The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse
Open accessThe DOI matches its Crossref record; 25 of the 25 references it cites were found in a registry or an index; 23 of the 50 sources its claims cite were checked for support.
Abstract
We therefore tested the ability of the DASH to predict future instances of deadly or persistent domestic abuse.
Overview
CoChat Summary
TL;DR
The DASH risk assessment showed moderate predictive validity for deadly domestic violence (AUC = 0.67) and persistent domestic abuse (AUC = 0.62) in Dyfed-Powys Police data. A shortened five-item DASH and a criminological-variable score improved prediction, with the combined DASH-CV reaching AUC = 0.84 for deadly violence.
Key findings
- DASH risk stratification predicted deadly violence versus control with AUC = 0.67 (95% CI [0.59, 0.75], p < 0.001) and persistent offending versus control with AUC = 0.62 (95% CI [0.58, 0.66], p < 0.001). High-risk ratings were 25.8% (16/62) in the Deadly group, 16.2% (69/427) in the Persistent group, and 5.3% (23/432) in the control group; chi-square tests were significant.
- Fifteen of the 27 DASH items (55.6%) did not significantly distinguish the Deadly and control groups. In a multiple logistic regression with all 27 items, only five were significantly predictive: Q27 criminal history, Q24 drugs/alcohol/mental health, Q26 breach of bail/injunction, Q16 use of weapons, and Q6 separation within the past year.
- A DASH short form (DASH-SF) using those five items (score 0–5) predicted deadly violence with AUC = 0.80 (95% CI [0.73, 0.87], p < 0.001), significantly greater than the full DASH (p < 0.001), and predicted persistent offending with AUC = 0.71 (95% CI [0.64, 0.75], p < 0.001), also greater than the full DASH (p < 0.001).
- A criminological-variable score (CV-score) from Police National Computer records predicted deadly violence with AUC = 0.82 (95% CI [0.76, 0.89], p < 0.001) and persistent offending with AUC = 0.70 (95% CI [0.66, 0.74], p < 0.001). The combined DASH-CV predicted deadly violence with AUC = 0.84 (95% CI [0.78, 0.91], p < 0.001) and persistent offending with AUC = 0.74 (95% CI [0.70, 0.78], p < 0.001), though for deadly violence it was not significantly greater than DASH-SF or CV-score alone.
Methods
- Archival analysis of 24,889 domestic abuse crimes and incidents in the Dyfed-Powys Police area from 1 January 2016 to 31 December 2019. Groups were Deadly offenders (N = 141), Persistent offenders (859 identified; random sample of 427 used), and a control group of one-time offenders (final N = 432).
- The DASH contains 27 main questions with Yes/No/Refused responses and risk stratification as Standard, Medium, or High. Cases with fewer than 24 of 27 DASH items answered were eliminated.
- Signal detection theory via ROC/AUC assessed predictive accuracy; binary logistic regression calculated odds ratios for individual DASH items; multiple logistic regression entered all 27 items to predict deadly violence; chi-square tested risk stratification across groups. A CV-score was built from number of previous offences and number of previous violent offences, and DASH-CV combined DASH-SF with CV-score.
Limitations
- The study used official police data only; the authors note that only 20–24% of domestic violence incidents are reported, which may contaminate the control group and likely makes the figures lower than true figures.
- The authors state the study was not powered to examine subgroup differences, and larger-scale studies are needed for different genders, heterosexual versus non-heterosexual relationships, cultural settings, older adults, and different forms of domestic abuse.
- Excluding cases with incomplete DASH assessments may have removed higher-risk and/or chaotic cases and does not represent true operational conditions. The authors also caution that DASH-CV was developed and tested in the same construction sample and needs testing in an independent sample to rule out inflation from sample-specific item selection.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
17 Valid · 3 Not valid · 20 Not checked · 74 Own results and interpretations
Some cited sources support this claim and some don’t — read each source’s verdict below.
In the paper
“However, Thornton [8,12] and Chalkley and Strang [11] found that males in the control sample had significantly more arrests and convictions than those in the deadly domestic offender's sample.”
§ The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse > 4. Discussion
What the sources say
- Partly supportedcites [8] Thornton, 2017, Police attempts to predict domestic murder and serious assaults: Is early warning possible yet?, Camb. J. Evid. Based Polic. sourceExperimental — accuracy not yet measured
The Findings Descriptive Analysis says male controls had significantly more arrests and convictions for violence than the domestic murder and serious-assault offenders. The claim omits that restriction: Table 7 shows nearly identical averages for all prior arrests and for all prior convictions and cautions, so it does not support the broader wording. (LLM, conf=0.88)
- Not matchedcites [12] Thornton, S. (2011). Predicting Serious Domestic Assaults and Murder in the Thames Valley. [Master’s Thesis, University of Cambridge].
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Disputedcites [11] Chalkley, 2017, Predicting domestic homicides and serious violence in Dorset: A replication of Thornton’s Thames Valley analysis, Camb. J. Evid. Based Polic. sourceExperimental — accuracy not yet measured
Table 5 (listed under Conclusion) shows that male controls averaged more prior arrests, but the differences were not statistically significant (p = 0.167 for all arrests; p = 0.061 for violent arrests). Deadly offenders averaged slightly more prior convictions and cautions overall (18.23 versus 17.79; p = 0.916), while the difference in violent convictions and cautions was also not significant (p = 0.091). (LLM, conf=0.90)
Checked by a reasoning model — may be wrong.
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“while an algorithm based upon police records was far more successful (AUC ≈ 0.75).”
§ The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse > Previous Research on DASH
What the sources say
- Supportedcites [9] Turner, 2022, Predicting domestic abuse (fairly) and police risk assessment, Psychosoc. Interv. source
In “DASH vs. Our Model,” the paper reports mean AUCs of .748 for IPV and .763 for non-IPV using variables that exclude DASH, compared with .567 and .551 for DASH risk grading. It also says removing DASH made a negligible difference to model performance, supporting the comparison with an algorithm based on other police data. (LLM, conf=0.94)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“They also found that the DASH assessment had little predictive accuracy (AUC ≈ 0.55), while an algorithm based upon police records was far more successful (AUC ≈ 0.75).”
§ The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse > Previous Research on DASH
What the sources say
- Supportedcites [9] Turner, 2022, Predicting domestic abuse (fairly) and police risk assessment, Psychosoc. Interv. source
In “DASH vs. Our Model,” the paper reports mean AUCs of .567 for IPV and .551 for non-IPV using DASH risk grading, compared with .748 and .763 for its predictive model. Those figures support the citing sentence’s approximate values of .55 and .75. (LLM, conf=0.94)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“The present results therefore differ from previous examinations of the DASH that failed to find that it had any predictive validity [8][9][10][11][12].”
§ The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse > 4. Discussion
What the sources say
- Partly supportedcites [8] Thornton, 2017, Police attempts to predict domestic murder and serious assaults: Is early warning possible yet?, Camb. J. Evid. Based Polic. sourceExperimental — accuracy not yet measured
The Discussion (page not available) reports that DASH and its predecessor did not accurately assess risk, with high false-negative and false-positive rates for deadly domestic assaults. That supports poor predictive performance in the setting studied, but not the broader assertion that DASH had no predictive validity for any outcome. (LLM, conf=0.86)
- Partly supportedcites [9] Turner, 2022, Predicting domestic abuse (fairly) and police risk assessment, Psychosoc. Interv. sourceExperimental — accuracy not yet measured
In “DASH vs. Our Model” (page unavailable), Turner et al. report that DASH contributes almost nothing to their predictive model and that officer risk grading performs poorly. But “almost nothing” is not “no predictive validity”: DASH risk grading alone had AUCs of .567 for IPV and .551 for non-IPV, both above chance. The cited paper supports poor predictive performance, not the absolute claim that DASH had no predictive validity. (LLM, conf=0.84)
- Not checked — no full text of the sourcecites [10] Turner, 2019, Dashing hopes? The predictive accuracy of domestic abuse risk assessment by police, Br. J. Criminol. source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Limiting studies to those that looked at severe/near fatal violence (as in the present study) produced an AUC of 0.66.”
§ The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse > 4. Discussion
What the sources say
- Supportedcites [20] Gubbels, 2019, Predicting domestic violence: A meta-analysis on the predictive validity of risk assessment tools, Aggress. Violent Behav. source
Table 3 (indexed under 5. Conclusion; page unavailable) lists “Severe/near fatal” as a category for the type of domestic violence predicted and gives its mean AUC as 0.661. Rounded to two decimal places, this supports the cited AUC of 0.66 for that subgroup. (LLM, conf=0.91)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
77 more claims · 4 with their cited sources checked
Create a free account to see all 114 claims and to verify the 12 references still open.
- Probably not a substantial claim. 16
- It may state two separate claims as one. 22
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse > •
“The DASH-SF was also a predictor of persistent offending (AUC = 0.71, 95% CI [0.64, 0.75], p < 0.001)”
§ The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse > 4. Discussion
“DASH-SF: AUC = 0.70, DASH-CV: AUC = 0.74”
Citations
Cites
The meaning and use of the area under a receiver operating characteristic (ROC) curve.
Hanley, J. A., & McNeil, B. J. (1982) · Radiology · doi:10.1148/radiology.143.1.7063747 · Cited by 21,856 · cited 1×
“Magnitudes of AUC were compared via the methods described by Hanley and McNeil [17,18].”
Hanley, J. A., & McNeil, B. J. (1983) · Radiology · doi:10.1148/radiology.148.3.6878708 · Cited by 7,016 · cited 1×
“Magnitudes of AUC were compared via the methods described by Hanley and McNeil [17,18].”
Comparing effect sizes in follow-up studies: ROC Area, Cohen's d, and r.
Rice, M. E., & Harris, G. T. (2005) · Law and Human Behavior · doi:10.1007/s10979-005-6832-7 · Cited by 1,608 · cited 4×
“The AUC is the widely accepted measure of predictive accuracy [13] as it is unaffected by base rates (and can therefore be used to compare performance for both rare and frequent events), different selection ratios, and non-normal distributions [14].”
InfluentialStrategy of prevention: lessons from cardiovascular disease.
Rose, G. (1981) · BMJ · doi:10.1136/bmj.282.6279.1847 · Cited by 1,302 · cited 1×
“The low risk paradox has been identified in many risk assessment schemes and is known as the “prevention paradox” [28].”
Psychological Science Can Improve Diagnostic Decisions
Swets, J. A., Dawes, R. M., & Monahan, J. (2000) · Psychological Science in the Public Interest · doi:10.1111/1529-1006.001 · Cited by 868
Campbell, J. C., Webster, D. W., & Glass, N. (2009) · Journal of Interpersonal Violence · doi:10.1177/0886260508317180 · Cited by 650 · cited 1×
“Looking at some of the most used and studied domestic violence risk instruments across the world, they showed that the DVSI [21] had an AUC of 0.61, the B-SAFER [22] an AUC of 0.60, the DA [23] an AUC of 0.66, the ODARA [24] an AUC of 0.69, and the SARA [25] an AUC of 0.64.”
InfluentialHilton, N. Z., Harris, G. T., Rice, M. E., Houghton, R. E., & Eke, A. W. (2008) · Law and Human Behavior · doi:10.1007/s10979-007-9088-6 · Cited by 178 · cited 1×
“Looking at some of the most used and studied domestic violence risk instruments across the world, they showed that the DVSI [21] had an AUC of 0.61, the B-SAFER [22] an AUC of 0.60, the DA [23] an AUC of 0.66, the ODARA [24] an AUC of 0.69, and the SARA [25] an AUC of 0.64.”
InfluentialA coffee can, factor analysis, and prediction of antisocial behavior: The structure of criminal risk
Kroner, D. G., Mills, J. F., & Reddon, J. R. (2005) · International Journal of Law and Psychiatry · doi:10.1016/j.ijlp.2004.01.011 · Cited by 173 · cited 1×
“Much research (see, for example, Kroner et al. [19]) has shown that criminological variables are strong predictors of future violence.”
The “Golden Thread”: Coercive Control and Risk Assessment for Domestic Violence
Myhill, A., & Hohl, K. (2019) · Journal of Interpersonal Violence · doi:10.1177/0886260516675464 · Cited by 168
Assessing the Risk of Domestic Violence Reoffending: A Validation Study.
Williams, K. R., & Houghton, A. B. (2004) · Law and Human Behavior · doi:10.1023/b:lahu.0000039334.59297.f0 · Cited by 137 · cited 1×
“Looking at some of the most used and studied domestic violence risk instruments across the world, they showed that the DVSI [21] had an AUC of 0.61, the B-SAFER [22] an AUC of 0.60, the DA [23] an AUC of 0.66, the ODARA [24] an AUC of 0.69, and the SARA [25] an AUC of 0.64.”
Influential
Research works printed but not linked: 1
- Thornton, S. (2011). Predicting Serious Domestic Assaults and Murder in the Thames Valley. [Master’s Thesis, University of Cambridge].
Cited by
(2026) · doi:10.62379/jishs.v4i4.4345 · Cited by 2
“Perkara yang tumpang tindih memerlukan keputusan jalur dominan, bukan kompromi yang mencampur indikator secara serampangan, sehingga elemen lintas jalur tetap dapat diperiksa sebagai faktor pemberat tanpa mengacaukan struktur pembuktian unsur (Caulfield et al., 2025).”
Metadata
- Authors
- Caulfield, K. M., Gray, N. S., Edwards, A., & Snowden, R. J.
- Affiliations
- Swansea University (GB)
- Swansea Bay University Health Board (GB)
- Dyfed-Powys Police Force, Llangunnor, Carmarthen SA31 2PF, UK
- Cardiff University (GB)
- Corresponding author
- Snowden, Robert J.
- Venue
- Forensic Sciences
- Publisher
- MDPI AG
- Published
- 2025-11-21
- Volume
- 5
- Issue
- 4
- Pages
- 64
- ISSN
- 2673-6756 (ISSN-L)
- OpenAlex
- W4416458455
- Cited by
- 0
- References
- 22
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Forensic Sciences (publishedVersion, cc-by)
- ORCA Online Research @Cardiff (Cardiff University) (acceptedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Dyfed-Powys Domestic Abuse Gold Group
- Swansea University
- Topics
- Social Sciences › Health › Intimate Partner and Family Violence
- Social Sciences › Sociology and Political Science › Homicide, Infanticide, and Child Abuse
- Social Sciences › Psychology › Clinical Psychology › Child Abuse and Trauma
- Keywords
- domestic violence, stalking, harassment, honour-related violence, risk assessment, prediction accuracy, signal detection theory, risk assessment validity
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse. https://app.cochat.ai/papers/10.3390/forensicsci5040064
- The paper
- Caulfield, K. M., Gray, N. S., Edwards, A., & Snowden, R. J. (2025). The Domestic Abuse, Stalking and Harassment and Honour-Based Violence (DASH) Risk Assessment Instrument in Predicting Deadly or Persistent Domestic Abuse. Forensic Sciences. https://doi.org/10.3390/forensicsci5040064
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.