how to combine standard errors of the mean

[elena]

Hi all,

I have 3 different Mean substitution rates and their standard errors
obtained using BEAST for three different datasets (not different runs
of the same dataset which you can combine using tracer). I would like
to combine them in a single estimate. In such situation I would need
the sample size of the distribution and in the case of BEAST I wonder
whether this value corresponds to the EFFECTIVE SAMPLE SIZE of the
posterior distribution. I would greatly appreciate any help!

Thanks very much

elena

[Hassan Ebrahimi]

Elena,

I am not sure but I think you want to calculate pooled mean and pooled variance of the three sample sets. In this case you should use the number of samples for each sets. The equations are not complicated but it's easier easier to use the calculator in the following web page :
http://home.ubalt.edu/ntsbarsh/Business-stat/otherapplets/Pooled.htm

The ESS ( EFFECTIVE SAMPLE SIZE ) has nothing to do with this.

I hope this helps,

Regards,

Hassan

---

From: elena <elegom...@gmail.com>
To: beast-users <beast...@googlegroups.com>
Sent: Monday, February 6, 2012 12:14 AM
Subject: how to combine standard errors of the mean

--
You received this message because you are subscribed to the Google Groups "beast-users" group.
To post to this group, send email to beast...@googlegroups.com.
To unsubscribe from this group, send email to beast-users+unsub...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/beast-users?hl=en.

[Tod W Reeder]

In your case, if your three different data sets represent three different genes, and you are wanting to get a single overall mean substitution rate, I would think it would be better to do this from the context of a single unpartitioned concatenated analysis so that the "mean" substitution rate is based on the same posterior distribution of trees.  Not sure how meaningful calculating a mean from a pooled set of different posterior distribution of   rates would be when each of the original estimates are based on different posterior distributions of trees.  But then, maybe I could be wrong and missing something here since I'm not totally clear as to your end goal.

---------------------------------------------------
Tod W. Reeder
Associate Chair & Graduate Advisor
Department of Biology
San Diego State University
San Diego, CA  92182-4614
619-594-7826 or 619-594-6375 (office)
619-594-5676 (fax)
tre...@sunstroke.sdsu.edu
http://sites.google.com/site/todreederslab/
---------------------------------------------------

--
You received this message because you are subscribed to the Google Groups "beast-users" group.
To post to this group, send email to beast...@googlegroups.com.

To unsubscribe from this group, send email to beast-users...@googlegroups.com.

[Andrew Rambaut]

Hi All,

This sounds like a case for using a hierarchical analysis (i.e., you want to learn about the population from which the rates are being drawn). Beauti 1.7 will be released very shortly (or can be obtained from the source code repository) and enables you to set these up easily. See for a description of the approach:

http://sysbio.oxfordjournals.org/content/52/5/649.short

Andrew

[elena]

Dear Tod,

Thanks for your response. I think that I did not explain my self
clearly, Indeed, my three different datasets correspond to three
different species and one gene and I would like to obtain a single
estimate for the mean.Rate of that gene pooled for all three species.
Does it make sense?

regards
e

On 5 Feb, 19:56, Tod W Reeder <tree...@sunstroke.sdsu.edu> wrote:
> In your case, if your three different data sets represent three different genes, and you are wanting to get a single overall mean substitution rate, I would think it would be better to do this from the context of a single unpartitioned concatenated analysis so that the "mean" substitution rate is based on the same posterior distribution of trees.  Not sure how meaningful calculating a mean from a pooled set of different posterior distribution of   rates would be when each of the original estimates are based on different posterior distributions of trees.  But then, maybe I could be wrong and missing something here since I'm not totally clear as to your end goal.
>
> ---------------------------------------------------
> Tod W. Reeder
> Associate Chair & Graduate Advisor
> Department of Biology
> San Diego State University
> San Diego, CA  92182-4614
> 619-594-7826 or 619-594-6375 (office)
> 619-594-5676 (fax)

> tree...@sunstroke.sdsu.eduhttp://sites.google.com/site/todreederslab/

[elena]

Dear Hassan,

The problem that I see is that the estimates that I have and that I
got from BEAST correspond to the mean.Rate and the standard error of
the mean (not the variance) so I need the sample size in order to
apply the formula you mention. For this reason I asked about the ESS
because intuitively I though the standard error of the mean calculated
by beast would be based on the ESS of the posterior distribution.
Hope it makes sense,

thanks!
e

On 5 Feb, 18:28, Hassan Ebrahimi <hx...@yahoo.com> wrote:
> Elena,
>
> I am not sure but I think you want to calculate pooled mean and pooled variance of the three sample sets. In this case you should use the number of samples for each sets. The equations are not complicated but it's easier easier to use the calculator in the following web page :http://home.ubalt.edu/ntsbarsh/Business-stat/otherapplets/Pooled.htm
>
> The ESS ( EFFECTIVE SAMPLE SIZE ) has nothing to do with this.
>
> I hope this helps,
>
> Regards,
>
> Hassan
>
> ________________________________

>  From: elena <elegomezd...@gmail.com>

> To: beast-users <beast...@googlegroups.com>
> Sent: Monday, February 6, 2012 12:14 AM
> Subject: how to combine standard errors of the mean
>
> Hi all,
>
> I have 3 different Mean substitution rates and their standard errors
> obtained using BEAST for three different datasets (not different runs
> of the same dataset which you can combine using tracer). I would like
> to combine them in a single estimate. In such situation I would need
> the sample size of the distribution and in the case of BEAST I wonder
> whether this value corresponds to the EFFECTIVE SAMPLE SIZE of the
> posterior distribution. I would greatly appreciate any help!
>
> Thanks very much
>
> elena
>
> --
> You received this message because you are subscribed to the Google Groups "beast-users" group.
> To post to this group, send email to beast...@googlegroups.com.

> To unsubscribe from this group, send email to beast-users...@googlegroups.com.