CHAPTER 5

Conclusions

The findings of this synthesis project suggest that a majority of state DOTs are currently customizing the methods in the HSM for use within their state. The DOTs are generally evenly split between developing/applying calibration factors to HSM SPFs versus developing state-specific SPFs. The most common HSM SPFs that are being calibrated are design-level SPFs for two-lane rural roadway segments and intersections, multi-lane rural roadway segments and intersections, and urban–suburban arterial roadway segments and intersections. Freeway segment and ramp SPFs are generally applied less often. For state-specific SPFs, however, network-screening-level SPFs are more commonly developed than design-level SPFs (see Table 84). As the focus of this NCHRP synthesis report was on state DOT practices, no information was sought on customization at a city or MPO level.

The primary drivers of whether responding DOTs choose to calibrate HSM SPFs or develop state-specific SPFs, as well as the type of SPF to apply (network-screening-level or design-level), are the level of precision needed from the SPFs, data availability, and resource constraints. These factors are also indicated as the biggest barriers to the development of state-specific SPFs. DOTs that developed and applied calibration factors to HSM SPFs generally indicated that doing so provides predictions that are accurate enough and that provide the best value for resources available. DOTs that developed their own SPFs generally indicated that they did so because additional precision was needed in the predictions and because sufficient resources were available.

For DOTs calibrating the HSM SPFs, almost all (71%, 15 of 19 respondents) indicated using the calibration factor defined in the HSM. The only other alternative noted was the calibration function, which is being used by one (of 19 or 5%) of responding DOTs. These calibration factors are generally developed using spreadsheet-based tools or manual calculations. Only 13% (four of 31) of responding DOTs indicated using the FHWA Calibrator Tool.

Half of the responding DOTs have not updated calibration factors after their original development. Very few DOTs (14%, or three of 22 respondents) indicated that they update their calibration factors very frequently (more often than every three years). State-specific SPFs are generally updated more frequently than calibration factors: only one-third (33%, or nine of 27) of responding DOTs indicated that they had not been updated since they were developed while 26% (or seven of 37) of responding DOTs indicated that their SPFs were updated very frequently (more often than every three years). The difference in these update intervals may be due to the difference in SPF type between calibrated and state-specific SPFs as DOTs are more likely to both calibrate design-level SPFs and develop their own network-screening-level SPFs.

With regard to regionalization, less than half of responding DOTs (43%, or nine of 21) indicated that they always or sometimes used different calibration factors to account for differences in safety performance across individual regions of their state. The literature review indicated that some DOTs tried regionalization but found no real differences and thus opted for a single calibration

factor. However, in some cases, insufficient sample size was a driver in the lack of regionalization. Approximately 26% (or seven of 27) of responding state DOTs indicated that regional differences were accounted for in state-specific SPFs.

The range of sample sizes range considerably for SPF calibration and development activities across individual projects; see Chapter 2 for a summary of individual studies. However, it seems that most DOTs attempt, when possible, to achieve minimum sample sizes as suggested in the HSM and research literature. Exceptions are generally for cases where a sufficient sample does not exist within a given agency.

A particularly noteworthy finding is the impact of the upcoming second edition of the Highway Safety Manual (HSM2) on SPF calibration/development efforts. The case examples revealed that states are waiting for the publication of HSM2 to pursue additional customization activities.

Finally, the results indicate that state DOTs do not borrow or apply calibration factors or SPFs from other states. A key reason is the differences across states in data collection and crash reporting.

Opportunities for potential future research include:

• The responses to several survey questions could not be differentiated for network-screening-level and design-level SPFs. Future survey work on this topic should be more careful to distinguish between these two on issues like SPF development cost, update interval, and so on.
• The wide range of calibration and SPF development updates suggests that states do not have sufficient guidance on when calibration factors or SPFs should be updated. Additional research to provide assistance to state DOTs would be beneficial.
• Development of a repository of state-specific SPFs and/or calibration factors is desirable. Such a repository would help states better understand what design features are being considered or how other states are applying SPFs, which could help create some consistency in the state-specific SPF development. This repository could be something as simple as a spreadsheet that is shared among state DOTs (e.g., through the AASHTO Committee on Safety) and updated regularly by individual state representatives or—preferably—a public-facing and searchable “living” online repository, similar to the FHWA CMF Clearinghouse. Within either form, the repository should have a standard format that state DOTs use to provide summary information of SPF calibration and/or SPF development effort for each SPF that is performed. The following information might be considered:
  • Facility type
  • Sample size, both number of unique sites and total mileage (for roadway segments)
  • Total number of crashes observed during study period
  • Analysis years
  • For calibration factors:
    • Calibration method
    • Type of SPF calibrated (network-screening-level or design-level)
    • Calibration factor
  • For SPFs:
    • Type of SPF developed (network-screening-level or design-level)
    • What adjustment factors were included
    • Statistical method used
    • SPF equation
    • Range of applicable traffic volumes
  • URL link to report documenting calibration and/or SPF development process