Hi everyone,
I wonder if you could give me your opinion on the following?
Hypothetically, say we are conducting a case-control study where we have 50 alcohol dependent people (cases) and 50 controls. 1:1 matching has been employed (i.e. a particular case-control pair has been individually matched on age and sex).
When comparing the case and control groups as regards potential risk factors I have read that Breslow and Day (1980) advocated the use of paired t-tests (or the Wicoxon signed ranks non parametric equivalent) when we have interval/ordinal type data.
Now when we have dichotomous data (e.g. 'smoking'), since we are viewing the data as comprised of 50 matched pairs....am I also correct in thinking that
a)A McNemar test could be employed to test the null hypothesis that the proportion of alcohol dependent people who are smokers is the same as the proportion of controls who are smokers?
b)A Cohen's Kappa could be calculated to measure the degree of agreement between the two groups as regards smoking?
Similarly, if we had a factor with k levels e.g. socioeconomic class then could Cohen's Kappa also be employed in this case?
I have seen Cohen's Kappa employed in the situation when we are assessing degree of agreement between, for example, two judges where each judge categorises the same group of subjects into one of k categories but I have never seen it employed in the case-control scenario (with 1:1 matching). I would appreciate your views.
Kindest Regards,
Kim
Dr Kim Pearce PhD, CStat, Fellow HEA
Senior Statistician
Haematological Sciences
Room MG261
Institute of Cellular Medicine
William Leech Building
Medical School
Newcastle University
Framlington Place
Newcastle upon Tyne
NE2 4HH
Tel: (0044) (0)191 208 8142
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|