Trends in gender homophily in scientific publications (data)

This dataset contains records of research articles extracted from the Web of Science (WoS) from 1980 to 2019---in total, 15,642 journals, 28,241,100 articles and 111,980,858 authorships across 153 research areas.

The main dataset (author_address_article_gend_v3.parquet), in Parquet format, contains all the authorships, where an authorship is defined as the tuple article-author. There are 12 variables per authorship (row):

ut: unique article identifier.

daisng_id: unique author identifier.

author_no: author number, as listed in the article.

country: author country (two-letter ISO code).

date: publication date.

gender: gender of the author ("male" or "female"), as provided by the Genderize.io API.

probability: probability of the gender attribute, as provided by the Genderize.io API.

count: number of entries for the author first name, as provided by the Genderize.io API.

jsc: journal subject category.

field: field of research.

research_area: area of research.

n_aut: number of authors in this publication.

journal: journal name.

alphabetical: whether the author list for this article is in alphabetical order.

With the previous dataset, a resampler was applied to generate null homophily values for each year. There are 4 datasets in R Data Serialization (RDS) format:

null_field.rds: null homophily values per country, year and field of research.

null_field_comp.rds: null homophily values per year and field of research (only for complete authorships).

null_research.rds: null homophily values per year and area of research.

null_research_comp.rds: null homophily values per year and area of research (only for complete authorships).

All these datasets have the same structure:

country: country (two-letter ISO code).

year: year.

variable: either field or research area name.

m: average homophily.

s: homophily std. error.

Finally, some supplementary files used in the descriptive analysis and methods:

File null_research_l2019.rds is an example of the output from the resampling algorithm for year 2019.

File wos_category_to_field.csv is a mapping from WoS categories to more general fields.

File jcr_if_2020.csv contains the percentiles of the journal impact factor for the JCR 2020.