[R-meta] Importing Correlations from PDF to table format

[Viechtbauer, Wolfgang (SP)]

Hi Kiet,

Thanks for providing this feedback -- this is likely to be quite useful for others looking into automating this process (including me)!

Best,
Wolfgang

>-----Original Message-----
>From: Kiet Huynh [mailto:kietduchuynh using gmail.com]
>Sent: Wednesday, 02 March, 2022 4:23
>To: Viechtbauer, Wolfgang (SP)
>Cc: James Pustejovsky; R meta
>Subject: Re: [R-meta] Importing Correlations from PDF to table format
>
>Hi Wolfgang,
>
>Thank you for your recommendation. Using both the tabulizer package and and rcalc
>function has done exactly what I was hoping for.
>
>I found the tabulizer package to be much more accurate than the pdftools package.
>The tabulizer package is mostly accurate, but sometimes it struggles with
>correctly identifying negative numbers in the correlation table. So I still have
>to do some data cleaning in R to fix incorrect values. Despite these issues, my
>process for coding meta-analysis is much more efficient and accurate now.
>
>Big thanks to you and James for your help!
>
>Kiet
>
>On Feb 28, 2022, at 11:04 AM, Viechtbauer, Wolfgang (SP)
><wolfgang.viechtbauer using maastrichtuniversity.nl> wrote:
>
>Hi Kiet,
>
>The rcalc() function from metafor could be used for this. It even computes the
>var-cov matrix of the elements in the correlation matrix for you:
>
>library(metafor)
>R <- matrix(c(1, .3, .5, .3, 1, .6, .5, .4, 1), 3, 3)
>R
>rcalc(R, ni=50)
>
>Best,
>Wolfgang
>
>-----Original Message-----
>From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces using r-project.org] On
>Behalf Of Kiet Huynh
>Sent: Monday, 28 February, 2022 19:45
>To: James Pustejovsky
>Cc: R meta
>Subject: Re: [R-meta] Importing Correlations from PDF to table format
>
>Hi James,
>
>Thank you for recommending these helpful packages. I was able to import the pdf
>correlation table into a dataframe format in R. Are you aware of any R code that
>could convert that correlation matrix dataframe into a meta-analysis type
>dataframe (i.e., a column for variable 1, a column for variable 2, and a column
>for correlation effect size)?
>
>Best,
>
>Kiet
>
>On Feb 25, 2022, at 10:58 AM, James Pustejovsky <jepusto using gmail.com> wrote:
>
>The pdftools package might be helpful:
>https://github.com/ropensci/pdftools <https://github.com/ropensci/pdftools>
>It has very low-level utilities for extracting text from pdf. You'd still have
>to do some data clean-up to get the correlations into the form needed for
>analysis.
>
>The tabulizer package is meant to provide tools customized for working with pdf
>tables:
>
>https://github.com/ropensci/tabulizer <https://github.com/ropensci/tabulizer>
>But it requires Java and it appears to be archived on CRAN. I'm not sure what
>its development status is. Caveat emptor, I guess.
>
>James