Show Summary Details
Page of

Statistics and the design of experiments and surveys 

Statistics and the design of experiments and surveys
Chapter:
Statistics and the design of experiments and surveys
Author(s):

Graham Dunn

DOI:
10.1093/med/9780199696758.003.0017
Page of

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). © Oxford University Press, 2016. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy and Legal Notice).

date: 25 August 2019

Research into mental illness uses a much wider variety of statistical methods than those familiar to a typical medical statistician. In many ways there is more similarity to the statistical toolbox of the sociologist or educationalist. It would be a pointless exercise to try to describe this variety here but, instead, we shall cover a few areas that are especially characteristic of psychiatry. The first and perhaps the most obvious is the problem of measurement. Measurement reliability and its estimation are discussed in the next section. Misclassification errors are a concern of the third section, a major part of which is concerned with the estimation of prevalence through the use of fallible screening questionnaires. This is followed by a discussion of both measurement error and misclassification error in the context of modelling patterns of risk. Another major concern is the presence of missing data. Although this is common to all areas of medical research, it is of particular interest to the psychiatric epidemiologist because there is a long tradition (since the early 1970s) of introducing missing data by design. Here we are thinking of two-phase or double sampling (often confusingly called two-stage sampling by psychiatrists and other clinical research workers). In this design a first-phase sample are all given a screen questionnaire. They are then stratified on the basis of the results of the screen (usually, but not necessarily, using two strata—likely cases and likely non-cases) and subsampled for a second-phase diagnostic interview. This is the major topic of the third section. If we are interested in modelling patterns of risk, however, we are not usually merely interested in describing patterns of association. Typically we want to know if genetic or environmental exposures have a causal effect on the development of illness. Similarly, a clinician is concerned with answers to the question ‘What is the causal effect of treatment on outcome?’ How do we define a causal effect? How do we measure or estimate it? How do we design studies in order that we can get a valid estimate of a causal effect of treatment? Here we are concerned with the design and analysis of randomized controlled trials (RCTs). This is the focus of the fourth section of the present chapter. Finally, at the end of this chapter pointers are given to where the interested reader might find other relevant and useful material on psychiatric statistics.

Access to the complete content on Oxford Medicine Online requires a subscription or purchase. Public users are able to search the site and view the abstracts for each book and chapter without a subscription.

Please subscribe or login to access full text content.

If you have purchased a print title that contains an access token, please see the token for information about how to register your code.

For questions on access or troubleshooting, please check our FAQs, and if you can't find the answer there, please contact us.