[R-meta] Differences in calculation of CVR in escalc()

[Viechtbauer Wolfgang (SP)]

I see. No, there is no reference explicitly for that. But this all follows directly from the proof that the sample mean and sample variance are independent for the normal distribution.

Best,
Wolfgang

-----Original Message-----
From: Samuel Knapp [mailto:samuel.knapp at tum.de] 
Sent: Thursday, 12 October, 2017 11:10
To: Viechtbauer Wolfgang (SP); r-sig-meta-analysis at r-project.org
Cc: Marcel van der Heijden
Subject: Re: [R-meta] Differences in calculation of CVR in escalc()

Thanks.

I actually meant a reference with a version of Eq. 12 from Nakagawa et 
al. (2015) without the correlation term, and including the argument 
against the use of the correlation term, as done in this paper.

Best,

Samuel

On 12/10/17 09:55, Viechtbauer Wolfgang (SP) wrote:
> I've added a note to help(escalc) about this.
>
> A reference showing that the sample mean and variance are independent for normally distributed data? You can find some references here:
>
> https://en.wikipedia.org/wiki/Normal_distribution#Properties
>
> Best,
> Wolfgang
>
> -----Original Message-----
> From: Samuel Knapp [mailto:samuel.knapp at tum.de]
> Sent: Thursday, 12 October, 2017 9:29
> To: Viechtbauer Wolfgang (SP); r-sig-meta-analysis at r-project.org
> Cc: Marcel van der Heijden
> Subject: Re: [R-meta] Differences in calculation of CVR in escalc()
>
> Hi Wolfgang,
>
> many thanks for your explanation. I think I understand your point now.
>
> However, it would be nice, if this difference in calculation would be
> made clear in the help of escalc. So far, only Nakagawa et al. (2015) is
> cited (even 3 times), which lead me to assume that calculations are done
> as shown in the given reference.
>
> Also, is there any reference with your argument?
>
> Best,
>
> Samuel