[R-meta] Forest Plot

[Christine Emmer]

Hi James, thank you for your ideas I like the idea of a forest-and-bridie plot, unfortunately there is a time limit for the thesis. But I will definitely consider the idea for the next analysis! All the best, Christine 

> Am 11.10.2018 um 13:18 schrieb James Pustejovsky <jepusto using gmail.com>:
> 
> In the spirit of “confess your unpopular opinion,” I think forest plots with so many effect sizes are uninformative. I would rather just see a histogram or density plot of the raw effect sizes, along with a contour-enhanced funnel plot. (And the raw data in supplementary materials, of course.)
> 
> One thing I have seen done in this situation is create a forest plot of *average* effects from each study, and then overlay dots with the individual ES estimates. This makes the chart more legible because there’s only one line per study, and it also helps to illustrate the clustered structure of the data. The result looks a bit like birds or bees buzzing around the trees in the forest. (Call it a forest-and-birdie plot?) Unfortunately I don’t know of functions to do this automatically. Would have to code it custom.
> 
> James
> 
>> On Oct 11, 2018, at 5:02 AM, Christine Emmer <christine.emmer using uni-mannheim.de> wrote:
>> 
>> Thank you very much for your help! Is it possible to split the groups of studies via R? Or should I just graphically cut the forest plot?
>> 
>> Best, christine 
>> 
>> 
>>> Am 11.10.2018 um 11:51 schrieb Gerta Ruecker <ruecker using imbi.uni-freiburg.de>:
>>> 
>>> Dear Christine,
>>> 
>>> Yes, this makes sense. For the technical point, you may split the forest plot into plots of groups of studies (without any further meaning, simply alphabetically), just to have all studies shown in a readable way on four or five pages in your thesis or paper, such as splitting a long table into pages.
>>> 
>>> Best,
>>> 
>>> Gerta
>>> 
>>> 
>>>> Am 11.10.2018 um 11:45 schrieb Christine Emmer:
>>>> Dear Wolfgang, and dear Gerta,
>>>> 
>>>> Thank you very much for your quick reply. I was using a three-level approach to account for dependencies of effect sizes from the same study. I still have to think about interpretation and what the distribution of heterogeneity means for my results (around 50% on sample level, and 50% on study level).
>>>> Does this make sense for you?
>>>> 
>>>> Best, Christine
>>>> 
>>>>> Am 11.10.2018 um 11:34 schrieb Gerta Ruecker <ruecker using imbi.uni-freiburg.de>:
>>>>> 
>>>>> Dear Christine,
>>>>> 
>>>>> With over 300 effect sizes, I think this is somewhat unavoidable ;-) With 800% enlargement you see the details.
>>>>> 
>>>>> Beyond the technical problem, another point is: Many effect sizes come from the same study and they do by no means seem to be independent. Rather, there are blocks defined by studies. You should think how to interpret this (what do different effect sizes from the same study mean?), and whether it really makes sense to put all these into the same meta-analysis (that assumes independent effect sizes) and even to show a pooled estimate.
>>>>> 
>>>>> Best,
>>>>> 
>>>>> Gerta
>>>>> 
>>>>> 
>>>>>> Am 11.10.2018 um 11:28 schrieb Viechtbauer, Wolfgang (SP):
>>>>>> Hi Christine,
>>>>>> 
>>>>>> Increase the height of the plotting device (argument height in function pdf() -- see help(pdf)). And then play around with the settings for 'cex' and possibly 'xlim' in forest(). If you run the forest() command like this:
>>>>>> 
>>>>>> sav <- forest(...)
>>>>>> sav
>>>>>> 
>>>>>> then you can see what the defaults are. The default settings may not be appropriate for forest plots witht this many effects.
>>>>>> 
>>>>>> Best,
>>>>>> Wolfgang
>>>>>> 
>>>>>> -----Original Message-----
>>>>>> From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces using r-project.org] On Behalf Of Christine Emmer
>>>>>> Sent: Thursday, 11 October, 2018 11:00
>>>>>> To: r-sig-meta-analysis using r-project.org
>>>>>> Subject: [R-meta] Forest Plot
>>>>>> 
>>>>>> ATTACHMENT(S) REMOVED: forest0.pdf
>>>>>> 
>>>>>> Hello,
>>>>>> 
>>>>>> I am currently working on my master’s thesis, a meta-analysis in social science. I included over 300 effect sizes, which I want to display by a forest plot.
>>>>>> Unfortunately, the plot is because of the high number of effect sizes not clearly arranged (see attached picture).
>>>>>> Maybe there is a simple solution?
>>>>>> 
>>>>>> Many thanks and best regards, Christine
>>>>>> _______________________________________________
>>>>>> R-sig-meta-analysis mailing list
>>>>>> R-sig-meta-analysis using r-project.org
>>>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
>>>>> -- 
>>>>> 
>>>>> Dr. rer. nat. Gerta Rücker, Dipl.-Math.
>>>>> 
>>>>> Institute of Medical Biometry and Statistics,
>>>>> Faculty of Medicine and Medical Center - University of Freiburg
>>>>> 
>>>>> Stefan-Meier-Str. 26, D-79104 Freiburg, Germany
>>>>> 
>>>>> Phone:    +49/761/203-6673
>>>>> Fax:      +49/761/203-6680
>>>>> Mail:     ruecker using imbi.uni-freiburg.de
>>>>> Homepage: https://portal.uni-freiburg.de/imbi/persons/ruecker?set_language=en
>>>>> 
>>> 
>>> -- 
>>> 
>>> Dr. rer. nat. Gerta Rücker, Dipl.-Math.
>>> 
>>> Institute of Medical Biometry and Statistics,
>>> Faculty of Medicine and Medical Center - University of Freiburg
>>> 
>>> Stefan-Meier-Str. 26, D-79104 Freiburg, Germany
>>> 
>>> Phone:    +49/761/203-6673
>>> Fax:      +49/761/203-6680
>>> Mail:     ruecker using imbi.uni-freiburg.de
>>> Homepage: https://portal.uni-freiburg.de/imbi/persons/ruecker?set_language=en
>>> 
>> 
>> 
>>   [[alternative HTML version deleted]]
>> 
>> _______________________________________________
>> R-sig-meta-analysis mailing list
>> R-sig-meta-analysis using r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis