[R-sig-ME] blme optimizer warnings

[Viechtbauer, Wolfgang (SP)]

A while ago, I looked at this article and replicated the results with metafor.

library(metafor)

meta$Variance <- 1/meta$Precision

res <- rma.mv(g, Variance, random = list(~ 1 | Study/Outcome, ~ 1 | Subscale), data=meta)
res

The variances given in the article are quite heavily rounded and one value is even given as 0. With meta$Variance <- 1/meta$Precision, we can recreate the variances to avoid this issue.

Best,
Wolfgang

>-----Original Message-----
>From: R-sig-mixed-models [mailto:r-sig-mixed-models-bounces using r-project.org]
>On Behalf Of Sijia Huang
>Sent: Thursday, 14 May, 2020 4:58
>To: Vincent Dorie
>Cc: r-sig-mixed-models
>Subject: Re: [R-sig-ME] blme optimizer warnings
>
>Thank you so much, Vincent!
>
>On Wed, May 13, 2020 at 7:54 PM Vincent Dorie <vdorie using gmail.com> wrote:
>
>> A couple of guesses here in addition to what Ben mentioned, but you
>> likely don't want a prior on the covariance of the random effects and
>> the weights should be on the scale of inverse variances. The following
>> replicates the numbers for the CCREM column from table 1:
>>
>> blmer(g ~ 1 + (1 | Study) + (1 | Subscale) + (1 | Outcome:Study:Subscale),
>>       data = meta, weights = Precision, control =
>> lmerControl(optimizer = "bobyqa"),
>>       resid.prior = point(1), cov.prior = NULL)
>>
>> On Wed, May 13, 2020 at 10:04 PM Sijia Huang <huangsjcc using gmail.com> wrote:
>> >
>> > Here it is. Thanks!
>> >
>> > A demonstration and evaluation of the use of cross-classified
>> > random-effects models for meta-analysis