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: piece-wise logistic regression

From   Steve Samuels <>
Subject   Re: st: piece-wise logistic regression
Date   Thu, 29 Apr 2010 10:56:50 -0400

A historical note: The two-piece model has long been called a "hockey
stick" model. My oldest reference with this name dates to 1976*,
although the model itself is older.  The location of the threshold is
an important parameter in the literature of this model.  I recall
using non-linear regression in NCSS  to find a CI for the location.  A
google search shows that more recently bootstrapping has been used.


*Hasselblad, V., Creason, J. P. & Nelson, W. C. (1976) Regression
Using "Hockey Stick" Functions (U.S. Environmental Protection Agency,
Research Triangle Park, NC), Report No. EPA-600/1- 76-024, pp. 1-13.
available at:

On Thu, Apr 29, 2010 at 4:38 AM, Maarten buis <> wrote:
> --- On Thu, 29/4/10, Bruno Schoumaker wrote:
>> It notably explains how stepwise regression can be used to
>> determine the number of knots and their location. Not
>> foolproof, but very useful when no clear idea about the
>> number and location of knots.
> A similar but slightly different approach is to use the
> same algoritm used by -fracpoly- and -mfp. A recent
> discussion of that algorithm, with an application to
> restricted cubic splines can be found in Royston and
> Sauerbrei (2007), but I can see no reason why this could
> not be applied to linear splines.
> Hope this helps,
> Maarten
> Patrick Royston and Willi Sauerbrei (2007) "Multivariable
> modeling with cubic regression splines: A principled
> approach" The Stata Journal, 7(1):45--70.
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
> --------------------------

Steven Samuels
18 Cantine's Island
Saugerties NY 12477
Voice: 845-246-0774
Fax:    206-202-4783
*   For searches and help try:

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