Large language models (LLMs) are increasingly deployed in politically sensitive contexts, raising concerns about their susceptibility to ideological biases. In this work, we examine how synthetic persona conditioning shapes ideological expression across seven open-weight instruction-tuned models (7B–72B parameters) using the Political Compass Test (62 statements) as a standardized behavioral probe. Across three studies involving 200,000 synthetic personas and more than 260 million model responses, we analyze implicit and explicit malleability, as well as theme-associated variation. We find that: (i) larger models exhibit broader implicit ideological coverage, increasing from 14%–35% for 7–8B models to up to 49% for 70B+ models; (ii) explicit ideological priming induces large and statistically significant shifts, with right-authoritarian cues moving all models in the intended direction and generally producing larger shifts than left-libertarian cues across model-axis comparisons; (iii) left-libertarian priming produces more heterogeneous responses, with counter-directional economic shifts in three of four 7–8B models and intended-direction shifts across all 70B+ models; and (iv) theme-associated semantic content in persona descriptions is linked to systematic and interpretable directional shifts in ideological space. While our results identify an upstream mechanism through which persona conditioning can alter model responses under a standardized ideological probe, we do not test whether such shifts affect users’ beliefs, decisions, or political behavior. Our findings are best understood as evidence of ideological malleability at the generation layer, highlighting the need to account for interactional factors when evaluating political neutrality, fairness, and safety in English-prompted, persona-conditioned language models.
As governments intensify efforts to decarbonize their energy systems in response to climate change, the Russian invasion of Ukraine has (re-)introduced energy security as a salient and potentially competing policy priority. This study investigates whether the renewed emphasis on energy security expands the coalition supporting the energy transition or, conversely, creates a new axis of conflict that complicates public support. We argue that the energy crisis may simultaneously erode and reinforce public backing for climate mitigation policies, contingent on whether citizens view domestic renewable energy and energy security as complementary or conflicting goals. Using original survey data from Germany and Switzerland, we find that public opinion is divided on the perceived trade-off between energy security and the energy transition. Despite country-specific differences, individuals prioritizing energy security are consistently more likely to perceive a conflict between the two objectives. The findings highlight the need for policy strategies that frame climate and security goals as mutually reinforcing to sustain public support.
In this paper, we examined reasons for ethnic differences in parental support at the transition to apprenticeship training in Germany and their consequences for the likelihood of finding a training position and the likelihood of premature training termination. We extended previous research by focusing on three different types of parental support: help with writing applications, information about open training positions, and efforts towards getting their children a position. Using longitudinal data from Starting Cohort 4 (SC4) of the German National Educational Panel Study (NEPS), we show that part of ethnic differences in these types of support can be explained by measures of ethnicity-related background characteristics. Moreover, the relations of receiving-country-specific resources to parental support propensities seem to be domain specific. Furthermore, we find that only minority trainees benefit from writing support in terms of finding training and training success. In contrast, if parents made efforts to get their children a position, both minority and majority school-leavers are less likely to start training and, in the case of majority trainees, to finish training successfully. These are important findings as they give new insights into the reasons for ethnic disadvantages in parental support during the transition to apprenticeships and into the relation of different types of parental support to the training search success and premature termination risks.
This paper examines child online protection (COP) initiatives in Indonesia, focusing on normative implications and methodological challenges for computational social science. It analyzes how international norms influence Indonesia’s legal framework and policy development, particularly in the context of China’s growing digital influence through platforms like TikTok. The study reveals significant gaps in awareness and implementation of COP measures, exacerbated by the popularity of Chinese-owned apps among Indonesian youth. It highlights the need for enhanced digital literacy, targeted regulations, and interdisciplinary research approaches. The paper offers insights into innovative data collection methods and ethical considerations for studying children’s online behaviors. It concludes with recommendations for strengthening Indonesia’s COP framework, emphasizing the importance of balancing technological advancements with child safety in the digital age, while addressing the unique challenges posed by Chinese technological influence.
International crises and Euroscepticism have made European issues prominent in citizens' lives. This article studies the role of three key European issues - migration, the environment, and EU integration - for political parties and citizens. The analysis centres on nine EU member states, combining party manifestos from the 2024 European Parliament elections with survey data. It finds a gap between the concerns of citizens and the political parties, an important consideration for election campaigns in general. Moreover, the analysis suggests that the salience in party manifestos has a modest direct influence and a stronger indirect impact on their appeal to citizens in most countries studied. Specifically, citizens concerned with migration and the environment evaluate parties based on the prominence they give to these issues during the campaign. The findings offer important avenues for further research on party issue emphasis and the measurement of issue salience via large language models (LLMs).
