introduction

This code reads in data from an interview-based research project, transforms that data, and creates the visualizations used in the research output (manuscript).

Created by Casey Helgeson (Orchid id 0000-0001-5333-9954).

This code was tested and confirmed to work with R version 4.0.3 (2020-10-10), nickname: "Bunny-Wunnies Freak Out."

inputData

Five data sources are supplied in the "inputData" subdirectory: "brat files" (folder), "response_scale.csv", "coauthorship.csv", "personnel_blinded.csv", and "by.institution".

• "brat_files" is a folder containing 29 .ann files generated through qualitative data analysis of interview transcripts using the Brat Rapid Annotation Tool (https://brat.nlplab.org/).

• "resonse_scale.csv" contains six interview responses from each study participants.

• "coauthorship.csv" contains two coauthorship indices calculated for each study participant.

• "personnel_blinded.csv" contains the rank (faculty versus postdoc) and months of project funding for each study participant.

• "by.institution.csv" contains information on which project institutions contributed to coauthored publications.

code

Three R files, "annotations.R", "response.R", and "count_inst.R" can be found in the "code" subdirectory.

"annotations.R" reads in data generated through qualitative data analysis of interview transcripts using the Brat Rapid Annotation Tool (https://brat.nlplab.org/). Specifically, the code reads in a set of .ann files generated by the Brat Tool (one file per transcript). The code also reads in a personnel file with information about each interview participant in order to subdivide the transcript data according to personnel categories. "annotations.R" may be useful to others who need to process and plot the results of qualitative data analysis using the Brat Tool. As per comments in the code, using "annotations.R" for your own data will require modifying a vector containing the names of the attributes and entities codes used in the text annotation.

"response.R" reads in interview participants' responses to six questions using a 1-5 quantitative response scale. The code tabulates and plots these data. The code also reads in two other files containing information about each participant to be plotted against interview responses. "response.R" is pretty specific to the particular analysis we have done for the associated research manuscript and is unlikely to be useful to others beyond reproducing our work or inspecting the details of our analysis for that manuscript.

"count_inst.R" reads in data indicating which institutions contributed to each project-supported publication. The code counts how many publications each institution co-authored and also how many publications were coauthored across each pair of institutions.

how to run the code

It is expected that a user trying to reproduce our results would do so from within an R session. Figures from the manuscript can be reproduced as per the bullet points below. Figures produced by running the code will be found in the "outputs" subdirectory. "annotation.R", "response.R", and "count_inst.R" can be run in any order and need not be run in the same R session.

• Figure 1: Set working directory to the "inputData" folder then run "annotations.R".

• Figures 2, 3, and S1: Set working directory to the "inputData" folder then run "response.R".

• Table S8: Set working directory to the "inputData" folder then run "count_inst.R".