found and then illustrate the impact they have on data quality and useability. Our conclusion, based on past research and our own analysis, is that longer, balanced and unlabelled scales offer the maximum flexibility and reliability in the majority of cases.