Standard Errors

Safety performance functions, development (calibration, crash modification factors [Part C], crash proportions), IHSDM and other software

Moderator: khardy

Standard Errors

Postby dengstro » Fri Apr 26, 2013 2:13 pm

Are there standard errors for the SPF's and CMF's contained in Part C of the HSM? How would that relate to calibration?
Posts: 12
Joined: Wed Feb 16, 2011 3:10 pm

Re: Standard Errors

Postby mdimaiuta » Fri May 03, 2013 1:48 pm


As noted in the HSM Introduction to Part C (p. C-17), “Some of the CMFs in Part C include a standard error value, indicating the variability of the CMF estimation in relation to sample data values.” All Part C CMFs are contained in Part D; where standard errors are known, they are shown in Part D, but not in Part C. So, some cross-referencing is necessary to find the standard errors.

An example of a CMF which does not have a standard error is CMF1r (Lane Width) in Chapter 10. The same CMF is shown in Chapter 13, p. 13-3 and 13-4. Footnotes in Table 13-2 and Figure 13-1 note that “Standard error of the CMF is unknown.”

An example of a CMF which does have a standard error is CMF2i (Intersection Left-Turn Lanes), for three-leg intersections with minor road stop control (Chapter 10). In Table 10-13 (p. 10-32), the CMF is shown as 0.56 for one approach with left-turn lane. Chapter 14, p. 14-21, Table 14-10 shows the same CMF with a standard error of 0.07.

There are no standard errors for the SPFs in Part C. However, each SPF does have an overdispersion parameter. As noted in the Introduction to Part C (p. C-11), “The overdispersion parameter provides an indication of the statistical reliability of the SPF. The closer the overdispersion parameter is to zero, the more statistically reliable the SPF. This parameter is used in the site-specific Empirical-Bayes (EB) Method to provide a weighting to Npredicted and Nobserved. Overdispersion parameters are provided for each SPF in the Part C chapters.”

This does not effect calibration, in that no matter which model is being calibrated, the procedure is the same - as described in Appendix A to Part C.

Mike Dimaiuta
Posts: 29
Joined: Tue Nov 30, 2010 1:50 pm

Return to Part C - Predictive Methods

Who is online

Users browsing this forum: No registered users and 1 guest