[R-meta] Stepwise meta-regression

[Michael Dewey]

Dear Hiam

Try
https://www.stata.com/support/faqs/statistics/stepwise-regression-problems/

I know it comes from a Stata FAQ but it is a convenient summary of some 
of the sources.

For a more recent non-technical summary

@article{babyak04,
    author = {Babyak, M A},
    title = {What you see may not be what you get: a brief, nontechnical
       introduction to overfitting in regression--type models},
    journal = {Psychosomatic Medicine},
    year = {2004},
    volume = {66},
    pages = {411--421}
}

Michael

On 01/06/2019 20:00, Hiam S. Chemaitelly wrote:
> Dear Michael,
> 
> I am sorry for the delayed reply.
> 
> Many thanks for sharing this very useful resource. It is very informative and insightful.
> 
> I am looking at how different a priori defined factors affect the prevalence of a sexually transmitted infection in a population. The data are based on a systematic review of studies reporting the prevalence of this infection in the population of interest.
> 
> I am sorry to trouble you, but can you please clarify what you meant by " You do presumably realise that it leads to models which do not replicate on new datasets and the confidence intervals it gives do not have their usual meaning?"
> 
> Many thanks for all of your time and support to this end.
> 
> Best regards,
> Hiam
> 
> Hiam Chemaitelly
> Senior Research Specialist, Epidemiologist
> Infectious Disease Epidemiology Group
> Weill Cornell Medicine-Qatar
> Education City, POBox: 24144, Doha, Qatar
> Phone: + (974) 4492 8443, Fax: + (974) 4492 8422
> Email: hsc2001 using qatar-med.cornell.edu
> 
> -----Original Message-----
> From: Michael Dewey <lists using dewey.myzen.co.uk>
> Sent: Friday, May 31, 2019 6:01 PM
> To: Hiam S. Chemaitelly <hsc2001 using qatar-med.cornell.edu>; r-sig-meta-analysis using r-project.org
> Cc: Laith Jamal Abu Raddad <lja2002 using qatar-med.cornell.edu>
> Subject: Re: [R-meta] Stepwise meta-regression
> 
> Does
> http://www.metafor-project.org/doku.php/tips:model_selection_with_glmulti_and_mumin
> help?
> 
> It might help if you had stated what your scientific question is which demands variable subset selection. You do presumably realise that it leads to models which do not replicate on new datasets and the confidence intervals it gives do not have their usual meaning?
> 
> Michael
> 
> On 29/05/2019 19:30, Hiam S. Chemaitelly wrote:
>> Dear R-sig-meta-analysis admin,
>>
>> I hope this email finds you well.
>>
>> I am interested in doing stepwise meta-regression for a small dataset, but I could not find a way to do it. I have been advised to contact this mailing list asking for help with this matter.
>>
>> My question is: is there a way to do stepwise meta-regression with the outcome being prevalence and independent variables including time as well as categorical variables? If not, what could be an alternative option?
>>
>> Looking forward to hearing from you.
>>
>> Best regards,
>> Hiam
>>
>> Hiam Chemaitelly
>> Senior Research Specialist, Epidemiologist Infectious Disease
>> Epidemiology Group Weill Cornell Medicine-Qatar Education City, POBox:
>> 24144, Doha, Qatar
>> Phone: + (974) 4492 8443, Fax: + (974) 4492 8422
>> Email: hsc2001 using qatar-med.cornell.edu
>>
>> Disclaimer: This email and its attachments may be confidential and are intended solely for the use of the individual to whom it is addressed. If you are not the intended recipient, any reading, printing, storage, disclosure, copying or any other action taken in respect of this e-mail is prohibited and may be unlawful. If you are not the intended recipient, please notify the sender immediately by using the reply function and then permanently delete what you have received.
>>
>> [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> R-sig-meta-analysis mailing list
>> R-sig-meta-analysis using r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
>>
>> ---
>> This email has been checked for viruses by AVG.
>> https://www.avg.com
>>
>>
> 
> --
> Michael
> http://www.dewey.myzen.co.uk/home.html
> Disclaimer: This email and its attachments may be confidential and are intended solely for the use of the individual to whom it is addressed. If you are not the intended recipient, any reading, printing, storage, disclosure, copying or any other action taken in respect of this e-mail is prohibited and may be unlawful. If you are not the intended recipient, please notify the sender immediately by using the reply function and then permanently delete what you have received.
> 

-- 
Michael
http://www.dewey.myzen.co.uk/home.html