[R-meta] Fwd: predicted intervals in metafor

[Viechtbauer, Wolfgang (SP)]

Dear Raynaud,

Not sure if you ever received an answer to this.

In essence, the results indicate no evidence for residual heterogeneity in the true effects. I suspect that there is little (or any) evidence for heterogeneity in these data in the first place (i.e., without including moderators in the model, the Q-test is probably not significant to begin with and I^2 might be zero or close to it).

But I see: "vi=se_cohen". Argument 'vi' is for the *sampling variances*. However, 'se_cohen' suggests that you are passing the *stanard errors* to the function. This would be incorrect. Either use "vi=se_cohen^2" or "sei=se_cohen".

Best,
Wolfgang

-----Original Message-----
From: Raynaud Armstrong [mailto:raynaud.armstrong using gmail.com] 
Sent: Friday, 08 June, 2018 19:24
To: Michael Dewey
Cc: Viechtbauer, Wolfgang (SP); r-sig-meta-analysis using r-project.org
Subject: Re: [R-meta] Fwd: predicted intervals in metafor

Hi Michael, 
Sorry, I had forgotten to reply to the subscription email. I believe I am now registered.
My apologies once again.

Ray

On Fri, Jun 8, 2018 at 1:12 PM, Michael Dewey <lists using dewey.myzen.co.uk> wrote:
Ray, please keep the list in the loop and please register so we do not have to approve each post.

Michael

On 08/06/2018 18:03, Raynaud Armstrong wrote:
---------- Forwarded message ----------
From: Raynaud Armstrong <raynaud.armstrong using gmail.com>
Date: Fri, Jun 8, 2018 at 1:02 PM
Subject: Re: [R-meta] predicted intervals in metafor
To: Michael Dewey <lists using dewey.myzen.co.uk>

Thanks Michael. *I got the following output for my meta-regression. My
heterogeneity statistics are all 0 - what does this mean?*

#meta-regression
metareg <- rma (yi=cohen, vi=se_cohen, mods = ~ mean_age + year +
country, data=metav)
Warning message:
In rma(yi = cohen, vi = se_cohen, mods = ~mean_age + year + country,  :
   Studies with NAs omitted from model fitting.
metareg

Mixed-Effects Model (k = 7; tau^2 estimator: REML)

tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.1825)
tau (square root of estimated tau^2 value):             0
I^2 (residual heterogeneity / unaccounted variability): 0.00%
H^2 (unaccounted variability / sampling variability):   1.00
R^2 (amount of heterogeneity accounted for):            NA%

Test for Residual Heterogeneity:
QE(df = 3) = 0.3265, p-val = 0.9550

Test of Moderators (coefficient(s) 2:4):
QM(df = 3) = 1.6838, p-val = 0.6406

Model Results:

             estimate       se     zval    pval     ci.lb     ci.ub
intrcpt      39.8696  65.5910   0.6079  0.5433  -88.6863  168.4255
mean_age      0.0456   0.0373   1.2223  0.2216   -0.0275    0.1187
year         -0.0203   0.0332  -0.6109  0.5413   -0.0852    0.0447
countryUSA   -0.1974   0.3959  -0.4987  0.6180   -0.9733    0.5785

---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Please let me know. Thanks.
Ray