Prompt Robustness Is Task-Dependent: Comparing Objective and Belief-Style Questions in LLM Evaluation

• URL: http://arxiv.org/abs/2607.05554v1
• Date: Mon, 06 Jul 2026 18:47:47 GMT
• Title: Prompt Robustness Is Task-Dependent: Comparing Objective and Belief-Style Questions in LLM Evaluation
• Abstract summary: We ask whether prompt robustness differs between objective questions with fixed answers and subjective questions that ask for opinions or values.<n>We evaluate four instruction-tuned model families on three objective datasets and three subjective datasets.<n>Using a binomial generalized estimating equation, we find significant effects of model, dataset, prompt category, and their interactions.
• Score: 1.7613484075466104
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Survey-style evaluations of large language models often treat a prompted response as a measure of a model's values or beliefs. This assumption is particularly fragile when responses are read as evidence of political values, social attitudes, or beliefs. We ask whether prompt robustness differs between objective questions with fixed answers and subjective questions that ask for opinions or values. We evaluate four instruction-tuned model families on three objective datasets (MMLU, ARC, and CulturalBench) and three subjective datasets (Political Compass Test, ValueBench, and World Values Survey). For each question/statement, we apply multiple types of prompt changes, such as variations in wording, framing, and format, and measure whether the model gives the same answer across variants. Using a binomial generalized estimating equation, we find significant effects of model, dataset, prompt category, and their interactions. The dataset type effect is also significant, and the interaction between dataset type and prompt category is large. These results show that prompt robustness depends on the question type, the prompt change, and the model.

Related papers

• Effects of Answer Format Variation on Gender Bias in Large Language Models [5.457386921498268]
  We study how changes in answer format impact the measurement of gender bias in large language models.<n>We find that answer format does substantially alter measured outcomes, including reversals in order rankings.
  arXiv  Detail & Related papers  (2026-08-18T08:41:05Z)
• ViTaB-A: Evaluating Multimodal Large Language Models on Visual Table Attribution [37.049064455196486]
  Multimodal Large Language Models (mLLMs) are often used to answer questions in structured data.<n>We study structured data attribution/citation, which is the ability of the models to point to the specific rows and columns that support an answer.
  arXiv  Detail & Related papers  (2026-02-17T18:01:35Z)
• On the Credibility of Evaluating LLMs using Survey Questions [0.42061757959666934]
  Recent studies evaluate the value orientation of large language models (LLMs) using adapted social surveys.<n>This paper identifies limitations in this methodology that, depending on the exact setup, can lead to both underestimating and overestimating the similarity of value orientation.<n>Using the World Value Survey in three languages across five countries, we demonstrate that prompting methods (direct vs. chain-of-thought) and decoding strategies (greedy vs. sampling) significantly affect results.
  arXiv  Detail & Related papers  (2026-02-03T21:45:43Z)
• Survey-to-Behavior: Downstream Alignment of Human Values in LLMs via Survey Questions [11.962443965910486]
  Large language models implicitly encode preferences over human values.<n>Can we reliably modify a model's value system in downstream behavior by training it to answer value survey questions accordingly?<n>We demonstrate that our simple approach can not only change the model's answers to in-domain survey questions, but also produces substantial shifts (value alignment) in implicit downstream task behavior.
  arXiv  Detail & Related papers  (2025-08-15T11:36:17Z)
• Revisiting LLM Value Probing Strategies: Are They Robust and Expressive? [81.49470136653665]
  We evaluate the robustness and expressiveness of value representations across three widely used probing strategies.<n>We show that the demographic context has little effect on the free-text generation, and the models' values only weakly correlate with their preference for value-based actions.
  arXiv  Detail & Related papers  (2025-07-17T18:56:41Z)
• MinosEval: Distinguishing Factoid and Non-Factoid for Tailored Open-Ended QA Evaluation with LLMs [15.278241998033822]
  Open-ended question answering (QA) is a key task for evaluating the capabilities of large language models (LLMs)<n>We propose textbfMinosEval, a novel evaluation method that first distinguishes open-ended questions and then ranks candidate answers.
  arXiv  Detail & Related papers  (2025-06-18T07:49:13Z)
• Pointwise Mutual Information as a Performance Gauge for Retrieval-Augmented Generation [78.28197013467157]
  We show that the pointwise mutual information between a context and a question is an effective gauge for language model performance.<n>We propose two methods that use the pointwise mutual information between a document and a question as a gauge for selecting and constructing prompts that lead to better performance.
  arXiv  Detail & Related papers  (2024-11-12T13:14:09Z)
• Contextualized Evaluations: Judging Language Model Responses to Underspecified Queries [85.81295563405433]
  We present a protocol that synthetically constructs context surrounding an under-specified query and provides it during evaluation.<n>We find that the presence of context can 1) alter conclusions drawn from evaluation, even flipping benchmark rankings between model pairs, 2) nudge evaluators to make fewer judgments based on surface-level criteria, like style, and 3) provide new insights about model behavior across diverse contexts.
  arXiv  Detail & Related papers  (2024-11-11T18:58:38Z)
• Political Compass or Spinning Arrow? Towards More Meaningful Evaluations for Values and Opinions in Large Language Models [61.45529177682614]
  We challenge the prevailing constrained evaluation paradigm for values and opinions in large language models. We show that models give substantively different answers when not forced. We distill these findings into recommendations and open challenges in evaluating values and opinions in LLMs.
  arXiv  Detail & Related papers  (2024-02-26T18:00:49Z)
• Answering Ambiguous Questions via Iterative Prompting [84.3426020642704]
  In open-domain question answering, due to the ambiguity of questions, multiple plausible answers may exist. One approach is to directly predict all valid answers, but this can struggle with balancing relevance and diversity. We present AmbigPrompt to address the imperfections of existing approaches to answering ambiguous questions.
  arXiv  Detail & Related papers  (2023-07-08T04:32:17Z)
• A Revised Generative Evaluation of Visual Dialogue [80.17353102854405]
  We propose a revised evaluation scheme for the VisDial dataset. We measure consensus between answers generated by the model and a set of relevant answers. We release these sets and code for the revised evaluation scheme as DenseVisDial.
  arXiv  Detail & Related papers  (2020-04-20T13:26:45Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.