1. SAFETY EFFECTS OF RESTRICTING THE SPEED LIMIT FROM 90 TO 70 KM/H Ellen De Pauw 1, Melissa Thierie 1, Stijn Daniels 1, Tom Brijs 1 1 Hasselt University, Transportation Research Institute, Belgium Contact information: Ellen De Pauw, Melissa Thierie, Stijn Daniels, Tom Brijs, Hasselt University, Transportation Research Institute (IMOB), Wetenschapspark 5, 3590 Diepenbeek, Belgium, tel: +32 11 26 91 57, ellen.depauw@uhasselt.be, stijn.daniels@uhasselt.be, tom.brijs@uhasselt.be ABSTRACT The Flemish government decided in 2001 to lower speed limits from 90 to 70 km/h at numerous regional roads. This study examines the effectiveness of this measure, through the application of a before-after study with a comparison group. A total of 61 road sections with a total length of 116 km were included. At all of these road sections the speed limit was restricted in 2001 or 2002. The comparison group consisted of 19 road sections (total length = 53 km) with an unchanged speed limit of 90 km/h during the total research period. Taking into account trend effects, the analyses showed a 5 percent decrease in the crash rates after the speed restriction. Crashes with serious injuries and fatalities showed a non-significant decrease of 9%. Separate analyses were executed for crashes that occurred at intersections and at road sections. Those analyses showed a higher effectiveness of the speed restriction at road sections, both for injury crashes and more severe crashes. Also a higher effectiveness was found for the severe crashes compared to the injury crashes. THE MEASURE - Road sections at Flemish regional roads with a speed limit of 90 km/h were selected to lower the speed limit to 70 km/h - Four main criteria were handled to select those sections. At least one criterion had to be fulfilled to implement the measure: •road sections without cycle paths or with paths next to the roadways and only a small inter space •road sections with obstacles close to the roadways with a high risk of collision •road sections outside the built-up area but with a high building density, and a high number of vulnerable road users •road sections on which several severe crashes occurred in the past DISCUSSION  No application of the empirical Bayes method: -Due to data availability -Subsequent weakness: regression to the mean not taken into account -However: RTTM rather unlikely  Higher effectiveness at road sections compared to intersections -Crashes that occur at intersections may be less influenced by speed and causation might rather be related to maneuvers (e.g. turning left). But this does not explain the increase  Results not strongly significant -Future extension of dataset and replication of research recommended  Results show limited effect size -Actual speed behavior might not correspond with legally imposed speed limits. Factors such as habits, non-acceptance of the new measure or inattentiveness might explain why the actual speed adaptation is lower than the required speed adaptation -Apart from speed behavior, other factors explain the occurrence of crashes. Those factors are not necessarily strongly influenced by the speed reduction measure DATA Locations - Research group: 61 road sections, with a total length of 116 km •located in 16 different municipalities in the province of Limburg (Flanders, Belgium) •the length of the sections ranged from 0,1 to 6,04 km •at 48 sections the speed limit restriction was set in 2002, 13 had an adaptation in 2001 •no other measures were performed during the research period that could have had an effect on driven speed (e.g. changes in traffic regulations such as the right-of-way rules and changes in infrastructure, such as narrowing or broadening roads) - Comparison group: 19 sections, with a total distance of 53 km •also selected from regional roads in the province of Limburg •had a speed limit of 90 km/h during the whole study period •no other measures that could have had an effect on speed were performed Crash data -Flemish geo-coded crash data are available from 1996 till 2007 •before period: from 1996 to 2000/2001 •after period: from 2002/2003 to 2007 - Two groups of crashes •injury crashes: all crashes with at least a slightly injured person •severe crashes: only crashes with serious injuries and fatalities •no inclusion of property-damage only crashes since these are not included in the federal crash data - Each crash was analyzed to determine whether it occurred at a road section or at an intersection Fig 1. Mean crash rates per km in the research and comparison group from 1996 to 2007, both for injury crashes and more severe crashes RESULTS Analysis per location - Slightly more locations show a decrease in crash rates, but also a lot of locations show an increase, especially at intersections - Only a limited number of locations show significant results Meta-analysis - Injury crashes •total: a non-significant decrease in crash rates of 5% (however upper bound is close to 1) •intersections: a significant increase of 11% •road sections: significant decrease of 11 % - Severe crashes •total: non-significant decrease of 9% •intersections: a non-significant decrease of 6% •road sections: a significant decrease of 36%  A higher effectiveness for more severe crashes compared to all injury crashes, which means that the speed restriction had a higher effect on crashes with serious injuries and fatalities, compared to crashes with slight injuries  A stronger effectiveness for crashes at road sections compared to intersections Analysis per location TotalIntersectionsRoad sections # eff < 1 (# signif) # eff >1 (# signif) # eff < 1 (#signif) # eff >1 (# signif) # eff < 1 (# signif) # eff >1 (# signif) Injury crashes 38 (2)23 (3)26 (1)35 (4)43 (1)18 (3) Severe crashes 33 (5)28 (2)33 (1)28 (1)41 (0)13 (0) Meta-analysis TotalIntersectionsRoad sections eff [95% CI] Injury crashes 0,95 [0,88; 1,03]1,11 [1; 1,21]0,89 [0,78; 1] Severe crashes 0,91 [0,76; 1,07]0,94 [0,68; 1,19]0,64 [0,43; 0,86] METHOD - Before-after study with a comparison group •the index of effectiveness (eff) shows the proportional change in crash rates after the implementation of the speed restriction, with control for the general trend effects •an index 1 shows an increase - Analyses per location and meta-analyses over all locations