We map the topic structure of psychology utilizing a sample of over 500,000 abstracts of research articles and conference proceedings spanning two decades (1995–2015). To do so, we apply structural topic models to examine three research questions: (i) What are the discipline’s most prevalent research topics? (ii) How did the scientific discourse in psychology change over the last decades, especially since the advent of neurosciences? (iii) And was this change carried by high impact (HI) or less prestigious journals? Our results reveal that topics related to natural sciences are trending, while their ’counterparts’ leaning to humanities are declining in popularity. Those trends are even more pronounced in the leading outlets of the field. Furthermore, our findings indicate a continued interest in methodological topics accompanied by the ascent of neurosciences and related methods and technologies (e.g. fMRI’s). At the same time, other established approaches (e.g. psychoanalysis) become less popular and indicate a relative decline of topics related to the social sciences and the humanities.

Media discourse is often seen as an important condition of people’s attitudes and perceptions. Despite a rich literature, however, it is not well understood how media exposure influences attitudes towards immigrants. In contrast to previous studies, we argue that people rely on ‘availability heuristics’ shaped by mass media. From that point of view, it is the specific content of media discourse on immigration that affects people’s concerns. We use structural topic models to classify media content of more than 24,000 articles of leading German newspapers from 2001 to 2016. Utilizing linear fixed-effects models allows us to relate a person’s concern towards immigration, as reported in the German Socioeconomic Panel, to prevalent topics discussed in print media while controlling for several confounding factors (e.g., party preferences, interest in politics, etc.). We find a robust relationship between topic salience and attitudes towards integration. Our results reveal that specific topics with negative contents (e.g., domestic violence) increase concerns, while others (e.g., scientific studies, soccer) decrease concerns substantially, underlining the importance of available information provided by media. In addition, people with higher education are generally less affected by media salience of topics.

Is the pursuit of interdisciplinary or innovative research beneficial or detrimental for the impact of early career researchers? We focus on young scholars as they represent an understudied population who have yet to secure a place within academia. Which effects promise higher scientific recognition (i.e., citations) is therefore crucial for the high-stakes decisions young researchers face. To capture these effects, we introduce measurements for interdisciplinarity and novelty that can be applied to a researcher’s career. In contrast to previous studies investigating research impact on the paper level, hence, our paper focuses on a career perspective (i.e., the level of authors). To consider different disciplinary cultures, we utilize a comprehensive dataset on U.S. physicists (n = 4003) and psychologists (n = 4097), who graduated between 2008 and 2012, and traced their publication records. Our results indicate that conducting interdisciplinary research as an early career researcher in physics is beneficial, while it is negatively associated with research impact in psychology. In both fields, physics and psychology, early career researchers focusing on novel combinations of existing knowledge are associated with higher future impact. Taking some risks by deviating to a certain degree from mainstream paradigms seems therefore like a rewarding strategy for young scholars.

Media discourse is often seen as an important condition of people's attitudes and perceptions. Despite a rich literature, however, it is not well understood how media exposure influences attitudes towards immigrants. In contrast to previous studies, we argue that people rely on 'availability heuristics' shaped by mass media. From that point of view, it is the specific content of media discourse on immigration that affects people's concerns. We use 'Structural Topic Models' to classify media content of more than 24.000 articles of leading German newspapers from 2001 to 2016. Utilizing 'linear fixed effect models' allows us to relate a person's concern towards immigration as reported in the German Socioeconomic Panel to prevalent topics discussed in print media while controlling for several confounding factors (e.g., party preferences, interest in politics, etc.). We find a robust relationship between topic salience and attitudes towards integration. Our results also reveal that specific topics with negative contents (e.g., domestic violence) to increase concerns, while others (e.g., scientific studies, soccer) decrease concerns substantially, underlining the importance of available information provided by media. In addition, people with higher education are generally less affected by media salience of topics.

The rise of populism concerns many political scientists and practitioners, yet the detection of its underlying language remains fragmentary. This paper aims to provide a reliable, valid, and scalable approach to measure populist rhetoric. For that purpose, we created an annotated dataset based on parliamentary speeches of the German Bundestag (2013–2021). Following the ideational definition of populism, we label moralizing references to “the virtuous people” or “the corrupt elite” as core dimensions of populist language. To identify, in addition, how the thin ideology of populism is “thickened,” we annotate how populist statements are attached to left-wing or right-wing host ideologies. We then train a transformer-based model (PopBERT) as a multilabel classifier to detect and quantify each dimension. A battery of validation checks reveals that the model has a strong predictive accuracy, provides high qualitative face validity, matches party rankings of expert surveys, and detects out-of-sample text snippets correctly. PopBERT enables dynamic analyses of how German-speaking politicians and parties use populist language as a strategic device. Furthermore, the annotator-level data may also be applied in cross-domain applications or to develop related classifiers.

Machine learning (ML) techniques have become one of the most successful scientific tools and changed the everyday life of people around the globe (e.g., search engines). A vast amount of digital data sources on human behaviour has emerged due to the rise of the internet and opened the door for computer scientists to apply ML on social phenomena. In the social sciences, however, the adoption of ML has been less enthusiastic. To investigate the relation of traditional statistics and ML, this paper shows how ML might be used as regression analysis. For that purpose, we illustrate what a typical social science approach might look like and how using ML techniques could contribute additional insights when it comes to estimators (non-linearity) or the assessment of model fit (predictive power). In particular, we reveal how epistemological differences shape the potential usage of ML in the social sciences and discuss the methodological trade-off of applying ML compared to traditional statistics.

The rise of populism concerns many political scientists and practitioners, yet the detection of its underlying language remains fragmentary. This paper aims to provide a reliable, valid, and scalable approach to measure populist stances. For that purpose, we created an annotated dataset based on parliamentary speeches of the German Bundestag (2013 to 2021). Following the ideational definition of populism, we label moralizing references to the virtuous people or the corrupt elite as core dimensions of populist language. To identify, in addition, how the thin ideology of populism is thickened, we annotate how populist statements are attached to left-wing or right-wing host ideologies. We then train a transformer-based model (PopBERT) as a multilabel classifier to detect and quantify each dimension. A battery of validation checks reveals that the model has a strong predictive accuracy, provides high qualitative face validity, matches party rankings of expert surveys, and detects out-of-sample text snippets correctly. PopBERT enables dynamic analyses of how German-speaking politicians and parties use populist language as a strategic device. Furthermore, the annotator-level data may also be applied in cross-domain applications or to develop related classifiers.

How does the arrival at a new work environment affect the performance of creative workers? Even though it is a shared experience in many careers, we know surprisingly little about the impact of transferring to an unfamiliar workplace. To test the direct and indirect effects of how creative workers fit new colleagues at scale, we need to have (i) information on career trajectories and (ii) assess the cultural fit (CF) to a work environment computationally. We address these challenges by collecting a unique dataset on early career researchers (ECR) and their transition to another university after graduating in physics and psychology between 2006 and 2015 in the US. Then, we measure the CF by utilizing unsupervised machine learning and comparing each ECR’s research to the work of their new colleagues. In a close approximation of both fields (> 2.5 million documents), our results not only show a strong impact of CF on creative performance but also that fitting-in moderates the trade-off between specialization and generalization strategies. Moreover, we reveal how high-status organizations impose social pressures that further augment the effects of CF.

Agent-based modeling (ABM) is a scientific method used in fields such as social science, biology, and ecology to simulate interactions of autonomous agents and study the resulting emergent phenomena. The relationships between agents, which structure the simulated interactions, are often represented by network graphs. Since empirical data on networks is rare, many agent-based models rely on artificially generated networks (Amblard et al., 2015). Consequently, generating a valid network structure at the beginning of a simulation is critical. Additionally, accessing and modifying the network during the simulation are steps that must be managed in almost any agent-based model. Pop2net is a Python package that combines many steps related to network generation and management for ABM using a bipartite approach. Bipartite networks consist of two distinct types of entities where edges are only formed between entities of different types. In Pop2net, relationships are represented as bipartite networks connecting actors and locations. When two actors are linked to the same location, they are considered connected through that shared location. In this way, locations serve as a contact layer between actors, representing places where interactions occur or contexts that facilitate actor connections. The aim of Pop2net’s bipartite approach to relations is to simplify the generation and management of realistic network structures in ABM.

The impact of YouTube’s recommendation algorithm has been subject to much debate. Case studies and anecdotal evidence have suggested that algorithmic recommendations may lure the platform’s users into informational ‘rabbit holes’ where they are exposed to biased content and misinformation. However, empirical research on the phenomenon has been challenging to conduct because the effects of the algorithm are inherently confounded by user and creator behavior. In this chapter, we study the net effect of algorithmic recommendations through a comparative network approach. That is, we collect the recommendation links between videos for various political and non-political issues. The resulting networks are then compared on various structural network characteristics. If the algorithm creates rabbit holes, one would expect the recommendation networks of videos on conspiratorial content to be more modular, less centralized and intra-topically dense. We analyzed 15,455 videos and 154,311 recommendation links on 40 different topics, each belonging to one of four main categories: news, science, conspiracy, and non-controversial. On the macro level, we find that conspiratorial video networks are small, sparse and centralized, with their density partly explained by lower view counts and network size. The sentiment of their videos is negative on average–like news networks, but unlike science and non-controversial networks. However, sentiment does not but view count does predict how often videos are recommended.

Academic employment is marked by high mobility and significant performance pressures, particularly during the early stages of professional advancement. As early-career researchers (ECRs) change institutions frequently, they need to navigate the culture of the hiring department and may benefit from fitting in or suffer if they do not. Our study examines how this initial cultural fit shapes job performance immediately following organizational entry. Building on a theoretical perspective that conceptualizes culture as a resource, we argue that ECRs’ ability to publish outstanding work is contingent on their initial fit. We test this argument among ECRs in physics and psychology in the United States, a context defined by pronounced cultural variation across departments. Using a novel natural language processing approach, we operationalize cultural fit as the semantic similarity between a researcher’s prior work and the collective output of their new department, analyzing over 2.5 million scientific documents. Our analysis reveals that higher initial cultural fit predicts superior publication performance and that this relationship is moderated by the specialization of academic fields. This study advances our understanding of academic careers and organizational culture theory by highlighting the importance of initial cultural alignment in shaping publication outcomes and introduces a scalable and fine-grained method to measure cultural fit within organizations.