Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, is already up and running.

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

Re: st: st: Handling age in Hazard Ratios

From   Austin Nichols <>
Subject   Re: st: st: Handling age in Hazard Ratios
Date   Mon, 4 Jun 2012 15:25:39 -0400

Clifton Chow <>:
Maximum, eh?
tw function ln(.9299)*x+ln(1.0007)*x^2, ra(0 80) xla(52)

On Mon, Jun 4, 2012 at 2:24 PM, Clifton Chow
<> wrote:
> I got it from Wooldridge, p. 193.  It's just the maximum.
>>  -------Original Message-------
> Where did you get the algebra suggesting that .9299/2*1.0007 represents the turning point?
>  di ln(.9299)/(2*ln(1.0007))
>>  From: Clifton Chow <>
>>  To:
>>  Subject: st: Handling age in Hazard Ratios
>>  Sent: 04 Jun '12 12:46
>>  I ran a proportional hazards model on the duration of employment and had as my covariates, age and age^2, respectively.  The coefficients and hazard ratios for both variables are below:
>>  Coefficient:  Age     = -.0727
>>                    Age^2 = .0007
>>  Hazard Ratio: Age = .9299
>>                     Age^2 = 1.0007
>>  I am trying to interpret the diminishing effect of the quadratic term by calculating the age at which the risk changes from a decrease to ian ncrease risk of job loss. I did this by dividing the age coefficient by 2 * age^2 coefficient.  However, when performing this calculation on the raw coefficient, the age of change is 52 (.0727/2*.0007) but in the hazard ratios, that age is 46 (.9299/2*1.0007). Does anyone know which convention to report? I think the difference of 5 years between the two sets of coefficients are important, right?

*   For searches and help try:

© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index