interpreting weight in meta-analysis of proportion

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

interpreting weight in meta-analysis of proportion

petretta
Dear all,

I use R 3.0 for Windows.

I performed a meta-analysis of the prevalence (single proportion)  
reported in 14 different studyes using the command:

res<-metaprop(case,n,sm="PFT", comb.fixed=FALSE, comb.random=TRUE,  
studlab<- paste(Study))

print(res)

A referee ask a brief explanation of the W-statistic reported in the  
results, in particular, why the summ of the individual weights of all  
the studies is 100%.

Any suggestion is welcome.


--
Mario Petretta
Department of Translational Medical Sciences
Naples University Federico II
Italy

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: interpreting weight in meta-analysis of proportion

Michael Dewey
At 16:30 16/04/2014, [hidden email] wrote:
>Dear all,
>
>I use R 3.0 for Windows.
>
>I performed a meta-analysis of the prevalence (single proportion)
>reported in 14 different studyes using the command:
>
>res<-metaprop(case,n,sm="PFT", comb.fixed=FALSE, comb.random=TRUE,
>studlab<- paste(Study))

Using which package?


>print(res)
>
>A referee ask a brief explanation of the W-statistic reported in the
>results, in particular, why the summ of the individual weights of all
>the studies is 100%.

What did you expect the weights to sum to, I wonder.

>Any suggestion is welcome.
>
>
>--
>Mario Petretta
>Department of Translational Medical Sciences
>Naples University Federico II
>Italy
>
>

Michael Dewey
[hidden email]
http://www.aghmed.fsnet.co.uk/home.html

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: interpreting weight in meta-analysis of proportion

petretta
In reply to this post by petretta
Prof. Dewey, sorry for the trivial question and many thank for the replay.


> Using which package?

In this case I used the meta package, but I know that for all but the
DerSimonian-Laird method the R function rma.uni of R package metafor is
called internally.

> What did you expect the weights to sum to, I wonder.

I think that, to better explain the influence of single study in pooling
the effect size, the weight are presented as percentage of the sum of
total weight of each study, but I ask for a confirm. Nevertheless, I ask
if it is possible to obtain for each study the value of the absolute
weight, other than the relative weight, or at least the absolute value of
the sum of the weights.

Sincerely

Mario Petretta
>Department of Translational Medical Sciences
>Naples University Federico II
>Italy
>




Message: 2
Date: Thu, 17 Apr 2014 12:43:06 +0100
From: Michael Dewey <[hidden email]>
To: [hidden email], [hidden email]
Subject: Re: [R] interpreting weight in meta-analysis of proportion
Message-ID: <[hidden email]>
Content-Type: text/plain; charset="us-ascii"; format=flowed

At 16:30 16/04/2014, [hidden email] wrote:
>Dear all,
>
>I use R 3.0 for Windows.
>
>I performed a meta-analysis of the prevalence (single proportion)
>reported in 14 different studyes using the command:
>
>res<-metaprop(case,n,sm="PFT", comb.fixed=FALSE, comb.random=TRUE,
>studlab<- paste(Study))

Using which package?


>print(res)
>
>A referee ask a brief explanation of the W-statistic reported in the
>results, in particular, why the summ of the individual weights of all
>the studies is 100%.

What did you expect the weights to sum to, I wonder.

>Any suggestion is welcome.
>
>
>--
>Mario Petretta
>Department of Translational Medical Sciences
>Naples University Federico II
>Italy
>
>

Michael Dewey
[hidden email]
http://www.aghmed.fsnet.co.uk/home.html

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

Re: interpreting weight in meta-analysis of proportion

Bernd Weiss
On 18.04.2014 13:02, [hidden email] wrote:

> Prof. Dewey, sorry for the trivial question and many thank for the replay.
>
>
>> Using which package?
>
> In this case I used the meta package, but I know that for all but the
> DerSimonian-Laird method the R function rma.uni of R package metafor is
> called internally.
>
>> What did you expect the weights to sum to, I wonder.
>
> I think that, to better explain the influence of single study in pooling
> the effect size, the weight are presented as percentage of the sum of
> total weight of each study, but I ask for a confirm. Nevertheless, I ask
> if it is possible to obtain for each study the value of the absolute
> weight, other than the relative weight, or at least the absolute value of
> the sum of the weights.
>

## This is an example from the examples-section
res <- metaprop(4:1, c(10, 20, 30, 40), comb.fixed=FALSE, comb.random=TRUE)

## Object res contains a lot of interesting information
## Open ?metaprop and read the section on "values"
str(res)

## Obtaining the random-effects weights
res$w.random

## Calculating relative weights manually
res$w.random/(sum(res$w.random))

HTH,

Bernd

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view
|

R: interpreting weight in meta-analysis of proportion

petretta
Fine !

Many many thanks!

Mario


-----Messaggio originale-----
Da: Bernd Weiss [mailto:[hidden email]]
Inviato: venerdì 18 aprile 2014 15.58
A: [hidden email]; [hidden email]
Oggetto: Re: [R] interpreting weight in meta-analysis of proportion

On 18.04.2014 13:02, [hidden email] wrote:

> Prof. Dewey, sorry for the trivial question and many thank for the replay.
>
>
>> Using which package?
>
> In this case I used the meta package, but I know that for all but the
> DerSimonian-Laird method the R function rma.uni of R package metafor
> is called internally.
>
>> What did you expect the weights to sum to, I wonder.
>
> I think that, to better explain the influence of single study in
> pooling the effect size, the weight are presented as percentage of the
> sum of total weight of each study, but I ask for a confirm.
> Nevertheless, I ask if it is possible to obtain for each study the
> value of the absolute weight, other than the relative weight, or at
> least the absolute value of the sum of the weights.
>

## This is an example from the examples-section res <- metaprop(4:1, c(10,
20, 30, 40), comb.fixed=FALSE, comb.random=TRUE)

## Object res contains a lot of interesting information ## Open ?metaprop
and read the section on "values"
str(res)

## Obtaining the random-effects weights
res$w.random

## Calculating relative weights manually
res$w.random/(sum(res$w.random))

HTH,

Bernd

______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.