How to interpret odds ratio

[John Page-2]

Hi everyone,

I am having problems with interpreting odds ration in Logistic Regression.
In his book David Howell(2002, p.593)wrote:
"If we used Sex as a predictor and coded Male=1, Female=2, then ...
suppose that Sex had been a predictor in the cancer study and that the
coefficient was 0.40. Exponentiating this, we would have 1.49. This would
mean that, holding all other variables constant, the odds of a female
improving are about 1.5 times greater than the odds of a male
improving."[end of quotation]

My question is why the conclusion is not just opposite? How can we
interpret the odds ration in favour of one sex rather than other?

I appreciate your comments.

Have a lovely weekend,

John

[hidden email]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Whanger, J. Mr. CTR]

John,

The interpretation of the odds ratio is dependent on the numerical
coding of the variable.  In this case, Female=2 and Male=1.  Using the
same example, if the coding were reversed such that Male=2 and Female=1,
then the coefficient would be negative and the odds ratio would be less
than one, again indicating the odds of improvement for a Male is less
than the odds of improvement for a Female given the independent
variables in the logistic model.

It might be useful to reverse the coding of the Sex variable to see how
the results differ.

Jim

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
John Page
Sent: Friday, May 02, 2008 10:41 AM
To: [hidden email]
Subject: How to interpret odds ratio

Hi everyone,

I am having problems with interpreting odds ration in Logistic
Regression.
In his book David Howell(2002, p.593)wrote:
"If we used Sex as a predictor and coded Male=1, Female=2, then ...
suppose that Sex had been a predictor in the cancer study and that the
coefficient was 0.40. Exponentiating this, we would have 1.49. This
would mean that, holding all other variables constant, the odds of a
female improving are about 1.5 times greater than the odds of a male
improving."[end of quotation]

My question is why the conclusion is not just opposite? How can we
interpret the odds ration in favour of one sex rather than other?

I appreciate your comments.

Have a lovely weekend,

John

[hidden email]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command SIGNOFF SPSSX-L For a list
of commands to manage subscriptions, send the command INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Hector Maletta]

Because in the Howell example being female is given a value of 2 whereas
being male receives a value of 1, therefore "increasing" the value of sex
from 1 to 2 raises the odds by 1.49. The most usual way is coding binaries
as 0 and 1, and then the coefficient affects only the category coded as 1,
compared to the one coded as zero.
However, this is the default contrast. You can choose other contrasts, for
instance comparing the effect of each category to the overall effect, or
other choices.
By the way, it is not a "ration", like those served at mealtimes in barracks
and jails, or allowed for civilians in times of shortage, but a "ratio" or
quotient of two probabilities.
Hector

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
John Page
Sent: 02 May 2008 11:41
To: [hidden email]
Subject: How to interpret odds ratio

Hi everyone,

I am having problems with interpreting odds ration in Logistic Regression.
In his book David Howell(2002, p.593)wrote:
"If we used Sex as a predictor and coded Male=1, Female=2, then ...
suppose that Sex had been a predictor in the cancer study and that the
coefficient was 0.40. Exponentiating this, we would have 1.49. This would
mean that, holding all other variables constant, the odds of a female
improving are about 1.5 times greater than the odds of a male
improving."[end of quotation]

My question is why the conclusion is not just opposite? How can we
interpret the odds ration in favour of one sex rather than other?

I appreciate your comments.

Have a lovely weekend,

John

[hidden email]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD