Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Testing for association

From   "Roger B. Newson" <[email protected]>
To   [email protected]
Subject   Re: st: Testing for association
Date   Mon, 26 Mar 2012 14:36:58 +0100

You really need to define what you mean by "individual association of each genotype with the response variable". Specifically, you have to specify what you are comparing each genotype with.

You can always define a binary variable for each genotype, and measure the association of each of these binary variables with the disease outcome. This method compares disease risks (or disease odds) between each genotype and all the rest. The main problem with this method is that it immediately trebles the number of multiple P-values, compared to the number that you would have if you just did a chi-squared for association of all genotypes with the disease outcome. This raises issues of multiple comparisons.

Alternatively, you might compare disease risks (or odds) between pairs of genotypes, using the -punafcc- package for case-control data, or the -punaf- package for cohort or cross-sectional data, to estimate risk ratios between 2 fantasy scenarios, such as a fantasy scenario in which all subjects are AA and a fantasy scenario in which all subjects are CC. This method may also raise multiple-comparisons issues. However, if you want to use it, then the -punaf- and -punafcc- packages are downloadable from SSC.

The packages -qqvalue- and -smileplot- are also downloadable from SSC, and can be used to calculate q-values and discovery sets allowing for multiple comparisons.

I hope this helps.

Best wishes


Roger B Newson BSc MSc DPhil
Lecturer in Medical Statistics
Respiratory Epidemiology and Public Health Group
National Heart and Lung Institute
Imperial College London
Royal Brompton Campus
Room 33, Emmanuel Kaye Building
1B Manresa Road
London SW3 6LR
Tel: +44 (0)20 7352 8121 ext 3381
Fax: +44 (0)20 7351 8322
Email: [email protected]
Web page:
Departmental Web page:

Opinions expressed are those of the author, not of the institution.

On 26/03/2012 14:20, Shittu, Aminu wrote:
Hi List members,
I have a genotype variable with 3 levels (AA, AC and CC); is there a way of testing for individual association of each genotype with the response variable ulcer status (positive and negative) of a patient? They are recorded in 2 separate columns though. I used "tab ulcerstatus genotype, col chi" and got the overall chi2 and p-value.
*   For searches and help try:
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index