Post-2020 decline in CSS within sociology and psychology: noise or trend

Determine whether the apparent post-2020 decline in the proportion of papers classified as Computational Social Science within sociology and psychology represents random noise attributable to data limitations in the Microsoft Academic Graph or a sustained downward trend in those fields.

Background

The study applies an ensemble text classifier to 11 million Microsoft Academic Graph records (1990–2021) to identify Computational Social Science (CSS) papers across sociology, psychology, political science, and economics.

In analyzing temporal trends, the authors note a recent decline in CSS papers within political science and a weak decline after 2020 in sociology and psychology. They caution that data limitations—such as shifting field boundaries and coverage issues around 2020–2021—complicate interpretation, prompting an explicit uncertainty about whether the observed post-2020 declines in sociology and psychology are noise or a real trend.

References

Due to data limitations it remains unclear whether the weak decline in CSS observed in sociology and psychology after 2020 is merely noise or indicative of a continuing trend.

— From Division to Unity: A Large-Scale Study on the Emergence of Computational Social Science, 1990-2021  (2412.08087 - Bao et al., 2024) in Section 3 (Results), Subsection "The reactions of social sciences to CSS"