Reference-Distribution Dependence in LLM-Based Synthetic Persona Data: Diagnosis and Post Hoc Adjustment of Demographic Distributions

• URL: http://arxiv.org/abs/2608.28668v1
• Date: Mon, 24 Aug 2026 07:45:48 GMT
• Title: Reference-Distribution Dependence in LLM-Based Synthetic Persona Data: Diagnosis and Post Hoc Adjustment of Demographic Distributions
• Abstract summary: We show that most of the observed error is attributable to the choice of reference rather than to the generator.<n>We recommend treating synthetic persona data as auxiliary material for small-scale survey design.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We diagnose how closely the demographic distributions in LLM-based synthetic persona data match external reference distributions. For the three variables examined, we show that most of the observed error is attributable to the choice of reference rather than to the generator. Using total variation distance (TVD), we compare the sex x age group x province joint distribution of 1,000,000 records from Nemotron-Personas-Korea (NPK) with Korean official statistics. Against resident-registration figures for April 2026, the time of use, the bias bound, defined as the largest possible difference in the share of any subgroup formed from the three variables, is 1.81 percentage points. This is comparable to the margin of error of a survey of roughly 2,900 respondents. This value is not a fixed property of the data. Matching the reference period and series to the generating reference identified here, the 2024 register-based census restricted to Korean nationals, lowers it to 0.56 percentage points. Over the 15 months between the best-matching month (January 2025) and the time of use, the resident-registration population structure itself moves more than twice the distance of NPK's minimum error. Raking and cell post-stratification, the two weighting schemes used in Korean survey practice, remove most of the reference-period dependence at a variance inflation of about 0.2% in both cases. After raking against the generating reference, the residual joint discrepancy lies at, and marginally above, the upper bound of what a perfect generator would produce when realizing 1,000,000 records (97.6th percentile of the Monte Carlo distribution). We recommend treating synthetic persona data as auxiliary material for small-scale survey design rather than as a substitute for survey data, and re-running both diagnosis and adjustment against official statistics current at the time of use.

Related papers

• From Foundation Embeddings to Cropland Maps: Label Efficiency, Temporal Transferability and Independent Human Validation [0.45998296599540095]
  Geospatial foundation models provide reusable representations of satellite imagery that support downstream mapping.<n>We evaluate whether annual AlphaEarth embeddings support binary cultivated-versus-non-cultivated mapping in Maine, USA.
  arXiv  Detail & Related papers  (2026-09-15T13:04:35Z)
• Marginal Fidelity Does Not Establish User Simulation in Demographic Synthetic Survey Panels: Response Contracts, Support Collapse and Conditioning Failure [51.736723807086385]
  Population-marginal agreement is evidence about an elicitation contract and an estimand obtainable without simulated respondents, not evidence of individual simulation.<n>On nine aligned model-battery pairs, a no-persona population-prevalence query averages 6.27 MAE versus 12.39 for committed panels and wins all nine comparisons.
  arXiv  Detail & Related papers  (2026-09-07T10:22:22Z)
• Can Large Language Models Represent Urban Publics? Behavioral Replication and Population Mismatch in an Affordable-Housing Experiment [1.7984459258809926]
  We show that large language models can approximate one aggregate contrast while failing to preserve the population structure.<n>Model evaluation in urban planning should test whether this spatial and social structure survives simulation, not only average effects.
  arXiv  Detail & Related papers  (2026-07-29T16:27:36Z)
• SAGA: A Sequence-Adaptive Generative Architecture for Multi-Horizon Probabilistic Forecasting with Adaptive Temporal Conformal Prediction [0.0]
  We propose SAGA, a decoder-only transformer that delivers individual-level prediction intervals with finite-sample marginal coverage guarantees.<n>We trained on the longitudinal Swedish LISA register over 1990 to 2022, comprising 2,143,817 individuals and 61,284,903 person-years.<n> SAGA reduces continuous ranked probability score by 31.9 percent at the ten-year horizon and mean absolute error by 37.7 percent at the twenty-year horizon.
  arXiv  Detail & Related papers  (2026-05-18T18:35:20Z)
• Unmasking Algorithmic Bias in Predictive Policing: A GAN-Based Simulation Framework with Multi-City Temporal Analysis [0.0]
  We show how racial bias propagates through the full enforcement pipeline from crime occurrence to police contact.<n>Experiments reveal extreme and year variant bias in Baltimores detected mode, with mean annual DIR up to 15714 in 2019, moderate under detection of Black residents in Chicago DIR equals 0.22.<n>We further demonstrate that a Conditional Tabular GAN CTGAN debiasing approach partially redistributes detection rates but cannot eliminate structural disparity without accompanying policy intervention.
  arXiv  Detail & Related papers  (2026-03-19T14:55:32Z)
• Denoising the US Census: Succinct Block Hierarchical Regression [48.50026671993762]
  We introduce BlueDown, a new post-processing method that produces more accurate, consistent estimates.<n>We obtain especially large accuracy improvements for aggregates at the county and tract levels on evaluation metrics proposed by the US Census Bureau.
  arXiv  Detail & Related papers  (2026-03-10T17:39:25Z)
• Evidence-based Distributional Alignment for Large Language Models [58.65469623911573]
  We propose Evi-DA, an evidence-based alignment technique that improves the fidelity and robustness of LLM-based distribution estimation.<n>Given a target country, Evi-DA retrieves related World Values Survey items and their answer distributions, predicts a coarse Welzel value signature for each option, and infers the country-conditioned answer distribution in a structured format.
  arXiv  Detail & Related papers  (2026-03-03T03:34:06Z)
• A Deep Generative Framework for Joint Households and Individuals Population Synthesis [0.562479170374811]
  We propose a deep generative framework to generate a synthetic population with household-individual and individual-individual relationships. Results for an application in Delaware, USA demonstrate the ability to ensure the realism of generated household-individual records.
  arXiv  Detail & Related papers  (2024-06-30T23:01:58Z)
• Exploring a new machine learning based probabilistic model for high-resolution indoor radon mapping, using the German indoor radon survey data [0.0]
  Radon is a carcinogenic, radioactive gas that can accumulate indoors. Radon concentration at the national scale is usually estimated on the basis of extensive measurement campaigns. We propose a model-based approach that allows a more realistic estimation of indoor radon distribution.
  arXiv  Detail & Related papers  (2023-10-17T10:51:05Z)
• Beyond Normal: On the Evaluation of Mutual Information Estimators [52.85079110699378]
  We show how to construct a diverse family of distributions with known ground-truth mutual information. We provide guidelines for practitioners on how to select appropriate estimator adapted to the difficulty of problem considered.
  arXiv  Detail & Related papers  (2023-06-19T17:26:34Z)
• Detecting Adversarial Data by Probing Multiple Perturbations Using Expected Perturbation Score [62.54911162109439]
  Adversarial detection aims to determine whether a given sample is an adversarial one based on the discrepancy between natural and adversarial distributions. We propose a new statistic called expected perturbation score (EPS), which is essentially the expected score of a sample after various perturbations. We develop EPS-based maximum mean discrepancy (MMD) as a metric to measure the discrepancy between the test sample and natural samples.
  arXiv  Detail & Related papers  (2023-05-25T13:14:58Z)
• Statistical and Computational Phase Transitions in Group Testing [73.55361918807883]
  We study the group testing problem where the goal is to identify a set of k infected individuals carrying a rare disease. We consider two different simple random procedures for assigning individuals tests.
  arXiv  Detail & Related papers  (2022-06-15T16:38:50Z)
• The Gap on GAP: Tackling the Problem of Differing Data Distributions in Bias-Measuring Datasets [58.53269361115974]
  Diagnostic datasets that can detect biased models are an important prerequisite for bias reduction within natural language processing. undesired patterns in the collected data can make such tests incorrect. We introduce a theoretically grounded method for weighting test samples to cope with such patterns in the test data.
  arXiv  Detail & Related papers  (2020-11-03T16:50:13Z)
• Magnify Your Population: Statistical Downscaling to Augment the Spatial Resolution of Socioeconomic Census Data [48.7576911714538]
  We present a new statistical downscaling approach to derive fine-scale estimates of key socioeconomic attributes. For each selected socioeconomic variable, a Random Forest model is trained on the source Census units and then used to generate fine-scale gridded predictions. As a case study, we apply this method to Census data in the United States, downscaling the selected socioeconomic variables available at the block group level, to a grid of 300 spatial resolution.
  arXiv  Detail & Related papers  (2020-06-23T16:52:18Z)

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.