Using offender crime scene behavior to link stranger sexual assaults: A comparison of three statistical approaches

M. Tonkin, T. Pakkanen, J. Sirén, C. Bennell, J. Woodhams, A. Burrell, H. Imre, J. M. Winter, E. Lam, G. ten Brinke, M. Webb, G. N. Labuschagne, L. Ashmore-Hills, J. J. van der Kemp, S. Lipponen, L. Rainbow, C. G. Salfati, P. Santtila

Research output: Contribution to JournalArticleAcademicpeer-review

Abstract

Purpose This study compared the utility of different statistical methods in differentiating sexual crimes committed by the same person from sexual crimes committed by different persons. Methods Logistic regression, iterative classification tree (ICT), and Bayesian analysis were applied to a dataset of 3,364 solved, unsolved, serial, and apparent one-off sexual assaults committed in five countries. Receiver Operating Characteristic analysis was used to compare the statistical approaches. Results All approaches achieved statistically significant levels of discrimination accuracy. Two out of three Bayesian methods achieved a statistically higher level of accuracy (Areas Under the Curve [AUC]=0.89 [Bayesian coding method 1]; AUC=0.91 [Bayesian coding method 3]) than ICT analysis (AUC=0.88), logistic regression (AUC=0.87), and Bayesian coding method 2 (AUC=0.86). Conclusions The ability to capture/utilize between-offender differences in behavioral consistency appear to be of benefit when linking sexual offenses. Statistical approaches that utilize individual offender behaviors when generating crime linkage predictions may be preferable to approaches that rely on a single summary score of behavioral similarity. Crime linkage decision-support tools should incorporate a range of statistical methods and future research must compare these methods in terms of accuracy, usability, and suitability for practice.

Original languageEnglish
Pages (from-to)19-28
Number of pages10
JournalJournal of Criminal Justice
Volume50
DOIs
Publication statusPublished - 1 May 2017

Fingerprint

Bayes Theorem
Crime
assault
Area Under Curve
offender
offense
coding
statistical method
Logistic Models
logistics
sexual offense
regression
human being
ROC Curve
discrimination
recipient
ability

Keywords

  • Bayesian analysis
  • Classification tree analysis
  • Comparative case analysis
  • Crime linkage
  • Logistic regression
  • Stranger sexual assault

Cite this

Tonkin, M. ; Pakkanen, T. ; Sirén, J. ; Bennell, C. ; Woodhams, J. ; Burrell, A. ; Imre, H. ; Winter, J. M. ; Lam, E. ; ten Brinke, G. ; Webb, M. ; Labuschagne, G. N. ; Ashmore-Hills, L. ; van der Kemp, J. J. ; Lipponen, S. ; Rainbow, L. ; Salfati, C. G. ; Santtila, P. / Using offender crime scene behavior to link stranger sexual assaults : A comparison of three statistical approaches. In: Journal of Criminal Justice. 2017 ; Vol. 50. pp. 19-28.
@article{97c241b6e00d45e8b55effe1edd09d86,
title = "Using offender crime scene behavior to link stranger sexual assaults: A comparison of three statistical approaches",
abstract = "Purpose This study compared the utility of different statistical methods in differentiating sexual crimes committed by the same person from sexual crimes committed by different persons. Methods Logistic regression, iterative classification tree (ICT), and Bayesian analysis were applied to a dataset of 3,364 solved, unsolved, serial, and apparent one-off sexual assaults committed in five countries. Receiver Operating Characteristic analysis was used to compare the statistical approaches. Results All approaches achieved statistically significant levels of discrimination accuracy. Two out of three Bayesian methods achieved a statistically higher level of accuracy (Areas Under the Curve [AUC]=0.89 [Bayesian coding method 1]; AUC=0.91 [Bayesian coding method 3]) than ICT analysis (AUC=0.88), logistic regression (AUC=0.87), and Bayesian coding method 2 (AUC=0.86). Conclusions The ability to capture/utilize between-offender differences in behavioral consistency appear to be of benefit when linking sexual offenses. Statistical approaches that utilize individual offender behaviors when generating crime linkage predictions may be preferable to approaches that rely on a single summary score of behavioral similarity. Crime linkage decision-support tools should incorporate a range of statistical methods and future research must compare these methods in terms of accuracy, usability, and suitability for practice.",
keywords = "Bayesian analysis, Classification tree analysis, Comparative case analysis, Crime linkage, Logistic regression, Stranger sexual assault",
author = "M. Tonkin and T. Pakkanen and J. Sir{\'e}n and C. Bennell and J. Woodhams and A. Burrell and H. Imre and Winter, {J. M.} and E. Lam and {ten Brinke}, G. and M. Webb and Labuschagne, {G. N.} and L. Ashmore-Hills and {van der Kemp}, {J. J.} and S. Lipponen and L. Rainbow and Salfati, {C. G.} and P. Santtila",
year = "2017",
month = "5",
day = "1",
doi = "10.1016/j.jcrimjus.2017.04.002",
language = "English",
volume = "50",
pages = "19--28",
journal = "Journal of Criminal Justice",
issn = "0047-2352",
publisher = "Elsevier BV",

}

Tonkin, M, Pakkanen, T, Sirén, J, Bennell, C, Woodhams, J, Burrell, A, Imre, H, Winter, JM, Lam, E, ten Brinke, G, Webb, M, Labuschagne, GN, Ashmore-Hills, L, van der Kemp, JJ, Lipponen, S, Rainbow, L, Salfati, CG & Santtila, P 2017, 'Using offender crime scene behavior to link stranger sexual assaults: A comparison of three statistical approaches' Journal of Criminal Justice, vol. 50, pp. 19-28. https://doi.org/10.1016/j.jcrimjus.2017.04.002

Using offender crime scene behavior to link stranger sexual assaults : A comparison of three statistical approaches. / Tonkin, M.; Pakkanen, T.; Sirén, J.; Bennell, C.; Woodhams, J.; Burrell, A.; Imre, H.; Winter, J. M.; Lam, E.; ten Brinke, G.; Webb, M.; Labuschagne, G. N.; Ashmore-Hills, L.; van der Kemp, J. J.; Lipponen, S.; Rainbow, L.; Salfati, C. G.; Santtila, P.

In: Journal of Criminal Justice, Vol. 50, 01.05.2017, p. 19-28.

Research output: Contribution to JournalArticleAcademicpeer-review

TY - JOUR

T1 - Using offender crime scene behavior to link stranger sexual assaults

T2 - A comparison of three statistical approaches

AU - Tonkin, M.

AU - Pakkanen, T.

AU - Sirén, J.

AU - Bennell, C.

AU - Woodhams, J.

AU - Burrell, A.

AU - Imre, H.

AU - Winter, J. M.

AU - Lam, E.

AU - ten Brinke, G.

AU - Webb, M.

AU - Labuschagne, G. N.

AU - Ashmore-Hills, L.

AU - van der Kemp, J. J.

AU - Lipponen, S.

AU - Rainbow, L.

AU - Salfati, C. G.

AU - Santtila, P.

PY - 2017/5/1

Y1 - 2017/5/1

N2 - Purpose This study compared the utility of different statistical methods in differentiating sexual crimes committed by the same person from sexual crimes committed by different persons. Methods Logistic regression, iterative classification tree (ICT), and Bayesian analysis were applied to a dataset of 3,364 solved, unsolved, serial, and apparent one-off sexual assaults committed in five countries. Receiver Operating Characteristic analysis was used to compare the statistical approaches. Results All approaches achieved statistically significant levels of discrimination accuracy. Two out of three Bayesian methods achieved a statistically higher level of accuracy (Areas Under the Curve [AUC]=0.89 [Bayesian coding method 1]; AUC=0.91 [Bayesian coding method 3]) than ICT analysis (AUC=0.88), logistic regression (AUC=0.87), and Bayesian coding method 2 (AUC=0.86). Conclusions The ability to capture/utilize between-offender differences in behavioral consistency appear to be of benefit when linking sexual offenses. Statistical approaches that utilize individual offender behaviors when generating crime linkage predictions may be preferable to approaches that rely on a single summary score of behavioral similarity. Crime linkage decision-support tools should incorporate a range of statistical methods and future research must compare these methods in terms of accuracy, usability, and suitability for practice.

AB - Purpose This study compared the utility of different statistical methods in differentiating sexual crimes committed by the same person from sexual crimes committed by different persons. Methods Logistic regression, iterative classification tree (ICT), and Bayesian analysis were applied to a dataset of 3,364 solved, unsolved, serial, and apparent one-off sexual assaults committed in five countries. Receiver Operating Characteristic analysis was used to compare the statistical approaches. Results All approaches achieved statistically significant levels of discrimination accuracy. Two out of three Bayesian methods achieved a statistically higher level of accuracy (Areas Under the Curve [AUC]=0.89 [Bayesian coding method 1]; AUC=0.91 [Bayesian coding method 3]) than ICT analysis (AUC=0.88), logistic regression (AUC=0.87), and Bayesian coding method 2 (AUC=0.86). Conclusions The ability to capture/utilize between-offender differences in behavioral consistency appear to be of benefit when linking sexual offenses. Statistical approaches that utilize individual offender behaviors when generating crime linkage predictions may be preferable to approaches that rely on a single summary score of behavioral similarity. Crime linkage decision-support tools should incorporate a range of statistical methods and future research must compare these methods in terms of accuracy, usability, and suitability for practice.

KW - Bayesian analysis

KW - Classification tree analysis

KW - Comparative case analysis

KW - Crime linkage

KW - Logistic regression

KW - Stranger sexual assault

UR - http://www.scopus.com/inward/record.url?scp=85017526859&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85017526859&partnerID=8YFLogxK

U2 - 10.1016/j.jcrimjus.2017.04.002

DO - 10.1016/j.jcrimjus.2017.04.002

M3 - Article

VL - 50

SP - 19

EP - 28

JO - Journal of Criminal Justice

JF - Journal of Criminal Justice

SN - 0047-2352

ER -