Re: Logistic Regression Book Wanted.
From: Marc Schwartz (MSchwartz_at_mn.rr.com)
Date: 01/23/05
- Next message: Ken Schmidt: "Name of simple statistical procedure"
- Previous message: Richard Ulrich: "Re: Logistic Regression Book Wanted."
- In reply to: ghirardelli: "Re: Logistic Regression Book Wanted."
- Next in thread: Richard Ulrich: "Re: Logistic Regression Book Wanted."
- Messages sorted by: [ date ] [ thread ]
Date: Sat, 22 Jan 2005 20:18:18 -0600
ghirardelli wrote:
> Marc Schwartz wrote:
snip
>> Two excellent books below. The first is an introductory and very
>> readable text, given a basic background as you indicate and would be a
>> good starting place. The second goes more in depth with respect to
>> model development strategies, variable selection approaches and model
>> validation mechanisms. It covers more than just logistic regression
>> and also focuses on the use of SAS, S-PLUS and R as the basis for the
>> implementation of many of the techniques discussed.
>>
>>
>> Modelling Binary Data
>> Second Edition
>> David Collett
>> Chapman & Hall/CRC
>> http://www.amazon.com/exec/obidos/tg/detail/-/1584883243/
>>
>>
>> Regression Modeling Strategies
>> Frank E. Harrell
>> Springer-Verlag
>> http://www.amazon.com/exec/obidos/tg/detail/-/0387952322
>>
>>
>> HTH,
>>
>> Marc Schwartz
>
>
> I have been interested by your remarks and would like to know
> whether the quoted David Collett book is easier to work with than the
> Hosmer & Lemeshow(Applied Logistic Regression) or the Christensen
> (Log-Linear models) one?
> Regards,
> Ghirardelli
Hard to say.
I have the first edition of H&L (1989). It was my first book on this
subject back in the early 90's and certainly found it to be readable,
though I was lucky in having a great mentor to aid me when I needed.
Of course that first edition is now dated with respect to some content,
though they published an updated edition in 2000. I have not seen the
new edition, so I cannot offer a comparison.
A direct comparison of Collett's latest with the first edition of H&L
would lead me to Collett, relative to the writing style and structure. I
think that Collett does a better job of building a foundation of
knowledge and then extending that foundation as the book progresses.
On the Christensen book, I do not have access to the updated version,
which has new coverage of LR, as indicated by the change in the name of
the book from the first edition. I would have to defer to others as to
whether or not the new content is sufficient to in effect, serve as a
stand alone reference on LR.
Ultimately, I think that "better" or "easier" is in the eye of the
reader. I have rarely found over the years, that a single book or
reference is sufficient. I will generally have more than one book for a
given subject, whether it is statistics or some other technical area of
expertise. Each reference will presume certain base knowledge (given the
target audience) and do a better or worse job of covering certain
material to various degrees of detail. I will find myself using more
than one over time when I need to look something up or refresh my memory
on something.
Given that, my recommendation would be to try to secure a copy of the
books at a local library (university or otherwise) and take some time
with each. This will give you an opportunity to perhaps review the
introductory chapter or two, get a feel for writing style and content,
without committing to a purchase of expensive reference texts.
If you find that one book tends to grab you more than others, stick with
it. Read it thoroughly and learn from it. You can always come back at a
later date to review other books, perhaps some that may focus on the use
of a particular statistical application (as with Frank's).
I hope that provides some insight that is helpful to you.
Best regards,
Marc
- Next message: Ken Schmidt: "Name of simple statistical procedure"
- Previous message: Richard Ulrich: "Re: Logistic Regression Book Wanted."
- In reply to: ghirardelli: "Re: Logistic Regression Book Wanted."
- Next in thread: Richard Ulrich: "Re: Logistic Regression Book Wanted."
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
