APPENDIX G. SUPPLEMENTAL INVESTIGATION OF AVERAGE LANE WIDTH, CENTER LINE MARKINGS, AND EDGE LINE MARKINGS

INTRODUCTION

The previous analyses (see Appendix F) could not answer questions regarding average lane width or pavement markings. Average lane width was not included in previous models because it provided counterintuitive results or was not significant. The previous analyses included segments where the absence of center line or edge line markings is not reasonable, such roads with higher ADT or multiple lanes. Therefore, additional analyses were conducted that used subsets of the Phase II database discussed in the previous appendix. These more focused analyses were designed to answer the following questions:

• Is operating speed sensitive to the average width of the lane?
• Does the presence of edge line markings influence operating speed?
• Does the presence of center line markings influence operating speed?

AVERAGE LANE WIDTH

The model coefficient for the average lane width variable in preliminary analyses (see Appendix F) had a counterintuitive sign or was not statistically significant. Therefore, the variable was not included in the evaluations that considered transportation professional variables or demographics variables. Because of continue interest in understanding how lane width affects operating speed or safety, the research team conducted additional evaluations.

Figure G-1. Several segments have average lane widths that are wider than the 11 to 12-ft values typically associated with lane widths. Reviewing the segment data revealed that several of the sites with very large average lane widths were associated with unmarked parking.

The revised, more focused analyses only included segments with an average lane width between 8 and 13 ft (about 70 percent of the full database). Various subsets of the reduced database were also created. The goal was to do an analysis uniquely for roads with curbs versus shoulders and for roads with one lane per direction versus roads with two lanes per direction. Another desired consideration is to investigate if roads without medians have different results. Opposing drivers are closer to each other when a median is not present; and therefore, may be more sensitive to the width of the travel lane or the presence of center line or edge line markings.

Source: TTI

WHEN SHOULD CENTER LINE AND EDGE LINE MARKINGS BE USED

The MUTCD (FHWA 2023) includes language on the use of edge line markings in Section 3B.10. The standard and guidance language focuses on higher speed (freeways and expressways) or rural roads. The option section includes items for streets and highways without including specific values for travel way or ADT. The following is the information included in the MUTCD:

• Standard.
  • Edge line markings shall be placed on paved streets or highways with the following characteristics: Freeways, Expressways, and Rural arterials with a traveled way of 20 feet or more in width and an ADT of 6,000 vehicles per day or greater.
• Guidance.

• • Edge line markings should be placed on paved streets or highways with the following characteristics:
    • Rural arterials and collectors with a traveled way of 20 feet or more in width and an ADT of 3,000 vehicles per day or greater.
    • On other paved streets and highways where an engineering study indicates a need for edge line markings.
    • Edge line markings should not be placed where an engineering study or engineering judgment indicates that providing them is likely to decrease safety for all road users.
• Option.
  • Edge line markings may be placed on streets and highways with or without center line markings.
  • Edge line markings may be excluded, based on engineering judgment, for reasons such as if the traveled way edges are delineated by curbs, parking, or other markings.
  • If a bicycle lane is marked on the outside portion of the traveled way, the edge line that would mark the outside edge of the bicycle lane may be omitted.
  • Edge line markings may be used where edge delineation is desirable to minimize unnecessary driving on paved shoulders or on refuge areas that have lesser structural pavement strength than the adjacent roadway.

The MUTCD includes language on the use of the yellow center line markings in Section 3B.02. The standard section specifies ADT and travel way values for two-way urban arterials and collectors. The following is the information included in the MUTCD:

• Standard.
  • Center line markings shall be placed on all paved undivided two-way urban arterials and collectors that have a traveled way of 20 feet or more in width and an ADT of 6,000 vehicles per day or greater. Center line markings shall also be placed on all paved undivided two-way streets or highways that have three or more lanes for moving motor vehicle traffic.
• Guidance.
  • Center line markings should be placed on paved urban arterials and collectors that have a traveled way of 20 feet or more in width and an ADT of 4,000 vehicles per day or greater. Center line markings should also be placed on all rural arterials and collectors that have a traveled way of 18 feet or more in width and an ADT of 3,000 vehicles per day or greater. Center line markings should also be placed on other traveled ways where an engineering study indicates such a need.
  • Engineering judgment should be used in determining whether to place center line markings on traveled ways that are less than 16 feet wide because of the potential for traffic encroaching on the pavement edges, traffic being affected by parked vehicles, and traffic encroaching into the opposing traffic lane.

A blog from 2016 (Moses 2016) explored the question of whether center line striping makes streets safer or slows drivers down. The author makes the point that striping can denote “my” territory from “your” territory. When streets do not have striping, they are generally

understood by most to be common areas. But when line striping is provided, “the general perception is that the roadway real estate has been ‘assigned’ to a particular variety of user, and users are more likely to become agitated or more aggressive when ‘others’ infringe on ‘their’ territory.” The center line marking of “my” versus “your” territory could be associated with lower speeds because drivers have a smaller area available or could be associated with higher speeds because drivers have greater confidence that “others” will not encroach into their space. This serves as another example of how the contextual expectations of a roadway can have an influence on a driver’s choice of speed.

DEVELOPMENT OF SUBSETS

The research team created several subsets of the main database in recognition that the influence of lane width, edge line markings, or center line markings on operating speed could vary depending upon the types of sites included. As an example, opposing drivers are closer to each other when a median is not present; and therefore, drivers may be more sensitive to the width of the travel lane on undivided roadways, especially on two-lane (as compared to four-lane) roads. Table G-1 summarizes the resulting subsets created.

To focus on the effects of lane width, the research team initially applied the following filters:

• Only include sites with lane widths (AvgLaneWidth) between 8 and 13 ft (inclusive).
• Only include sites with either curbs or with shoulders. In other words, remove the segments with a mixture of curbs and shoulders (i.e., CurbGut_DirOfTravel_1yes = 90).

After removing the larger average lane widths and the segments with a mix of shoulders versus curbs, the data were subdivided by number of lanes per direction and presence of shoulder versus curbs resulting in the following subsets:

• 3838 records with 1 or 2 lanes per direction and curbs (CurbGut_DirOfTravel_1yes = 1).
• 1606 records with 1 lane per direction and curbs (CurbGut_DirOfTravel_1yes = 1).
• 2232 records with 2 lanes per direction and curbs (CurbGut_DirOfTravel_1yes = 1).
• 526 records with 1 lane per direction and shoulders (CurbGut_DirOfTravel_1yes = 0).
• 72 records with 2 lanes per direction and shoulders (CurbGut_DirOfTravel_1yes = 0).

To focus on the effects of center line or edge line markings, the research team applied the following filters resulting in a subset with 1598 records:

• Only include sites with lane widths (AvgLaneWidth) between 8 and 13 ft (inclusive).
• Only include sites with median type of none (Median = none).
• Only include sites with curbs or with shoulders. In other words, remove the segments with a mixture of curbs and shoulders (i.e., CurbGut_DirOfTravel_1yes = 90).
• Remove sites with a posted speed limit of 40 mph because all remaining 40-mph streets had a center line.

The above 1598 subset was further reduced to create the following:

• 979 records by removing sites with on-street parking or with a bike lane.
• 295 records by only including segments with an ADT greater than 7200 veh/day or edge to edge width greater than 24 ft. The ADT and edge to edge ranges were selected as being 120 percent of the values included in the MUTCD warrant for center line markings being placed on paved undivided two-way urban arterials and collectors.
• 66 records by removing segments with an ADT greater than 6000 veh/day or edge to edge width greater than 20 ft. The ADT and edge to edge ranges were selected as the values included in the MUTCD warrant for center line markings being placed on paved undivided two-way urban arterials and collectors.

Table G-1. Key characteristics of subsets.

MODEL RESULTS

The research team used the variables considered in the transportation professional model. The parameter estimates, Type 3 test results, and summary statistics for each subset are provided in the following tables:

• Table G-2 provides the parameter estimates for analysis using the 3838 subset.
• Table G-3 provides the Type 3 test of fixed effects results for analysis using the 3838 subset.
• Table G-4 provides the parameter estimates for analysis using the 2232 subset.
• Table G-5 provides the Type 3 test of fixed effects results for analysis using the 2232 subset.
• Table G-6 provides the summary statistics for the 2232 subset.
• Table G-7 provides the parameter estimates for analysis using the 1606 subset.

• Table G-8 provides the Type 3 test of fixed effects results for analysis using the 1606 subset.
• Table G-9 provides the summary statistics for the 1606 subset.
• Table G-10 provides the parameter estimates for analysis using the 1598 subset.
• Table G-11 provides the Type 3 of fixed effects results for analysis using the 1598.
• Table G-12 provides the summary statistics for the 1598 subset.
• Table G-13 provides the parameter estimates for analysis using the 979 subset.
• Table G-14 provides the Type 3 test of fixed effects results for analysis using the 979 subset.
• Table G-15 provides the summary statistics for the 979 subset.
• Table G-16 provides the parameter estimates for analysis using the 526 subset.
• Table G-17 provides the Type 3 test of fixed effects results for analysis using the 526 subset.
• Table G-18 provides the summary statistics for the 526 subset.
• Table G-19 provides the parameter estimates for analysis using the 295 subset.
• Table G-20 provides the Type 3 test of fixed effects results for analysis using the 295 subset.
• Table G-21 provides the summary statistics for the 295 subset.
• Table G-22 provides the parameter estimates for analysis using the 72 subset.
• Table G-23 provides the Type 3 test of fixed effects results for analysis using the 72 subset.
• Table G-24 provides the summary statistics for the 72 subset.
• Table G-25 provides the parameter estimates for analysis using the 66 subset.
• Table G-26 provides the Type 3 test of fixed effects results for analysis using the 66 subset.
• Table G-27 provides the summary statistics for the 66 subset.

Table G-2. Parameter estimates for analysis using the 3838 subset.

Table G-3. Type 3 test of fixed effects results for analysis using the 3838 subset.

Table G-4. Parameter estimates for analysis using the 2232 subset.

Table G-5. Type 3 test of fixed effects results for analysis using the 2232 subset.

Table G-6. Summary statistics for the 2232 subset.

Table G-7. Parameter estimates for analysis using the 1606 subset.

Table G-8. Type 3 test of fixed effects results for analysis using the 1606 subset.

Table G-9. Summary statistics for the 1606 subset.

Table G-10. Parameter estimates for analysis using the 1598 subset.

Table G-11. Type 3 test of fixed effects resultsfor analysis using the 1598 subset.

Table G-12. Summary statistics for the 1598 subset.

Table G-13. Parameter estimates for analysis using the 979 subset.

Table G-14. Type 3 test of fixed effects results for analysis using the 979 subset.

Table G-15. Summary statistics for the 979 subset.

Table G-16. Parameter estimates for analysis using the 526 subset.

Table G-17. Type 3 test of fixed effects results for analysis using the 526 subset.

Table G-18. Summary statistics for the 526 subset.

Table G-19. Parameter estimates foranalysis using the 295 subset.

Table G-20. Type 3 test of fixed effects results for analysis using the 295 subset.

Table G-21. Summary statistics for the 295 subset.

Table G-22. Parameter estimates for analysis using the 72 subset.

Table G-23. Type 3 test of fixed effects results for analysis using the 72 subset.

Table G-24. Summary statistics for the 72 subset.

Table G-25. Parameter estimates for analysis using the 66 subset.

Table G-26. Type 3 test of fixed effects results for analysis using the 66 subset.

Table G-27. Summary statistics for the 66 subset.

ANALYSIS AND FINDINGS

Lane Width (AvgLaneWidth)

Table G-28 summarizes the estimates for the AvgLaneWidth variable for the subsets. The average lane width was found to be not significant for the subsets that included segments with 2 lanes per direction (subsets 3838, 2232, and 72). The additional travel surface space for motorists provides additional room for drivers to move away from other vehicles which could encourage higher speeds.

The average lane width was found to be significant at the 0.05 level for those roads without a median and 1 lane per direction (subsets 1606, 1598, and 979). Those roads could have had either shoulders or curbs. For the larger subset (subset 1606) where all median types were considered, average lane width was only significant at the 0.10 level when the edge treatment was curbed. The speed difference between an 8-ft and 13-ft lane on two-lane undivided streets without bike or parking lanes is about 1.3 mph. These road segments could have curbs or shoulders and are in the posted speed limit range of 25 to 35 mph.

Table G-28. Estimates for AvgLaneWidth Variable for the subsets

Center Line (CenterTreat_1yes) and Edge Line (Elmark_DirOfTravel) Markings

For investigating whether the presence or absence of center line markings is associated with different operating speeds, a subset was created that only contained two-lane roads with no medians, and an average lane width between 8 and 13 ft. This subset only included roads with

speed limits of 25, 30, and 35 mph. The 40-mph roads were removed because all 40-mph roads had center line markings. Additional subsets were created by restricting the presence of bike or parking lanes and by restricting the ADT and travel way width. Table G-29 provides the number of segments and ADT range for those segments with and without center line markings by subset. As a reminder, the difference in the subsets is whether roads with bike lanes or parking lanes were or were not included and based on restricting the ADT and edge to edge width present. The 1598 subset included roads with bike lanes or parking lanes, the 979 subset removed segments with bike or parking lanes, while the 66 and 295 subsets do not have bike and parking lanes and represent roads with smaller ADTs or edge to edge width. Table G-30 provides similar material by edge line marking presence. The MUTCD language includes language regarding the use of center line markings with respect to ADT and travel way width. The standard is to place the markings on two-way urban arterials and collectors with a travel way of 20 ft or more and an ADT of 6,000 vehicles per day.

Table G-29. Number of segments and ADT range by presence of center line markings.

Table G-30. Number of segments and AADT range by presence of edge line markings.

Table G-31. Number of segments and edge-to-edge width range by presence of center line markings.

Table G-32 provides the estimates for the CenterTreat_1yes variable for the relevant subsets while Table G-33 provides the estimates for edge line markings. For all subsets, the center line and edge line variables were significant, and the absence of the markings are associated with slower speeds. The results show a larger difference in operating speed when focusing on road segments with the ADTs and edge-to-edge widths near the values in the MUTCD warrants. Using the 295 subset, the absence of center line markings is associated with a 5.8 mph lower speed. Using the 295 subset, the absence of edge line markings is associated with a 1.97 mph lower speed.

Table G-32. Estimates for CenterTreat_1yes variable for select subsets.

Table G-33. Estimates for Elmark_DirOfTravel variable for select subsets.

DISCUSSION

The response variable used in this evaluation, ffspd, reflects the 67^th percentile speed for the entire year and not the traditional 85^th percentile speed (i.e., daytime only, free flow, etc.). Therefore, the results of these investigations should not be used for speed prediction; rather, the results provide an appreciation of which roadway characteristics may be influencing the speed drivers are selecting on a two or four-lane arterial roadway segment.

The analyses discussed in this appendix focused on learning when average lane width, center line markings, or edge line markings influence operation speed. The key findings were: