Chung-Hong Chan

As computational methods become increasingly central to social science research, there is a growing demand for research software infrastructures that are sustainable, grounded in principles of open science, and embedded in the community. In response, we are introducing the Methods Hub as an open-source, community-driven collection of computational tools designed to support social science research in solving complex data-related tasks. The Methods Hub integrates three complementary components: First, quality-curated content, including computational methods and accompanying tutorials, which are specifically presented from a social science research perspective. Second, the methods and tutorials are hosted on a web portal enabling researchers to discover, apply, learn, and publish computational methods. Third, interactive, browser-based execution backends allow users to directly run, test, and adapt the code of methods and tutorials. These components are guided by a set of core principles: a long-term infrastructure perspective, reproducibility requirements, and a focus on community. The Methods Hub builds a long-term infrastructure that extends beyond the scope of single projects. Further, reproducibility functions as a quality assurance measure while advancing the practices of open science. Finally, the Methods Hub is built on public, community-based submissions by treating research software as citable academic output and providing visibility, accessibility, and formal recognition. We are also broadening the community by fostering learning and ease of access. Looking ahead, the Methods Hub is exploring the potential of AI as both a research tool and to further strengthen the infrastructure, while maintaining transparency, reproducibility, and quality standards for computational social science research.

The utility of topic modeling for social science research can be shown by the number of overview articles on the method written by communication or political science scholars: for example, Bernhard...

The use of automated coding procedures to scale up content analysis has risen over the last years. Using a mixed-method approach, we examine researchers’ justifications to scale up content analysis and assess the methodological adjustments—or lack thereof—when employing supervised machine learning for 38 large-scale content analyses. Almost all of the included studies displayed deficiencies in study design, primarily related to the uncritical use of frequentist statistics on datasets containing the entire statistical population, or employing supervised machine learning without methodological adjustments to account for misclassifications. Our findings question the need for large datasets and automated coding in the first place.

In recent decades, disruptive media events, such as major terrorist attacks, have gained increasing relevance in news coverage around the world. Despite the growing importance of such globally broadcast media events, little research to date has examined cross-national variation in event coverage or the predictors of this variation. This study examines news coverage about the 9/11 terror attacks in the United States across 51 countries to analyze whether the topical focus and emotional tone of news coverage about the attacks varied according to a country’s proximity to the United States and the dominant role perceptions of its journalistic culture. We show that these macro-level predictors are associated in varying degrees with the country-level topical focus and emotional tone of reporting over the 30 days following this salient event. Moreover, our analysis also suggests that temporal developments may have uniformly structured much of this worldwide coverage.

Replication crises have shaken the scientific landscape during the last decade. As potential solutions, open science practices were heavily discussed and have been implemented with varying success in different disciplines. We argue that computational-x disciplines such as computational social science, are also susceptible for the symptoms of the crises, but in terms of reproducibility. We expand the binary definition of reproducibility into a tier system which allows increasing levels of reproducibility based on external verfiability to counteract the practice of open-washing. We provide solutions for barriers in Computational Social Science that hinder researchers from obtaining the highest level of reproducibility, including the use of alternate data sources and considering reproducibility proactively.