The M-FLAME© Methodology is the world's leading edge technology in firefighter selection. Designed as a post-offer, pre-employment liability assessment, the M-FLAME© is an actuarial methodology for the prediction of each of the 18 liabilities in firefighting. The M-FLAME© statistically combines a candidate's background/historical data, interview and observational data, and psychometric data to compare the pattern of responding to that of officers who have demonstrated job performance liabilities.
Actuarial decision-making has always proven superior to impressionistic judgement in the prediction of discrete outcomes. Matrix Inc. has made actuarial methods possible through the unique program of obtaining accurate, objective supervisory ratings of actual job performance for every officer at various intervals throughout their career. These serial feedbacks involve 54 variables ranging from the type and frequency of distinct categories of misconduct (e.g., excessive force, sexually inappropriate behavior, racially offensive conduct, at-fault motor vehicle accidents, etc), the citizen response to misconduct (e.g., formal complaints, claims filed, lawsuits, etc.) and the departmental response to misconduct (e.g., written reprimands, suspensions, terminations, arrest of officer, etc.).
The M-FLAME© methodology battery includes several standardized measures collected from each candidate.
The data includes 26 demographic variables, 78 biographical/historical variables (deemed relevant to police conduct from scholarly research and federal case law), 22 structured interview variables, 16 observational variables, and over 300 psychometric variables. A conservative, flexible statistical methodology analyzes a given candidate's pattern of responding to provide an understanding of the exact probabilities (and known error rates) of risk a candidate poses across each of the 18 liabilites in firefighting.
The statistical model includes factor analytic data reduction techniques, and intercorrelational matrices of the factors involved and the 18 outcome liabilities. Those factors statistically significantly related to each outcome are loaded into linear regression models to dichotomous and continuous forms of each outcome variable as well as predictive discriminant analyses. Mean predictions of forward and backward stepping regressions and mean correct classification values in predictive discriminants (after randomized hold-outs for cross validation) provide the foundation of understanding the liabilitiy potiential a candidate poses to public safety and security. (see Evolving Normative Prediciton Model)
An automated report production facility allows for the near instantaneous availability of the M-FLAME© Report. Several sample reports are available here: (you will need Adobe Reader to view these sample reports)
If you would like more information on how Matrix, Inc. can help you address questions like these, then please contact us.