Pith. sign in

REVIEW 5 major objections 5 minor 45 references

Uncovering Hidden Violent Tendencies in LLMs: A Demographic Analysis via Behavioral Vignettes

T0 review · 5 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Probing token likelihoods reveals that instruction-tuned LLMs can rank violent answers above pacifist ones even while their generated text stays polite, and that this hidden preference shifts with persona demographics.

desk verdict The VBVQ application to LLMs is genuinely new and worth a look, but the central 'hidden violent tendencies' claim rides on an unvalidated likelihood proxy that its own appendix shows is not robust. read the letter →

arxiv 2506.20822 v1 pith:MY7BFQXL submitted 2025-06-25 cs.CL cs.AI

Quintin Myers , Yanjun Gao This is my paper · ORCID
classification cs.CLcs.AI
keywords largelanguagemodelsviolencedetectiondemographicbiassequencelikelihoodbehavioralvignettespersonapromptingsafetyalignmentfairnessevaluation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Instruction-tuned LLMs are being proposed as tools for detecting online violence, and this paper asks whether they can be trusted with morally ambiguous conflicts. The authors take a validated social-science instrument, the Violent Behavior Vignette Questionnaire (VBVQ), and run its ten everyday-conflict scenarios through six LLMs under persona prompts that vary race, age, and US location. They claim that the models' surface-level text, which is polite and nonviolent, hides an internal preference for violent answers that only appears when one measures the sequence likelihood the model assigns to violent reference responses. They further report that this hidden preference shifts with demographics, most strikingly that every model judged 15-year-old personas less violence-prone than older ones, the opposite of the age-crime curve. If the claim is right, safety alignment can mask rather than remove violent tendencies, and bias can hide in probability space where ordinary output inspection will not catch it.

What carries the argument

The load-bearing instrument is sequence likelihood probing. For every vignette-plus-persona prompt, the model's log-likelihood is computed for each of three reference answer categories—PACIFIST, NON-VIOLENT, VIOLENT—averaged over human-written labels and four LLM-generated paraphrases, then softmax-normalized into a probability distribution over the three categories. The Top-Rank Rate is the share of generations in which a given category receives the highest normalized likelihood, and pairwise differences in Prob(VI) supply the demographic heatmaps. This machinery does the work of separating surface text from internal preference: a model can emit polite language while still assigning the highest probability to a violent reference string, which is the paper's operational definition of a hidden violent tendency.

What would settle it

For any vignette-persona pair in which VI ranks top by sequence likelihood, generate 100 free-form continuations and count the proportion that describe violent actions; if that proportion is roughly zero or matches the other categories, the 'hidden tendency' is an artifact of the short reference probes rather than a real behavioral preference.

Watch

Extended reading notes

Core claim

The central claim is that LLMs, despite appearing aligned through categorical choices and free-form text, harbor latent violent response tendencies that can be elicited with demographic personas. The paper's evidence is the Top-Rank Rate: for each vignette-persona prompt, the authors compute the softmax-normalized probability the model assigns to human-written and paraphrased reference responses in three categories—PACIFIST, NON-VIOLENT, and VIOLENT—and tally which category ranks first most often. Under this measure, models like the DS-Qwen-32B variant show a majority preference for the violent category, while Llama-3.1-8B favors the pacifist category, even though their sampled text outputs are invariably calm. The demographic heatmaps of pairwise differences in Prob(VI) then reveal the headline patterns: the age reversal (15-year-old personas rated least violent), race patterns read as political-correctness overcorrection, and inconsistent location effects. The paper reads these as evidence of hidden bias and overcorrection from safety training, concluding that LLMs are not yet prepared to ethically detect violence in real-world settings.

Load-bearing premise

The whole argument rests on the premise that how likely a model finds a few pre-written answer strings reveals its true violent tendency—a premise the paper's own appendix weakens by showing the headline demographic patterns shift when those strings are replaced.

Editorial extensions

If this is right

  • If the divergence is real, standard safety evaluations that judge generated text alone will miss latent violent preferences; probability-space auditing belongs in the safety toolkit.
  • If the age reversal is a true model behavior, deploying these LLMs in youth-facing violence-prevention contexts could systematically under-estimate the violence risk of adolescent personas.
  • If the race results reflect overcorrection, then well-intentioned alignment can perpetuate positive stereotypes (e.g., Asian passivity) even as it removes overt racial slurs.
  • Because the paper's own Appendix A.1 shows the demographic heatmaps shift substantially when the LLM-generated paraphrases are excluded, any real deployment must re-validate the reference strings and report both human-label-only and paraphrase-inclusive results.
  • The paper explicitly recommends that community violence intervention programs not yet use LLMs for violence detection, a direct operational implication of its findings.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editor's inference: A natural extension would be to test the likelihood-probing proxy against an actual behavioral benchmark—generate long continuations under the same personas and count how often the sampled text describes violent actions; if that rate tracks the Top-Rank rate, the 'hidden tendency' is not hidden at all but an artifact of short-reference likelihood normalization.
  • Editor's inference: The age reversal invites a controlled perturbation study: vary the perceived vulnerability of the persona (for example, '15-year-old accompanied by a police officer' versus '15-year-old alone') to see whether the lower violent probability for teenagers is a genuine statistical belief or a safety-driven refusal pattern.
  • Editor's inference: A practical fairness metric suggested by this design would be to track Top-Rank Rate across demographic subgroups at model release, treating probability-space demographic gaps as a bias flag even when text outputs look aligned.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper adapts the Violent Behavior Vignette Questionnaire (VBVQ), a validated social science instrument, to probe six instruction-tuned LLMs under zero-shot persona-based prompting that varies race, age, and US location. It measures surface-level text generation via BERTScore against human-labeled PA/NV/VI reference categories, and 'internal tendencies' via the softmax-normalized sequence likelihood of reference strings. The authors claim (1) that surface-level output often diverges from internal violent tendencies and (2) that these tendencies vary demographically, often contradicting criminological expectations. The main evidence for both claims is the sequence-likelihood metric, with pairwise differences in Prob(VI) aggregated across models.

Significance. If the central claims were established, this would be a significant contribution to AI safety and fairness, with implications for high-stakes deployment of LLMs in violence detection. The use of a validated social science instrument and the systematic variation of demographic cues are strengths, as is the inclusion of multiple models from different geopolitical contexts. The paper also makes a good-faith effort to report robustness analysis in Appendix A.1. However, the main findings rest entirely on an unvalidated and apparently unstable likelihood-based proxy for 'internal violent tendency.' The paper's own appendix shows that the headline age and race patterns shift substantially when GPT-4o paraphrases are removed from the reference set, which undercuts the robustness of the central claims. The significance of the paper therefore depends on whether the proxy can be validated and whether the findings survive pre-registered robustness checks.

major comments (5)
  1. [Section 3, 'Probing LLM response tendencies via sequence likelihood'] The central metric is a softmax-normalized sequence likelihood computed over three reference strings (PA, NV, VI). This metric is not length-normalized, and no evidence is provided that it reflects the model's 'internal preference' for violent responses. Sequence likelihoods systematically decrease with response length, and the three reference categories likely differ in average length, so the softmax probabilities are a within-set ranking that is not calibrated across categories. The paper should at minimum length-normalize the likelihoods and validate the proxy against an external behavioral measure (e.g., the model's own free-form response choices or human judgments).
  2. [Appendix A.1, 'More heatmap visualization for Prob VI difference'] The robustness analysis in Appendix A.1 shows that the headline findings are not stable under a reasonable change in reference strings. When GPT-4o paraphrases are excluded, the age pattern for DS-Qwen and DS-Qwen-32B flips (the 15-year-old persona becomes the most violent, contrary to the main text's finding #1), and Mixtral-8x22B exhibits large race effects (a 44% drop from WHITE to NATIVE and a 34% increase from WHITE to HISPANIC) that are absent in the paraphrase-inclusive analysis. The main text presents the paraphrase-inclusive results as primary without quantifying this instability. The central demographic claims are therefore not robust.
  3. [Section 3, 'Prompting with varying demographics'] The paper states that pairwise differences in Prob(VI) control for other demographic variables, but the controlling procedure is not described. Since race, age, and location are varied jointly in every prompt, the reader cannot determine whether the reported pairwise differences isolate a single demographic dimension or are confounded by correlated variation. A regression or factorial design with explicit controls is needed to support the demographic attribution claims.
  4. [Figure 2 and Section 4, 'Overall results'] The pairwise differences in Prob(VI) displayed in Figure 2 are presented without confidence intervals or significance tests. The Kruskal-Wallis tests in Table 2 are computed on BERTScore F1, not on the sequence-likelihood probabilities that underlie Figure 2. Given the small number of models and the use of stochastic sampling, the reported differences may be within sampling noise. The paper should report uncertainty estimates for the Prob(VI) differences.
  5. [Section 3, 'Vignette dataset' and Appendix A.1] The GPT-4o-generated paraphrases are used as reference strings for all models, including GPT-4o-mini. Since GPT-4o is a sibling model to GPT-4o-mini, the likelihoods of these paraphrases may be systematically higher for GPT-4o-mini than for other models, biasing cross-model comparisons. The paper does not analyze sensitivity to the paraphrase source or demonstrate that the paraphrases are equivalent probes across intent categories, despite the Appendix A.1 results showing that the choice of reference strings materially changes the conclusions.
minor comments (5)
  1. [Section 4, 'Overall results'] The paper says 'all LLMs consistently choose the most passive options' when given the 10 categorical options, but no supporting table, figure, or quantitative result is provided. Please include the actual selection rates or a supplementary table.
  2. [Table 3 and Section 3] Table 3 lists five LLMs, while the abstract and text mention six. Clarify that GPT-4o-mini was excluded from the sequence-likelihood analysis due to API limitations and that Llama-3.1-8B was substituted, and clearly state which models appear in each analysis.
  3. [References] Several citations are incomplete or missing from the reference list: 'Wang et al.' (Self-consistency improves chain of thought reasoning) and 'Zhang et al.' (BERTScore) lack venue/year details, and 'CDE' and 'noa' appear as incomplete citations.
  4. [Throughout] There are typographical and grammatical errors, including 'tendancy,' 'shows has the highest,' and 'a drop 44%.' A careful proofread is needed.
  5. [Section 3, 'Probing LLM response tendencies via sequence likelihood'] The definition of Top-Rank Rate as the 'proportion of generations' is ambiguous, since sequence-likelihood computation does not involve sampling multiple generations in the same way as open-text generation. Please clarify the unit of analysis.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reference standards are external to the target models, and the robustness caveats in Appendix A.1 are validity concerns rather than definitional recycling.

full rationale

The paper's derivation chain is: use an externally validated social-science instrument (VBVQ) to obtain human-labeled PA/NV/VI reference responses; generate additional paraphrases with GPT-4o; then score each target LLM's sequence likelihood against these reference strings, softmax-normalize the three likelihoods, and compare demographic subgroups. At no point is a parameter fitted to the outcome being 'predicted,' nor is any conclusion derived from a quantity that is defined in terms of the conclusion itself. The human labels and GPT-4o paraphrases are external to the six evaluated models, so the 'hidden violent tendency' claim is not constructed from the target models' own outputs. The paper contains no self-citations, no imported uniqueness theorem, and no ansatz smuggled in via citation. The main caveat is the paper's own Appendix A.1: when paraphrases are excluded and only human-written labels are used, several headline heatmap patterns shift substantially (e.g., Mixtral's race effects and the Qwen age pattern). That is a robustness and construct-validity limitation, not circularity: the sequence-likelihood probe is an explicit operationalization of 'tendency,' and the softmax over exactly three reference classes produces a within-set ranking rather than a calibrated probability. Those are measurement concerns, but they do not amount to the derivation reducing to its inputs by construction. The use of GPT-4o-generated paraphrases when evaluating GPT-4o-mini is also a potential independence/contamination concern, but it is not a circular reduction because the paraphrase set is not fitted to GPT-4o-mini's behavior. Overall, the central empirical chain is self-contained; there is no circular step to quote.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No model parameters are fitted to data, but the analysis depends on several unvalidated measurement assumptions: the VBVQ-to-LLM transfer, the likelihood-as-preference proxy, and the GPT-4o paraphrase set. These assumptions carry the weight of the headline findings.

assumptions (5)
  • domain assumption The Violent Behavior Vignette Questionnaire (VBVQ), validated on human adults, is a valid probe of LLM violent response tendencies.
    The paper transfers a human psychometric instrument to LLMs without evidence that models interpret the vignettes and response categories the same way humans do.
  • domain assumption Softmax-normalized sequence likelihood of the PA/NV/VI reference strings reflects the model's latent violent preference.
    This proxy is not validated against behavior or human judgments; likelihoods are known to be sensitive to reference wording and length.
  • ad hoc to paper The four GPT-4o-generated paraphrases of each human-labeled response preserve the original intent and are appropriate for averaging.
    No human validation of paraphrase fidelity is reported, and Appendix A.1 demonstrates that including paraphrases changes the main qualitative findings.
  • domain assumption Criminological and sociological findings (e.g., age-crime curve, racialized violence stereotypes) are the correct benchmarks against which LLM demographic variation should be judged.
    The paper interprets deviations from these expectations as bias or overcorrection, but LLM training objectives may produce different norms.
  • domain assumption Persona prompts such as 'Native-American, 15 year old male from Miami' are interpreted by the model as intended demographic identity cues.
    Models may treat demographic labels differently from humans, and the paraphrase-inclusive vs human-label sensitivity suggests persona effects are not robust.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Uncovering Hidden Violent Tendencies in LLMs: A Demographic Analysis via Behavioral Vignettes." pith.science (2026). https://pith.science/paper/MY7BFQXL

@misc{pith2026250620822,
  author       = {Pith},
  title        = {Pith review of: Uncovering Hidden Violent Tendencies in LLMs: A Demographic Analysis via Behavioral Vignettes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MY7BFQXL}},
  note         = {Machine review of arXiv:2506.20822}
}
read the original abstract

Large language models (LLMs) are increasingly proposed for detecting and responding to violent content online, yet their ability to reason about morally ambiguous, real-world scenarios remains underexamined. We present the first study to evaluate LLMs using a validated social science instrument designed to measure human response to everyday conflict, namely the Violent Behavior Vignette Questionnaire (VBVQ). To assess potential bias, we introduce persona-based prompting that varies race, age, and geographic identity within the United States. Six LLMs developed across different geopolitical and organizational contexts are evaluated under a unified zero-shot setting. Our study reveals two key findings: (1) LLMs surface-level text generation often diverges from their internal preference for violent responses; (2) their violent tendencies vary across demographics, frequently contradicting established findings in criminology, social science, and psychology.

Figures

Figures reproduced from arXiv: 2506.20822 by the authors.

Figure 1
Figure 1. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Directional differences in predicted violence probability (Prob VI) across Race (top), Age (Lower Left) and Location (Lower Right). Each heatmap shows average differences in predicted VI intent between demographic pairs (A → B), aggregated across five LLMs. Positive values indicate higher predicted violence when switching from group A to B while controlling for other variables. categorical and free-form outputs, whi… view at source ↗
Figure 3
Figure 3. Comparison of ∆ Prob(VI) across LLMs for Age, Race and Location [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Comparison of ∆ Prob(VI) across LLMs for Age, Race and Location, using only the human written labels (without their LLM generated paraphrases). Patton et al. (2013) who identified "internet bang￾ing." Internet banging is a set of online behaviors by those perceived to …
Figure 5
Figure 5. Figure 5: Free-form responses from four large language models (LLMs) to the same morally ambiguous scenario [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 34 canonical work pages

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    https://cde.ucr.cjis.gov/LATEST/webapp/#/pages/explorer/crime/crime-trend CDE

    work page

  4. [4]

    Michelle Alexander. 2012. The New Jim Crow : Mass Incarceration in the Age of Colorblindness , revised edition edition. The New Press, New York

    work page 2012

  5. [5]

    Ayman Alhelbawy, Poesio Massimo, and Udo Kruschwitz. 2016. https://aclanthology.org/L16-1257/ Towards a corpus of violence acts in A rabic social media . In Proceedings of the Tenth International Conference on Language Resources and Evaluation ( LREC `16) , pages 1627--1631, Portoro z , Slovenia. European Language Resources Association (ELRA)

    work page 2016

  6. [6]

    Pierpaolo Basile, Marco Degemmis, Marco Polignano, Giovanni Semeraro, Lucia Siciliani, Vincenzo Tamburrano, Fabiana Battista, and Rosa Scardigno. 2024. Llamantino against cyber intimate partner violence. In Proceedings of the 10th Italian Conference on Computational Linguistics (CLiC-it 2024), pages 52--58

    work page 2024

  7. [7]

    CDC. 2025. https://www.cdc.gov/community-violence/about/index.html About Community Violence

    work page 2025

  8. [8]

    Rosalind S. Chou. 2015. Asian American Sexual Politics : The Construction of Race , Gender , and Sexuality , reprint edition edition. Rowman & Littlefield Publishers, Lanham, Maryland

    work page 2015

Show all 45 references
  1. [9]

    Chou and Joe R

    Rosalind S. Chou and Joe R. Feagin. 2014. Myth of the Model Minority : Asian Americans Facing Racism , Second Edition , 2 edition edition. Routledge, Boulder

    2014

  2. [10]

    Joe R. Feagin. 1991. https://doi.org/10.2307/2095676 The Continuing Significance of Race : Antiblack Discrimination in Public Places . American Sociological Review, 56(1):101--116

    1991 doi

  3. [11]

    Frey, Desmond U

    William R. Frey, Desmond U. Patton, Michael B. Gaskell, and Kyle A. McGregor. 2020. https://doi.org/10.1177/0894439318788314 Artificial Intelligence and Inclusion : Formerly Gang - Involved Youth as Domain Experts for Analyzing Unstructured Twitter Data . Social Science Comput...

    2020 doi

  4. [12]

    Isabel O Gallegos, Ryan A Rossi, Joe Barrow, Md Mehrab Tanjim, Sungchul Kim, Franck Dernoncourt, Tong Yu, Ruiyi Zhang, and Nesreen K Ahmed. 2024. Bias and fairness in large language models: A survey. Computational Linguistics, 50(3):1097--1179

    2024

  5. [13]

    Kimberly Glasgow and Ronald Schouten. 2014. Assessing violence risk in threatening communications. ACL 2014, page 38

    2014

  6. [14]

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, and 1 others. 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948

    2025 arXiv

  7. [15]

    Pengcheng He, Xiaodong Liu, Jianfeng Gao, and Weizhu Chen. 2020. Deberta: Decoding-enhanced bert with disentangled attention. arXiv preprint arXiv:2006.03654

    2020 arXiv

  8. [16]

    Tiancheng Hu, Yara Kyrychenko, Steve Rathje, Nigel Collier, Sander van der Linden, and Jon Roozenbeek. 2025. Generative language models exhibit social identity biases. Nature Computational Science, 5(1):65--75

    2025

  9. [17]

    Yibo Hu, MohammadSaleh Hosseini, Erick Skorupa Parolin, Javier Osorio, Latifur Khan, Patrick Brandt, and Vito D ' Orazio. 2022. https://doi.org/10.18653/v1/2022.naacl-main.400 C onfli BERT : A pre-trained language model for political conflict and violence . In Proceedings of t...

    2022 doi

  10. [18]

    Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, and 1 others. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276

    2024 arXiv

  11. [19]

    Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, Lélio Renard Lavaud, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas ...

    2023 arXiv

  12. [20]

    Albert Q Jiang, Alexandre Sablayrolles, Antoine Roux, Arthur Mensch, Blanche Savary, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Emma Bou Hanna, Florian Bressand, and 1 others. 2024. Mixtral of experts. arXiv preprint arXiv:2401.04088

    2024 arXiv

  13. [21]

    Minju Kim, Heuiyeen Yeen, and Myoung-Wan Koo. 2024. Towards context-based violence detection: A korean crime dialogue dataset. In Findings of the Association for Computational Linguistics: EACL 2024, pages 603--623

    2024

  14. [22]

    Grieser, Shug Miller, James Shepard, Javier Garcia-Perez, Nick Deas, Desmond U

    Shana Kleiner, Jessica A. Grieser, Shug Miller, James Shepard, Javier Garcia-Perez, Nick Deas, Desmond U. Patton, Elsbeth Turcan, and Kathleen McKeown. 2024. https://doi.org/10.1007/s43681-024-00623-2 Unmasking camouflage: exploring the challenges of large language models in d...

    2024 doi

  15. [23]

    Jeffrey Lane. 2018. The Digital Street . Oxford University Press, New York

    2018

  16. [24]

    Lee and Ramiro Martinez

    Matthew T. Lee and Ramiro Martinez. 2009. https://doi.org/10.1108/S1521-6136(2009)0000013004 Immigration reduces crime: an emerging scholarly consensus . In William F. Mcdonald, editor, Sociology of Crime , Law and Deviance , volume 13, pages 3--16. Emerald Group Publishing Limited

    2009 doi

  17. [25]

    Stowell, and Matthew T

    Ramiro Martinez, Jacob I. Stowell, and Matthew T. Lee. 2010. https://doi.org/10.1111/j.1745-9125.2010.00202.x Immigration and Crime in an Era of Transformation : A Longitudinal Analysis of Homicides in San Diego Neighborhoods , 1980–2000* . Criminology, 48(3):797--829

    2010

  18. [26]

    Patrick E McKight and Julius Najab. 2010. Kruskal-wallis test. The corsini encyclopedia of psychology, pages 1--1

    2010

  19. [27]

    Nunes, Chantal A

    Kevin L. Nunes, Chantal A. Hermann, Sacha Maimone, Maya Atlas, and Brian A. Grant. 2021. https://doi.org/10.1177/0033294120939308 The Violent Behavior Vignette Questionnaire ( VBVQ ): A Measure of Violent Behavior for Research in Forensic and Non - Forensic Settings and Popula...

    2021 doi

  20. [28]

    Nunes, Chloe I

    Kevin L. Nunes, Chloe I. Pedneault, Chantal A. Hermann, and Julia M. Fraser. 2023. https://doi.org/10.1037/vio0000470 Do violent attitude measures assess evaluative attitudes toward violence? Psychology of Violence, 13(4):348--359. Publisher: Educational Publishing Foundation

    2023 doi

  21. [29]

    Patton, William R

    Desmond U. Patton, William R. Frey, Kyle A. McGregor, Fei-Tzin Lee, Kathleen McKeown, and Emanuel Moss. 2020. https://doi.org/10.1145/3375627.3375841 Contextual Analysis of Social Media : The Promise and Challenge of Eliciting Context in Social Media Posts with Natural Languag...

    2020

  22. [30]

    Eschmann, and Dirk A

    Desmond Upton Patton, Robert D. Eschmann, and Dirk A. Butler. 2013. https://doi.org/10.1016/j.chb.2012.12.035 Internet banging: New trends in social media, gang violence, masculinity and hip hop . Computers in Human Behavior, 29(5):A54--A59

    2013 doi

  23. [31]

    Eschmann, Caitlin Elsaesser, and Eddie Bocanegra

    Desmond Upton Patton, Robert D. Eschmann, Caitlin Elsaesser, and Eddie Bocanegra. 2016. https://doi.org/10.1016/j.chb.2016.05.052 Sticks, stones and Facebook accounts: What violence outreach workers know about social media and urban-based gang violence in Chicago . Computers i...

    2016 doi

  24. [32]

    Eschmann, Sadiq Patel, and Shantel Crosby

    Desmond Upton Patton, Patrick Leonard, Caitlin Elaesser, Robert D. Eschmann, Sadiq Patel, and Shantel Crosby. 2019. https://doi.org/10.1177/0044118X17720325 What’s a Threat on Social Media ? How Black and Latino Chicago Young Men Define and Navigate Threats Online . Youth & So...

    2019 doi

  25. [33]

    Desmond Upton Patton, Kyle McGregor, and Gary Slutkin. 2018. https://doi.org/10.1542/peds.2017-2438 Youth Gun Violence Prevention in a Digital Age . Pediatrics, 141(4):e20172438

    2018 doi

  26. [34]

    Ellie Pavlick, Heng Ji, Xiaoman Pan, and Chris Callison-Burch. 2016. The gun violence database: A new task and data set for nlp. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1018--1024

    2016

  27. [35]

    Pyrooz, Scott H

    David C. Pyrooz, Scott H. Decker, and Richard K. Moule Jr. 2015. https://doi.org/10.1080/07418825.2013.778326 Criminal and Routine Activities in Online Settings : Gangs , Offenders , and the Internet . Justice Quarterly, 32(3):471--499. Publisher: Routledge \_eprint: https://d...

    2015

  28. [36]

    Sourav Saha, Jahedul Alam Junaed, Maryam Saleki, Arnab Sen Sharma, Mohammad Rashidujjaman Rifat, Mohamed Rahouti, Syed Ishtiaque Ahmed, Nabeel Mohammed, and Mohammad Ruhul Amin. 2023. Vio-lens: A novel dataset of annotated social network posts leading to different forms of com...

    2023

  29. [37]

    Sapolsky

    Robert M. Sapolsky. 2018. Behave: The Biology of Humans at Our Best and Worst , reprint edition edition. Penguin Books, New York, New York

    2018

  30. [38]

    Mohammad Shokri, Emily Klapper, Jason Shan, and Sarah Ita Levitan. 2025. Finding common patterns in domestic violence stories posted on reddit. In Proceedings of the The 7th Workshop on Narrative Understanding, pages 15--25

    2025

  31. [39]

    Shulman, Laurence D

    Elizabeth P. Shulman, Laurence D. Steinberg, and Alex R. Piquero. 2013. https://doi.org/10.1007/s10964-013-9950-4 The Age – Crime Curve in Adolescence and Early Adulthood is Not Due to Age Differences in Economic Status . Journal of Youth and Adolescence, 42(6):848--860

    2013 doi

  32. [40]

    Darrell Steffensmeier, Jessie Slepicka, and Jennifer Schwartz. 2025. https://doi.org/10.1146/annurev-criminol-111523-122451 International and Historical Variation in the Age – Crime Curve . Annual Review of Criminology, 8(Volume 8, 2025):239--268. Publisher: Annual Reviews

    2025 doi

  33. [41]

    D'Alessio

    Lisa Stolzenberg and Stewart J. D'Alessio. 2008. https://doi.org/10.1177/0022427807309441 Co- Offending and the Age - Crime Curve . Journal of Research in Crime and Delinquency, 45(1):65--86. Publisher: SAGE Publications Inc

    2008 doi

  34. [42]

    Self-consistency improves chain of thought reasoning in language models

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc V Le, Ed H Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models. In The Eleventh International Conference on Learning Representations

  35. [43]

    Magdalena Wolska, Matti Wiegmann, Christopher Schr \"o der, Ole Borchardt, Benno Stein, and Martin Potthast. 2023. Trigger warnings: Bootstrapping a violence detector for fan fiction. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 569--576

    2023

  36. [44]

    Bertscore: Evaluating text generation with bert

    Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q Weinberger, and Yoav Artzi. Bertscore: Evaluating text generation with bert. In International Conference on Learning Representations

  37. [45]

    Caleb Ziems and Diyi Yang. 2021. https://doi.org/10.18653/v1/2021.findings-emnlp.82 To protect and to serve? analyzing entity-centric framing of police violence . In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 957--976, Punta Cana, Dominican Re...

    2021 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.