GCBI-alyzer
A Gender Citation Balance Index tool
Using tools from data science, one can estimate the likelihood that a person with any given first name self-identifies as 'woman' or 'man'. For example, a person named Jacqueline is 98% likely to self identify as 'W' and a person named Ileri is 67% likely to self-identify as 'M.' With this approach it has been estimated that authorship of papers published in broad-scope neuroscience journals is 55.3% M-first-author/M-last-author, 10.2% M/W, 26.2% W/M, and 8.3% W/W (Dworkin et al, 2020). For the Journal of Cognitive Neuroscience, the current break down (as of late 2021) is 40.7% M/M, 11.5% M/W, 32% W/M, and 15.9% W/W (Postle & Fulvio, 2021), which can be taken as a proxy for the gender breakdown of active research teams in our field. These latter values provide the base rate with which one can compute Gender Citation Balance Indices (GCBI) for the reference section of any paper that may be published in JoCN. The GCBI is a value that can range from -1 to >0, with values of 0 indicating that the proportion of X/X papers in one's reference section perfectly matches the base rate of JoCN authorship (see "About this tool" section below for more information). JoCN is encouraging authors to use this "GCBI-alyzer" to calculate the GCBIs for the reference section of their manuscript, and to include this along with other metadata (e.g., acknowledgment, grant funding).
Instructions:
Please copy and paste your reference list with Crossref DOIs in the box below, and click the "Submit" button. Copying and pasting from a pdf document is not recommended, as this is known to cause problems with the text supplied to the parser that prevents some references from being categorized. Disabling AdBlock and other similar browser extensions may be necessary if you are experiencing failures, as these are known to interfere with the queries. Please note that it may take several minutes to obtain a result, especially for large reference lists. If you experience any technical difficulties, please see the "Understanding Failures" section toward the bottom of this page.
Crossref DOIs have the format: https://doi.org/10.1101/2020.08.19.257402. If you need to acquire the Crossref DOIs for your references, please retrieve them from crossref.org here. Note that some references may not have a Crossref DOI associated with it. In such a case, that reference simply will not be categorized. You may copy and paste your manuscript's reference list with DOIs or the crossref query output into the box below.
*update 7/24/24: The tool had been experiencing some inconsistencies and in some cases had been unable to complete its queries. We have made some adjustments 'under the hood', and it seems to be stable at this time. Please note that very long reference lists make take a little longer to complete than previously. If you encounter further difficulties, please contact Jacqueline Fulvio at jcogneuro@gmail.com for assistance.
Please input your reference list.
-------------------------------------------
About this tool
This tool is intended to provide authors planning to submit manuscripts to the Journal of Cognitive Neuroscience (JoCN) Gender Citation Balance indices (GCBIs) which quantify the reference gender category breakdown of their reference list relative to the authorship gender category breakdown of the journal.
The tool obtains the metadata associated with each DOI from the crossref.org API. The first names of the first and last author of each reference are extracted from the metadata and queried for gender using the genderize.io API. Genders are assigned when the probability of the name belonging to a man or woman is >= 0.70 and authorship gender categories (i.e., 'MM', 'WM', 'MW', 'WW') are assigned according to the gender assignments of the first and last authors. Finally, the GCBIs are computed for each authorship gender category as the observed proportion of the categorized DOIs from the submitted reference list minus the expected proportion for that category based on JoCN authorship divided by the expected proportion for that category. Because the GCBIs depend upon different expected proportions for each of the four categories, the maximum positive (i.e., over-citation) value that could be obtained varies.
Interpreting the GCBIs
The GCBIs are reported separately for each authorship gender category to 3 decimals places. Negative GCBIs indicate that the corresponding authorship gender category was cited less often than expected given JoCN authorship, whereas positive GBCIs indicate that the corresponding category was cited more often than expected. The greater the magnitude, the greater the over-/under-citation.
The tool will be updated at regular intervals to report GCBIs with respect to JoCN authorship within the last decade or so. The first version of the tool was launched in October 2020, with the expected proportions based on the JoCN authorship during the January 2009 - July 2020 time frame of 40.8% M/M, 10.8% M/W, 33.5% W/M, and 14.9% W/W (Fulvio et al, 2020). The current version updated in November 2021 includes the JoCN authorship during the January 2009 - December 2021 time frame. We emphasize that the GCBIs are specific with respect to JoCN authorship and therefore may not reflect gender balance in citation rates for other fields.
Understanding failures
Occasionally, reference gender categorization may fail, which will be reflected in the value returned in the "Number of DOIs that could not be categorized" output fields above. If you find that a large proportion of the DOIs submitted could not be categorized due to problems with the DOI, we suggest that you first check that your DOIs are properly listed. You can verify that your DOIs are correct using the CrossRef Simple Text Query linked above.
Beyond formatting, categorization failures may arise for other reasons such as:
(1) The metadata for the reference did not contain the full first name for one of both of the first and last authors. This tends to be more likely for older publications, and may also occur for certain journals.
(2) The gender query returned a probability <0.70 such that a gender could not be assigned based on our criterion.
The tool is known to work with the Chrome, Firefox, and Safari web browsers. At this time, we are making additional tweaks so other issues may be encountered. If you are experiencing technical difficulties, please send your reference list input and information about your web browser to Jacqueline Fulvio (jacqueline.fulvio@wisc.edu) who can provide assistance.
We welcome your feedback!
We thank all of those who have assisted in troubleshooting by awaiting outputs from the tool! This tool is still in development, and we continue to ask for your input, especially to assist our testing and troubleshooting!
For more information about the background research, our article accompanying this work can be found at the following link: Fulvio, J.M., Akinnola, I., Postle, B.R. (2021). Gender (im)balance in citation practices in cognitive neuroscience. Journal of Cognitive Neuroscience. https://doi.org/10.1101/2020.08.19.257402.
Our one-year follow-up post-launch of the tool reporting the impact of its usage as part of the JoCN review process can be found at the following link: https://old.postlab.psych.wisc.edu/files/9316/3563/1172/JOCN-2021-0273_Proof_hi.pdf
For direct correspondence, please contact us via email:
Dr. Brad Postle, Editor-in-Chief, Journal of Cognitive Neuroscience, postle@wisc.edu
Dr. Jacqueline Fulvio, Research Scientist, Postle laboratory, jacqueline.fulvio@wisc.edu
page last updated: July 24, 2024
Thanks for stopping by!