discriminative analysis

discriminative analysis - Nonfatal Workplace Violence risk...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Nonfatal Workplace Violence risk factors: Data From A Police Contact Sample Mario Scalora David o'Neil Washington Thomas Casady Sarah Newell Alicia Deal CJ 737 04-05-06 Research Questions/Purpose Identify risk factors relevant to liability of Nonfatal assaults in the workplace If risk factors are known then employees can utilize several sources of legal liability to prevent workplace violence Are potential risk "foreseeable" across sources (either internal or external) to the work site? Effect-Danger Ratio Effect-danger ratio: interaction b/t work-site contextual factors & aggressor estimates of consequences of harm may explain the nonfatal workplace violence The intensity of violence is related to the aggressors' subjective estimates of the ability to perform harmful behaviors while permitting as little danger to themselves as possible Ex. Aggressors w/ less subsequent contact w/ victims-more likely to engage in more intense forms of violent behavior (robbery) Not supported in this research Sample: Data & Method N=281 cases Lincoln, Nebraska law enforcement agency 18 month period ('97-'98) Lincoln police department computer database local law enforcement databases incident repots & victim statements police contacts: ranged from obscene phone call complaints to physical assaults & location codes: referred to businesses where such acts occurred Info. Collected from: Targeted Excluded: cases not directly involving employees as victims or any harassing or threatening behavior Independently coded 20% of 281 cases in an overlapping fashion to assess coding reliability Reliability: .90 & Kappa coefficient: > or equal .91 Variables Dependent Variable: Internal threat: conflicts b/t employees or domestic struggles (does not include business relationships) External threat: threats posed by persons either not employed by or significantly related to persons employed w/I the workplace (customer or stranger) Independent Variables Perpetrator-victim relationship (stranger or nonstranger: acquaintance/ consumer) Mention or use of a weapon Whether threats had been made The intensity level of violence Prior police contact Motives of the incidents Monetary (monetary dispute or billing issue) Relationship (relationship or domestic disputes) Sexual (sexualized language or attempted sexual contact) Results: Overall No fatal workplace violence during the investigated time span Motives: "perceived mistreatment" (32%) 13.2% monetary dispute Perps acquainted w/ victim & seeking retaliation or reconciliation of differences Relationship: 21% Sexualized motives from non-intimates: 12.5% No motive: 21% Nearly half were physically assaulted 61.5%: detectable or medically verified injury 1/3: internal or other more serious injuries Bruises (10.4%); lacerations (17.7%) Overall 31.3%: victims were threatened before incident Prior threat not related to assaultive behavior Perps of threats sign. less likely to have engaged in assaultive behavior May have some of the threats triggered a response on their own accord prior to further escalation vics of physical aggression (p<.0001) threatened w/ weapon (p<.0001) suffer more serious injury Less likely to report lesser forms of violence? Male vics (48%) were more likely to: Equal #s of men & women reported prior threats Related to assaultive behavior: Perps who had prior contact w/ police (48% ); p<.0001 Table 1: Demographic characteristics Perp Sex: Male female Perp mean age Relationship to vic: Stranger Nonstranger Services recipient Prior police contact Mental ill suspected Motives: Job dissat Relationship Monetary dispute Customer dispute Victim: Male female Mean age Prior threat Use of weapon Witness present Location: retail Nonretail Government External threat Assault (%) 75.7 24.3 26.97 55.7 44.3 47.1 60 17.1 External threat: Nonassault (%) 19.3*** 80.7 31 77.3** 22.7 5.7*** 26.1*** 5.7* Internal threat: Assault (%) 82.4 17.6 33.27 Internal threat Nonassault (%) 77.8 22.2 34.93 68.6 7.8 33.3 43.1 48.9* 26.7** 13.3* 73.3** 10 75.7 75.7 24.3 32.59 10 8.6 65.7 40 45.7 14.3 5.7 19.3*** 19.3*** 80.7 33.10 40.9*** 1.1* 15.9*** 43.2 51.1 5.7 51 49 30.68 21.6 15.7 60.8 21.6 72.5 5.9 17.8*** 82.2 32 55.6*** 0.0** 33.3** 40 48.9 11 Demo. Results: External Sources N=189 (67.2%) 79.3% (n=150) involved parties w/ "no legitimate relationship" to the workplace (robbery) or customer service disputes Remaining 39: "job dutyrelated" (mental health worker assaulted by patient) Age did not sign. differ b/t assault & nonassault Men more likely to be assaulted (p<.0001) 32.3% nonstranger Location Overall cases: not related to assaults Monetary disputes (7.6%): no greater risk of assault More likely to engage in assaultive behavior if: mentally ill (10.8% overall); p<.02 perps w/ prior records (41.1% overall); p<.0001 Acquainted w/ victim; p<.0004 by service recipients (nearly half) p<.0001 presence of weapons (4.4%) p<.02 Prior threats p<..0001 Customer service disputes Overall 44.7% p<.0001 Retail: 41/8% Nonretail: 48.7% Government9.5% Witnesses present: 38% overall cases; p<.0001 Differential Analysis: External Threats o o o o o Examined assault vs.. nonassault reliably differentiated among the groups X2 (12)=131.308, p,.0001, r2 canonical=.583 Correctly reclassified 86.7% of the external cases, 84.3% assault, 88.6% nonassault (Table 3) Canonical discriminant: nonassault: -1.049; assault: 1.318 (Figure 1) o Provides differentiation b/t assault & nonassault groups o o Variables more likely among assaults: o Men more likely to be victims o Customer service dispute o Perps w/ prior criminal history Threats less likely to precede assaults Table 2: Standardized structure weights customer service dissat. motive Coworker relationship Domestic violence motive Victim gender Witness present Sexual contact motive Prior threats made Prior police contact (perp) Stranger relationship to vic Mental illness suspected (perp) Use of weapon Retail vs./ nonretail Perp gender Monetary dispute External Threat .577 .577 .501 -.374 -.311 .307 -.199 .158 .154 -.027 .016 .087 Demo. Results: Internal Sources 32.7% (n=92) 2/3 domestic violence component (former boyfriend or ex-husband threatening female worker) Women vics: overall 64.6% Perp: mostly men Age didn't differ b/t groups Not related to assaults: Location: nonretail (61.5%); retail (30.2%) yet higher likelihood in nonretail settings Assaults more likely if: Witnesses present: 47.9% overall cases p<.007 Male victims p<.001 Perps w/ prior criminal records(59.4%); p<.05 Relationship motive (57.3%); p<.003 Job dissat. Motive: (24%); p<.002 Assaults less likely if: Mentally ill (overall 16.7%) p<.014 Prior threats p<.001 Discriminant Analysis: Internal Threats Examined assault vs.. nonassault Classified correctly: 83.3% overall, 82.2% nonassault, 84.3% assault (Table 4) Canonical discriminant (figure 2): Provides reasonable differentiation b/t assault & nonassault groups nonassault: -.940 Assault: .829 Reliably different: x2 (10)=52.058. p<.0001, R2 canonical=.443 variables more likely among assaults: Coworkers more likely to engage in assaults Male victims Presence of weapons Witnesses present Prior threats less likely among assaults Table 2: Standardized structure weights Customer service dissat. motive Coworker relationship Domestic violence motive Victim gender Witness present Sexual contact motive Prior threats made Prior police contact (perp) Stranger relationship to vic Mental illness suspected (perp) Use of weapon Retail vs./ nonretail Perp gender Monetary dispute Internal Threat .501 -.455 .414 .320 -.419 .230 -.292 .331 -.229 .064 Discussion Generally consistent w/ past research: Men more likely to be assaulted by strangers & threatened w/ a weapon History of prior violence among perps risk factor for assaults (both external & internal) Higher assaults among non-retail locations vs.. retail Prior threat was inversely related to physical aggression & more common among nonassaultive threats among both internal & external yet a substantial # still accumulated in assaults-suggests an escalating process 2/3 internal involve domestic violence-need for employers to assistance & support, better training, preventive strategies Implications Yet differences included: Men more likely to be assaulted than women across both internal & external threats even though women more likely to be overall victims Most intense levels of violence were in the presence of witnessestherefore can't assume "safety in #s as minimizing risk Findings on motives suggests a need for review of training practices on how to diffuse conflict More assaults among men & greater prior threat in nonassaults suggests different reporting thresholds for men & women Mental illness: identifiable signs more prevalent in internal situations, were more related to assault among external situations Need for employee training & employee assistance or other supportive services Expansion of community-policing approach to include workplace violence Limitations Consisted only of info. provided from police department records The documentation of risk factors studied was sometimes difficult given the variable levels of detail related to documentation of key factors across police reports Ex. Data regarding factors such as motivations & mental illness were generally limited to 3rd party descriptions, perps self-reported statements Limited in details reported by the officers Critique No mention of missing data, linearity & normality (scatter plots), outliers Doesn't state how they coded the variables: i.e. Male: 0; female-1 Barely discussed discriminant analysis: Did not discuss group means, ANOVA, Wilks' Lambda, eigenvalues, Box's M (homoscedasticity), structure matrix, log determinants (homogeneity), w/i groups matrices, functions entered into model, etc. Did they use rotation method, if so, what kind? Why only use discriminant analysis for assault & nonassault analysis- could've used it also for external & internal threat groups Doesn't really discuss the 2 figures (canonical discriminants) No mention of other contributing factors such as the presence of drug/alcohol use Argue that prior threats may not be as common among assaultive behavior as nonassaultive behavior, yet are prevalent enough to escalate into violence--they only make this statement & do not back it up w/ evidence or provide examples, etc. ...
View Full Document

This note was uploaded on 06/30/2011 for the course CJ 737 taught by Professor Tittertington during the Spring '05 term at Sam Houston State University.

Ask a homework question - tutors are online