Data reuse and the open data citation advantage.

Area of Science:

• Bioinformatics
• Genomics
• Data Science

Background:

• Attribution for data reuse is crucial for rewarding data creators and ensuring research provenance.
• Previous studies suggest publicly available datasets receive more citations, but lacked statistical power to control for confounding factors.
• Understanding data reuse patterns over time and across datasets is limited.

Purpose of the Study:

• To quantify the citation benefit of making gene expression microarray data publicly available while controlling for known citation predictors.
• To investigate the temporal patterns and variability of data reuse by third-party investigators.

Main Methods:

• Multivariate regression analysis of 10,555 studies generating gene expression microarray data.
• Inclusion of covariates: publication date, journal impact factor, open access, author count, author publication history, author country, institution citation history, and study topic.
• Compilation of 9,724 instances of third-party data reuse using GEO or ArrayExpress accession numbers.

Main Results:

• Studies with publicly available data received 9% more citations (95% CI: 5% to 13%) after controlling for covariates.
• The citation benefit was highest for papers published in 2004-2005 (approx. 30%).
• Third-party data reuse accumulated for at least six years, while authors typically reused their own data within two years.

Conclusions:

• A robust citation benefit exists for open data, though smaller than previously estimated.
• Third-party data reuse has a direct, long-lasting impact on citation rates.
• A substantial fraction of archived gene expression microarray datasets are reused, with increasing intensity since 2003.