Psychometric prediction of police misconduct Dr. Robert Davis Matrix Incorporated, Baton Rouge Dr. Gill Sitarenios MHS Inc. International Association of Chiefs of Police (IACP) September, 2005

1
Psychometric prediction of police misconduct
Dr. Robert Davis Matrix Incorporated, Baton
Rouge Dr. Gill SitareniosMHS
Inc.International Association of Chiefs of
Police (IACP)September, 2005
2

• 500 item self-report measure.
• Assesses personality traits and attitudes
  pertinent to the law enforcement culture
  and profession.
• Used in risk assessment for the pre-employment
  selection of law enforcement agents.
• 4 sets of scale scores are provided
• 4 Validity Scales
• 10 POST Dimensions
• 15 Empirical Scales
• 18 Outcome Scales

3
Sample Items
On the streets, respect is so important that
anyone who threatens or hits an officer must be
punished on the spot. Suspects expect to be
roughed up a little during an arrest. Members
of certain ethnic groups are more likely to
commit crimes than others. An officer should
not complain about the behavior of a fellow
officer. If an order given by a superior
officer violates the written rules, I would
follow the written rules.
4
Validity Scales

• VirtuosityResponds in socially desirable manner
• Test AttitudeCooperation, etc.
• Non-Disclosure
• Overly defensive, withholding information
• Deceit
• Lying, misrepresentation
• .

5
POST Dimensions

• 10 scales were rationally derived by mapping the
  M-PULSE items onto the California POST
  Dimensions.
• These scales were assessed psychometrically and
  revised until adequate levels of reliability
  (internal consistency) were achieved.

6
POST Dimensions (contd)

• Social Incompetence
• Lack of Teamwork
• Rigidity
• Unreliability
• Reckless Impulsivity
• Lack of Integrity/Ethics
• Emotional Instability/Stress Intolerance
• Poor Decision-Making Judgment
• Passivity-Submissiveness
• Substance Abuse

7
POST Dimension Reliabilities
8
Empirical Scales

• The M-PULSE items were analyzed to determine if
  item groups form meaningful clusters.
• FA, Principal Components Analysis, Varimax
  Rotation, factor loading gt .30.
• Resulted in 20 Empirical Scales 5 factors with
  3 subscales each.

9
Empirical Scales (contd)

• Negative Self-Perception
• Negative Emotions
• Negative Actions
• Negative Attitude
• Negative Perception Of Others
• Cynical Beliefs
• Oppositional Actions
• Alienated Attitudes
• Negative Perceptions Of Law Enforcement
• Inappropriate Attitudes About Use of Force
• Overly Traditional Officer Traits
• Suspiciousness

10
Empirical Scales (contd)

• Unethical Behavior
• Lack of Personal Integrity
• Perceived Lack of Integrity for the Department
• Amorality
• Unpredictability
• Risk Taking
• Competitiveness
• Novelty Seeking

11
Outcome Scales

• Outcome data was available for about
• 500 police officers in the database.
• 18 specific outcomes were tracked, such as
  inappropriate weapon use, excessive force, and
  criminal conduct.

12
Outcome Scales (contd)

• Criminal Conduct
• Potential for Termination
• Procedural and Conduct Mistakes
• Inappropriate Use of Weapon
• Chemical Abuse/Dependency
• Property Damage
• Excessive Force
• Interpersonal Difficulties
• Motor Vehicle Accidents
• Off-Duty Misconduct
• Misuse of Vehicle
• Potential for Resignation
• Racially Offensive Conduct
• Potential for Reprimand/Suspension
• Sexually Offensive Conduct
• Lawsuit Potential
• Unprofessional Conduct
• Discharge of Weapon

13
Outcome Scales

• As well as the 18 components, an overall BAD
  cop or not classification was given to each
  officer.
• Discriminant Function Analyses were used to
  determine best set of items to predict outcomes.

14
Outcome Scale Results
15
Outcome Scales The following graph shows the
results for the M-PULSE Inventory outcome
scales.
Positive Mid-Range Negative
16
Key Points

• Built specifically for police use (not retrofit)
  
• Scientific
• Shows reliability and validity in large data
  sets and predicts bad behavior.
• Objective
• Thats the point! Existing procedures are
  generally unable to quantify with any precision
  or metrics the extent to which they are related
  to policing outcome.
• And it gets better!
• M-PULSE methodology includes interview data and
  provides an actuarial prediction.

17
Thank you for Attending!
For more information on
Please visit the MHS booth in the exhibit
area, Or send e-mail correspondence to RD_at_mhs.com