Guidelines and warnings

This package attempts to infer gender (or more precisely, sex assigned at birth) based on first names using historical data, typically data that was gathered by the state. This method has many limitations, and before you use this package be sure to take into account the following guidelines.

1) Your analysis and the way you report it should take into account the limitations of this method, which include its reliance of data created by the state and its inability to see beyond the state-imposed gender binary. At a minimum, be sure to read our article explaining the limitations of this method, as well as the review article that is critical of this sort of methodology, both cited below.

2) Do not use this package to study individuals: it is at most useful for studying populations in the aggregate.

3) Resort to this method only when the alternative is not a more nuanced and justifiable approach to studying gender, but where the alternative is not studying gender at all. For instance, for many historical sources this approach might be the only way to get a sense of the sex ratios in a population. But ask whether you really need to use this method, whether you are using it responsibly, or whether you could use a better approach instead.

Blevins, Cameron, and Lincoln A. Mullen, “Jane, John … Leslie? A Historical Method for Algorithmic Gender Prediction,” Digital Humanities Quarterly 9, no. 3 (2015). http://www.digitalhumanities.org/dhq/vol/9/3/000223/000223.html

Mihaljević, Helena, Marco Tullney, Lucía Santamaría, and Christian Steinfeldt. “Reflections on Gender Analyses of Bibliographic Corpora.” Frontiers in Big Data 2 (August 28, 2019): 29. https://doi.org/10.3389/fdata.2019.00029.

Description

Data sets, historical or otherwise, often contain a list of first names but seldom identify those names by gender. Most techniques for finding gender programmatically rely on lists of male and female names. However, the gender associated with names can vary over time. Any data set that covers the normal span of a human life will require a historical method to find gender from names. This R package uses historical datasets from the U.S. Social Security Administration, the U.S. Census Bureau (via IPUMS USA), and the North Atlantic Population Project to provide predictions of gender for first names for particular countries and time periods.

Installation

You can install this package from CRAN:

The first time you use the package you will be prompted to install the accompanying genderdata package. Alternatively, you can install this package for yourself from the rOpenSci package repository:

If you prefer, you can install the development versions of both packages from the rOpenSci package repository:

Using the package

The gender() function takes a character vector of names and a year or range of years and uses various datasets to predict the gender of names. Here we predict the gender of the names Madison and Hillary in 1930 and again in the 2000s using Social Security data.

See the package vignette for a fuller introduction and suggestions on how to use the gender() function efficiently with large datasets.

vignette(topic = "predicting-gender", package = "gender")

To read the documentation for the datasets, install the genderdata package then examine the included datasets.

library(genderdata)
data(package = "genderdata")