arXiv subjects

Anna-Carolina Haensch

Publications and source records attributed to Anna-Carolina Haensch.

At least 19 recordsLinked to original sources

arXiv2026

ChinaHeritaQA: A Culturally-Grounded Visual Question Answering Dataset for World Heritage Sites in China

We introduce ChinaHeritaQA, a multimodal benchmark dataset for evaluating the cultural reasoning abilities of vision-language models (VLMs) on UNESCO World Heritage sites in China. The dataset comprises 2,279 in-the-wild images paired with 14,133 bilingual (Chinese/English) multiple-choice QA pairs spanning seven cognitive dimensions, from basic identity recognition to historical periodization and architectural analysis. Guided by a UNESCO-aligned heritage ontology and verified through rigorous human annotation, the dataset ensures linguistic quality and factual consistency. Evaluations of state-of-the-art VLMs reveal that while top models exceed human performance on average, substantial task-level variation emerges: models excel at visual recognition but struggle with culturally grounded reasoning. Performance also varies by dynasty and region. ChinaHeritaQA reveals that strong visual retrieval does not extend to cultural and historical understanding. We release the dataset to support future research on culturally aware multimodal learning.

cs.CV↗

arXiv2026

In-Context Learning for the Imputation of Public Opinion Data with Large Language Models

Large language models have been widely evaluated as simulators of individual survey responses. In practice, however, fully unobserved responses are rare; the dominant problem is partial non-response. Imputation aims to restore the overall structure of a survey dataset by filling in these missing values. It has its own well-defined evaluation criteria and differs fundamentally from prediction. We propose to impute missing survey data through in-context learning (ICL). We systematically evaluate ICL design choices across different missingness mechanisms (MCAR, MAR, MNAR) on 150 opinion variables spanning 15 waves of the American Trends Panel. Compared to well-established statistical methods for data imputation like MICE PMM, our ICL approach consistently reduces absolute error across all missingness mechanisms, with the largest gains under non-random missingness (MNAR). Notably, the best-performing specification (gpt-oss-120b with 100 in-context examples) achieves near-nominal aggregate coverage (approaching the 95% level) with confidence intervals two to five times narrower than MICE PMM. We publish a Python package with an sklearn-like API to enable easy deployment of our method using local and proprietary LLMs.

cs.CL↗

arXiv2026

Survey Response Generation: Generating Closed-Ended Survey Responses In-Silico with Large Language Models

Many in-silico simulations of human survey responses with large language models (LLMs) focus on generating closed-ended survey responses, whereas LLMs are typically trained to generate open-ended text instead. Previous research has used a diverse range of methods for generating closed-ended survey responses with LLMs, and a standard practice remains to be identified. In this paper, we systematically investigate the impact that various Survey Response Generation Methods have on predicted survey responses. We present the results of 32 mio. simulated survey responses across 8 Survey Response Generation Methods, 4 political attitude surveys, and 10 open-weight language models. We find significant differences between the Survey Response Generation Methods in both individual-level and subpopulation-level alignment. Our results show that Restricted Generation Methods perform best overall, and that reasoning output does not consistently improve alignment. Our work underlines the significant impact that Survey Response Generation Methods have on simulated survey responses, and we develop practical recommendations on the application of Survey Response Generation Methods.

cs.CL↗

arXiv2026

Capabilities and Evaluation Biases of Large Language Models in Classical Chinese Poetry Generation: A Case Study on Tang Poetry

Large Language Models (LLMs) are increasingly applied to creative domains, yet their performance in classical Chinese poetry generation and evaluation remains poorly understood. We propose a three-step evaluation framework that combines computational metrics, LLM-as-a-judge assessment, and human expert validation. Using this framework, we evaluate six state-of-the-art LLMs across multiple dimensions of poetic quality, including themes, emotions, imagery, form, and style, in the context of Tang poetry generation. Our analysis reveals a critical "echo chamber" effect: LLMs systematically overrate machine-generated poems that mimic statistical patterns yet fail strict prosodic rules, diverging significantly from human expert judgments. These findings underscore the limitations of using LLMs as standalone evaluators for culturally complex tasks, highlighting the necessity of hybrid human-model validation frameworks.

cs.CL↗

arXiv2026

Too Open for Opinion? Embracing Open-Endedness in Large Language Models for Social Simulation

Large Language Models (LLMs) are increasingly used to simulate public opinion and other social phenomena. Most current studies constrain these simulations to multiple-choice or short-answer formats for ease of scoring and comparison, but such closed designs overlook the inherently generative nature of LLMs. In this position paper, we argue that open-endedness, using free-form text that captures topics, viewpoints, and reasoning processes "in" LLMs, is essential for realistic social simulation. Drawing on decades of survey-methodology research and recent advances in NLP, we argue why this open-endedness is valuable in LLM social simulations, showing how it can improve measurement and design, support exploration of unanticipated views, and reduce researcher-imposed directive bias. It also captures expressiveness and individuality, aids in pretesting, and ultimately enhances methodological utility. We call for novel practices and evaluation frameworks that leverage rather than constrain the open-ended generative diversity of LLMs, creating synergies between NLP and social science.

cs.CL↗

arXiv2025

Safer Prompts: Reducing Risks from Memorization in Visual Generative AI

Visual Generative AI models have demonstrated remarkable capability in generating high-quality images from user inputs like text prompts. However, because these models have billions of parameters, they risk memorizing certain parts of the training data and reproducing the memorized content. Memorization often raises concerns about safety of such models -- usually involving intellectual property (IP) infringement risk -- and deters their large scale adoption. In this paper, we evaluate the effectiveness of prompt engineering techniques in reducing memorization risk in image generation. Our findings demonstrate the effectiveness of prompt engineering in reducing the similarity between generated images and the training data of diffusion models, while maintaining relevance and aestheticity of the generated output.

cs.CV↗

arXiv2025

Beyond Correctness: Evaluating and Improving LLM Feedback in Statistical Education

Large language models (LLMs) have been proposed as scalable tools to address the gap between the importance of individualized written feedback and the practical challenges of providing it at scale. However, concerns persist regarding the accuracy, depth, and pedagogical value of their feedback responses. The present study investigates the extent to which LLMs can generate feedback that aligns with educational theory and compares techniques to improve their performance. Using mock in-class exam data from two consecutive years of an introductory statistics course at LMU Munich, we evaluated GPT-generated feedback against an established but expanded pedagogical framework. Four enhancement methods were compared in a highly standardized setting, making meaningful comparisons possible: Using a state-of-the-art model, zero-shot prompting, few-shot prompting, and supervised fine-tuning using Low-Rank Adaptation (LoRA). Results show that while all LLM setups reliably provided correctness judgments and explanations, their ability to deliver contextual feedback and suggestions on how students can monitor and regulate their own learning remained limited. Among the tested methods, zero-shot prompting achieved the strongest balance between quality and cost, while fine-tuning required substantially more resources without yielding clear advantages. For educators, this suggests that carefully designed prompts can substantially improve the usefulness of LLM feedback, making it a promising tool, particularly in large introductory courses where students would otherwise receive little or no written feedback.

stat.OT↗

arXiv2025

Can we trust LLMs as a tutor for our students? Evaluating the Quality of LLM-generated Feedback in Statistics Exams

One of the central challenges for instructors is offering meaningful individual feedback, especially in large courses. Faced with limited time and resources, educators are often forced to rely on generalized feedback, even when more personalized support would be pedagogically valuable. To overcome this limitation, one potential technical solution is to utilize large language models (LLMs). For an exploratory study using a new platform connected with LLMs, we conducted a LLM-corrected mock exam during the "Introduction to Statistics" lecture at the University of Munich (Germany). The online platform allows instructors to upload exercises along with the correct solutions. Students complete these exercises and receive overall feedback on their results, as well as individualized feedback generated by GPT-4 based on the correct answers provided by the lecturers. The resulting dataset comprised task-level information for all participating students, including individual responses and the corresponding LLM-generated feedback. Our systematic analysis revealed that approximately 7 \% of the 2,389 feedback instances contained errors, ranging from minor technical inaccuracies to conceptually misleading explanations. Further, using a combined feedback framework approach, we found that the feedback predominantly focused on explaining why an answer was correct or incorrect, with fewer instances providing deeper conceptual insights, learning strategies or self-regulatory advice. These findings highlight both the potential and the limitations of deploying LLMs as scalable feedback tools in higher education, emphasizing the need for careful quality monitoring and prompt design to maximize their pedagogical value.

stat.OT↗

arXiv2025

Systematic Evaluation of Uncertainty Estimation Methods in Large Language Models

Large language models (LLMs) produce outputs with varying levels of uncertainty, and, just as often, varying levels of correctness; making their practical reliability far from guaranteed. To quantify this uncertainty, we systematically evaluate four approaches for confidence estimation in LLM outputs: VCE, MSP, Sample Consistency, and CoCoA (Vashurin et al., 2025). For the evaluation of the approaches, we conduct experiments on four question-answering tasks using a state-of-the-art open-source LLM. Our results show that each uncertainty metric captures a different facet of model confidence and that the hybrid CoCoA approach yields the best reliability overall, improving both calibration and discrimination of correct answers. We discuss the trade-offs of each method and provide recommendations for selecting uncertainty measures in LLM applications.

cs.CL↗

arXiv2025

Mexico 2021: Psychological Intimate Partner Violence Against Women and the Role of Childhood Violence Exposure -- A Machine Learning Approach

In 2021, psychological violence was the most prevalent form of intimate partner violence (IPV) suffered by women in Mexico. The consequences of psychological IPV can include low self-esteem, depression, and even potential suicide. It is, therefore, crucial to identify the most relevant risk and protective factors of psychological IPV against women in Mexico. To this end, we adopt an ecological approach and analyze the role of a wide range of factors across four interrelated levels: Individual, relationship, community, and societal. We construct a multidimensional data set with 61,205 observations and 59 variables by integrating nationally representative data from the 2021 Mexican Survey on the Dynamics of Household Relationships with nine additional sources. For model estimation and factor selection, we combine model-based boosting with stability selection. Our findings reveal that women who were exposed to violence in childhood and whose partners were exposed to violence in childhood face a heightened risk of psychological IPV. These findings highlight the critical yet often overlooked role of childhood violence exposure for psychological IPV risk. Additionally, we confirm the role of three protective factors previously identified by Torres Munguía and Martínez-Zarzoso (2022) using 2016 data, now validated with 2021 data: women who had their first sex later in life and under consent, who have autonomy in decision-making regarding their professional life and use of economic resources, and who live in a household where housework is done only by male members face a lower risk of suffering psychological IPV.

stat.AP↗

arXiv2025

AIn't Nothing But a Survey? Using Large Language Models for Coding German Open-Ended Survey Responses on Survey Motivation

The recent development and wider accessibility of LLMs have spurred discussions about how they can be used in survey research, including classifying open-ended survey responses. Due to their linguistic capacities, it is possible that LLMs are an efficient alternative to time-consuming manual coding and the pre-training of supervised machine learning models. As most existing research on this topic has focused on English-language responses relating to non-complex topics or on single LLMs, it is unclear whether its findings generalize and how the quality of these classifications compares to established methods. In this study, we investigate to what extent different LLMs can be used to code open-ended survey responses in other contexts, using German data on reasons for survey participation as an example. We compare several state-of-the-art LLMs and several prompting approaches, and evaluate the LLMs' performance by using human expert codings. Overall performance differs greatly between LLMs, and only a fine-tuned LLM achieves satisfactory levels of predictive performance. Performance differences between prompting approaches are conditional on the LLM used. Finally, LLMs' unequal classification performance across different categories of reasons for survey participation results in different categorical distributions when not using fine-tuning. We discuss the implications of these findings, both for methodological research on coding open-ended responses and for their substantive analysis, and for practitioners processing or substantively analyzing such data. Finally, we highlight the many trade-offs researchers need to consider when choosing automated methods for open-ended response classification in the age of LLMs. In doing so, our study contributes to the growing body of research about the conditions under which LLMs can be efficiently, accurately, and reliably leveraged in survey research.

cs.CL↗

arXiv2025

Algorithmic Fidelity of Large Language Models in Generating Synthetic German Public Opinions: A Case Study

In recent research, large language models (LLMs) have been increasingly used to investigate public opinions. This study investigates the algorithmic fidelity of LLMs, i.e., the ability to replicate the socio-cultural context and nuanced opinions of human participants. Using open-ended survey data from the German Longitudinal Election Studies (GLES), we prompt different LLMs to generate synthetic public opinions reflective of German subpopulations by incorporating demographic features into the persona prompts. Our results show that Llama performs better than other LLMs at representing subpopulations, particularly when there is lower opinion diversity within those groups. Our findings further reveal that the LLM performs better for supporters of left-leaning parties like The Greens and The Left compared to other parties, and matches the least with the right-party AfD. Additionally, the inclusion or exclusion of specific variables in the prompts can significantly impact the models' predictions. These findings underscore the importance of aligning LLMs to more effectively model diverse public opinions while minimizing political biases and enhancing robustness in representativeness.

cs.CL↗

arXiv2025

United in Diversity? Contextual Biases in LLM-Based Predictions of the 2024 European Parliament Elections

"Synthetic samples" based on large language models (LLMs) have been argued to serve as efficient alternatives to surveys of humans, assuming that their training data includes information on human attitudes and behavior. However, LLM-synthetic samples might exhibit bias, for example due to training data and fine-tuning processes being unrepresentative of diverse contexts. Such biases risk reinforcing existing biases in research, policymaking, and society. Therefore, researchers need to investigate if and under which conditions LLM-generated synthetic samples can be used for public opinion prediction. In this study, we examine to what extent LLM-based predictions of individual public opinion exhibit context-dependent biases by predicting the results of the 2024 European Parliament elections. Prompting three LLMs with individual-level background information of 26,000 eligible European voters, we ask the LLMs to predict each person's voting behavior. By comparing them to the actual results, we show that LLM-based predictions of future voting behavior largely fail, their accuracy is unequally distributed across national and linguistic contexts, and they require detailed attitudinal information in the prompt. The findings emphasize the limited applicability of LLM-synthetic samples to public opinion prediction. In investigating their contextual biases, this study contributes to the understanding and mitigation of inequalities in the development of LLMs and their applications in computational social science.

cs.CY↗

arXiv2025

"It Listens Better Than My Therapist": Exploring Social Media Discourse on LLMs as Mental Health Tool

The emergence of generative AI chatbots such as ChatGPT has prompted growing public and academic interest in their role as informal mental health support tools. While early rule-based systems have been around for several years, large language models (LLMs) offer new capabilities in conversational fluency, empathy simulation, and availability. This study explores how users engage with LLMs as mental health tools by analyzing over 10,000 TikTok comments from videos referencing LLMs as mental health tools. Using a self-developed tiered coding schema and supervised classification models, we identify user experiences, attitudes, and recurring themes. Results show that nearly 20% of comments reflect personal use, with these users expressing overwhelmingly positive attitudes. Commonly cited benefits include accessibility, emotional support, and perceived therapeutic value. However, concerns around privacy, generic responses, and the lack of professional oversight remain prominent. It is important to note that the user feedback does not indicate which therapeutic framework, if any, the LLM-generated output aligns with. While the findings underscore the growing relevance of AI in everyday practices, they also highlight the urgent need for clinical and ethical scrutiny in the use of AI for mental health support.

cs.CL↗

arXiv2024

Cracking the Code: Evaluating Zero-Shot Prompting Methods for Providing Programming Feedback

Despite the growing use of large language models (LLMs) for providing feedback, limited research has explored how to achieve high-quality feedback. This case study introduces an evaluation framework to assess different zero-shot prompt engineering methods. We varied the prompts systematically and analyzed the provided feedback on programming errors in R. The results suggest that prompts suggesting a stepwise procedure increase the precision, while omitting explicit specifications about which provided data to analyze improves error identification.

cs.SE↗

arXiv2024

The Potential and Challenges of Evaluating Attitudes, Opinions, and Values in Large Language Models

Recent advances in Large Language Models (LLMs) have sparked wide interest in validating and comprehending the human-like cognitive-behavioral traits LLMs may capture and convey. These cognitive-behavioral traits include typically Attitudes, Opinions, Values (AOVs). However, measuring AOVs embedded within LLMs remains opaque, and different evaluation methods may yield different results. This has led to a lack of clarity on how different studies are related to each other and how they can be interpreted. This paper aims to bridge this gap by providing a comprehensive overview of recent works on the evaluation of AOVs in LLMs. Moreover, we survey related approaches in different stages of the evaluation pipeline in these works. By doing so, we address the potential and challenges with respect to understanding the model, human-AI alignment, and downstream application in social sciences. Finally, we provide practical insights into evaluation methods, model enhancement, and interdisciplinary collaboration, thereby contributing to the evolving landscape of evaluating AOVs in LLMs.

cs.CL↗

arXiv2024

Vox Populi, Vox AI? Using Language Models to Estimate German Public Opinion

The recent development of large language models (LLMs) has spurred discussions about whether LLM-generated "synthetic samples" could complement or replace traditional surveys, considering their training data potentially reflects attitudes and behaviors prevalent in the population. A number of mostly US-based studies have prompted LLMs to mimic survey respondents, with some of them finding that the responses closely match the survey data. However, several contextual factors related to the relationship between the respective target population and LLM training data might affect the generalizability of such findings. In this study, we investigate the extent to which LLMs can estimate public opinion in Germany, using the example of vote choice. We generate a synthetic sample of personas matching the individual characteristics of the 2017 German Longitudinal Election Study respondents. We ask the LLM GPT-3.5 to predict each respondent's vote choice and compare these predictions to the survey-based estimates on the aggregate and subgroup levels. We find that GPT-3.5 does not predict citizens' vote choice accurately, exhibiting a bias towards the Green and Left parties. While the LLM captures the tendencies of "typical" voter subgroups, such as partisans, it misses the multifaceted factors swaying individual voter choices. By examining the LLM-based prediction of voting behavior in a new context, our study contributes to the growing body of research about the conditions under which LLMs can be leveraged for studying public opinion. The findings point to disparities in opinion representation in LLMs and underscore the limitations in applying them for public opinion estimation.

cs.AI↗

arXiv2023

30 Years of Synthetic Data

The idea to generate synthetic data as a tool for broadening access to sensitive microdata has been proposed for the first time three decades ago. While first applications of the idea emerged around the turn of the century, the approach really gained momentum over the last ten years, stimulated at least in parts by some recent developments in computer science. We consider the upcoming 30th jubilee of Rubin's seminal paper on synthetic data (Rubin, 1993) as an opportunity to look back at the historical developments, but also to offer a review of the diverse approaches and methodological underpinnings proposed over the years. We will also discuss the various strategies that have been suggested to measure the utility and remaining risk of disclosure of the generated data.

cs.CR↗