Skip to main content

Improving Motor Carrier Safety Measurement (2017) / Chapter Skim
Currently Skimming:

5 Extensions of and Implementation of the Item Response Theory Model to the Safety Measurement System
Pages 97-104

The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.


From page 97...
... That is, the deterministic algorithm used to produce Behavior Analysis and Safety Improvement Categories (BASICs) does not take into account the sources of variation that are inherent to the data collection process (e.g., whether or not a carrier gets inspected on a particular trip)
Read the entire page →
From page 98...
... Instead, we propose using model output to quantify the uncertainty, and include the uncertainty as part of the ranking approach. Consider an example with 10 carriers and an assumption that the Federal Motor Carrier Safety Administration (FMCSA)
Read the entire page →
From page 99...
... On the other hand, if overall carrier safety in the industry improves, there might be interest in allocating resources to other safety initiatives. In this case, an absolute approach that would send fewer letters and initiate fewer investigations might be desired.
Read the entire page →
From page 100...
... corresponds to a hypothetical carrier (Carrier 1) with a bad safety score, qi and not a lot of uncertainty.
Read the entire page →
From page 101...
... produces an updated crash rate for a carrier given that carrier's estimated latent risk of a crash and its VMT, which can then serve as an absolute metric on which decisions for interventions can be based. While this has the look of a model of future crash risk, we are not proposing this as a model of crash risk, but as a construct that is dependent on qi , the latent safety parameter, that is an absolute measure of ˆ ­ arrier (Carrier 2)
Read the entire page →
From page 102...
... It would be worthwhile to also explore a hybrid measure that combines the proposed relative and absolute measures. STRATIFICATION VERSUS MODEL-BASED ADJUSTMENT Another important issue is being sure to compare sufficiently similar carriers and not to inappropriately compare different types of carriers.
Read the entire page →
From page 103...
... . IMPLEMENTATION ISSUES Clearly, the proposed Bayesian IRT model, which involves use of 20–30 million observations and hundreds of variables to estimate hundreds of model parameters is something that requires very specific expertise, usually found in academic statisticians who carry out research on these specific models.
Read the entire page →
From page 104...
... 104 IMPROVING MOTOR CARRIER SAFETY MEASUREMENT that the IRT model, like SMS, is sensitive to outlying values in MCMIS, FMCSA should consider institution of edit routines to identify discrepant submissions, and imputation procedures to fill in for input values that are missing or fail the edits.
Read the entire page →

Key Terms

  • carrier safety
  • motor carrier
  • safety measurement
  • safety scores
  • hypothetical carriers

  • hypothetical carrier
  • absolute approach
  • safety administration
  • bad safety
  • crash rate

  • inspection rate
  • crash risk
  • poisson regression
  • newly proposed
  • proposed ranking

  • future outcome
  • federal motor
  • irt model
  • carrier characteristics
  • true safety

  • assign safety
  • current safety
  • latent safety
  • safety parameter
  • safety initiatives

  • safety improvement
  • worst safety
  • mean safety
  • safety improvements
  • improve safety


    • This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
    More information on Chapter Skim is available.