Code to Replicate Analsis for: "And, not Or: Quantity, Quality in Scientific Publishing"

Matthew J. Michalska-Smith and Stefano Allesina

---

Data Collection

A python script for extracting publication records associated with Scopus IDs is provided in Code_Analysis/Retrieve_papers.py. This Script requires only a ;-delimited file containing columns labeled 'scopus1', 'scopus2', and 'scopus3' containing the (up to three) synonymous Scopus IDs affiliated with each author to be included in the analysis, and a Scopus API Key. The latter can be obtained at (https://dev.elsevier.com/sc_apis.html).

This script will produce a series of files (one for each author) in the Data folder to be utilized in the further analysis. Each of these files will be a ;-delimited file with the header:

Author;PaperID;Citations;Year;Journal

In order to facilitate replication, we also provide the actual data used in our analysis with the author and paper identifiers anonymized. These files have six digit codes replacing the ScopusIDs of the authors and the PaperIDs of each of their publications and do not include journal information, but are otherwise identical to the output of Retrieve_papers.py.

Basic Pairwise Analysis

This part of the analysis involves running the script PairwiseComparision.R, in the folder Code_Analysis. There are several global parameters set at the beginning of the file:

Parameter	Default Value	Definition
INTERVAL_START	1980	earliest year to consider publications from
INTERVAL_END	2006	latest year to consider publications from
MINPAPERS	20	the minimum number of publications an author must have within the interval to be included
QUALITY_TRANS	"log"	the transformation (if any) to perform on citations to make them a quality metric

This will produce a data file located at

Figures/PairwiseComparison_[QUALITY_TRANS]_[INTERVAL_START]-[INTERVAL_END].RData

to be used in the figure generation scripts.

Figure and Table generation

Figures and tables associated with the analysis can be found in the Code_Figures folder. These two scripts DensityPlot.Rmd and KStestPlot.Rmd plot the density distributions of the concordance values and statistical comparisons between the distributions, respectively. The latter also produces the tables found in the text.