[R-meta] [EXTERNAL] Re: question about effect size estimates using Berkey

[Van Meter, Anna]

Dear Michael,


Thank you - this is very helpful!


A few follow-up questions for the group:

*I was using the Berkey method to account for the fact that individuals might be represented in multiple prevalence groups. Is that still necessary with this approach that allows the heterogeneity to be different within subgroup, or is it better to estimate with the Konstantopolous (yi, vi) approach?

*If it is still better to use Berkey (which I suspect is the case), does the struct="DIAG" command account for the block diagonal or is struct="UN" better?

*I had been using ML as the estimator, but this example is using REML. I'm sure this topic has been addressed before, so if someone could point me to information about which is best in this scenario, I would appreciate it.

*Finally, when I run a simple model with only the prevalence type (threegroup) as a moderator:

resmvberkeyhybrid<-rma.mv(yi, berkeyV, mods = ~ newgroup, random = ~ newgroup | articleno, struct="DIAG", data=kidtall1, digits=3)

I get separate coefficients for two of the three groups, but when I include other moderators in the model:

resmvberkeyhybridmods<-rma.mv(yi, berkeyV, mods=cbind(newgroup, yearpub_center, USA, multiinformantyn, broad, age_center, lifetime), random = ~ newgroup | articleno, struct="DIAG",  data=kidtall1, digits=3)

I get only one coefficient for threegroup. I'm not sure why this would change.


Many thanks!


Best,

Anna


________________________________
From: Michael Dewey <lists using dewey.myzen.co.uk>
Sent: Monday, December 24, 2018 8:21 AM
To: Van Meter, Anna; r-sig-meta-analysis using r-project.org
Subject: [EXTERNAL] Re: [R-meta] question about effect size estimates using Berkey

External Email. Use Caution.

Dear Anna



I am not an authority on multivariate meta-analysis but a similar

phenomenon is well known for univariate so I suspect that the situation

is that when you use a moderator you are using all the data-set to

estimate variances whereas when you subset you are just using that subset.



https://urldefense.proofpoint.com/v2/url?u=http-3A__www.metafor-2Dproject.org_doku.php_tips-3Acomp-5Ftwo-5Findependent-5Festimates&d=DwIGaQ&c=vq5m7Kktb9l80A_wDJ5D-g&r=OFmh4IthkcBq02H-yJ16iDBLgPO6gowYXXX_q2XHqM4&m=SXa8XpjoFTkt_9HG0Tvy95QYkD_jMndIHm8F_pTNA74&s=oJkLUH50LltjfymgGZSoxa1WShj4GH0nAc2WyPDlZV8&e=



Explains the situation for the univariate case.



Michael



On 23/12/2018 16:40, Van Meter, Anna wrote:

> Hello,

>

>

>

> I have a question about inconsistent results that I am getting when trying to calculate effect sizes using the Berkey method. The data are prevalence rates from different studies and some studies include multiple prevalence rates, which may reflect the same people. For example, one study could report the prevalence for bipolar I and for the full bipolar spectrum, which would include those with bipolar I, plus others who have other subtypes of bipolar disorder. There are three potential prevalence categories � bipolar I, bipolar I & II, all bipolar.

>

>

>

> I have created two binary dummy codes to represent which subtypes are included in each sample � inc2yn (bipolar I & II) and nosyn (all bipolar). There is also a code that says which of the three categories an effect size belongs to (threegroup) and is coded 1 (bipolar I), 2 (bipolar I & II) or 3 (all bipolar).

>

>

>

> Id refers to individual effects sizes, articleno is the study, so Ids are nested within articleno.

>

>

>

> There are 27 effect sizes from 18 studies. There are 13 bipolar I effects, 7 bipolar I & II effects, and 7 all bipolar effects.

>

>

>

> Initially, I ran the following code to get estimates for the three effect sizes:

>

> resmvberkey0<-rma.mv(yi, berkeyV, data=kidtall1, mods= cbind(inc2yn, nosyn), slab = paste(reference),random = list((~ 1 | Id), ~ 1 | articleno), method="ML")

>

>

>

> Then, to get an estimate for bipolar I & II (for example). I would use the following command:

>

> predict(resmvberkey0, transf=transf.ilogit, newmods = cbind(1,0))

>

>

>

> Later, when I was making a forest plot, I wanted to get estimates from a model that did not include moderators (it seems that you cannot include moderators in the addpoly command). This led me to use the subset command to get estimates for each effect size separately:

>

> resmvberkey0bp1_2<-rma.mv(yi, berkeyV, data=kidtall1, subset=threegroup==2, random = list((~ 1 | Id), ~ 1 | articleno), method="ML")

>

> predict(resmvberkey0bp1_2, transf=transf.ilogit)

>

>

>

> The rates I get by manipulating the moderators to estimate for a single category are different from the rates I get when I use the subset command:

>

> moderator result for bipolar I = .11%

>

> subset result for bipolar I = .15%

>

>

>

> moderator result for bipolar I & II = .18%

>

> subset result for bipolar I & II is .17%

>

>

>

> moderator result for all bipolar is 1.51%

>

> subset result for all bipolar is 3.56%

>

>

>

> Any thoughts about why these results would be so different � and why � would be greatly appreciated.

>

>

>

> Thank you,

>

> Anna

>

>

> Anna Van Meter, PhD

>

> Assistant Professor, The Feinstein Institute for Medical Research

>

> Adjunct Assistant Professor, Ferkauf Graduate School of Psychology, Yeshiva University

>

> The Zucker Hillside Hospital, Division of Psychiatry Research

>

> 75-59 263rd Street

>

> Glen Oaks, NY 11004

>

> 718.470.5813

>

>

>

> The information contained in this electronic e-mail transmission and any attachments are intended only for the use of the individual or entity to whom or to which it is addressed, and may contain information that is privileged, confidential and exempt from disclosure under applicable law. If the reader of this communication is not the intended recipient, or the employee or agent responsible for delivering this communication to the intended recipient, you are hereby notified that any dissemination, distribution, copying or disclosure of this communication and any attachment is strictly prohibited. If you have received this transmission in error, please notify the sender immediately by telephone and electronic mail, and delete the original communication and any attachment from any computer, server or other electronic recording or storage device or medium. Receipt by anyone other than the intended recipient is not a waiver of any attorney-client, physician-patient or other privilege.

>        [[alternative HTML version deleted]]

>

>

> _______________________________________________

> R-sig-meta-analysis mailing list

> R-sig-meta-analysis using r-project.org

> https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_mailman_listinfo_r-2Dsig-2Dmeta-2Danalysis&d=DwIGaQ&c=vq5m7Kktb9l80A_wDJ5D-g&r=OFmh4IthkcBq02H-yJ16iDBLgPO6gowYXXX_q2XHqM4&m=SXa8XpjoFTkt_9HG0Tvy95QYkD_jMndIHm8F_pTNA74&s=_IybPi_-FLjrvka2ICpDICE9sqABp-xdoKU8bjdO2-Y&e=

>



--

Michael

https://urldefense.proofpoint.com/v2/url?u=http-3A__www.dewey.myzen.co.uk_home.html&d=DwIGaQ&c=vq5m7Kktb9l80A_wDJ5D-g&r=OFmh4IthkcBq02H-yJ16iDBLgPO6gowYXXX_q2XHqM4&m=SXa8XpjoFTkt_9HG0Tvy95QYkD_jMndIHm8F_pTNA74&s=BGjriTMX2UobNycy1XbI-y6N2tFKMj4-aiwzBFTx-js&e=



The information contained in this electronic e-mail transmission and any attachments are intended only for the use of the individual or entity to whom or to which it is addressed, and may contain information that is privileged, confidential and exempt from disclosure under applicable law. If the reader of this communication is not the intended recipient, or the employee or agent responsible for delivering this communication to the intended recipient, you are hereby notified that any dissemination, distribution, copying or disclosure of this communication and any attachment is strictly prohibited. If you have received this transmission in error, please notify the sender immediately by telephone and electronic mail, and delete the original communication and any attachment from any computer, server or other electronic recording or storage device or medium. Receipt by anyone other than the intended recipient is not a waiver of any attorney-client, physician-patient or other privilege.
	[[alternative HTML version deleted]]