USF Libraries
USF Digital Collections

Investigation of fatal crashes in Florida

MISSING IMAGE

Material Information

Title:
Investigation of fatal crashes in Florida final report
Physical Description:
1 online resource (xi, 73 leaves) : ill. ;
Language:
English
Creator:
University of South Florida -- Center for Urban Transportation Research
Florida -- Dept. of Transportation. -- Safety Office
United States -- National Highway Traffic Safety Administration
Publisher:
Center for Urban Transportation Research
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Traffic accidents -- Research -- Florida   ( lcsh )
Traffic accidents -- Statistics -- Florida   ( lcsh )
Genre:
bibliography   ( marcgt )
statistics   ( marcgt )
technical report   ( marcgt )
non-fiction   ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 48-49).
Statement of Responsibility:
Center for Urban Transportation Research, University of South Florida, College of Engineering ; for Florida Department of Transportation Office of Research.
General Note:
Title from e-book t.p. (viewed Aug. 31, 2011).
General Note:
"This report was prepared for the State Safety Office, Department of Transportation, State of Florida in cooperation with the National Highway Traffic Safety Administration, U.S. Department of Transportation and/or Federal Highway Administration, U.S. Department of Transportation."--P. i.
General Note:
"February 2001."

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029078008
oclc - 748588607
usfldc doi - C01-00138
usfldc handle - c1.138
System ID:
SFS0032248:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200361Ia 4500
controlfield tag 001 029078008
005 20110831161802.0
006 m o d
007 cr bn|||||||||
008 110831s2001 flua obst 000 0 eng d
datafield ind1 8 ind2 024
subfield code a C01-00138
035
(OCoLC)748588607
040
FHM
c FHM
043
n-us-fl
090
HE5614.3.F6 (ONLINE)
0 245
Investigation of fatal crashes in Florida
h [electronic resource] :
b final report /
Center for Urban Transportation Research, University of South Florida, College of Engineering ; for Florida Department of Transportation Office of Research.
260
Tampa, Fla. :
Center for Urban Transportation Research,
[2001]
300
1 online resource (xi, 73 leaves) :
ill.
500
Title from e-book t.p. (viewed Aug. 31, 2011).
5 FTS
"This report was prepared for the State Safety Office, Department of Transportation, State of Florida in cooperation with the National Highway Traffic Safety Administration, U.S. Department of Transportation and/or Federal Highway Administration, U.S. Department of Transportation."--P. i.
FTS
"February 2001."
FTS
504
Includes bibliographical references (leaves 48-49).
650
Traffic accidents
x Research
z Florida.
Traffic accidents
Florida
v Statistics.
2 710
University of South Florida.
Center for Urban Transportation Research.
1
Florida.
Dept. of Transportation.
Safety Office.
United States.
National Highway Traffic Safety Administration.
776
i Print version:
t Investigation of fatal crashes in Florida.
d Tampa, Fla. : Center for Urban Transportation Research, [2001]
w (OCoLC)46381374
773
Center for Urban Transportation Research Publications [USF].
4 856
u http://digital.lib.usf.edu/?c1.138
FTS
y USF ONLINE ACCESS
951
10
SFU01:001960096;
FTS



PAGE 1

Investigation of Fatal Crashes in Florida Final Report February 2001 Cen ter for Urban Transportation Research Univ ersity of South Florida College of Engineering 4202 E. Fowle r Avenue, CUT 100 Tampa, Florida 33620-5350 (813) 974-3120, fax (813) 974-5168 Fo r Florida Department of T ransponation Office of Research BC158

PAGE 2

DISCLAIMER TIUs. report was prepared for the State Safety Office, Department of Transportation, State of Florida in cooperation with the National Highway Traffic Safety .Administtotion, U.S. Department of T.ra.osportation and/or Federal Highway .Administrarion, U.S. Department of Transportation. The conclusions and opinions expressed in this report are those of the subgrantee, and do not necessarily represent those of the State of Florida, Department of Transportation, State Safety Office, U.S. Department ofT ransportati.on, or any other agatcy of the Stace or Federal Go v ernment i

PAGE 3

ACKNOWLEDGEMENTS The CUTR project team consisted of the following: Patricia A. Turner, Seruor Research Associate, Principal Investigator, C UTR Xuehao Chu, Senior Research Associate, Principal Investigator, CUTR Nevine Georggi, Research Associate, CUTR George Yanez, Gnduate Research Associate, CUTR The authors would like to re<:ognize the following individuals for assistan ce during the p roject 2nd during the preparation of this report. They p rovided information essential for the completion of this report George E. Rice, Jr., Florida Department of Transportation, Safety Office, FDOT Project Manager Theodore Petritseh, Florida Department of Transportation, Safety Office Patrick Brady, Florida Depmment of Transportation, Safety Of6ce FrankJullan, U.S. Department ofT=sportation, Federal Highway Administration Bobby Norbum, U.S. Department ofT=sportation, Federal HighWlly Administtation J. Cluis Richter, U S. Department of Transportation, Federal Highway Administration Simon Washington, School of Civil 2nd Environmental Engineering. Geozgja Institute of Technology Jennifer Ogle, School of Ci.U 2nd EM-ironmental Engineering. Geozgja Institute of Technology ii

PAGE 4

TABLE OF CONTENTS Executive Summary ................................................................................................................ ............ ............................... viii Chap terl. Introduction ...... ........................................................................... .. ... .......... . ................... .............................. 1 1 1 Study Purpose ........................................................................ .... ........................................ .................... ........... 1 1 .2 Research Objectives .................................................................................... .............................................. ....... 1 1.3 Research Approach ........................................................ .. ...................................................................... ......... 2 1.4 Report Organization ...... .................................................. ................................................................................ 3 Chap te r 2. Methodology ................................................................................................................................................... 4 2.1 Introduction ....................................................................................................................................................... 4 2. 2 S rate Grouping ............................................................................................................................................. ..... 4 2 3 Data Sources ................................................ ..................................... ................................... .............................. 5 2.3.1 Highway Statistics .................................................................................................................................. 5 2.3. 2 F ARS ...................................... ................................................................................................................. 6 2. 4 Level-One Analysis ................................................................... .............................................. ......................... 7 2. 4 .1 Data Prepantion .................................................................................................................................... 7 2. 4 2 Analytic Me tbods ................................................................................................................................... 8 2 5 Exposure Analysis ........................................................................................................................................ .... 9 2 .5.1 Exposure Meas urc ...................... .................. ........................ ...... ........................................................... 9 2.5.2 Analytic Methods ................................................................................................................................. 11 2 6 Multi-Factor Analysis .................................................................... ................................................................. 11 Chapter 3. Level-One Analysis .. .................................................................................................................................... 13 3 1 Introduction .... ........ .... ..................................................................................................................................... 1 3 3.2 P erson Type .................. .............. .... ................................................................................................................. 13 3.3. Driver Analysis .......................... ........ ........................................................ ...... ................................................. 14 3 .3.1. High Over-Representation .......................................... ....................................................................... 14 3.3.2. A.teas of Special Concern ............................................................ ...................................................... 14 3 .4. Passenger Analysis ...................................................................................................... .................................... 16 3.4.1. High Over-Representation ....................................................................................... ...................... .... 16 3.4 .2. A.tea of Special Concern ..................................................................................................................... 16 3.5. NonMotorist Analysis .......................................................... .................................................................... ..... 17 3 5.1. High Over-Representation ................................................ ................................................................. 17 3.5.2. A.teas of Special Concern ............................................................................................. ...................... 18 3. 6. SWilJXlai)' .. .. .. .. .. .. .. .. .. .. .. .. ...... .. .................................................................................................................. ....... 18 ofFa/41 Cr41hls in Fhrida: Final &pori U1

PAGE 5

TABLE OF CONTENTS (CONT'D) Chap ter 4 Exposure Analysis .............................. .................. .......... ....................... .............................. ... ... ............. ..... 20 4.1 Introduction ................................ .......................... .......... ............................. ............................................ ..... 20 4.2 Driver Analysis .................. ................ ...... .......................... . . ......... ....... ............. .... ...................... ,, .............. 20 4.2.1 Changes from Level-One Analysis ........ .................... .......... ........................ .............. ....................... 20 4. 2.2 Reduced Over-Re presentation ........ ...................... ............................................................................ 20 4. 2.3 No Changes in Over Representation .................................... ........................ .... ........ ....................... 21 4.2. 4 Changes from Undert o Over-Representation .... ............ .................. ...... ...................................... 21 4.3 Passenger Analysis ...................................................................................... .................. .................................. 22 4.3.1 Chmges from Level -One Analysis .................. .... .................. ........................ ................................... 22 4.3.2 Changes from Underto OverRepresentation ........... ....... ................................... ........ ....... ........ 24 4 .4 NonMotorist Analysis .................... .............................................................................................................. 24 4.4. 1 Changes from Level-One Ana!ysis ......................... ............ ........ ............................ ........................ 24 4.4.2 Changes from Under-to OverRepresentation ............... ..... .......................................................... 25 4.5 Summary .......... ............ .............................. .......... ...... ........................ .... .......................................................... 25 Chapter 5 Multi-Factor Analysis ........................ ...... .. ................. : ...... .... .............................................. ......................... 27 5.1 Introduction ........................................................................ .......... ...................................... ............ .... .... ........ 27 5.2 Over-Rep resentation ofNon-Motorist ........................ ....... ................. ........................... .......................... 27 5 3 Non-Motorist Activitie in Roadway ......................................... ..... ............................................................ 30 5 4 SUflliil.3.l'Y ..................... ................................................................................... 33 Chapter 6. Results .............. .......... ............................ ........................................ .... ............................................................ 34 6.1 Introduction ........ .... ........................ ........................................................................................ ...... .... .... ......... 34 6.2 Snmmaty ............................................ ..... . .................................... ........ .......... ........... .......... . ............. ............. 34 6 .3 Over-Represented Areas ................ ........ ...................................... .... .......................................... ...... ...... ....... 36 6.3.1 Behavioral Aopects ............ .............. .... ............ ................ ........ ...... .......................... ............................ 36 6.3.2 E nvironmental A$pectS .. .... ................ ...... .......... .......... .............................................. ........................ 38 6 .3.3 Vehicle A$pects .......... .... ...... .................................. ............ ...................................... ............ .... ............ 39 6 .3.4 Engineering A$p
PAGE 6

TABLE OF CONTENTS (CONT'D) B ibli ography -................. ............ ................ .. ............... .. .................................. 48 APPWOIX A: State Grouping an.d F AR.S VaDables ..................... .................. ................ . .... ......... ................ ....... SO APPENDIX B : Reoults rom Level-One Analysis ........................................ ..... ....................... .. .............................. 5 8 APPENDIX C: Results from Exposure Analysis ............................................... ........................................................ 6 2 APPENDIX 0 : Resulu from Multi-Factor An:al)'1is ... ..................... ........................... ........ ......... .. ...... ..................... 67 APPENDIX E : Suaunuy of Results .............. ....... ........ ...... ...... .... .... .......... ........ ..... ... .......... ................. ..................... 71

PAGE 7

LIST OF TABLES Table 1.1 Table 2.1 Table22 Table 2.3 Table24 Table 3.1 Table 3.2 Table 3 3 Tabl e 4 1 Table 4.2 Table 4.3 Tablc4.4 Table4.5 Table 5.1 Table 5 2 Ta ble 6.1 Table 7.1 Table7.2 Top 10 Highway Safety Concerns Identified By Aorida Safety Expetts .................................. ... 1 Fatal Crashes, Vehicles Involved, and Person s Invo lved by Stllte Group,19941 998 .............. 7 Driven by Group and Vehicles Invo lv ed ....................................................................................... 10 Drive.rs by Group and Fault in Two Vehicle CiliShes ............................................... ,. ................... t 0 Comparison of Rd>tive Exposure Between Approaches ............................................................. 11 Over -rted to Having Activities in Roadway .. ....................................................................................................................... 31 Summary of Overoll Study Results ................................................................................................... 35 Top Three Problem Areas Based on Degxee of Ove.r-represenllltion and Index Values ....... 43 Indexing Problem Areas ..................................................................................................................... 44 VI

PAGE 8

LIST OF FIGURES Figure 1.1 Fatality Rate per 100 Million Vehicle Miles Traveled: Florida vs. US, 1 975-1998 .... ...... ....... ...... ........ 1 Figure 1.2 Distribution of Fatalities by Person Types in Florida and U.S. 1994-1 998 .. .... ........... ....... ................. 3 l""stigati41r ojFt1141 Cr4Jhu in Fhrida: Fmal &port

PAGE 9

STUDY PURPOSE Executive Summary Investigation of Fatal Crashes in Florida_ In 1998, Florida's fatality rate per !00 million vehicle miles tta v eled was 2.6 compored to the oationwide average of 1.6. Because Florida's &tal eruh rates and fatality utes consistendy outrank national avenges, the Florida Department of Transportation (FD01) contracted with the Center for Urban Trans portation Research (CUTR) at the University of South Florida (US F) to help identify highway safety probl em areas that may contribute to less safe travel on Florida roadways This research was conducted as port of a larger pooled-fund study to detennine why fatal crash utes are higher in the Southeastern Ucited States (US) compared to rest of the nation and what could be done to reduce fatal crashes in the region. The research objective was to identify problem areas related to highway safety in which Florida is over-represented relative to other states and the nation as a whole. The research results provide poucy makers and highwy safety advocates with a better understanding of the factors that may contribute to higher fatatity rates in Florida and may be used to help identify specific problem areas where the fDOT can focus safety improvement measures. RESEARCH APPROACH CUTR researchers conducted several activities to accomplish the research objective. Data were collected and analyzed through the applic..Oon of a multiple-step process used to identify highway safety problem areas ucique to Florida. Data Collection. Researchers obtained data on all &tal crashes in the Ucited States that occurred from 1994 to 1998 from the Fatal Analysis Reporting System (FARS) database. The dat1 included motor vehicle ttaffic crashes from all states during the five-year period resulting in a &tal injury to a vehicle occupant or non motorist within 30 days of the crash. Data Analysis. First, states were grouped based on fatali
PAGE 10

data based on the degree of over -representatio n (DOR), data quality, and statistical significance. Researchers synthesized the results and ranked problem ueas for further research consideration. RESEARCH FINDINGS Level-One Analysis. This analysis measu.ced the degree of over-representation of fatalities in Florida relative to each of the state groups and included several components for each variable. Researchers computed the proportions of fatalities the different values of a given variable represent within Florida and each of the other state groups and tested for statistical significance. Researchers calculated the degree of over-representation for every value of a given variable relative to each of the three state groups The Level-One analysis identified a number of highway safety problem areas that are over-represented in Florida relative to the state groups and the nation as a who le. Some of the problem areas are common to drivers, passengerS, and non-motorists, including roads with 4 lanes, roads with 6 lanes, regulatory sign$, and divided roads with no barriers. Some areas are common to drivers and passengers only, including shoulders, urban local roads, and vehicles turning left just prior to the crash. Some ue common to passengen and non motorists only, including intersections Some are common to drivers and non motorists only, includin g urban non interstate highways. Still others are unique to the individual person types. Unique to non-motorists ue U.S. highways, roads with 26-45 mph speed limits, drivers with at least one non-speed related moving violation conviction, and having activities in roadway such as walking, riding, standing, etc. Unique to driverS ue tn.ffic signals and being struck. And unique to passengers ue municipal roads and drivers 75 years or older. However, these resul ts only illusttate areas that are potenti.J problems because any over-reptesentation may be eliminated once exposure is controlled or other factors are inuodueed. Exposure An.Jysis. This step applied quasi-induced exposure to driven direedy and to passengers and non motorists indirecdy through varying degrees of modification to account for the differences in exposure across the state groups. The Exposure analysis revealed that the many highway safety areas of concern identified through the Lcvelwere no lon,ger so or the degree of concern is significandy teduced once exposure is con s idered. Results are presented using four different scenarios. Scenario 1: EVgh degrees of over-representation under Level-One ans/ysis with no stsltistically signiiiC1Ult reduction in the degree of over-representation. This scenario includes highway nfety problem ueas that were highly over-zepresented in the Level-One analysis and u a result of the Exposure analysis, show no statistieally significant reduction in the degree of over-representation. Most highly over repreented among these are shoulders and vehicles turning left just prior to the crash for both drivers and passengers and U.S highways for non-motoristS Among the other potential problem ueas, dark but lighted conclitions, municip.J roads, regulatory signs, and drivers 75 years or older are common to both drivers and passengers; motorcycles, 9-clock impact point, being suuck ue unique to drivers; clivided roads with no barrie< and 3-clock impct are unique to passengers; and intersections, state highways, and udnn non-interstate highways are unique to non-motorists IX

PAGE 11

Scenario 2: in Level-One snalysis but once exposure is conaoUed. PotentW problem areas identified under this scenario include those that were identified in the Level-One analysis 2S being under-represented but show significant degrees of over-representation once exposure is controlled. Common to both driven and passengers axe aashe s involving one drunk driver, roads on grade, and non-junctions. Unique to drivers is head -on crashes, unique to passengers is the problem of drivers involved in one crash in the 3 years prior to the crosb in question Scenario 3. High degree$ of over-represenradon under Level-One analysis and higher overrcp.tcsenudoD after exposure. PotentW problem areas identified with high degrees of over -repr esentation under the Level -One analysis and show even higher degrees of over -representation once aposure is conttolled are .included in this scenario. These include dark conditions and state highways, both of which are unique to nonmotorists. ScMJUio 4. High degrees of over-representadon under Level-One snalysis but Exposure twalysis is DOt sppUcab/e. This scenario includes problem areas identified with high degrees of over-representation in the_ Level One analysis but to which Exposure analysis is not These include non -mot orists walking, riding, or having other activities in roadwo y, drivers with at least one suspension or revocation in the 3 yean prior to the aash in question, and drivers with at least one non-speed related moving violation conviction in the 3 years prior to the crash in question. All these are unique to non-motorists. Multi-Factor Analysis The Exposure onalysis was not applicable to cettain problem areas identified in the Level-One analysis for passengers or non -m otorists. For passengers, these include personal characteristics, such as age. For non-motorists these include eharactedstic:s that drivers non-mototis[s do no[ share. The multi-factor analysis attempts to determine under what specific situations problem areas identified in the Level-One analysis a.te more over-teptesent<:d than under other $ituations. Two issues related to non-motorists were examined through multi-factor analy$:iS. One issue is related to nonmotorist in general. Level -One analysis revealed that non-motorist as a person type are over-represented in Florida relative to the state gtoups and the nation as a whole. The question to be answered was (
PAGE 12

. : pe>ent pro.u!e pollC)'mk.ers and highway safety advocates with a better undu11anding of the factors "that may eon tribute to le .. safe: highwa y tnvel in Flotida. Finally, the research provides basis for future research to ew:nine 'Oibethet and how roadwa y engineering practices and legislative policies related to highway safety problem ueas differ from those pnctices and policies in states with safer highwa y rmvel.

PAGE 13

Chapter! INTRODUCTION 1.1 STUDY PURPOSE The eight southeastern states comprising the Federal Highway Administt:1tion s (FHWA) Atlanta Resource Center (formerly known as Region IV), namely Alabama, Florida, Georgia, Kentucky, :Mississippi, North and South Carolina, and Tennessee, consistently outrank other regions with respeet to number of &tal crashes and &tal crash rates. Approximately one-forth of the Nation's fatalities occur in the southeastern region where the fatality rate is about 20 percen t above the national mean rate. In 1998, the FHW A Atlanta Resource Center commissioned a study to investigate crash data, detemUne causative factors and make recommendation$ for reducing fatal aashes as put of the agency's 10-yeat Stt:ltegic P l an tp reduce fatalities and injuries in the southeastern region by 20 percenL The eight states voluntarily agreed to participate in a regional pooled-fund study coordinated by Georgia Institute of Technology. As FJgUie 1.1 shows, Florida's fatality rate per 100 million v ehicle miles tt:1veled is 2.1 compared to the national rate of 1.6 in 1998 Because Florida's &tal crashes and fatality rates outranks national averages, the Florida Department of Transportation (FD01) contracted with the Center for Urban Transportation Research (CUTR) at the Univer$ity of South Florida (USF) to develop a technical scope of work that was consistent with the overall objectives of the pooled fund study Figure 2.1 Ftttslity Rste per 100 MiDion Veblcle Miles TrsYded: FloridJl vs. US, 1975-1998 --us 4 J.S 3 l! 2.5 ,2 2 l.S 1 o.s 0 1975 1980 1985 1990 1998 Yev Sourc:e: FARS d2ta. 1 Fot more information on the pool .funded $tudy and te$ults, contact Simon \'Vasbington, Ph.D School of Civil and En.Uoom .. w Eapeaiog, Georgi> Io..titute oTecl>nology, Albnu, Georg;., 30332"' Sjny>o Wasbine;ron@c;c. g;u,m sdy, or phoae (404) 89+6476 1

PAGE 14

1.2 RESEARCH OBJECTIVE The pwpose of the reseaxch was to identify problem ;u;eas related to highway safety in which Flotids is over-represented relative to other stttes and che rution a s a whole. The reseatch results provide policy makers and highway safety advoC' eUmined individ.W variables from che F ARS dstaSet to identify a preliminary list of highway safety problem areas where Florida may be over-represented relative to the other state gtoups and the nation as a whole. The degree of overrepresentation (DOR) was calculated as follows: DOR =( FLPtoportion ofFatalities -t)xtOO Step I Compamon Group's Proportion of Fatalities Exposure Analysis applied the concept of quasi-induced exposure to the problem areas identified in Step 1 to control fo r differences in exposure between Florida and each of the state groups and the nation as a whole. The degree of over-representation was calculated based on che following: lThe Dist:Ji(t of Colu.mbia. and the other 49 nates inro three groups of diff'ete:nt kvels of bjgbway safety. For the Dilaic.t of Columbia ia Kfcm:d ro a.s Jttk. 2

PAGE 15

FL Proportion of Faulities DOR FLProportion ofE.cposure 1 100 Step 2 = Comp>Dson Group's Proportion of Fatalities x Compui.sou Groups Ptoportion ofExpoMe The last step, Multi-Factor analysis, introduced additional va.ciables to examine ptelimirwy problem a.teas to which the quasi induced exposute approach was not applicable. The statistical prognm, SPSS, and Microsoft Excel wue used to analyze the data based on the DOR formulas, data quality, and statistical significance. Resea.tchers synthesized the zesults and ranked problem ueas fot furthet research consideration. Figure 1.2 Distribution of Fstslities by Person Types in FJo.tids and U.S., 1994-1998 .. -Sou:rce: T abulattd &om F ARS, 1998, www.nbta.dor goylpoQplc laesa I& rs htmL NH'l'SA. Fmal Report Prepustion. Ruea.tchets documented the tesults of the analyses and ptesented them both gnphically and in writing This report constimtes the project delivuable. 1.4 REPORT ORGANIZATION The remsinder of this teport is divided into six chaptus. Chapter 2 details the teseatch methodology used to identify problem ueas. Chapters 3-5 summarize the results from the thtee steps of the methodology. Cbaptet 6 id
PAGE 16

2.1 INTRODUCTION Chapter2 METHODOLOGY Researchers used a multi-step process to help identify problem areas that may contribute to Flo ri da's relatively low level of highw:ty safecy. The process began with grouping states according to average fatality rates and proceeded with determining areas where Florida is over-represented relative to otha: stat e groups occurs when the number of fatalities in certain situation is disp ro portionately higher in Florida than in the defined state groups. This pxocess was done in two sequential levels. The first l evel of analysis, or Level-On e analysis, identified potential pxoblem axeas without taking into consideration any differences in exposuze between F1orida and other state groups The sec o nd le v el of analysis varied, depending on the natuxe of problem axeas identified from Level-One analysis. In mos t cases, the second level introduced differences in exposure and i s refened to as Exposure analysis I n the other f<:W cases for which exposux e could not be measured, the second level introduced additional variables to determine under what situations these problem axeas are most highly over-xepxesented This third level of analysis is xeferred to as Multi-Factox analysis. The analyses wexe conducted separately for thxee pers o n types : drivers, passengers, and non mo t orists (pedestrians and bicyclists). These thxee types of persons were selected due to differences in how exposure is measmed and likely conuibuting factors In these person types were separated because the y differ in t enn of share s b etween Florida and the nation as a who le. For example, Florida had 3,282 non-motorists died from highway crashes during the period from 1994 to 1998, representing 23.7 percen t of all highway traffic fatalities in this period. In comparison, the nation as a whole had 14. 9 percent of all highway ttaf6c fatolities as non-motorists in the same period. 2.2 STATE GROUPING The District of Columbia and the other 49 states were di.nded into thxee groups of different levels of highway safety (See Table A 1 in Appendix A). Hereafter, the District of Columbia is aJ.o referred to as a s tate for convenience The rationale was that the compariaon of Florida with multiple groups of states with different lev el s of highway safety would provide more insight into highway safety problem areas where Florida is over-repr esented "This grouping is based on fatality rates, ie., the number of fatalities per tOO million vehicle miles of travel (Vl-.fl). These C.tality rates were defined as averages over the period I 993-97. Data on the number of fatalities and the amount of vehicle travel by indi.ndual states were from the Highway Statistics for the years 1993 thxough 1997 from the website of FHWA's Office of Highway Infoanation The period from 1993-1997 was used for grouping rather than from 1994-1998 because Highway Statistics f o r 1998 was not available when the states were grouped at the beginning of the project. The grouping was little changed when the period from 1994 to !998 was used. During the period from 1993 to 1997, Florida's fatolity rate was lower than II states but higher than 39 states. It was decided to keep the states with higher fatalicy rates as a single group (Less Safe). 4

PAGE 17

However, it was decided to divide the states with lower fatality rates further into two groups: states with their average fttality rate more than 25 percent lower than Florida's (Safest) ond states with their average fatality rate less chan 25 percent lower than Florida's (Safer). Note chatFlorida's rate at 2.14 i$ about 25 percent higher than the national rate at 1. 7 1. 2.3 DATA SOURCES F ARS data &om 1994 to 1998 ,. ... the main data source for the analyses. This was supplemented with data on vehicle travel from Highway Statistics produced by the Federal Highway Administration. Data on vehicle travel were used to calc:uhte the average number of fatalities per unit of vehicle travel for individual state>. For each database, the following briefly describes: the sources of the data items and potential enou and biases in the data. 2.3.1 Highway Statistics Data Sources. Data on dally vehicle-miles of travel (DVMI) in Highway Statistics are derived from the Highway Perfoimance Management System (HPMS) In concept, travel is a calculated value that is a product of the annual average dally traffic (AAD'I) and the centerline length of the section for which the AAOT is reported. AADT is required to be reported for each section of Interstate, NHS, and other principal arterial; as a travel can be computed for these functional systems on a 100 percent basis. Fot minor a.rtesial, rural major collector and urban collector systems, travel j s calc:uhted from samp l es using the AAOT, c enterline length reported for each sample section and the HPMS sample expansion factor for each section. The D\111.IT will be adjusted for the functional systems where sample dat a are used if the universe and expanded sample length do not e qual. For the most part, states us e unknown methods to estimate travel for the rural minor collector and rural/urban local functional systems. For these systems, travel values are developed by the States using their own procedures and are provided in HPMS. Some states use supplemental traffic counts outside of the HPMS procedures; others employ estimating techniques, such as fuel use, to detemline uavcl on these systems. Eaprs and Travel estimates reported via the HPMS should be of reasonabl e quality particularly for the higher order functional systems. AADT and travel data are edited by the HPMS software for unusual values and for unusual changes to preoiously reported values. FHWA routinely works with State data providers to modify reported AADT values that do not appear to be reasonable before incorporating them into a fin>l master lile. Although AADT is required to be updated annually in the HPMS, counts are required to be updated on a 3-year cycle For any reporting year, MDT for uncounted sections is to be derived by factoring the latest yeais count for those sections. States that follow the HPMS sampling instructions in developing traffic counting progwns have adequate counting and classification tools to prepare qwlity AADT and travel estimates for HPMS The consistency of the sampling and counting procedures should also provide comparable traffic data. 1n practice, FH\V;. is aware that not all states rigorously follow the recommended sampling. counting, and estimating procedures. Reporting of AADT based on actu21 traffic counts, on all lntentate and principal arterials on a 3-year cycle is a required but not necessarily followed protocol for HPMS rep orting. The calc:uhtion and application of various adjustment factors to 24-or 48-bour coverage counts to enable them to tepresent AADT i s as much art as science Classification counts,

PAGE 18

which arc needed to adjust pneumatic tube counts collected for three or more we vehicles as well as for other HPMS items, are dieult to collect and to apply on a statewide basis. Equipment used to obtain count information is only accurate within certain limits and can suffer from malfunctions and breakdowns, which ean affect the reliability of traffic counts. 2.3.2FARS Oat Sowces. The National Highway Traffic Safety Administration (NHrSA) has a cooperative agreement with an agency in each state to provide information on aU qualifying fatal crashes in the state. These agreements are managed by Regional Contracting Officer's Technical Representatives located in the10 NHrSA Regional Offices. State employees, called "FARS analysts," arc responsible for gathering, translating, and transmitting their state's data to NHrSA in a standard format Each F ARS analyst attends a formal training program. and also receives on-the-job training. The number of analym varies by state, depending on the number of fatal crashes and the ease of obtaining data. FARS data are obtained solely from the state's existing documents, which generally include some or aU of the following: Police Accident Reports (PARS) State vehicle registration .6Jes State driver licensing files State Highway Department data Vital StatistiC. Death certificates Coroner/Medical examiner reports Hospital medical records Emergency medical service reports Analysts use these documents t o code FARS data items on tlu:ee standard FARS forms. The Accident Form asks for information such as the time and location of the crash, the first harmful event, whether the crash was a hit-and-run crash, whether a school bus was involved, and the number of vehicles and people involved. The Vehicle and Driver Forms call for data on each ctash involved vehicle and driver such as vehicle type, initial and principle impact points, most harmful and driven' license status. The Petson Form. contains data on each p erson invoh,ed in the crash, including age, gender, role in the crash (driver, passenger, non-motorist), injury severity, and .restraint use. The data collected within FARS do not include any personal identilj.ing infotmation, such as addresses, or social s ecurity numbers. Each analyst enters data into a local microcomputer data file, and fo.rw.uds weekly updates to NHrSA's central computer database. Data are automatically checked when entered for acceptable range values and for consistency, enabling the analyst to make corrections immediately. To be included in FARS, a crash must involve a motor vehicle traveling on a traffic way customarily open to the public, and tesult in the death of a petson (either o.n occupant of a vehicle or a non-occu pant) within 30 days of the crash. Data Items. The FARS file contains descriptions of each fatal crash repor:ted. Each ease has more than 100 coded data elements that characterize the crash, the vehicles, and the people involved. The final data for each year arc organized into three files: the Accident File, the Vehicle and Driver File, and the Person File Table 2.1 shows the size of each file by state gtouping. The variables, for the 6

PAGE 19

:1 i period und e r study, unique to each of these ftles are listed separately in Tables A-2 -A-4 in Appendix A. Table 2.1 Fat """ 37,697 ..... Not r 8,779 An uo j "LOO 17,845 999R1 Died prior 0 21 12 4 37' 239 1,511 1.305 725 3,780. Total .... 17< ,.. 507,569 i Errors and Biases. Because FARS is a census of all fat al crashes, there are no sampling errors. However, nonsampling errors can still exist. There are mainly three forms of non-sampling errors involving FARS data The first is non -response error, which results when information was not collected for a certain variable or a porticular value within this vari a ble These wer e not collected ei t her because t hey were missing in one of the state dat a sources, such as Police Accident ReportS, or because they are not included in the state sources for data collection. Another form of non-sampling error in F ARS is coding error. In this case, correct data were obtained from individual state sources, but errors were made in coding state data into F ARS. The third form of non-sampling error is measurement error, which can result, for example, from the investigating office incorrectly estimating the traveling speeds of the vehicles involved in a f at al crash. Some of these errors are not visible in the data files, while others reflect themselves as missing or being coded as uoknown. Potentially, these errors could pose a serious p r oblem to analyses conducted in the study if they account for a large share of the crashes each state group and these shares differ berween Florida and othe r state groups 2.4 LEVEL -ONE ANALYSIS 2.4. 1 Data Preparation Data were prepared in several steps. First, the F ARS files from each of the five y ears were appended toge t her into a single ftle for each file type, resulting in three expanded files: the expanded Accident file, the expanded Vehicle and Driver file, and the expanded Person fde. Second, the variao les in each expanded file for further analysis w ere selected. They include 30 from the Accident File, 46 from the Vehicle and Driver File and 18 from the Person File. Except those variables for identification, variables in T abies A-2 A-4 not included for further analysis are shaded. Jnwstigation ofFauJ Crashes in Florida: Final RtpOrt 7

PAGE 20

1bird, the remaining variables in the expanded Accident Fde 1.1ld the expanded Vehicle and Driver File were meq:ed into the expanded Person Fde. The meu.ndard teu of equality between two proportions. Suppose one wants to test the statistical significance of difference in the proportions of fatalities in urban areas between Fl. and Less Safe: Pf'L and PLS. The difference PFL-PLS approaches the normal fomi when Vnl'Ft, VHJPLS, VFLQFt, and VLSQLS are all at least 10. The test statistic is then given by: where P = (VFL Pfl. + VLSPLS) I (Vfl. + VLS), Q = I P, and the first subscript in the symbols indicating a particular value in a variable has been dropped for simplicity. For this study, the testing was done at the 0.05 significance level The third component involved calculating the degree of over-representation for every value of a given vuia:ble relative to each of the thtee state gxoups. The degree of in Florida relative to state group j in the type of fatalities associated with value i of a variable is given by o, = l"'"'latl of Fatal Cra!h" i FionJ&< Final F.Jpm 8

PAGE 21

100 [(P, FL I Pij) -1), j = Safest Safer, or Less Safe. A positive wlue of 00 indicates that Florida is over-represerued xelative to state group j in the type of fatal crashes associated with value i of the variable considered. A negotive value of o,, on the other bond means that Florida is under represented 2.5 EXPOSURE ANALYSIS Exposure analysis applied quasi-induce d exposure to drivers directly and to passengers and non motorists indirectly through varying degrees of modification. One problem with I..evel-One analysis is that it does not take into account differences in exposure across the state groups Suppose Level One analysis indicated that 54 percent of fatalities in Florida happened in urban areas, compared with 35 pettent in the Safer states This would indicate that Florida is over-represented in urban fatalities relative to the Safer states. The degree of over representation is 100 (54 I 35-1) = 54.29 perc:ent However, this measure of over teptesentati.on overestimate the true v).]ue i proportionally more travel in Florida is done in urban areas than in the Safer states Suppose further that 60 percent of travel in Florida is urban and 30 percent in the Safer states. How should one adjust the calculation of the degree of over-representation taking into account the diffe r ence in exposure? One could start the adjustment with computing the relative exposure between Florida and the Safer states: 30/60 = 0.5. This relative exposure risk is then used in adjusting the degree of over-representation as follows: 100 (O.SO (54 I 35) !]} = -22.86 percent Florida was aetnally under-represented in urban fatalities relative to the Safer states once F1otida's to urban travel is considued. 2.5.1 Exposure Measure The commonly used measure of vehicle miles of travel i s useful when exposure is measured at aggregated levels such as functional classification Data for vehicle miles of travel, however, are unavailable for disaggregated situations, which most FARS variables descn'be An altema tiv e was needed. This research used the quasi-induced exposure measure As originally designed, the approach applies to dtive.rs and derives exposure estimates &om the distribution of driven in the set of twO-vehicle crashes for which fault can be reasonab l y attributed to one and only one driver. These aze often referred to as the "clean" crashes An underlying assumption of this approach is that not .. at-fault ch:ivers oonstitute a representative random s2mple of those on the road. This measure bas been in the development since 1967. Stamatiadis and Deacon (1997) reassessed the quasi induced exposure measure in terms of its underlying assumptions and concluded: 1) the quasi induced measure provides an accurate reflection of exposure to multiple vehicle cta$hes; 2) the quasi induced measure is an acceptable surrogate for vehicle miles of travel when estimates aze made for conditions during which the mix of toad users is fairly constant; and 3) the qiWi -induced measure is powerful tool for measw:ing relative expos me of drivers when real exposure data are unav2ilable. In addition to applying the induced exposure to drivers, reseazchers modified the procedure to apply to passengers and non motorists respectively. The steps in the procedure are discussed separately for each person type. Drlym. Severo! steps wue involved measuring the quasi induced exposure for drivers. F'ust, teseucbers e
PAGE 22

severity were included in the ftle. As shown in Table 2.2, the file included 9,933 Florida drivers. (Note that the number of drivers can be smaller than the number of vehicles within a given state group because of absence of drivers, for example.) Researchers then assigned at-fault or not-at-fault to the drivers. A driver was innocent if the investigating officer did not indicate any violation. Table 2.3 shows the number of drive.,; by fault. Drivers involved two-vehicle crashes in which none of the drivers or both drivers were at fault were deleted from the file. This resulted in 16,935 drivers remaining, as indicated in Tabl e 2 3. Table 2.2 Drivers by Group and Vehicles Involved Total Table 2.3 Drivers by Group and Faull in Two-Vehicle Crashes Next, researchers deleted F ARS variables that were not feasible for measuring quasi-induced exposure. Finally, researchers computed exposure for each value of every F ARS variable remaining in the fsle. Specifically, this step determined the total number of not-at-fault drivers in Florida (NFI) in the remaining file and the number o f not-at-fault drivers in Florida (AFI) that is associated with a particular value in a variable (e.g. urban). The level of exposure for Florida is measured by En= Nn/ An. Similarly the level of exposure for group j is mea$ured byE; N;/ A;. Passengers. The measure of exposure was modified slighdy for passengers from that used for drivers The same set of clean 2-vehicle crashes was used. In the case of drivers, the exposure measure was based on the distribution of not-atfaultdrivers in those clean crashes. For passengers, the exposure measure was based on the distribution of the passengers of those not-at fault drivers. Non-Motorists. For non -motoristS, the measure of exposure was modified from that for drivers more significantly, though the basic concept remained the same. Thefollowing steps were used in calculating this exposure measure. The ftrst step created a file of drivers from the vehicle file who were involved in single-vehicle crashes. The second step assigned at-fault or not-at-fault to the driver. A driver was innocent if the investigating officer did charge the driver with any violations. The third step merged this new variable measuring who is at fault into the person fde and assigned Investigation of F((la} Crashes in Florida: Final RepOrt 10

PAGE 23

fault or innocence to the non-motorists. Because there were no variables indicating viol2tion charges on the non-motorists were assumed to be not-at-fault whenever the driver was at fault. The fourth srep reduced the person file to include onl y those drive.s and non-motorists involved in single vehicle crashes with at le= one non-motorist. The fifth step computed exposure for each value of relevan t F ARS variables. The computation is similar t o tlie cases for drivers and passengers. However, both no t-at-fault drivers and non-motorists were considered in measuriris exposure. This is because both the amount of traffic and t he number of nonmo torists exposed to this traffic are important in determining the exposure for non-motorists . 2 5.2 Analytic Methods Once exposure was measured for both Florida and another sta t e group, the analysis procee d ed with comp uting the relative exposure EFt EFt IE;. The relative exposure was then used to adjust the degree of over-representation from Level One analysis: 0;; 100 [En, ; (P;n/ Pij)1], j Safest, Safer, or Less Safe. Again, a positive value of 0;; indicates that Florida is over-represented rel2tive to state group j in the type of fatal crashes associated with value i of the variable considered A negative value of 0 ;;, inilicates that Florida is under-represented. Table 2.4 compares relative exposure estimates using vehicle miles of tnivel and the induced approach with respect to drivers. What is shown in the table is Efl,; G Safest, Safer, Less Safe, or All) for urban and rural areas, respective ly. The comparison uses land use between urban and rural areas because land use is the only variable in F ARS for w hich rel2tively reliable statistics are available for bot h induced exposure and vehicle miles of travel. Using vehicle miles of travel, urban exposure in the Safest states, Safer states, Less Safe states, and t h e natio n as a whole is only 89 percent, 76 percent, 72 percent, and 82 percent of that in F l orida, respectivel y At the same time, rural exposure in these state groups is 30 percen t 70 pe r cent, 81 percent, an d 51 percent more than in Florida, respectively. Underth e induced approach, these percen tages change to 82, 59, 61, and 71 for urban exposure and 28, 66, 64, and 46 for rural exposure. In additio n to evidence from the literature, this further indicates the value of the quasi induced measure in controlling exposure. Table 2.4 Compari s on of Relative Exposure Between Approaches ------Use!! Safestj[ S:&ferj ( Less'L All J -H Urban !L o.su o.76 I 0.72 i l 0.82! I I Rural !I 1.30 I 1.70 I I 1.81 il mduced !1. 0.61l[ ___ J[ Rural i I 1.28 il 1.66 I 1.64 il 1.46! Note: Resulu Oll the V}.iT were ba.sed on Hi W2. while results: on the lnduud : "PP the mrtbodology nncd eulic::r. th y r 2.6 MULTI-FACTOR ANALYSIS The Exposure analysis was no t applicable to certain problem areas involving passengers and non motorists identified in the LevelOne analysis. Fo r passengers, these include personal characteristics, such as age. For non-motorists, these include characteristics that drivers and non-motorists do not share. Tbe Multi-factor analysis attempts to determine under what more specific situations problem areas identified in the Level One analysis are more over-represented than under other situations. ln'Vt!stigation of Fat41 Or ashes in Flori
PAGE 24

2.7 SUMMARY In sum, a multi step process was used to identify highway safety problem may contribute to less safe travel on Florida's roadways. For comparative purposes States were separated into three comparison groups based on fatality rates during the five-year period. Level-One analysis calculated the degree of over representation by comparing Florida's proportion of fablities to a comparison group's proportion of fatalities. Problem areas identified in the Level-One analysis were furthered analyzed using Exposure analysis. The concept of quasi-induced exposwe was used to account for the differences in exposure between Florida and each of the comparison groups. Finally, Multi factor analysis was used to determine under what specific situations problem areas identified in the Level-One analysis were more than other siruations. 12

PAGE 25

3.1 INTRODUCTION Chapter3 LEVEL-ONE ANALYSIS This chapter presents the r esults from the Level-One analysis for each of the three persoo types: drivers) passengers and In addition, dle chapter presents results on whether or not and by how much traffic fatalities among each of Ute three person types is over .. represented in florida relative t o the other state groups. While a luge number ofFARS variables and values within each we
PAGE 26

During the pe
PAGE 27

relative to the other sta te groups. Compared to the Safest states, this problem is over-repre=ted by 7.3 percent in Florida. Compared to the other state groups or the nation as a whole, however, this problem is either not over-represented (relative to the Safer states) or slighitd under-represented (relative to the Less Safe states or the nation) in Florida. Table .3.2 Driver Results from Level-One Analysis j l PtobkmSiae I State Group : FAR$ Vuiable Number j j Sbue I Sa! esc I S:acr ;t Leu Safe AD ( 1 1 Dark but lighted l l 1,294 'I 17.8% 18.4 i l 188 ;j 117. 8 4&0 : J Ma.ntlcto(Collision : I ""81 : I 2,625 L 59. 8', J Wd 1 I 87s i f 12.0% ; ( 141.7 1 2252 : I 295.3 I 163.9 HOff roadw-ly :1 1,35011 18.6A ll 65.7 I 47.4:J -l&G 1 --: ) Functiooai.Ciassi.6ruion 1 1 Uibaoother (cccway il.. 9 .,., ;J 122.4 I 226.4: I 8S1 S I t15.S ilv.wn1oa1 :I 822; 1 11.3% : I 101.1 I 199.8 :1 $7.$ .I !.od v .. I I V bon !I l,601 i l 49.w.:l 15.9 l J SM' [ 950 I 39.1 I !I Mwbcip>li'J I 1,421 l l 19.5. : I 29.4 183.7 i l 65.0 I 65.7 :I l,6$0 : I 50 .6. SOl I 100.2 I 4M:J 59.3 : I Traffic Cooud Device Jl 910 : I 12.5% 81.9 i l l$4.1 !60S I 100. 0 'I R<=yclel j 833 i l 11.4%J 7.1 I 6t.a' [ 31. 3 --. ( blitial lmpct Point i 19 o'clock point J [ $&0 I 29.1 : ( 67 1 1 1 Rok .. ColliOon il SINd< : I 2,247 I lO.?o/o I 674 )I 92.61 67.4 I N-sp..ding Mo..;,g I I Atkutonein3yeus 1 ,783 !I US% ; I 68.2i l 82.3[ v.obbons ; I Maneuw-er Pnot \ I T""""g kit tometAae 75 yean or older i I 9S8 L u .2% 1 42.S 52.7 I 76.4 I -47. 9 . 15

PAGE 28

3.4 PASS:ENGBR ANALYSIS Table 3.3 presents the Level-One analysis results for passengers. The followi,r)g discussion focuses on selected areas of concern otganized into two categories: those in which Florida 1s over tepresented by at least 100 percent and a few special areas not included in the first category. High Over-Representation As Table 3.3 indicates, the degree of ove.c .. reptesentation is at least 100 percen t Cot four are,-as including: roads with four lanes, roads with six lanes, regulatory signs, and vehicles tuming left just prior to the crash. The number of passenger fatalities related to these areas ti:om 1994 to 1998 is 907, 331, 965, and 576, respectively. These represent 27.9 percent, 10.2 percent, 29.7 perecnt, and 17.7 percent of all passenger fatalities, respectively. The range in the degree of ova-representation relative to the state groups is 142.7 percent to 272.3 p ercent for roads with four lanes 954.0 percent to 1760.6 percent for roads with six lanes, 135.6 percent to 167. 5 percent fo r regulatory signs, and 132.7 percent to 144.3 percent for vehicles tuming left prior to the crash. Compared to drivers, the degree of over-representation among passengers is higher with respect to vehicles tuming left but is lower with respect to regulatory signs. Areas of Special Concern The very old drivers and passengers are an area of special concern in Florida. A total of 481 passengers died in traffic crashes in Florida from 1994 to 1998 who were 75 years or older at the time of crashes. This represents 14.8 percent of all traffic fatalities among all ages of passengers The degree of over-representation ranged from 36.1 percent relative to the Safest states to 92.2 percent relative to the Less Safe states. In addition, a total of 363 passengers died in traffic crashes whose drivers were 75 years or older Older drivers are more highly over-represented in Florida than older passengers. Again. the lack of restraints use among passengers does not seem to be an area of coneem in teans of over-representation. A total of 2,043 passengers died in traffic crashes in Florida from 1994 to 1998 without wearing any restraint. This represents 62.9 percent of all passenger &talities. Wbile it is a significant problem in an absolute sense, it is not a large problem relative to the other state groups. This problem is over-represented by 14.9 percent relative to the Safest states and by 2.7 percent relative to the Safer states It is, however, under-represented to the Less Safe states by 5.5 percent. The degree of overrepresentation is higher among passengers than among drivers. 16

PAGE 29

! Table 3.3 Psssenger ResultS/fum uvel-One Ansly$is FUS Variable FARSValua I PJobkm. Size State: Group Numba Shu< w ... SaCco Lc::s St.fe All I" Condition Dan ., 19.4o/t 13.3 72.7 132.$ s.o or Mgk t .. ns : I 67 .So/o : j 31.J .... 44.S ... lo 4boC< 9()1 27. ?. 142 7 l1?.2 Z7Z.> 160.5 6bnc$ lll 10..20/e 9S4.0 1898.0 1 760.6 699.3 I"""'" 1<1 2,859 88.0% 24.0 2?.S 4S.2; 27.9 10 lnter&ection 1,198 36.9% S7.9 74.8 1t6..8 fill) 283 ... 7l.2 142.6 172.0 99.0 Urbank:ul llO 10.8'% 83. 7 50.6 194.6 74. 2 u .. i u""" ; 1 616 49.1% 15.9 64 0 IOO.l 41.$ : 606 18.7o/t 1 2.9 161. 7 51. 6 51.3 '""'' U-45,.. 1,642 SO.S% I 564 i l 106.4 I 61. 9 <16.4 I ml Trtf6eoignal 603 18.6% 97.8 161.2 180.2 \16.7 0&0 2.9.7% 100.$ 13.5.6 167.6 B0.6 Flow I ""'""' 1 ,630 50.2o/o 142.4 IOO.S 72.S 95.2 3 o'de>Ck point lSO 26. 9% 51.7 43.9 84.6 51.1 T'llO yean: ot """" 69() 267% 36. 8 2?.3 20.1 27.8 Tumitlg 516 17.7% U27 160.2 144.3 125.4 7S ycau or olcler l6l 51.4 9<>. 0 ll&Z 74.0 l'wcnp/>(}' 1s reus or older 481 lUo/o 36.1 64.l , 92.2 SI.O f1oDd.J. and ,,h. 3.5 NON-MOTORIST ANALYSIS Table 3.4 present !he results of Level -One 2n2lysis for non-motorists. The following discussion foci.ses on selected axe>.s of concem organized into two categories: those in which Florida is over represented by at least 100 percent and a few special areas not included in the first category. 3.5.1 High Over-Repte$entation As shown in Table 3.4, the degree of over-representation is at least 100 percent !or three areas including roods with six Jones, wban noninterstate freeways, and regulatory signs The number of non-motorist f>talities related to these area> from 1994 to 1998 is 413, 518, and 699, re$peetively. 17

PAGE 30

These represent 12.6 percent, 15.8 percent, and 21.3 percent of all non-motorist fatalities respectively. The range in the degree of over-representation relative to the state groups is 586 7 percent to 3008. 2 percent for roads with six lanes, 235.5 percent to 928.7 perCent for urban non interstate freew.oys, and 474.4 percent to 533.2 percent for gUlato.ty signs. Comp ared to both driven and p a ssengers, th e degree of over-representation among non-mototists is higher with respect to regulato.ty signs. Table JA NonMotorists ResJJits from Level-One Analysi s PtOblcmSiu PARS Variable FARSValuu State Group Numb Slwe Stit6t S>f"' All Number of Lanes 10.3 Puoc:OowW Cb.ssifia.boa II Urban other &ccw2y H SIS ,._ . '. il 928.7 1\ye>wncMp u.s. IUgh ... y 2&9% 2.o..l 80.7 99.3 Sure 980 29.9% 26.4 S3.3 24.0 :18.8 1 SpeodUmit 26-46 mph 2.310 70 .4% 21.0 45. 1 30.1 zs. 4 (", rnl Device RWo< 1.602 4&8o/: 9().6 78.2 l 40.4 63. 0 Other Moving Vt<>b.tions 7SI 23.00' 23.7 73.0 81.0 38.0 Suspensioas/'lkvOCI.tioat l A m 13..9% 22.2 S4. 0 53.1 3().8 Pcuon F
PAGE 31

these axe common to drivers and passengers only, including shoulders, urban local roads, and vehicles turning left just prior to the crash. Some are common to passengers and non-motorists only, including intersections. Some are common to drivers and non-motorists only, including urban noninteutate highwa)'S. Still others are unique to the individual person types. Unique to nonmotorists are U.S. highways, roads '>ith 26-45 mph speed limit$, dri v ers with at least one nonspeed related movitag violation and having activities in roadway. Unique to drive.rs a.re traffic signals and being struclc And unique to passengers are municipal roads and drivers 75 years or older. However, these results only show axeas that potentially are problems because any over-representation may be eliminated is controlled or other factors are introduced Exposwe analysis was used to a.cco\Ult for differences in exposure between Ao.r:i-da and the compuison states and is described in the next chapter. 19

PAGE 32

Chapter4 EXPOSURE ANALYSIS 4 1 INTRODUCTION This chapter presents the results from the Exposure analysis for drivers, passengers, and non motorists Two sets of areas of concern were considered : those that have heen identified through the Level-One analysis and those in which Florida i s changed from being under-to over-represented The focus on the first set is how Exposure analysis changes the clircc.tion and degree of over representation for those areas of concern identified in the Level-One analysis. For the second set the following triteria were used in the sereening: 1 The number of tnf6c fatalities related tO unknown values within a F ARS v ariable tepresent no more t1wt 5 percent of the total in Florida; and the share of these unknown tr.U6c fatalities is no more than 1 0 percent higher than any of the state groups. 2. The product between the number of tr.U6c fatalities related to any value within a F ARS variable and its share among all traffic fatalities is at least 10. 3 The difference in the s hares of a particular value within a FARS variable between Florida and a state group is signi.6candy different at the 0.05level 4 2 DRIVER ANALYSIS 4.2 1 Changes From Level-One Analysis Table 4 1 shows how Exposure analysis changes the direction and degree o f over-representation for those areas of concem identifie d as over .. representcd in the Level-One analysis The first two columns list the F ARS variables and values. The next four columns repeat the r esults from Level One analysis for ease of comparison. The last four columns show the degree of over-representation as a result of the Exposure analysis A bold number indicates that the difference between the proportions for the two groups in question is not statistically significant a t the 0 05 level This statistical significance was detennined by applying a test for differences in proportions to the clean crashes (process d e scribed in Chapter 2). Changes in the degree of over-representation in Table 4 1 may be discussed in rw o categories: those with reduced degrees of over representa tion and those with more or le>S the same degree of overr epresentation. 4.2.2 Reduced Over-Representation Among those that changed from being over-represented to being under-represented are urban local stteets and drivers with at least one convictions of non-speed related moving violations Among those with reduced degrees o f over representation are angled erashes, roads with 4 or 6 lanes, l evel tends, intersections, urban non-interstate freeways, wban areas, roads with speed limits between 26 mph and 45 mph, ttaffic signals, regulatory signs, and roads divided but without barriers. In either case, these results support the notion that drivers in Florida were ove.r-xepresented rela.tive to the state groups in these conditions becau se they bad larger exposure Once over exposure is taken int o account, Florida drivers are actUally under-repre s ented, only s lightly over-represented, or over represented with a much lower degree in these conditions. Those with an adjusted degree of over representation of over 25 are still potential problems, including ttaf.fic signals and regulatory signs 20

PAGE 33

this page IS m1ss1ng

PAGE 34

this page IS m1ss1ng

PAGE 35

Table 4.3 from Exposure Analy sis in Level-One Analysis Results for Passengers .. I E.xposw< .wlysis FARSVWbk FARSValut i : 1.25 HI-, All M2S us HI All LipCono61< L
PAGE 36

4.3.2 Changes from Underto Over-Representation In a d dition to those ueas from Level -One Vtalysis that show no reduction in the degree of overrepresentation from exposure control or relatively high degxces of over-representation even with exposure control in Table 4.3 the areas in Table 4 4 are -also pote:ntially se.z:ious concerns 1n each of these areas, Florida appears to be under-represented, highly so in most of these cases, under Level One analysis However. such under-representation appears to be the result of by passengers in Florida to conditions represented by these areas of concern Once exposure is controlled, Florida becomes overrepresented relative to the state groups Some of these areas of concern are h i ghly over-represented (with a degree of over representation 25 or higher), including: crashes involving one drunk driver roads with grades non-junctions, and drivers with at l east one crash in 3 years prior to the cur:ren t one Note that many of these areas of concem are likely to be related. For example. roads with 21anes, non-junction, and head-on collisions may b e more likely t o be seen or occur in rural areas Table 4.4 CbJlD&es from UndtUto Over-Representation for Passtmgcrs Sik l.c'fc.l-Oac A.oaJydt Expotwe Allal'ytt. FAllS Varisbk Numbu Slwe ..... u.s H1 All M1.5 u.s HI : All lo.. 859 ""'" -13 7 9 -14..S 71.6 50.7 24.6 $2.4 of "" 16. lY. -'9.0 -57. 8 .56.$ 19. 7 22.6 2 4 6 23.3 3.7 ........ 2W... 1 696 31 3 -36. 7 &a .,, S t 1.6 5 3 P
PAGE 37

this page IS m1ss1ng

PAGE 38

this page IS m1ss1ng

PAGE 39

5.1 INTRODUCTION ChapterS MULTI-FACTOR ANALYSIS The obje<:RS crashes and persons involved and are relatively reliabl y measw:ed. In addition, only those variables tha t show a high variation in degrees of over rep x escntation across its values are included. A total of 10 variables are included The first row under the headings gives the ovetall prob lem of non-motorists from Level-One anal y sis. Tills is included here for ease of comparison. The second column li sts the included F ARS values within each F ARS variable. For a given F ARS variable, only those values with reasonably large frequency are included. The third column gives the number of non motorist fatalities in Florida related to the particular value of a given FARS variable The fourth column gives the shar e of these non-motorist fatalities among all highway traffic fatalities in Florida related to the particular value of a given FARS ,..,;.bte. For example, there were 183 nonmotorist deaths on curved segment s of roadway in Florida during 1994-1998 representing 8.6 percen t of all traffic fatalitie s on curved segments during the same period. The nat four columns present the degree of ovu-representalion with respect of the number of non-motorist faWities belweett Floz:ida and Safest srates1 the Safer states, and Less Safe states, and aU states in the nation. For eumple, the share of non-motori s t fatalities on curved roads is about 80 percent higher in Florida than the Safest states, 151 percent higher than the Safer st1tes, 170 percent highe.r than the Less Safe states, and 109 percent higher than all states combined Several areas of concan stand out from this analysis Each i s discussed in some detail below 27

PAGE 40

Tsble 5.1 0ver-Rep res entsdon o f NonMotorist Fstwdes in Genersl F AitS Variable PARSValut: I SU.CC' Gtoup Numb !I Slw< s.c ... s .r .. I Leu Sato All PersonTJPC Non-Motori.ns 3.282 37 102 59 123 [ 75 179 ... Cu,. ed 180 = itSt I t ro I tO? 1 o.1 ugh i[?U I ts.w. :lu I 17 [<. I t H t Hf 1 n2 I t2t [97 I t ,014 [32 I t s I I S [34.3'1 I P<8 l30.t% [142 [211 I t87 I t4 3 9BO 193 169 169 111 y Rood I taa lt<.t % 127 [61 [75 [45 Ins l u.3% 1 -27 lo Is _,. io..\co 12.190 I s [ 1 0 5 [120 [ 7 6 ',;.,w 1362 I t9 .2% 1 J t I .a Iss l 7 1699 't7A% [tas 1364 [288 1 ISO rFJow 11,621 156 1115 1138 I 82 1.602 17 166 [56 130 I t-S y..., Old I at 137.9% 1 -J 111 58 l t 9 : 6-15" 1290 H t O IS< 163 l3 t I t2.2% .. I too lt26 179 [U.S% I s1 ItO& lt12 174 I s 196 ,lt04 1 3 7 '"""' 1 7 172 [6 9 l t9 '""" '""'"<>llr .. lh< o.otitvtJ. .. .,.,.; ... "'""' Alignmen!. N on-motorist fataliti .. account for 26 4 percent of all fatalities on straigh t alignmen t, compared to 8 6 percent on curved alignment. Nonmotorist fatalities are highly over represented relative to other state groups on both straight an d curved alignment. However, nonmotorl$t f atalities on curved alignment ate much highly over re p resented relative to other state groups than on straigh t lipment. In fact, the degree of over representa tio n is more than twice as high on curved alignment than on straight alignment r el ativ e to each of the s tate groups. Relative to all states, for examp l e, the degree of o verrepresentation is 44 on straight alignment versus 1 09 on curved alignment 28

PAGE 41

Light Condition. Florida s over-representation with respect to non-motorist fatalities is present under each of the major light conditions: day light, dark, and dark but lighted The most serious problem appears to be telated to dark conditions. First, a total of 1,141 non-motorists died from 1994-ihrougb 1998 in crashes occuning under dark conditions, which is mote than any other light condition.. Second, fatalities under dark conditions tepresent a much larger prob lem than fatalities that O<:CUI under day light conditions. The share of non-motorist fatalities out of :ill trUI'ic fatalities undet day light conditions is 15.7 pexc:ent, compued to 28 5 percent under dark condition s Third. the degtee of over-representation is more than 100 percent relative to each of the tluee stat e groups, indicating that the share of nonmotorist fatalities under dark conditions in each of these state groups is l ess than half of what is observed in Florida. Road Profile. Non-motorist fatalities account for 24.5 percent of :ill fatalities on roads without grade compared to 17.9 petcent on roads with gr a de. Non-motorist fatalities are highly over-represented relative to other state gtoups on roads either with or without grade. However, non-motorist f.a.talities on roads 'With grade are much highly over-represented re l ative to other state groups than on roads withou t grade. In fact, the degree of over-representation is more than twice as high on roads VL-"ith grade than on roads without grade relative to most of the state groups. Relative to all states, for example, the degree of over-representation is 43 on roads without grade versus 91 on roads with grade. Land Use. The problem of non-motorist fatalities, as expected, is larger in urban areas than in rural areas. In fact, the number of non-motorist fatalities accounts for 30.5 percent of :ill highway fatalities in urban areas compared with 15.5 pereent in rural areas. Also, a total of 979 nonm otorists died in rural areas during the study period, compared to a total of 2,303 in urban areas However, the degree of over-representation in nonmotorist fa t alitie s is much higher in rural areas than in urban areas In fact, over-representation is more than 11 times higher in rural areas relative to the Safest states, 3 times relative to the Safer states, and 2 times telative to the Less Safe states. In addition, the degree of over-representation of non-motorist fatalities in rural areas is mote than 100 pereent relative to the Safest and Safer states and close to 100 pereent relati v e to the Le ss Safe states These results on the discrepancy of over-representation between urban and rural areas have important implications related to potential reasons for Flotid-a's ovetteptesentation in oonmototist fatalities. The fact that non-motorist fatalities are more over represented in rural than in urban areas provides evidence that Florida's ove.rtepresentation in non motorist fatalities may not result &om a lack of sidewalks or being more urban than other states. Roadway Ownership. The problem of non-motorist fatalities appears to be the most serious on both U.S. highways and state highways, compared to :ill types of roadway own er ship. A l arge number of non-motorists died on these highways, with 948 and 980 on U.S. and stat e highways, respectively. Non-motorist fataliti e s also represent higher shares of in trU!'ic fa t alities on these highways than in general, with 30 1 percent and 25.7 percent on U.S. and state highways, respectively Finally, non-motorist fatalities on these two types of toadway o'Wilership are very highly overrepresented relative to the other state groups. In contrast, nonmotorist fatalities on interstates and municipal roads are not over-represented in Florida, while they are relatively modestly over represented on county roads. Trafiie Control Device. The problem of non-motorist fatalities related to regularory signs is relatively small, compared to no control device or other control devices. This is true in teemS of either the number of fatalities or the share out of :ill trUI'ic fatalities related to a particular type of control device. However, over-29

PAGE 42

representation is dramatically higher Wlder regulatoty signs than other no device or other devices. Rchtive to all states, for example, nonmotorist fatalities are over-represented in Florida by ISO percent with regulatoty signs, compared to 38 percent without any control devices and 0 with traffic sigtuls:Traffic Way Flow. About 1,600 non-motorists died on non-divided roads and divided roads without bac:iers, representing 23.0 percent and 24.2 percent of all fatalities on these two road types, respectively. The degtee of ovet--cepre.seo.tati:o.n on divided roads v.'ithout banien is much higher than on non-divided roads. Relative to all states, for example, the degree of over-representation is 82 on divided roads without ba.a:iers, compared to 30 on non-divided roads. Person Age. Non-motorist fatalities are over-represented in Florida across all age groups. However, they are most over-represented among those 16-64 years old. 5.3 NON-MOTORIST ACTIVITIES IN ROADWAY As indicated in the Levti-One analysis, having activities in the roadwlly is related to a significant portion of the problem of non-motorist fatalities in Florida. In fact, a total of 1,516 non-motorists died in Florida from 1994 to 1998. This represents 46.4 percent of all non-motorist fatalities in Florida during the study period. This share is 131 percent higher than that in the Safest states, 65 percent higher than in the Safer states, and 24 percent higher than in the Less Safe states. This section reports on a Multi factor analysis of this problem by introducing additional F ARS ,..nables. The purpose is to identify wider wbat conditions this problem is mote se.cious than other conditions. Table 5.2 shows the results of the Multi-factor analysis The following discusses several conditions related to the problem of non-motorist fatalities having activities in the roadway in Florida. Before the individual areas are discussed, it is interesting to point out a pattern related to the results on this problem In dividing the other states into the three groups of state with different levels of highway traffic safety, it was hoped that the degree of over-representation for a particular problem would be higher relative to the most safe states than relative to the less safe states. In most of the analyses carried out in this problem, this hoped pattem does not appear, except this case. For most of the FARS variables and values included here, this pattern does seem to hold. J()

PAGE 43

Tllbl e 5.2 of NoD-Motorist FaUllitie s Rehted to HaviDg A ctivities in Roadway PARS Variable PARS VaNe Probkm Sile Numbn Shan sae ... Sdu Less I All Pcqon fttt' Je(aVlCt:$ VI J t.s J 6 46 .2' ISO 79 S2 8J o., LiW> 357 l m l t30 I 1112 D.d< 637 l55 .8o/ ISS 1 42 130 143 I 411 1<7.0\' I toe lm 21>.= 596 1 <20\'o 1110 122 ISS 529 1 523% 1222 lt41 1128 1 1 08 238 157.6% 1 242 It 50 1447 I 52 u"'"" 56 151. 9% ISl 214 1 44 .2% 1151 ItO< 1 190 LondUse u"'"" 1,015 1113 1103 I tOO Runt SOl l51.2o/o 1 81 IS< 131 !S? [53.3% 1 93 l[n : n [74 u .s. 441 147.2% 1 1 21 1 83 1 34 153 428 1141 ; 1 60 119 1 64 84 144.1% 1101 I4G 1 1 4 154 328 1 204 1 113 [14 1 1<1 Tnffic : 1043 1121 I n I 176 uaffic us )7.3% sso 405 201 268 321 146.8% 1343 1 264 110 167 755 131 65 24 1 68 Onided 'ilh no """" m 46.00'/o 199 124 : 108 98 1 ,102 1 <5.8% 112? 1 69 I 1 72 414 47.2% 1 62 1113 I SOG'Io 1110 166 I 6-1S. 1 1 27 1 96 [211 I n [139 Old 1 1 58
PAGE 44

Ught Conditions. Within Florida, the darku the condition is, the more dangerous for non-motorists having octivities in roadwo.y. From 1994 to 1998, a total of 637, 477, and 357 nonmotorists died while they were having certain activities in roadW>.y under dark, dark but lighted, and day light conditions, respectively. They rep.tesent 55.8 petcent, 47.0 petcent, and 36.2 percent of all non-motorist fatalities under th e se conditions, respectively. Relative to the state groups, non-motorist fatalities having activities in roadway are tepresented under each of the light conditions However, the degtee of over-rq>resentation appears to be highest with dark but lighted conditions and lowest with day light conditions. Functional Oassi.ficatlon. In urban areas, the problem of non-motorists having activities in roadW>.y is most serious on arterials. especially principal arterials. \Vhile it is over-represented in Florida relative to the state groups across all rypes of classification, the degree of over representation is fat higher on arterials than on other types of roadway chssi6eations. Relative to all states, for the degree of over-representation is 150 for principal arterials and 1 24 for minor arterials, compared to 72 for interstates, 50 for non-int erstate freewo.ys, and 90 for local roads Note that the problem was not considered for rural areas by functional classification because of missing data. Land Use. The problem related to non-motorists having acti vities in the roadwo.y is far more serious in urbin a:r:eas than in rural are-as. In fact, the degre,e of over representation of the problem in urban areas more than doubles that in rural areas with twice as many non-motorist fatalitie$ in occurring in urban areas than in rural areas (1,015 in urban versus 501 in rural). Roadway Ownership. The problem of non-motorists h.--ing activities in the roadway is most serious on muoidpal roads in teans of over-representation. While it is ovet-l'epresented in Florida relative to the state groups across all of ownuship, the degree of over-representation is far higher on municipal roads than other types of ov."Dership. Relative to all states, for example, the degree of over-representation is 141 for municipal roads, com pared to 74 for interstates, 53 for U.S. highways, 64 for state highways, and 54 for county roads. Tf2flic-W a y Flow. The problem of non-motorists having activities in the roadway is similar between non divided roads and divided roads without barrius in terms of either magnitude (755 versus 737) or as a percent of all non-motorist fatalities (46. 4 percent versus 46.0 percent). The problem is highly over-represented on both types of r02d However, this problem is far more over-represented on divided roads withou t barriers and on non-.y (45 8 percent for males and 47.2 percent for funales). However, the problem of non-motorists having activities in the roadway is much serious among females in teans of over representation Relative to all states, for example, the de gree of over-representation is 72 among males versus 113 among females. Person Age. The share of non-motorist fatalities related to this problem does not seem to vary with person age in any systematic way. It is l owest at 41.5 percent those 65 to 74 years old and highest at 50.6 percent among those under 6 years old. However, the problem does app= to be more over-represented among the very old and the very young segments of the population. Relative to the most safe states (Safest), for example, the degr e e of over-representation is 263 percent for those under 6 years old, 196 percent for those 6 to 15 y ear s old, 213 percent for 65 to 74 years old, and 333 percent for those 75 years or older JZ

PAGE 45

5.4 SUMMARY This chaptu examined two issues telated to thtough a Multi -fact.Qr analysis. One issue is reltted to nonmotorist in general. Level-One analysis revealed that non-motorist fatalities as a peiSon type a.re in Florida relative to the thr e e stat e groups The question to be answered was: "In what areas of concern is the degree of over representation particula.rly high?" The Multi-factor analysis revealed that non-znototist fatalities are over-tepresented by over tOO petcent relative to e2ch of the three state groups under dark conditions, on U.S. highways, and under regulat ory signs In addition, this problem is f:u: more overrepresented in rural :u:eas than in urban :u:eas, indicating that nonmotorist fatalities in Florida may not be the r es ult of lacking pedestrian faciliti e s. Also the problem is far more over presented on curved alignment than on straight alignment, on roads with grade than without grade, on U.S. and state highways than on other types of roadway ownership, and among 16-64 ye:u:s old than the very young or the old The other issue is related to nonmototists in terms of a s peci.6c area of concern Lev el-One analysis revealed that non-motorists having certaill activities in roadway represent the most serious problem to non-motorists in florida The question to be answered was: w hat siruations is the degree of ovu .. repre.Seo.tation of the problem p:u:ticulu!y high?" The Multi factor analysis r eve aled that non-mototist5 ha\-ing activities in roadway ue over-represented by over 100 percent relative to each of the three state groups under dar k but lighted conditions, on 4-lane or 6 lane roads, on urban arterials, in w:ban areas on municipal roads .under traffic signals, and roads that are divided but with no b:u:tier. In addition, the probl em is far more over represented among females than among males and among the young and old than among the other age groups JJ

PAGE 46

6.1 INTRODUCTION Chapter6 RESULTS This chaptcr summarizes the results from the multi-step analysis ond focuses on highway safety problem areas that are ova-represented in Florida relativ e to other state groups So m e possibl e reasons for why Florida is over-represented in a partic:ular area are also discussed. The chapter concludes with suggestions for future r e search to examine if differences in legislative policies and engineering practices contribute to o1.rer-representation of these problem ueas in Florida. 6.2 RESULTS SUMMARY Table 6 1 summarizes problem areas fo un d to be over-represented in Florida b y person type The key criterion for including a problem area .in Florida was that the d e gree of over-representation relative to an states was at least 25 percent. Problem areas were assigned into four categories: behavior, environment, vehicle, and engineering Those included in the behavior ca t egory relate to the persons involved, including age, gender, history of crashes and traffic convictions, etc. Those in the environment category relate to the driving environment, including weather, time of day, e tc The vehicle category relates to vehicle characteris ti cs, in cluding age, movements just prior to the cra.sh, and the nature of the aasb, etc. The engineering category relates to roadway types, l ocations on roadway, traffic control device, etc. The first column lists the problem areas within each category that appeared to be ovarepresent e d in Flotida The next chtee groups of columns summ.arize the results for driven, passengers, and non m o torists, respecti vely. Three pieces of information are shown for a given pet>on type Two define the size of the problem area: the number of deaths related to the probl e m area and the pucent share of these deaths out of the total number of traffic deaths for the given person type and factor. For example, when junction factor is the number of drivers killed in crashes occurring a t non-junctions from 1 994 to 1998 in Florida is 4,456. The share of fatalities at non-junction locations is 61 percent The third piece of in formation is the degree of over-representa tion of this problem area. The degree of over-representation shown is relative to the nation as a whole and measures by how much Florida' s share of fatalities at non-junctions is higher than the national share. For example, the problem area of driver fatalities on non junctions as a share of an traffic fatalities in Florida is 38 percent highet than driver fatalities on non-junctions as a share of an traffic fatalities in the nation. Problem ueas are defined and an interpretation of the Exposure analysis is fumished in th e o ext section. Im'
PAGE 47

Table 6.1 Summ;uy of Oversll Srudy Results Category ProblemAr
PAGE 48

6.3 OVER-REPRESENTED PROBLEM AREAS The 21 problem "''"" highlighted in Table 6.1 are discussed in the following that describe major features of behavior, envitonment:al., vehicle and engineering aspects of fatal crashes found to be over-represented in Florida when compared to other states. A definition of the problem areas is offe.ed along with an interpretation of the analysis. 6.3.1 Behavior11! Aspects Florida is over-represented in six areas that <:an be categorized as behavioral. Common to drivers and passengers are fatal crashes involving one drunk driver. Two prob lems areas unique to passengers are: persons 75 years or older and drivers with one prior crash in lut three years The remaining three problem areas are common to non motorists. These include: nonmotorists having activities.in .toadway, d.civers with at least one non-speed related violation conviction in du:e:e yeus, and drivers with at least one suspension or in three yeus. Some discussion is offeted for each of the six problem areas. Crashes with one drunk driver. The number of drunk drivers in a fatal crash was a variable derived by NHTSA. Data &om the vehicle .61e are analyzed and if there is sufficient information to conclude that a driver was drunk. i.e., if the BAC is positive, or if the police r eponed alcohol involvement, then the driver is counted as a drunk driver. A driver being charged with an :alcohol violation alone does not result in the driver being counted as a drunk driver. From 1994 to 1998 in Florida, 2,406 drivers and 859 passengers died from tmf6c crashes involving one drunk driver This represents 33 percent of all driver fatalitie> and 26 percent of all passenger fatalities during the period. These shares are slightly lower than those for the nation as a whole. This under-representation, however, 2-ppean to be the result of Florida's under-exposwe to travelll1g by drunk drivers. After exposure is controlled, driver and passenger fatalities from crash.,. with one drunk driver become highly over-represented. In fact, the degree of over-representation is 72 percent with respect to drivers and 52 percent with respect to passengers Florida law' states that it is illeg:alto operate a motor vehicle if blood alcohol concentration (BAC) exceeds 0.08 g/dl. Florida observes a per se" law': BAC at or above 0.08 g/dl is per se illeg:al. Florida drivers can restore driving privileges during the 6-month of Administrative License Suspension only if spec:W hardship is demonstrated. Unlike some states, Florida lacks legislation that imposes mandatory jall time or community serrices for first offenses. Differences in legislative laws and enforcement practices between Floricja and other states may contribute to the over-representation. Thus, it may be -.,-orthwbile to document the differences and to gain awareness of countermeasures that the safer and safest states have implemented that have directly reduced crashes involving one drunk driver. Further, it may prove meaningful to review srudies that have documented crash reduction rates in states where BAC level was reduced &om 0.10 'Fku:id2 duptu 316 s.ectioa. 193. State$, wbifot a "per S<'' DU!Iaw, maioW. ohat i.s illegal to opente vehick i! the requisite BAC kg>! limit i.s c:xcec:dcd. E'Vidcnce of a pcaon'' BAC ac or above tbe pteK:a'bed limit is illegal ln other words, need be no fioding ofimpairmeslt. < bnp:lfwww.!astcall.orgltopiq/pers.htm> ojFa/41 Crashes in Florida: Final R'}f1 J6

PAGE 49

to 0.08 gm/ dl in addition to befote and aftet studies that document changes in crash ntes with implementing administrative license revocation laws. Pa&&engen age 75 years or older. A total of 481 p2ssengers :age 75 ot oldet died in tnffic crashes in Florida from 1994 to 1998 teptesenting 15 petcent of all p2ssengct fatalities. The shate of passenget fatalities among persons 75 years or oldet is more than 50 percent highet than the nation's share. Because the quasi-induced Exposwe analysis is not applicable to problem ateas chatacteti.zed pwely by passenget chancteristics, no measwe of ovet reptesentation i s available. As a tesult, any ovet-tepresent:ation is likely to once exposure is taken into accowlt Even though this particulat problem area may not be over-represented in Florida, steps should be taken to understand different charactedstics of these etasbes and attempts should be made to reduce the crash rate involving passengers over 75 years old For example, pattems may be found whete driver,; in these ctashes are within specific :age g;roup and/or rugher instllnces of night driving in these cnshes. In a recent report (Dissanayake and Lu, 2000), sevetal cdtical issues/ concetns of o lder drivers nationwide were ranked in order of most cdtical and they are: location and size of ttaffic signs and lettering, nighttime visibility, perception-teaction time, gap acceplllnce, deficiencies in driving knowledge narrow lanes, driving in congestion, maneuvering curves, and freeway driving Several factors such as driver demogtaphics and history, types of roads, time of cwhes, first harmful events, S>fety equipment use, etc., could be investigated further so that countermeaswes ditectiy rugeting the problem can be identified. Drivers with at least one prior crash in three years. From 1994 to 1998 in Florida, a total of 271 passengets died in traffic cnshes whose drivers had been involved in at least one etash in the previous three years. These fatalities repreSOS\t 8 petcent of all passenger deaths from ttaffic CtaShes. This share is significantly lowet than the national shate This under-representation, however, appeats to be the result of Florida's under-exposwe to the: problem. Once e.xposw:e is conttolled, the problem area becomes over-represented with a DOR of 59 percenL Fwther investigation of the ptoblern area may be beneficial to bettet understand driver characteristics related to prior driving convictions such as :age, gender, etash history, and crash characteristics such as speed l evels and lighting conditions. Non-motoristS having activities in roadway. The problem telated to non-motorists having activities in the toadway appears to be significant in Florida. This refers to non-motorists u-a!king. riding with or :against ttaffic, playing, working, sitting, lying, stllnding, etc. in the roadway From 1994 to 1998, a total of 1,516 non-motorists were suuck by a motor veiUcle whlle having activities in the roadway. representing 46 percent of all non-motorist fatalities in Florida. The share of nonmototists killed while having these activities in toadway is 83 petcent higher in Florida than the national share. While the quasi induced exposute apptoach does not apply to this problem, the Multi-factor analysis suggests a numbet of situations where this problem is highly ovet-reptesented. These include dark but lighted conditions, 4-lane roads, wban arteDals, wban areas, municip.I roads, female, and among the very young and old. For each situation, the degree of over-representation relative to the nation is over 100 percent. In other words, Florida's share of fatalities involving non-motorists having activities in roadway is more than two times the national share in each of these situations. J7

PAGE 50

Seven! roadway design and tnffic opemions features improve or impact non-motorists safety including: signs and markings, signalization sidewalks and paths, refuge islands, capacity, pedestrian level of service and lighting. Florida may want to review some of the "best pi.ctices" from other states relating to road\Vay design and uaffic operations and improving non-motorists safety Drivers with prior convicd.ons. Drivers with at least one suspenslonS/revocation6 or one non .. speed moving violation conviction in three years prior to the crash in question appear to be highly involved in non-motorist fatalities in Florida. A total of751 non-motorists died in uaf6c crashes in which the driver had at leas t one non-speed moving violation conviction in the previous 3 years, compared with 452 fatalities with respect to suspension or revocation (these represent 23 percent and 14 percent of all non-motorit fatalities, respectively). Both these $hares are over-represented in Florida being 38 percent and 31 percent higher than the national share, respectively. Although the quasi-induced exposure approach is not applicable in thi. situation, every different aspe<:t of the problem should be cons idered and potential solutions should be sought out. In case of drivers with prior conviction, it may be beneficial to understand legislative differences between Florida and other states, particularly Califomia, identified from the analysis as one of the afest in this area. For instance, the point system differs significantly between the state In California, license suspension occurs after 4 points in 12 months (sus pensio n for 60 days), whereas in Florida it is 12 points in 12 months (suspension for 30 days). 6.3.2 Environmental Aspects Two problem areas in this category are related to lighting condition at the time of the crash Crashes occuctiog under duk but lighted conditions are a problem area for motor-vehicle occupants, while crashes occuctiog undu dark condition are a problem area for non-motorists. The literature suggests some possible factors contributing to high crah occurrence at night. Glare from a variety of sources such as headlights, street lights, and building lights. Dirty windshields presents probl ems in turns of both visual discomfort and reduced visual efficiency. Fatigu e and lack of alertness could also be considered factors affecting the drive< Lack of congestion may encourage speeding Low visibility of warning signs and pedestrians Headlights cannot foUow curves, dips, hills in the road Alcohol;. more prevalent at nighttime Dork but lighted condition. From 1994 to 1998, a total of 1,294 drivers and 631 passengers died in Florida in crashes occuctiog under dark but lighted conditions, representing 18 percent of all driver fatalities and 19 percent of all passenger fatalities, respectively Fatalities among vehi cle occupants under these conditions are over-represented in Florida relative to the nation as a whole. The share of drive< and passenger fatalities in crashes occurring under dark but lighted condition ;. 48 pucent and 45 percent higher in Florida than in the nation, respectively. s Suspension means tc:mpowy -o;i.thdtawal of a licen$CC:'S privilege to drive a motor vehicle. 'Revocation means tbat a lieecset's privilege to drive 2 motor vehicle is tcaninated. A ntw license may be obtained only a.s. pitt
PAGE 51

Over-representation of crashes occurring under dark but lighted condition among drivers and passengers in Florida does not appear to be the: res ul t of o v er-exposure. Without exposure control. one would expect over-representation of such conditions in Florida because Florida has better weather conditions and appears to be more urbanized than the nation as a whole. More travel is likely to be done at night when the weather is good. In addition, more r:n vel at night is likelY to be completed under lighted conditions in urban areas than in rutlll areas. Once exposure is controlled, one would expect such over-representation t o disappear. However, the over-representation of duk but lighted conditions for drivers and passengers does not disappear when exposure is controll ed. Warrants for roadway lighting and standards of illumination can be compared between different state groups so that safety improvements leading to the reduction of nighttime could be achieved Dark condition. From 1994 to 1998, 1,14 1 non-motorists died in crashes occurring under dark conditions in Florida, representing 35 percent of all non-motorist fa t alities during this period. Relative to the nation as a who l e, fatalities among non-moto:dsu in crashes occun:ing unde r da.rk conditions are over-represented by approximately 25 percent. Ovu-zepresentation of crashes occuning under dark conditions among non-motorist fatalities in Florida does not appear to be the result of over-exposure. When exposure is not controlled as in the Level-One antlysis, the share of non-motorist fatalities oc:cu.ning under dark conditions among all non-motorist fatalities is 18 percent higher in Florida than in the nation When exposure control is introduced, the degree of ovu-representation increases to about 28 petcent. Possible factors conttibuting to this overr:eprescntatioo is the prevalence of alcohol use for both pedesu:Wts and drivers involved Low visibility of pedestrians by motorists may be anothe< factor. Recommended safe walking tips include: wearing reflective clothing. crossing at inte

PAGE 52

Exclusive left-cum lanes 2t sign:ali2ed intenections : the highest crash rates occur in p..,.U,sive-only left-tum approaches, followed by protected/p..,.Ussive. Level One analysis with xespe:ct to the vehicle maneuver ptior to the crash indicated that rumiog left was a problem area with a total of 806 driver fatalities and 576 passenger fatalities (11 and 12 percent, respectively). Comparing with the other states, a significant over-representation of 107 percent and 125 percent was noted. Exposure analysis increoses over-representation to 120 percent and 155 percent, re-spectively. Warrants for installing left nuning lanes and design standards can be compared between Florida and other state groups tht are safer in this particular area. If a cortelation between the differences in warrants and standards and over-representation of crashes while vehicle is tuming left in Florid is found, then appropO.te countemteasures could be identified and implemented. Head-on collisions. From 1994 to 1998 in Florida, a total of 980 drivers died in head-on traffic crashes, representing 22 percent of .U driver deaths involving collisions with another motor vehicle in transport. 1his share is significantly lower than the national share (45 percent). The under representation, however, appears to be the resul t of Florida's under exposure to this problem. Once exposure is controlled, the problem area becomes over represented with a. degree of over representation of 44 percent. Several factors may contribute to head-on collisions' including excessive speed, lighting problems, absence of medians and slippery s urfaces. Warrants for median instillation and design standards, speed limits, and lighting warrants and standards can be compared between Florida and other state groups. Initial impact: 9-clock point. From 1994 to 1998 in Florida, 1,195 drivers died in traffic crashes after their vehicle was struck directly on the driver side, representing 18 percent of .U driver fatalities involving collisions with another motor vehicle in transport. 1his share of driver fatalities is 36 percent higher than the national share. When exposure is controlled, the degree of over representation increases to 45 percent, though the difference between these two m e asures of over representation is not significant 21 the 0.05 level Initial impact: 3-clock point. From 1994 to 1998 in Florida, 780 pssenger fatalities occurted after the vehicle was struck directly on the passenger side, rep r esenting 27 percent of all passenger deaths involving collisions with another motor vehicle in transport. 1his share of passenger fatalities is more than 51 percent higher than the national share. When exposure is taken into account, the degree o over representation decreases to 39 petcent, though the difference is not signific-ant at the O.OS!evel. 6.3.4 Engineering Aspects Nine of the problem areas where Florida is over-represente d are included in the engineering category. Four of these areas are common to drivers and passengeu: non-junction, regulatory signs, shoulders and roads at grade. Crashes on U.S. roadway s and state highways are common t o non'Tt>fli< &g;oecook ITE, 1999. Tsbl 7 1 4 P'3' 206. 40

PAGE 53

mototists as weD as roadmys classified as urban non-interstate freeways. Traffic signals are unique to dtivers while divided highways with no ban:iers are unique to Relation to junction-non-junction. A non-junction crash is defined as a non intersection, non intersection xdated or driveway access crash From 1994 to 1998 in Flotida, 4,456 driver$ and 1,819 passengers died in crashes at non junctions, representing 61 percent of dtivers and 56 percent of passengers of all fatalities \1/hen the junction factor is considered. These shares are significantly lower than the national shares. However, when exposure is controlled, over representation by 38 percent and 29 percent is noted for drivers and passengers, respectively. Traffic control devices-regulatory signs. Regulatory signs, a type of traffic control device, contribute to safety by conveying essential control information to drivers, such as the legal requitements to stop, yield to the right of way to another road user, travel in the correct direction, e tc. From 1994 to 1998 in Florid., 2,339 drivers and 965 passengers were killed in crashes that took place where regulatory signs are used. These numbers represent 32 percent and 30 percent of driver and passenger fatalities, respectively. After controlling exposure, there is the significant over representations of 64 percent and 45 percent respectively compared to the nation as a whole. Traffic control device-traffic signals. The term traffic signal applies to the control of traffic at atgrade street locations, ramp metering, lane-use control, flashi t\g beacons, rail-road highway grade crossings, and moveable bridges. The F ARS code "traffic signals" includes: Traffic control signal (on colors} without pedestrian signal Traffic control signal (on colors) with pedestrian signal Traffic control signal (on colors} not known if pedestrian signal Flashing traffic control signal Flashing beacon Flashing highway traffic signal, rype unknown, or other Lane use conuol signal Other highWlly traffic signal Unknown highway traffic signals From 1994 to 1998 in Flotida, 910 dtivers died in traffic crashes that occurred at with traffic signals, representing 13 percent of all dtiver fatalities. After exposure is controlled the problem area is over-represented by 29 percent. Rcbtion to roadways-shoulders. This vatiable represents the number of cras hes that occurred on the shoulders of the roadway, which is that portion of the roadway contiguous with the traveled way for accommodation of vehicles stopped for emergencies, and for lateral support of base and surface courses From 1994 to 1998 in Florid., a total of875 drivers and 283 passengers were killed in crashes occurring on the shoulder, representing 12 percent and 9 percent of all dtiver and passenger fatalities, respectively. The share of fatalities related t o this problem area is higher than the national share. After exposuxe is conttolled, the problem area becomes ovetteptesente:d ..vith a DOR of 333 and 338, respectively Traffic way Bow-divided with no barriers. In general, highWllys can be divided into two or more roadways by leaving an intervening space or by a physical ban:ier or clearly indicated dividing section so constructed as to impede vehicular traffic. From 1994 to 1998 in Florida, a total of 1,630 41

PAGE 54

passengers died on divided roadways with no barriers, representing one-b.Jf of all passenger fat>lities during this period Exposure analysis lowers the over-representation, but still it is 71.5 percent. Roadway ownership--state highways. A total of 980 non-motorists died in Florida beto.-een 1994 and 1998, represent a 30 percent share, which is significantly higher than the national average. The results &om the Exposure analysis show the degree of o ver-representation o percent Roadway ownership-U.S. highways. A total of 948 non-motorists fat>lities in Florida between 1994 and 1998, represent a 29 percent share, which is significantly higher than the national average. The results &om the Exposure onalysis show the degree of over representation of 69 percent Roadway profile-roads at grade. The nwnber of fatalities in cnshes that occurred on roads at grade is 810 drivers and 332 passengers, representing 11 percent and 1 0 percent of all driver and passenger fatalities, respectively. These percentages are not over-represented with respect to other stotes as the Lev el-One analysis reveals. Once exposure is controlled, the problem area becomes over-repr esented with a DOR of32 and 82 for drivers and passengers, r especti vely. Functional aasslfieation-urban non-interstate freeway. With 518 non-motorist fat>lities and 1 6 percent share this category is highly over represented acc ording to the Level-One analysis. Once exposure is controlled, the problem of urban n o n interstate freeways related to f>tal crashes is still o.-er-represented with a degree of over-tepresentation of 30 percent. [.,.rfi&-" of Fatal C=hrt m Florida: Final Rrport 42

PAGE 55

Chapter7 FUTURE RESEARCH DIRECTION 7.11NTRODUCTION Possible fututt studies should focus on those problem uea.s summarized in Table 7 1 in which Florida is highly over-represented rehtivc to othet states ofter differencu in expo sure are ppropriately conttolled. The top thtee problem orcas with the highest degrees of over repre&entotion for drivers were: shoulders, vehicles turning left just prior to the cnsh, and croshes with one drunk driver The top thtee problem areas for passengers were: shouldeu, vehicl es turning l eft just prior to the cra.sh, and rood segments at grade. The top thtee problem areas for non-motorists were: haviog ac:Uvities in the roadway, state highways, 1Uld U.S. highways. Alternatively, the choice of such studies should toke into account not only the degree of over representation but also the size of the problem The index for each perso11 type a11d the total index in Table 7.2 are aeated just for this :alternative pproach to selectis>g future studies Specifically, the index for a given pers011 type the potenli>l reduetio11 in the number of &Wities related to a cen::ai:D area of concem Cot that penon type if &he overtcptc:SC:I).t:ation in this area of c::once.m wue elimin>ted For a specific state group j, the index is c:alculated as R; = V [Q, I ( 100 + Q,)], where Vis the number of &talities related to a given problem uea and Q, is the degree of over-representation of this probkm area in Florida relative to state g<. The top thtcc problem areas with the highest index v:alues for drivezs -..-ere: non-junction cruhes, crashes involving one drunk driver and regulat ory signs The top thtee problem areas for pauengera were: roads dmded without barrier, nonjunctions, and vehicles turning left just prior to the crash. The top thtee problem areu for non-motorist$ were: having ac:Uv:ities in roadway, state highways, ond U.S bigbways. The top thtcc pxoblem.,.... for all person rypes combined were the same u those fox drivers ...-ben the index is used 7.1 Top 'Three Problem Areas Bued oo Dqpu oF Over-representatio and Index Vlllues 1 Penon Type i l Criteria I Top Thtee Problem Areu Joa-s i iDOR I Sbou!llnda : I Noo-junction One drunk driver : I Regulatory sign 4J .

PAGE 56

: I Problem hess [ Drinrs _j Pa .. eneus i j Non-Motoriaul Tow I DOlt :llndex' il DOR' .jlndex' lndex' ,(N on-Jutlction$ ___ I 381 1,227. Q I 409, C I 1 1,63 6 : I One Drunk Drivet I nJ! 1 ,007 1 2!.1 2941 I @I I Regulatory Sign mj 45 [ 2991 I jshould-. 333; 1 6731 m[ 218[ I [TumingWt 120 1 350 1 I [ _190 [Having Activities in Roa dwsy : I Cl I 83; 1 [Divided with no Battier : I [_!21 682= 1 il .[om but Lighted 35 !I m [ 331 157! 1 1 I :[state Highway il C I I 7111 jus. Highway I Jl I I : I 69 [ 387[ 387 I wtiallmpact Pniot9 o'clock point I 45 I 311C I I = I I 371 JRoada ou Grade I 321 1 150' 1 I I 346 IHHd-on Co(lis;on mCI : I I I Dark CooGven wioh atleut ooe swpmaoo or revocation io 3 yean 4 [omrenwithooepciorcnshin3yeau _CCL 59[ 101 1 I Sowcc:: Compiled CtOJn te:Nits an C lupten 3-5. 1. DOR 1tandt Cor "Degree of OTerR.eptcsenlSoon. The m.eaJ\1fe &om the ExpoaU[(; ao.al}'ab wu wed it &he relative exposu.re between Florida and the ution u 1 wbole is diif'erenc. OthCl"WUCt the me:a.Nlc from the Levd-One a.oalysil wu wed. 2. The irukx indicates potttn:W ttduction in the ownbet o &tal.itie tdued to utUin uea of otw< ...,)yUt bccow< quu;.induced _. c:anftOt be 'CI:IC&SW:ed for them. The rem ainder of the section defines a stmtegy for future reseuch to explor e why and bow the problem areas identified in this reseuch are over-represented in Flori da compared to other st&tes. 44

PAGE 57

7.2 FUTUilE RESEARCH OBJECTIVE Chapter 6 summarized specific problem areas wit!Wl the behavior, enviro_omental, vehic;le and engineering C>tegories identified by the analysis as over-represented in Florida when compared to other "safer" states. Future research should be directed to understand why. Florida is over represented in these problem areas. The basis for the research is to dete.tmine if certain roadway engineuing practices and legislative policies Contribute to Florida s ove.rareptesentat:ion in these problem areas when compared to other state groups. In other words, the proposed research will answer the question; "do differences. in engineering practices and legisl ative policies directly correlate to the over-representation?" The research findings can be instrumental in pinpointing pecific ueas where the State can adopt safety improvement measures that are suecwfully practiced in other state groups identified as ustfer" or "safest,.. 7.3 SUGGESTED RESEARCH APPROACH The top tlu:ee problem areas with the highest degtees of overrepresentation foe drivers were: shoulders, vehicles tuming left just prior to the crash and crashes with one drunk driver The top three problem areas for passengers were: shoulders, vehicles turning left just prior to the crash, and road segments grade. The top th:r:ee problem areas for nonmotorists were: having activities in roadway, state highways, and U.S highways. Taking a closer look into these problem areas Florida's design standards and Ieg;slative policies will be compared to those of other state groups. From this comparison, comprehensive reasons why these p roblem areas are ovecrepresented in Florida will be better understood. Also, the role these differences play in over:all highway safety in Florida will be defined. Consequendy, safety professionals in Florida can make infom:ted decisions on different approaches to adopt traffic safety countermeasures to address these problem areas. 7.4 PROPOSED RESEARCH METHODOLOGY Prior to identifying differences in these practices and policies, the problem areas highlighted in Table 6.1 will be explained in sufficient detail. A total of five tasks are proposed in this research plan. The first task is necessary to pro..;de a more comprehensive understanding of safety issues associated with each problem area Noteworthy here is that problem areas (or variabl e s) f:all into two main types: human and structural. Legislative policy differ
PAGE 58

Task 1: Problem Ate as Investigated For evaluativ e pwpose., problem areas are grouped into four major categories. Definitions from FARS and from national and stat e manuals will be provided for each problem-uea. The following problem areas will be investigated: Behavior category Crashes with one drunk driver Nonmotorists having activities in the road Drivers with at least one nonspeeding moving violation in 3 years Dtivers with at least one su s pension or revocation in 3 years Environmen t category Dark but lighted condition Dark condition V ebicle category Tuminglefr Heack>n Collision Engineering category T roffic w:ay flow-divided with no barriers Traffic conuol device-mffic signals Relation to roadw:ar.-shoulders Task 2: Policy and Programs Differences in legislative laws and enforcement practices b etween Florida and other states may contzibute to the over-representation. Legislative policies that regulate drinking and driving penaltie s in Flori d a and other states will be compared as well as definitions of different traffic violations that relate to over .. represented variables th at describ e driven' behavior and history lntemet and other resources will be used to compare state statu tes pertaining tO drunk ddving. convictions, suspensions and r evo<:wnent the telationship between crosb tates and driver uac viol ation history, the effect s of selective legislative policies on the teduction of uaffic crashes law enforcement practices that help reduc e cra s h tales, and successful intetagency coordinated safe ty p t ograms. Task 3: Design Standatds and Practices Design standards and implementation practic.,. identified in Task 1 will be used in this task as elements to be c ompared between Florida and othet safet and safest state groups. Several comptehensive subtasks will be accomplished so that meaningful comparisons can be presented. As pteviously mentioned, a survey will be adminisuated to other state groups to compile information on their design s tandards and practices. Survey preparation tasks include : identifying key persons within agencies to survey, designing the survey may require solicitation of input from uaffic safety engineers and consulting national and state manuals, pre-testing of survey insuument, revising survey b ased o n ptetest results, administering the survey, sending remindetS and conducting follow-ups through mW outs, &.xes and em.US. 46

PAGE 59

Task4: Data Analysis Florida's design standards and legislative Jaws will be compared with states within each of the tluee state groups: less safe states, safer states, and sd'est states The purpose of the Iaitet comparison is to examine if diffetences in standards and Jaw between Florida and other states show any correlation with o ver all highway safety. Furthermore, if data pennitting, multivariate analysis may be carried out to examine the role of differences in these standards and law in overall highway safety wlille holding other charactetistics constant. There are two basic approaches to analyze the survey results. One approach involves simple comparisons of those standards and law between Florida and other states in areas that are determined to be over-represented in Florida The focus would be on whether those standards and law in Florida ue lower or less restrictive than those in other states. If they some conclusions may be drawn about the role of these lower standards or less restrictive law in the areas of overreptesentation in Florida. The alternative approach involves multivariate rc:gxession analysis of overall highway safety at the state level by including not only differences in those standards and-law but al,o differences in other factors across states. This approach is commonl y used in the economics literature, =mining the effects of policies related to eat belt law, alcohol law, annual vehicle inspection Jaw, driver licensing law, etc. The first approach is simple to understand. It directly deals with the areas of over-representation in Florida. However, any conclusions from this approach are not conclusive simply because it ignores many other factors that may have also played a large role in the over-representation of these areas in Flori&. The second approach, on the other hand is more technical It deals with overall highway safety, such as fatality r.ates per unit of travel Its ad<>antage is that conclusions from this approach are likely to be more conclusive simply because it accounts fot othe.r: factors that can be mea sured. The applicability of this second apptoach depends on the data collected from the survey, including the number of responses and the complexity of standards and law. Task 5: Findings and Conclusions The final task will dircuss the findings and implications of this research The report will also include p..Ccical conclusions a$$essing the extent to which differences in legislative policies and engineering practices between Florida and other stlltes inOuence highway safety and suggest possible changes existing practices and legislative policies. SUMMARY This chapter p t esented the top tluee problem areas for each person type based on degrees of over representlltion and the size of the problem (mdex) These results provide direction for potential research studies that should include these and other problem areas discussed in Chapter 6. The chapter also defines a stt:ategy for directing future research to explore why and how problem areas ue over-represented in Florida compared to other states. The research methodology includes 6ve proposed tasks lind mainly focuses on examining the differences in engineering practices and legislative policies between Flori& and the other state groups. 47

PAGE 60

Biblio graphy Cirillo, Julie Anna. Access Conttol &lfety Effectivenes s of Highwoy Design Features Volume t. Fedeal Highw.y Administntion, 1992. Clla)', Richud E., Yacov Y. Haimes, and James H Lambert. Comparison of Two Models Evaluating Automobile Safety Featw:e s of Engitumng. March/Apdl 1999. Dijkstn, Atze, and Fred C.M Wegman SAFESTAR European Effort To Establish Safe Road Standards. 1RN""s 201 March-April1999. Diuanayake, SWl.anada and J. John Lu. Effect of Lighting Condition on Highway Crash Involvement of Older Drivers Southeastern Tnnsportation Center, Uni. National Academy Press Wshington, D.C. 1999. A Policy on Geometric Design of Highways 2nd Stteets 1990 American Assocition of Sttte High\Wy and Transportation O.fficWs, 1990 41

PAGE 61

this page IS m1ss1ng

PAGE 62

Appendix A: State Groupings arid F ARS Variables so

PAGE 63

this page IS m1ss1ng

PAGE 64

The tables A2A4list the FARS variables unique to the Accident, Vehicle and Drivet, and Person files, respectively The shaded variables :u:c not inchlded in OU< amlysis. these variab les :u:e foU< identification numbe
PAGE 65

this page IS m1ss1ng

PAGE 66

this page IS m1ss1ng

PAGE 68

this page IS m1ss1ng

PAGE 69

this page IS m1ss1ng

PAGE 70

this page IS m1ss1ng

PAGE 71

this page IS m1ss1ng

PAGE 72

this page IS m1ss1ng

PAGE 73

uwlftd !he eoac:spoading S-tate group is noc m.ci;tic::ally difJemu: t t the O .OS kvd. ofFara/CnzJh
PAGE 74

Appendix C: Results from Exposure Analysis

PAGE 75

'C-1. PARS Varlsbk Value M25 L2S HI All M25 L2S HI All ll.igb 18. 19 liS ... .. 29; 72 3S Mulner of Collision .10 40 l I Nurt'lbtt or l..alx$ 4br.et 138. 2()9 291 ISS .; 15 3 ...... ., 1,739 1,91 8 702. .I) 13 129; 9 1 .... 61. 22! .10 .. zs: 2 11 56 124 66 15 14 50 19 ,.. 142 225 29S 164 298 4$1 594 333 Off 66: ., 29 4 -83: 122: 226 851 176 -46. 2 -24 doal 101, 65 200 87 ., -n: l -5 JAM Use 16 59 : 95 )9 -s .(\ 19: I 29 184. 6S' 66 22 19 61: 24 :!G-45 .10 100 59 o I I 82 134 161 100 24. 24. 8)' 2'J 198, 212 161 74 75 j n .. yF!ow >buOn=ott J>DO< Cn< two is natisti<::ally o/Fal41 C""h" ;, Fhrida: Filtal Rtpt 6J

PAGE 76

TabkC-2. r from TT J n" .. Probl em Size LC'rcl-Onc ADalysia Expos..ue Analylit: Va.ri.tbk PARS Value. Swe M25 1.25 H I All M2S L1! I HI 'I All Drutlk DriTcn On 2,406 .n.tor. -9 l : 6.4 -7.6 -7.4 81.5 : 72.$ 74 0 72.2 980 ; 22.3% -42.2 -49.1 -49.2 ...... 47.0 l$. 2 ...., ofl.Mie; 2""" 4,171 S7.lo/ .?r. -31.6 -34. 6 -29 .0 21.2; 15. 3 15. 2 16.3 I P..S. Gnd< 810 1 11.1% 9 68 58.4 >0.2 43.4 29.0 l1A 61.2"'1 .2 0 ..... l9.S' 416 l4.? l7.7 LondU.e Run! 3,67G 50.5'"1. no 14. 3 22.0 11.1 15. 2 Spe6% -'" .?A. 6.8 l.5 -4-l 3.2 ;,,,.,;,o..n7 I '=L. Ab ..,. tlul ......... '""'"""' 64

PAGE 77

Tabk C-3. Chant,et from ExjMsurr Ana!Jm in Ltvti.Oitl Ana!Jm &wits for Pasnngm i I FARS V.o.bk I 'FARSVw I I r Ligbt Condition 65

PAGE 78

C-4 Ch Umkr-f() .D ; . Probkm SQ:e Levd-OuAAalyab PARS V a riable FARSVaJue Num b er Slwc M2S L2S HI All M2S. us HI All One 85? 26.4% -13. 7 1 H .S .6 71.() 50 7 2-4. 6 524 Hcod-on 16.1% .. : I a -51. 8 56.S S2.4 19. 7 2.4.6 2M' Number afl..oru::$ 2bnes 1,696 52..2% l l. l -,0. 7 -38.8_ 3.3.7 5.1 1.6 S l 3.7 : Cnd< l32 11).2% -5?.8 60 9 74.4 8$.1 95. 4 820 ronlo NOn junction 1 819 $6.()1', -17.6 -20. 9 -24. 0 3 25. 3 l29 3 1 .2 28.6 UndUsc Runt 1.633 so.Jt. -11.2 27 6 -lH -221 ll.2 10.7 Nwnb of Cmh" :=,"""in l Z71 8.3% -39. 6 -41.9 632 54 1 00.5 58.8 ; 'Oay IS StaQS not one at the .OS I C'\'el Table C-5. Changer from E!f}omn Anafyris in Level-One Anab ris or Non-mof()nm lllttli!f11io ofFaial Cr4Shu;. Fhrida: F inal Rrport 66

PAGE 79

Appendix D: Results from Multi-Factor Analysis 67

PAGE 80

Tablt D-1 011t1'-F.#prrstnlation ofNon-Motqrift Fatalitits in Gtneral -! : I Problem Size I i PARSVuW>le FARSValue I Number E l Safest I Safer I I Pe..on Type : I Non-Motom" 1 3)82' 1 : j Alignment [Stt>igh t .I 3,094:1 26.4% I 231 75\ l 791 44 : I i l I 183 .[ 86% 1 80 [ 1s 1L.E..1 109 [ Light Co n dition i l Day Light I 986 ,j 1s.1% 1 uJ 91 , !17 [ 46 l o:uk I 1 ,141:[ 2s.s% 1 1 031 112J 121 1 ;! D:uk but Lighted I 1,014 , 34.3% [ 10 1 32 181 15 I Pro61e [Levd 1 : 1 24.5% 1 241 n[ 43 ; [ Gnde I 250 ![ 17.9%. [ 70; 1 118 i [ 115 91 I R.d:ation to Jwu;tion I Non-:Juo.ction I 2,337 i l 27.0%; [ 6!.[ 108 i [ 1271 1 80 ; I Intersection I 629 : [ 15.6% I 12[ 130 I 74'[ 42. I l Intersection [31 -17. Related J I.andUse i [ uro.n I 2,303 I 30.5% d 10 I 38: 1 44 ; 1 20 j[Runt I 116 ; [ 126 : I 951 102 i -i Roadway I Int=ttte Gl !1.5 % : I Owne,.hip : I us. Higbw.y 9481. 30.1%_1 142 I 211 i l 1871 143 I Sure Highw.y 98o 1 25.7% 1 931 169 i [ 1 691 Ill County Road 1ss I 14.1% 1 27[ 61J 75[ 45 < I MW>icipolity 725 ,[ 26.3% I -27[ 0 ; [ s[ -16 I T m f6c Conttol ; DevKe }one 28.1%[2[ : I [ Traffic sign>! I 362[ 19.2% 1 -31 [ 48 i [ ss 1 -7' I : [ RegulatOJ:)' sign I 699: [ 17.4% 1 185 [ 364 [ 288[ 150. I Traffic-Way Flow j [Notomded I 1,627 ; I 23.0% I 56 ; I ma m l 82 i l I Divided with no : baaict 1 30 :[Penon Age Yean Old I 81 1 37.9%, [ 3 ; [ 33 : I ss[ 19 [6-15 Yean Old I 290' [ 42.2% 1 10 1 54! [ 63 I 31 ' I 1 [16-24 Years Old I 329d 12.2%:[ 64[ 100 I 126 I 79' [25-64 Yean Old I 1,879 I 265% 1 57: [ 108 [ 112[ 74 < [65-74 Yean Old I 289' [ 22.9% 1 s [ 96i[ 104[ 37. 175 Yean or Older I 4n l _22.2% 1 7 I 72 ; [ 69 1 19. . Sowu: Computed br CU'IR &om FAilS, 1 1998. A number 1n bo6d that che dilfercn t.'\ dwe$ between Flori& lhc oottppOd.in& sutc $fOUP is aot a:tuistieaD.y
PAGE 81

I Numbe t of lane I !IGendu ;! 6 9

PAGE 82

i Table D-2. ofNon MoMist Fatalitiu Rllaltd to Actititiu in RoadrP'[)' PARS Variable PARS Value P'oblem Size s.-ce Numbet Sbare Safe" Safer Le .. saie All Person Age Undu6 41 50.6% 263 186 : 110 166 : IearsOl d .. 6 15 Yea.rs 127 43 8% 196 211 77 139 Old 1624 Years 158 48 0% 93 38 ; 35 so Old 25..64 Years 879 46.8% tt4 : 60 45 64 Old 65 74Years 120 41.5% 213 86 61 109 Old 75 Yean or m ; 156 : Older 191 46 .2% 76 158 cotm:pOOdiag :;'tlbol 70

PAGE 83

Appendix E: Summary of Results 71

PAGE 84

Tablt E 1 1 ,5 16 46% 8) with u least one m01'ing violacioo ' 743 23% 38 io. 3 )"<2lS I 7 5 Ol olde.r z 481 15% 51 l .witb u le2st o.ne SWpe:rutOn Ot teVOC2QOn m 4 5 2 1 4 % 31 c 1,2?4 3S 631 19%! 33 5 J:l lrh.k 1 ,141 35% 28 kft 806 11% 120 576 12% ISS Collision-Head o o 980 ; 13% 44 :E > 1 195 18% 45 780 24% 3? ... 4,456 61% 38 1 819 56% 29 r Sigos 2,339 3 2 % 64 ?65 30% 45 [Th 910 U% 29 toRoadw>y-87S 1 2 % 333 283 9%! 338 "' !! I t 1 6 30 soo.r.; 72 !i, c 71 al S utc 980 30% US. I 948 29% 69 Rood 8 10 11% 32 332 10% 82 518 16% )0 Nota' 1.00i\ -.of 0\'c: Rqmsa:ltttion. .. 2. 1bece problem a.reas wett under the ExpNtt uWysis 72

PAGE 85

TabkB-2. Problem Areas 1 unc Druolt Driver Left lJl with DO Baa:ier IDuk but' .... I U.S. Point9 oclock point Gndc Colli$ion Condition : 3 clock Point T least one non-$peeding in 3 yew' older at least one suspension oc in 3 ycors' prior cra.sh in 3 r= DORt Index : DOR Index2 lndeic2 38 1,227 29 409 1,636 72 1,007 52 294 1,301 64 913 333 673 120 !I 35 335; 45 37 1 29 205 338 218 891 !55 72 33 82 !57 150 39 219 i 51 162 59 101 790 83! 688 : 688 28 250 205 31 10 7 682 4 92 407 387 371 346 299 250 2193 205 205 162: 107 101 Tf.'-vu .. Not<" I. OOR >Wids for "Degcd. 2. The index i:ndiates the pottntial reduction tn the number of &talitics rc:b.ted 10 a certain :uea of concem if the overrcprescnution in this uea of c:on<:em wue di.minacecl The index is c:alcuhtc:d u R; -= V [Qi I (100 + QJ] for lbte group j, where V is dte nwnba of fatalities tt.Wod to a given problem area uul Q; is the degree: of ovc:Nepresc:ntation of th.U probkm attain Florida. relative to nate group j total index fWD$ cho"for individual penc>n C)-pes for a given problem area. 4. These problc:m :ueu were not ew.lu:arcd under the Exposure analysis because quasi-induced exporute aflftot be meuUd for them. 7J