Targeted funding and changing research landscapes: Dataset

This repository contains the data and script to replicate our result of the Research on Research Instite (RoRI) PORTFOLIO project, as reported inTraag, V., Brady, E., Simón-Martínez, C., Rafols, I., & Schneider, J. Targeted Funding and Changing Research Landscapes. RoRI Working Paper No. 18, February 2026. 10.6084/m9.figshare.31282411The data collection and analysis is described more in-depth in the report.DataAll data is available in the data directory. The base map of the science landscape is available in base_map.txt. The last 10 columns provide additional information about the various topics (also referred to as micro clusters). Without these 10 columns the file can be directly opened in VOSviewer. You can provide additional weight and score columns to VOSviewer for visualisation. Note that the id column refer to the identifier of the micro cluster which is also used in the other files.Funding schemesThe funding schemes themselves are available from the file funding_schemes.csv and include the following columns:funder: The funding agency of the funding scheme (SNSF or RCN).funding_scheme: The name of the funding scheme.is_targeted: Whether the funding scheme is considered targeted or not.min_start_year: The first starting year in the funding scheme.max_start_year: The last starting year in the funding scheme.min_end_year: The first end year in the funding scheme.max_end_year: The last end year in the funding scheme.in_targeted_analysis: Whether the funding scheme is also included in the targeted analysis or not.Publications in topicsTargetedThe number publications per topic for the various targeted funding schemes is available from the file topic_pubs.csv and contains the following columns:funder: The funding agency of the funding scheme (SNSF or RCN).funding_scheme: The name of the funding scheme.pub_year: The year of publication.micro_cluster_id: The micro cluster identified (matches the id column in the base_map.txt)type: The type of publications: before, during, after or grant.n_pub: The total number of publications.Non-targetedThe number publications per topic for all non-targeted funding schemes together is available from the file non_targeted_topic_pubs.csv and contains the following columns:funder: The funding agency of the funding scheme (SNSF or RCN).start_year: The starting year of the grantpub_year: The year of publication.micro_cluster_id: The micro cluster identified (matches the id column in the base_map.txt)type: The type of publications: before, during, after or grant.n_pub: The total number of publications.BaselinesThe baselines are available from three different files:norwegian_topic_pubs.csvswiss_topic_pubs.csvbaseline_topic_pubs.csvThese files contain the number of publications per topic per publication year in for respectively Norway, Switzerland and the EU + US (see the paper for the definition). The file contains the following columns:pub_year: The year of publication.micro_cluster_id: The micro cluster identified (matches the id column in the base_map.txt)n_pub: The total number of publications.SourceThe statistical model that can be run on the combination of the various files is available in the file src/model.stan. It is written in the Stan language and can be run in various languages. We ran the model in Python using cmdstanpy. You can setup a new conda environment using the specification in environment.yml usingconda env create -f environment.ymlOnce the environment is setup you can run the model using python model.py with the following arguments:usage: model [-h] --funder {RCN,SNSF} (--targeted_funding_programme TARGETED_FUNDING_PROGRAMME | --non_targeted_start_year NON_TARGETED_START_YEAR) [--stan_file STAN_FILE] --baseline {Swiss,Norwegian,EU+US} base_dir