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: large coefficients in logistic regression

From   Ronan Conroy <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: st: large coefficients in logistic regression
Date   Tue, 30 Aug 2011 08:58:32 +0100

On 2011 Lún 29, at 22:27, Sabrina Helmut wrote:

> Dear all,
> I have a general question regarding large coefficients in logistic regression. Is it possible that the estimated coefficients for a specific variable in logistic regression is highly dependent on the dimension of this variable. So, in my case the independent variable ranges from -0.0009 and 0.1197 which results in a coefficient from logistic regression that is 48.5. 

People sometimes overlook the interpretability of their measures of effect size, such as regression coefficients, odds ratios etc. 

For variables such as age and blood pressure, I frequently find that dividing by ten gives a more interpretable measure - the change in the outcome variable for a decade increase in age or for a ten-millimetre increase in blood pressure.

You might think about the natural units of your predictor variable. If it doesn't have any, I recommend using quantiles. With sufficient data you might calculate the effect size for a decile increase, but you might equally opt for an effect size based on quartiles or tertiles. 

The fact that you are ill at ease with the odds ratio suggests that you need to do some further thinking until you get a measure of effect size that communicates the relationship you have found. 

Ronán Conroy
[email protected]
Associate Professor
Division of Population Health Sciences
Royal College of Surgeons in Ireland
Beaux Lane House
Dublin 2

*   For searches and help try:

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