Skip to main content

Archived Comments for: Scratch lottery tickets are a poor incentive to respond to mailed questionnaires

Back to article

  1. Possible bias in study design

    David J Torgerson, University of York

    2 May 2006

    Whilst the study is described as a randomized controlled trial, unfortunately it does not appear to be one. The randomization procedure appears to be flawed. The authors created a list of participants based on their first names with those with first names beginning A at the top of the list and those with Z at the bottom. They then took the first 250 and allocated them to an intervention and so on. This is not randomization: it is quasi-alternation and can lead to bias. Your study may be biased as we can observe far fewer women in the control group than in the other groups. Presumably this is because Norwegian female names favour letters closer to A than Z. Also there will be a bias towards non-Norwegian names or certain religious group, which favour certain letters in the alphabet and will be represented in greater numbers in some of the groups. Unfortunately, due to these biases the study results may not be reliable.

    Competing interests