decimal.places

How many decimal places are used in abstracts? Code for paper in F1000: "Missing the point: are journals using the ideal number of decimal places?", published here: https://f1000research.com/articles/7-450/.

The key files are:

1. make.data.R, finds eligible articles from specified journal lists, then reads the abstracts from Pubmed and calculates the number of decimal places making a dataset ready for analysis.

2. Analysis.Ready.RData, Analysis.Ready.xlsx and Analysis.Ready.txt, the main dataset for analysis produced by make.data.R in R, Excel and tab-delimited format, respectively.

3. decimal.places.stats.Rmd, Rmarkdown file to create the analyses shown in the paper. Was run using R version 3.4.4.

4. MultinomialCIsBayes.R, calculates multinomial Dirichlet confidence intervals for the three categories of percents (too few, just right, too many) using WinBUGS. Requires WinBUGS (version 1.4) and the R package R2WinBUGS.

5. decimalplaces.R, calculates the number of decimal places and significant figures.

6. journal.meta.txt & journal.meta.ii.txt, tab-delimited lists of the eligible articles searched for percents. Split into two because of the time needed to make the data. journal.meta.RData & journal.meta.ii.RData are R-data versions of same.

The code requires the following R packages:

• diagram
• R2WinBUGS
• rentrez
• stringr
• XML

To replicate the results in the paper, use: decimal.places.stats.Rmd combined with Multi.results.RData and Analysis.Ready.RData. If Multi.results.RData does not exist, it will be created by the Rmarkdown code.

To recreate the entire analysis data set use: make.data.R optionally combined with journal.meta.RData and journal.meta.ii.RData to skip the step to create the lists of papers.

Archived data and code as at time of initial publication (version 1.0): http://doi.org/10.5281/zenodo.1213574 and after comments from peer reviewers (version 1.1): http://doi.org/10.5281/zenodo.1300056.