Academia Mentions of Prejudice Terms in Papers, 1900 to 2024

Key points from the chart

The chart tracks how often six pairs of prejudice related terms appear in academic paper abstracts each year, from 1900 to August 2024.
Most of the terms were rare or absent before the late 20th century and grew sharply afterwards.
Racism, sexism and transphobia terms reach their highest points between about 2020 and 2023, and all six series are below their peaks in the 2024 data.
Antisemitism is the exception, with its highest values in the early 1940s.
Each panel has its own scale, so line heights cannot be compared between panels. Terms for racism are the most frequent overall; terms for transphobia are the least frequent.

The six panels in text

Title: “Mentions of Prejudice in Academic Papers (1900 to August 2024)”. Six line charts in a grid of three rows and two columns. Each shows the relative frequency of a pair of terms in academic abstracts by year (the share of all words), with the x axis running from 1900 to 2024. The peaks described below are read from the chart by eye; it prints no exact values beyond the axis ticks.

Panel (terms counted together)	Line colour	Y axis range	Pattern on the chart
racism + racist	teal	0 to 1.7 x 10⁻⁵	Occasional small spikes before the mid 1960s, a first local peak around 1970, a gradual rise to about 2000, then a sharp climb to the highest point around 2020 to 2021. Lower again by 2024.
sexism + sexist	red	0 to 2.9 x 10⁻⁶	Close to zero until about 1970, then a steep rise to a high around 1980. A slow decline to a low in the late 2000s, followed by a second rise to a high in the early 2020s. Lower in 2024.
homophobia + homophobic	gold	0 to 1.5 x 10⁻⁶	Scattered isolated spikes before the 1970s, growth from the mid 1980s to a peak in the late 1990s, then a high but uneven level through the 2010s. Lower by 2024.
transphobia + transphobic	magenta	0 to 4.1 x 10⁻⁷	Effectively zero until the late 1990s, then steep growth from about 2010 to a peak in the early 2020s. Slightly lower in 2024.
islamophobia + islamophobic	purple	0 to 1.1 x 10⁻⁶	One isolated spike in the late 1940s, otherwise near zero until about 2000. A steep rise through the 2010s to a high in the late 2010s. Lower by 2024.
antisemitism + antisemitic	green	0 to 5.8 x 10⁻⁶	The only series with substantial early use. Its highest spike is in the early 1940s, with other spikes in that decade. Since about 1950 it has stayed at a lower, fairly steady level.

Footer notes on the chart: “Chart: @DavidRozado”, “Source: SSORC Abstracts (2024-08-06)”, “No rolling mean smoothing function applied” and “At least some of the early isolated spikes are due to what appears as occasional corrupted date stamps in the source data”.

What the creator says about the trend

David Rozado, who made the chart, has tracked prejudice related terms across news media, academic papers and social media. In a September 2024 post updating his analysis of academic abstracts, he writes that “the most recent data indicate a slight decline in mentions of prejudice in scholarly papers over the past few years, following a peak around 2021” [1]. In the same post he reports that several terms related to diversity, equity and inclusion (DEI) continued to grow in academic papers [1].

He offers one possible explanation, that attention may be shifting from terms such as “racist” and “sexist” towards more positively framed terms such as “fairness” and “inclusion”, but adds that it is difficult to confirm this given the natural ebb and flow of term popularity and the risk of cherry picking examples [1]. His earlier post on academic papers, published on 10 August 2022, is titled “The Academic Literature and its Increasing Emphasis on Prejudice and Social Justice” [2].

Notes on this infographic

Different scales. The y axes run from 0 to 1.7 x 10⁻⁵ for racism, but only 0 to 4.1 x 10⁻⁷ for transphobia. Similar looking peaks therefore stand for very different frequencies.
Partial 2024. The data end on 6 August 2024, so the 2024 values cover only part of the year.
Early spikes. As the chart’s own note says, some isolated early spikes may be caused by incorrect dates in the source data, so single early peaks should be treated with caution.
Source name. “SSORC” is printed on the chart. It most likely refers to S2ORC, the Semantic Scholar Open Research Corpus, but the chart does not spell this out.
Versions. Rozado’s Substack post shows a similar six panel chart that groups more terms in some panels (for example racism, racist, racists, xenophobia and white supremacy) [1]. This page describes the version shown above, which counts two terms per panel.

Questions about prejudice terms in academic papers

Are mentions of racism in academic papers declining?

Mentions of racism and racist in academic abstracts peaked around 2020 to 2021 and were lower again by 2024 on David Rozado's chart, which uses data up to 6 August 2024. Rozado wrote in September 2024 that "the most recent data indicate a slight decline in mentions of prejudice in scholarly papers over the past few years, following a peak around 2021" [1]. He also reports that several DEI related terms continued to grow [1].

When was antisemitism mentioned most in academic papers?

Antisemitism and antisemitic reached their highest frequency in academic abstracts in the early 1940s, the only one of the six prejudice term pairs on the chart with substantial early use. Since about 1950 the series has stayed at a lower, fairly steady level. The chart's own note warns that some isolated early spikes may come from corrupted date stamps in the source data, so single early peaks should be treated with caution.

Which prejudice term is most common in academic papers?

Terms for racism are the most frequent of the six pairs on the chart, with a y axis running up to 1.7 x 10⁻⁵ of all words in abstracts. Terms for transphobia are the least frequent, with an axis running only to 4.1 x 10⁻⁷. Because each panel has its own scale, similar looking peaks stand for very different frequencies.

Where does the data on prejudice terms in academic papers come from?

The chart's data come from "SSORC Abstracts (2024-08-06)", according to its footer. SSORC most likely refers to S2ORC, the Semantic Scholar Open Research Corpus, but the chart does not spell this out. The chart was made by David Rozado, and no rolling mean smoothing is applied to the lines.