[R-meta] 3 candidate random structures
Timothy MacKenzie
|@w|@wt @end|ng |rom gm@||@com
Wed Jul 28 18:12:18 CEST 2021
Dear Colleagues,
I just wanted to clarify that I fully understand that the 3 models that I
suggested need to be empirically tested with my data. However, my question
is that:
Given that I'm new to the `rma.mv()` function in metafor, I was wondering
if the candidate syntax that I have used (below) matches the general
definition of my data structure (i.e., studies>groups>outcomes>time)?
Thank you for your assistance, Tim
list(~ outcome | study, ~ time | interaction(study,outcome)), struct
=c("HCS","HAR")
list(~ outcome | study, ~ time | interaction(study,gr,outcome)), struct
=c("HCS","HAR")
list(~ outcome | interaction(study,gr), ~ time |
interaction(study,gr,outcome)), struct =c("HCS","HAR")
# Data structure:
study gr time outcome
1 1 0 1
1 1 0 2
1 1 1 1
1 1 1 2
1 1 2 1
1 1 2 2
1 2 0 1
1 2 0 2
1 2 1 1
1 2 1 2
1 2 2 1
1 2 2 2
2 1 0 1
2 1 0 2
2 2 0 1
2 2 0 2
On Mon, Jul 26, 2021 at 1:06 PM Timothy MacKenzie <fswfswt using gmail.com> wrote:
> Dear Meta-Analysis Experts,
>
> Several of my studies (15 out of 50) have a data structure similar to what
> I'm showing below. That is, each such study has multiple experimental
> groups ("gr"), tested several times, on 2 or more outcomes.
>
> I'm considering 3 different random structures (below). But I was wondering
> which one better aligns to such studies' data structure?
>
> Thank you for your expertise,
> Tim
>
> list(~ outcome | study, ~ time | interaction(study,outcome)), struct
> =c("HCS","HAR")
>
> list(~ outcome | study, ~ time | interaction(study,gr,outcome)), struct
> =c("HCS","HAR")
>
> list(~ outcome | interaction(study,gr), ~ time |
> interaction(study,gr,outcome)), struct =c("HCS","HAR")
>
> # Data structure:
> study gr time outcome
> 1 1 0 1
> 1 1 0 2
> 1 1 1 1
> 1 1 1 2
> 1 1 2 1
> 1 1 2 2
> 1 2 0 1
> 1 2 0 2
> 1 2 1 1
> 1 2 1 2
> 1 2 2 1
> 1 2 2 2
> 2 1 0 1
> 2 1 0 2
> 2 2 0 1
> 2 2 0 2
>
[[alternative HTML version deleted]]
More information about the R-sig-meta-analysis
mailing list