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: hosmer lemshow goodness of fit statistics

From   Ari Samaranayaka <>
To   <>
Subject   Re: st: hosmer lemshow goodness of fit statistics
Date   Thu, 9 Jun 2011 10:31:06 +1200

As Steve says small Pvalues indicate lack of fit. Reference for HL goodness fit test is Hosmer Jr., D. W., Lemeshow. S. 2000. Applied Logistic Regression. 2nd ed. New York: Wiley. According to my experience this test is very unreliable because Pvalue is very sensitive to the groupings. In other words you decide the Pvalue. This fact is acknowledged by the Hosmer and Lemeshow in a different publication. However this test is still widely used for logistic regression (I would say almost the standard) at least within the discipline of epidemiology. I like others Stata users' comments on this test although this is not specifically about Stata.

On 9/06/2011 9:20 a.m., Steven Samuels wrote:
No, just the reverse: small p-values indicate lack of fit.  But Ann is asking for guidance on cut-points, and I know of none.


On Jun 8, 2011, at 6:34 AM, Yusvita Triwiadhian S. wrote:

Hi Ann

as i know,it depends on your alpha that you are using. if significancy
<  alpha, it means the model fits the data well.

On Wed, Jun 1, 2011 at 2:00 AM, Fitzmaurice, Ann E.
<>  wrote:
Good evening

I have run a series of logistic regressions and obtained the HL GOF statistic, I have read in articles that "the HL GOF fits the data well" or that "the HL GOF statistic fits the data adequately"

Does anyone know if there are cut points for the definition of good fit,  or adequate fit and if so what is the reference (s)

*   For searches and help try:

Dr. Ari Samaranayaka
IPRU, Dept of Preventive and Social Medicine
University of Otago

Find out about the 11th World Conference on Injury Prevention and Safety Promotion by visiting

*   For searches and help try:

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