' '-.A'.- . IWImnmmmmmmm L 1 00870 8921 This is to certify that the dissertation entitled DRIVER CHARACTERISTICS AND THEIR RELATION TO TRAFFIC CONFLICT OCCURRENCE presented by Khaled M. Abdulghani has been accepted towards fulfillment of the requirements for Doctor of Philosophy degree". Civil Engineering Major professor/ Date May 17, 1982 MS U i: an Affirmative Action/Equal Opportunity Institution 0-1277 1 .15“ £425 ' w » . , ‘ v _,' ' . 4 . ~ *7 I'Clfl‘“, MSU LIBRARIES -_—. RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. ' .I i'.. -.. a Q». . J‘g‘, n“ ’, )tiva .1 '35 ’,.fi DRIVER CHARACTERISTICS AND THEIR RELATION TO TRAFFIC CONFLICT OCCURRENCE by Khaled M. Abdulghani A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Civil and Sanitary Engineering 1982 l 1‘) . . . n;? "| . LM - ;~ e, ‘ IF- '.1 ' ,1. .I . ..‘ ‘ D‘- ‘5 ‘u I— '\ ." 3 : . ~“y ' 0 it. .‘ n ,‘K‘, c“ ‘ u F—n.» J A 4w 4w w-v—ngffl '3 i l ' u .1 .1 »- ABSTRACT DRIVER CHARACTERISTICS AND THEIR RELATION TO TRAFFIC CONFLICT OCCURRENCE by Khaled M. Abdulghani A traffic conflict is defined as a hazardous situation in which an accident is avoided by evasive action by the driver of one of the vehicles. Research on the use of conflicts as a traffic engineering tool has been focused on attempts to correlate conflicts and accidents so that conflicts, which occur more frequently than accidents, can be used as a basis for predicting future accidents. Several stratifica- ‘ticnas and/or refinements to conflict measures have been introduced to acfliieve better correlation, but the use of driver characteristics has ruot been used in these research studies. The objectives of this study were to determine the driver charac- teri stics that are overrepresented in conflict involved drivers when cxmnpared to all Saudi Arabian drivers; the relationship of these char- acteristics to types of conflict; and the interrelationships between types of conflict and traffic control at the intersection. A final objective was to determine the possibility of predicting future con- flicts by knowing the mix of driver characteristics at an urban intersection. Conflict involved drivers observed at six different locations were found to possess similar characteristics, indicating that there is a pattern which defines hazardous drivers. Age was the single vari- able which explained the greatest variation in conflict involvement, but nationality, occupation, past accident experience and driver training “>- ...I u- were also significantly correlated with conflicts. Thus, predicting future conflict by knowing the mix of driver characteristics may be possible, but larger samples and more research is needed to achieve better correlation of variables in the multiple regression model. The relationships between driver characteristics and conflicts found in this study were similar to relationships known to exist between driver characteristics and accidents. Thus, it appears that conflicts may be used to identify hazardous intersections when past accident records are not available. Another conclusion reached in this study was that there are sig- nificant differences in the conflict rate depending on the training received by the driver. It appears that more strict driver licensing and enforcement of traffic regulations would decrease the conflict rate, and presumably the accident rate, in Saudi Arabia. Dedicated to my father, Rashad, and my mother, Faigah, who sacri- ficed their life for the education of myself and my ten brothers and sisters. My father has been continuously concerned about our education and provided the impossible to overcome the obstacles in our paths. His courage in going against the wish of the elders in sending my sister Roshdia abroad to study medicine and make her dream come true left me with great admiration. My mother's closeness to me and my brothers and sisters, and her continuous prayers that God assist us and be with us were replied. It is for her outpouring feelings and upbringing that gave us recognition among friends. On behalf of my brothers and sisters, may I express our deep grati- tude for the great price you have paid which we can never repay as long as we live. iii 11:2”!- "H u ’ . an I. .. F‘w- .. or ‘. u . :'_. "o' u‘; ~ “Q m, I ‘ i ' Q ~ ‘n6 . A‘ «I_ .' -. . ‘ \ ‘I F i. a . §.. ‘1‘ ‘~q .' a A I. N ‘ ACKNOWLEDGMENTS I wish to express my deep appreciation to my major professor, Dr. William C. Taylor, for his continuous guidance and assistance from the initial process of proposal development to the completion of this research, and throughout my entire doctoral program. I am deeply indebted to him for sharing his knowledge and experience with myself and to him and his wife Norma for being the best of friends in our home away from home. I wish to thank Dr. James D. Brogan for his invaluable guidance during my Master's program, and in the conduct of this research. I also want to express my gratitude to Dr. Tapan K. Datta whose exper- ience with this area of research provided constructive suggestions and guidance. I am grateful to Dr. Gail C. Blomquist who was readily available to make contributions on research problems. Dr. Koroos Mehjoob-Behrooz deserves recognition for his guidance in the statisti- cal design and analysis. The author wishes to extend his appreciation to Dr. Fahad H. Al-Dakhil, Vice Rector of UPM, for his assistance in eliminating all obstacles during the data collection phase. Recognition must also be given to Captain Abdulqadir Talha, Head of Dammam's Traffic Authority, vvho participated in providing manpower for this research. Special appreciation is extended to Ms. Vicki Brannan who helped in the typing of this thesis. My beloved wife Amera has shown unlimited devotion and understanding iii in during the field studies and the entire period of this research. She was very devoted in her support of me, and in the upbringing of our son Yasser and our daughter Nasreen whose closeness generated a new vigour in me. My brothers Sami and Essam deserve my special gratitude for their encouragement and support throughout my doctoral program. I would like to thank all the unnamed friends who contributed morally in the development of this research. iv . in. A. 41-0 . "‘ a. L;) .0. I.) 't‘ . u '. l . ‘ 'o . ‘ D . -. s .. . u s .‘ .2“- Ink 4. LI ) v an .E I . o '1 .._‘ i p '5 1'1'1 "T v. List of List of CHAPTER 1.1 1.2 1.3 CHAPTER 2.1 2.2 2.3 2.4 CHAPTER 3.1 3.2 3.3 3.4 3.5 CHAPTER 4.1 4.2 4.3 4.4 TABLE OF CONTENTS Tables Figures 1 Introduction The Problem Objectives of This Study 2: LITERATURE REVIEW Exposure Measures Accident Patterns Driver Characteristics Traffic Conflicts 3: METHODOLOGY Location Choice of Intersections Observer Training Experimental Design Schedule of Activities 4: DATA ANALYSIS Observer Reliability Representation of Driver Characteristics Characteristics of Hazardous Drivers Test for Driver Characteristics That Explain Conflict Involvement Page vii ix 12 13 18 23 23 23 24 26 28 28 29 32 35 '; 'A-«r. 1i 1‘ ‘ ‘- -s a o p" I “ 5 . . A \ [AF Ho 11v 5 ' 1 my .’ \ .0. fl 4 ""l - V A»; .a“ . i ' 'Ov-v Page 4.5 Evidence of Patterns Among Conflict Involved Drivers 47 4.6 Conflicts by Type 47 4.7 The Model 52 CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS 58 5.1 Conclusions 58 5.2 Discussion and Recommendations 59 BIBLIOGRAPHY 61 APPENDICES 65 vi .‘ll .\1 4‘ Table MN .33 10. 11. 12. 13. 14. 15. 16. 17. LIST OF TABLES Two motor vehicle accidents at urban intersections. Percentage accident rate by type of collision. Percentage of faulty driving techniques. Observed conflict counts for observer reliability. Conflict observation, reliability test. Conflict observation, reliability test. Chi-square test for population assumption. Sample sizes for studied intersections. Chi-square test for driver characteristics in sample 1 and sample 2 using all intersection data. Chi-square test for driver characteristics in sample 1 and sample 2 using signalized intersection data. Chi-square test for driver characteristics in sample 1 and sample 2 using unsignalized intersection data. Chi-square test for driver characteristics in sample 1 and sample 2 using unsignalized intersection data for drivers with two conflicts. Percentage increase in representation of conflict involved drivers over intersection driving public. Chi-square test for conflict involved drivers at signalized intersections. Chi-square test for conflict involved drivers at all intersections. Frequencies of conflicts by type at unsignalized intersections. Frequencies of conflicts by type at signalized intersections. vii 44 44 45 45 46 48 48 50 3.! U Table 18. Characteristics with similar distribution to conflict involvement by type. viii IV. Figure LIST OF FIGURES Percentage of driver age in samples 1 and 2. Percentages of driver age in sample 1 and turning conflict sample. . Percentage of driver age in sample 1 and in right angle conflict sample. Percentage of driver age in sample 1_and in side swipe conflict sample. Percentage of driver age in sample 1 and in rear end conflict sample. Age of conflict involved drivers with two or more accidents in past two years. Relationships between conflict index and age of conflict involved drivers. Relationships between conflict index and age of conflict involved drivers with and without accident history. ix Page 36 37 38 39 40 41 56 57 ‘ (In oc'oq. . '. . ,3 r o .Di . u. -\ . a 4 ‘2‘ “n4 ., ".1-é‘ .1;. . . 0 i a; ‘ 2‘ s r‘ "(u-i: ‘v fl '. ‘ h. \ ’I ‘ u 4 I - \ ~‘ . ‘ . v ' h‘ 7 “-5 “s CHAPTER 1 1.1 Introduction The prediction of intersection accidents has long been a goal of transportation research. Much of this research has been focused on attempts to develop techniques for predicting accidents as a basis for implementing safety standards to prevent or minimize these accidents. Traditionally, accident records have been used as a measure of the hazardousness of intersections. The fact that an accident may be the result of a driver's error, a defective vehicle, an inappropriate road- way condition, inadequate or improper traffic control devices, inade- quate design, or a combination of these factors, however, makes it dif- ficult to accurately predict future accidents based on past accidents. Also, the fact that accident records may be distorted, incorrect, not readily available or not statistically reliable, adds to the difficulty of developing adequate models from these records. Researchers have acknowledged the need for new methods to predict the number of accidents expected at any given intersection. Most of the research work in this area has attempted to use some measure of traffic exposure as a predictive variable. The most common exposure measures have been the volume, or the volume to capacity ratio. How- ever, neither of these measures capture the influence of geometric features on accident rates. One of the newest methods proposed as a basis for accident predic- tion is the Traffic Conflict Technique. A traffic conflict is defined as a hazardous situation in which an accident is avoided by evasive ‘ ' ‘ a a. ' .'. All. in V... 0.,“ J I! "‘7 up“ '5 K i - § 1 . . . F- ‘.\! (‘7. 2 action by one or both drivers to avoid a collision. A conflict is identi- fied by the occurrence of evasive actions (such as braking or weaving to avoid an impending accident) or a traffic violation. A traffic violation is defined in accordance with the uniform traffic code, and is considered a traffic conflict even if no other vehicle is in close proximity to the violator. Twelve conflict types have traditionally been used at intersections and are defined as follows (53): 1. Left turn, same direction: A left turn, same direction con- flict occurs when an instigating vehicle slows to make a left turn, thus placing a following, conflicted vehicle in jeopardy of a rear-end collision. 2. Right turn, same direction: A right turn, same direction con- flict situation occurs when an instigating vehicle slows to make a right turn, thus placing a following, conflicted vehicle in jeopardy of a rear-end collision. 3. Slow vehicle, same direction: A slow vehicle, same direction conflict situation occurs when an instigating vehicle slows while ap- proaching or passing through an intersection, thus placing a following vehicle in jeopardy of a rear-end collision. 4. Lane change: A lane change conflict situation occurs when an instigating vehicle changes from one lane to another, thus placing a following, conflicted vehicle in the new lane in jeopardy of a rear- end or side-swipe collision. 5. Opposing left turn: An opposing left turn conflict situation occurs when an oncoming vehicle makes a left turn, thus placing the conflicted vehicle in jeopardy of a head-on or broadside collision. 6. Right turn cross traffic, from right: A right turn cross 1.. “ . . 1‘ ‘flo "- 4“, h. .. , ‘ :r I". | \‘D -," d! u traffic from right situation occurs when an instigating vehicle approach- ing from the right makes a right turn, thus placing the conflicted vehicle in jeopardy of a broadside or rear-end collision. 7. Left turn cross traffic, from left: A left turn cross traffic from left conflict situation occurs when an instigating vehicle approach- ing from the left makes a left turn, thus placing a conflicted vehicle in jeopardy of a broadside or rear-end collision. 8. Left turn cross traffic, from right: A left turn cross traf- fic from right conflict situation occurs when an instigating vehicle approaching from the right makes a left turn, thus placing the con- flicted vehicle in jeopardy of a broadside collision. 9. Thru cross traffic, from left: A thru cross traffic from left conflict situation occurs when an instigating vehicle approaching from the left crosses in front of a conflicted vehicle, thus placing it in jeopardy of a broadside collision. IO. Thru cross traffic, from right: A thru cross traffic from right conflict situation occurs when an instigating vehicle approaching from the right crosses in front of the conflicted vehicle, thus placing it in jeopardy of a broadside collision. 11. U-Turns: A U—Turn conflict situation occurs when an instigat- ing vehicle makes a U-Turn in the vicinity of an intersection in front of a conflicted vehicle, thus placing it in jeopardy of a collision. 12. Disobeying red light or stop sign: This traffic violation occurs when an instigating vehicle violates the uniform traffic code. Whenever a vehicle is involved in more than one observed conflict, each type of conflict is recorded for that vehicle. For analysis purposes, the conflicts are often grOuped into four . a.» “una- . t' v _"’3 ”I.“ - 0 l .I p 7 .0 fi ( 1‘. ’I g 4“- .-\ .‘H‘ “.:‘ . I. 7“. .v‘ 2‘- .1 H, gr. ‘5. l 5.;- \ ‘ . ' v P1? 'v. a... 'n. . "a‘ “ categories: 1. Right angle conflicts: the sum of conflict counts for: a. Thru cross traffic, from left. b. Thru cross traffic, from right. c. Disobeying red light or stop sign. 2. Rear end conflicts: the sum of conflict counts for: a. Left turn, same direction. b. Right turn, same direction. c. Slow vehicle. 3. Turning conflicts: the sum of conflict counts for: a. Opposing left turn. b. Right turn cross traffic, from right. c. Left turn cross traffic, from left. d. Left turn cross traffic, from right. e. U-Turns. 4. Side swipe conflicts: the sum of conflicts counts for lane change conflicts. Research on the use of this technique has attempted to correlate conflicts and accidents so conflicts, which occur more frequently than accidents, can be used as a basis for predicting future accidents. If this correlation can be verified, conflicts can be used to identify potential hazards and operational deficiencies, and to evaluate traffic improvements. Although the traffic conflict technique has been studied for more than ten years, there are still many components that have not been thoroughly investigated. For example, insufficient research has been conducted on characteristics of conflict involved drivers (those who contribute heavily to conflicts). This research is directed toward A. F ow ‘ u ':' «VJ '.¢J«p _ ..,." i .0 .. ~- |‘_ . p . .- u A-po‘ liv . . v Q 1 I: Q ""2 r. W 5:}; 'Ia u ‘r o ‘ ‘J .- . a 3' ’s ‘ . 5 ‘3 .‘I 1 i . . ‘6. ' ‘ .- . u \ “ \ o «Q F ~ I .Q i 5 the study of the characteristics of conflict involved drivers, and the relationships of these characteristics to the number and type of conflicts. Although the scope of this research does not include the testing of conflict-accident relationships, an accident is simply a conflict where the evasive action was too little or too late, and accidents and conflicts should be statistically related. Glauz and Miglez (53) argued that most attempts to find strong correlations between conflicts and accidents have been unsuccessful for one or more of the following reasons: 1. Not all intersection accidents are reported, and the reported accidents may have happened at different times than when con- flicts were measured. 2. Not all intersection accidents are the types representative of the conflicts measured. 3. Not all accident data used by previous researches was statis- tically repeatable. 4. Conflict definitions have not been precise enough to ensure interobserver reliability, and there have been various inter- pretations of these definitions. 1.2 The Problem The objective of all highway safety programs is to reduce the num- ber of accidents and fatalities which occur each year. Previous acci- dent studies examined such factors as the characteristics of the roadway, the vehicle, and the environment and how each of these contribute to the occurrence of accidents. However, insufficient attention has been given to one of the basic elements of traffic accidents, the driver. 5.. l I Hun .‘ ,. . I" vf\ I '6 It ‘nt;p| “:I'l\ . a ‘co‘v- l n.« a" A L...” 3 d “we: IOIv. "in”. l P ml ‘ n u. 1 Ii... ’ .Y" 'n'rb . :...r‘f -.,‘ = l . ‘ l on. .~ ‘ .T l I I 00 n I s. q i v S b s >“ ‘ O z I" ‘ v - .“‘ 6 Recent studies have shown that there exists a 'pattern' among drivers involved in traffic accidents, and that certain driver charac- teristics are overrepresented in persons involved in accidents. Al- though several authors have looked at the possibility of predicting future accidents using individual characteristics, the literature con- tained little information on the potential for using characteristic 'patterns' as a basis for predicting future accidents. In the literature review, no studies were found of the character- istics of conflict involved drivers or attempts to identify those char- acteristics that are overrepresented in conflicts. 1.3 Objectives of This Study This study was designed to achieve the following objectives: 1. To determine driver characteristics that are overrepresented in conflict involved drivers when compared to all drivers. 2. To determine the relationship between drivers' characteris- tics and types of conflicts. 3. To determine the contribution of each conflict type to total conflicts at urban intersections. 4. To examine the possibility of predicting the number of con- flicts and their types by knowing the mix of driver characteristics. The importance of this research lies in the potential use of the results to establish a new basis for predicting future conflicts, and ultimately, future accidents. ’;nr bkfi I 'l 'ql. I p z m A. 'r-‘c wt :‘l':bb U Q tun-‘- j.-: 1cu: : fl. ' . Jr‘ ‘0- ' U'Ub. I h I ‘~ , b . h:- " ‘n- ‘ c '. ‘GPr; r‘ y~ ‘9-6. p .: at Q ‘. ' § \" V. . . - ya. I'. h.) \ V ,f‘ 6.; ...‘. . ~' u..‘* a . I: ‘V p- \ .- .'d1= In I x , N‘ Q n. .3, a‘ v: ‘- ~‘o‘ ‘ \ ‘U _. ‘ a . ‘ A "‘ "~ 5 -. .é: ‘1‘ . CHAPTER 2 LITERATURE REVIEW The distribution of traffic accidents in time and space, factors which contribute to or are correlated with accidents, and the potential value of traffic control devices in reducing accidents have all been the subject of extensive research. A large number of accident prediction models using various surrogate measures (mostly exposure related) have been used to explain these variations in traffic accidents. In contrast, a relatively small number of published articles have discussed the use of driver characteristics in accident prediction models even though these variables generally are strongly correlated with accident frequencies. The literature review was conducted to determine whether sufficient evidence of a relationship between accidents and conflicts, and acci- dents and driver characteristics,exist. The relationships discussed by other authors was then used to develop the hypothesis to be tested in this study. 2.1 Exposure Measures Intersection accidents account for more than 50 percent of the total accidents in many nations, which gives the 'intersection' a world wide priority in finding accident prediction procedures and remedial measures (3). Traffic accidents are the most direct measure of safety deficien- cies at intersections. However, postponing potential safety improve- ments until an accident history has evolved delays the implementation of accident reduction measures. The lack of reliable accident records, due to incomplete records or errors in coding either important accident 7 -r V! ‘ ll;( 1" 0 ~- .‘ M: vh' . u‘bvli 'l‘ u, . ,— .s . \ a... U H... II .F‘: "'5 1. IA, an I o». "“2 r! I"...‘ D‘fi‘l I -.‘.A .rl‘ ,- "~ '-.. . . A. ' HP‘A 1. in ‘ u. ‘- s- q s \ A ‘. '3» h a \A u- ”s I. ‘ ‘. I"... . lu‘ \ V .. lyi- \ '9 . .. .. ‘ rt‘ 1 8 facts or the precise accident location, contributes to the insufficien- cies in using accidents alone as a basis for safety improvements. Several individuals and agencies have tried to find a solution to this problem by introducing new methods to estimate the expected number of accidents at intersections. Some of the early studies like Jorgen- sen (4) and Hall (5) calculated the expected number of accidents on the basis of the number of vehicle-kilometers within the intersection. How- ever, the risk per kilometer travelled at intersections was not a satis- factory measure, since the number of accidents at intersections is dependent on more factors than volume and the width of the intersection. This lack of success led to more sophisticated methods of predict- ing accidents at intersections based on a combination of measures of exposure. These studies used mathematical functions based on the stream flows passing through an intersection as the measure of exposure, and compared accident occurrences with this calculated exposure. Accidents and flows were compared for many locations and regression equations derived which correlated with the observed accident occurrences. Thorson (8), for example, defined exposure as the sum of all vehicles entering the intersection. He theorized that the number of accidents that occurred at each intersection divided by this sum would provide a basis for comparing the hazardousness of intersections of different types. However, this too was proved to be unacceptable as it does not consider all the factors that effect accidents. Several researchers defined exposure as a probability distribution based on the number of opportunities of being involved in an accident at the intersection. Mathewson and Brenner (6) and Breuning and Bone (7) were among the first to point out the difference between exposure on a road situation, which is related to distance travelled, and - III'U b 97" In I}: an 1 un'd 32"» "vll 9 exposure at intersections, which is related to the maneuvers through the intersection. They suggest that an exposure index at intersections should take into account the flows on each crossing, merging, or diverg- ing path. These studies led to several measures of exposure being used to predict collisions at intersections. The following paragraphs summarize briefly what has been accomplished with these models; the traffic con- flict technique will be discussed in more detail later in this chapter. Early research on exposure was reported by Feuchtinger (13), Raff (14), Thorson (8), and Breuning and Bone (7). Each found that the number of accidents per vehicle was lower for locations with a high measure of exposure than for locations with a low measure of exposure. In each of these studies, however, measures of exposure were found by observing traffic at different locations rather than by taking observations at one location over different time periods. Thus, this phenomenon may be explained by the fact that locations with higher flows generally are built to higher design levels than those with lower flows. Thus none of these studies can claim to have shown conclusively the effect of flow on the ratio of accidents per vehicle. McDonald (15) and Thorson (8) found that accident risks do not vary much between heavily travelled intersections, even though the flows varied widely. For the intersections they studied, there does not seem to be a strong case for the use of the sum of entering traffic as a measure of exposure. Grossman (10) calculated the number of locations where one traffic stream crossed another traffic stream in an intersection, added the flows for each stream, and summed this over all locations and used it as the measure of exposure. This measure was used to compare different .- . at I I‘ - u \ 'fl .4 0 :6 I 'I 9 u r. .' MI. -. “‘2ro; 'v-., ' 'Iaqn - '-d l p A h o .1 '4 a. ’v . “ n _. s .11 4‘ O“ .I F ~4- ... I 10 intersection layouts. However, these models have not calibrated well with accident experience, indicating that not all traffic stream cross- ings are equally hazardous. A second group of researchers used the product of conflicting flows as a measure of exposure. This measure takes into account the number of times vehicles from different directions wish to occupy the same area of road space simultaneously. Surti (11) proposed using this pro- duct from a number of points (depending on type of intersection) where potential collisions may take place as a basis for an accident exposure indes. The essential parameters for this model are: number of con- flict points, ADT, and time. Time is a measure of the period of expo- sure to conflict and can be arbitrary since it only has a relative value. The purpose of Surti's work was to determine an "exposure index" for each of the conflict points to develop a general equation for the entire intersection. Surti (16) followed his initial work by using the same points con- cept and adding merging maneuvers to test the model. He considered, however, the same likelihood of accidents for different maneuvers. Chapman (17) agreed with the use of the product rather than the sum of flows as an exposure measure, but added that the square root of the product may be a better predictor of accident occurrences. McDonald (15) was among the leading researchers to use the square root of the product of flows as a measure of exposure. He studied intersections between a divided highway and a cross street and obtained a relation between the number of accidents per year 'AV', the average daily traffic entering from the divided highway 'Vd', and the cross street traffic ll. Vc . ’( v D» ‘r e‘_‘ 1r an. 0r: fiff‘. v.p.. h :p; ... ‘v .p. n ,‘n.. ‘1 ‘C. A : A. ‘.‘ 11 A = .OOO783 vd'455 vc°633 Thrope (21),in an investigation of accident rates at signalized intersections,related the number of accidents to the square root of the product of entering traffic, but did not justify the use of such a measure for signalized intersections, where the two traffic streams are separated by time. Webb (20) developed graphs (for urban, semi-urban, and rural sites) for signalized intersections, showing the average expected number of accidents per year according to a non-linear function of the average daily main road and minor road traffic. This study, like those of Raff (14) and others, was based on single data points from many locations rather than many data points for a single location. Thus the cause- effect relationship cannot be determined. Gwynn (12) conducted one of the most comprehensive studies of the relationship between accident rates (or involvement) and hourly volumes. Although this was a highway (not an urban intersection) study, it demon- strated that accident rates and traffic volumes do not parallel each other. This may be related to the fact that there is a ten- dency for a disproportionately high accident rate to occur during the hours with the lowest volumes (early morning hours). This is followed by a decrease in accident rate with an intermediate traffic flow, an in- crease in accidents with an increase in volume up to a point of conges- tion, and then the accident rate again falls off, even though the volume of traffic continues to increase. This study illustrated the fact that a stable relationship between exposure and accidents probably does not exist unless other conditions are eliminated from the analysis . .np :‘ :8} ‘ ‘v d \ , \uvu .‘g . A; .' all .0. HI — 1 r1 ; v. :"e 3,. :.... r. 'Ouuu‘n' .,‘ :V“ i 1.C Q... I.e “‘9. , . 12 (such as a high number of drivers under the influence of alcohol between midnight and 5:00 a.m., congested traffic). Thus, it is not surprising that attempts to develop accident predictive models based on exposure measures have not been successful. 2.2 Accident Patterns Intersection accident patterns have proven to be very stable over time and in many different traffic situations. Intersection accidents account for a high and relatively constant percentage of total accidents in many nations. Smeed (3) reported that in Great Britain over 50% of all injury accidents in 1970 occurred in urban intersections. In France, intersection accidents accounted for more than 45% of total injury acci- dents in 1972; in Denmark 38% in 1973; in Ireland 44% in 1973; in Holland 44% in 1971; and in the United States an average of 41% of all reported accidents between 1971 and 1979 occurred at intersections. Accident statistics published by the National Safety Council show little variation in the number of accidents when classified by type of collision over a period of ten years (see Table 1). The statistics also show a constant pattern in the type of accidents, with angle accidents having the highest frequency, followed by rear-end accidents and left turn accidents. The effect of intersection geometry on accidents has also been studied extensively. In one two-year study to find relationships be- tween motor vehicle accidents and the geometric and traffic features of highway intersections, Stanford Research Institute analyzed 558 inter- sections (57). The results were similar to the National Safety Council statistics as they found that angle accident rates are the highest for all types of intersections followed by rear-end accidents and weaving 13 and turning accidents. Table 2 presents the percentages of collisions by type for different geometrics and traffic control at intersections. Thus, if a particular intersection is experiencing an abnormal accident history, some other factors must be contributing to these accidents. 2.3 Driver Characteristics There have been fewer studies reported on the relationship between accidents and the driver. Yet the characteristics of drivers, their capabilities and their driving skills are important factors that affect accidents. The National Safety Council concluded that an average of 88 percent of all accidents at urban intersections were a result of im- proper driving (58). Table 3 presents the percentages of different faulty driving techniques that caused accidents over a ten year period compiled by the National Safety Council. In an attempt to find an easily measurable, yet reliable, surrogate measure of an accident prone driver, Spicer (26) examined the difference in car following behavior (as measured by the time gap between vehicles) between drivers with accident records and the general driving public. This study concluded that the smaller the following gap, the higher the probability of prior involvement in accidents. In the same study, Spicer also analyzed the relationship between intersection accidents and drivers age. He concluded that a crossing maneuver involving a young driver on the major road and an older driver on the minor road is more likely to result in accidents than any other combination of age groups. Siebrecht, Schumacher, and Lauer (22) made a comparison of two samples of drivers. The first was taken in 1950 and the second three years later from the vehicle license files in Iowa. A follow-up mail mnvey was sent to all individuals in the samples. The survey included I. \‘.Ill.|z lI Ill 1|. 1“. III, 1|!" l' I: ll" I‘ll II .N. CJEQT...) 3?...LC>< M.\mu— m~\fi.— h\$~ C\3— n\3~ Qkfimu N\3~ -N\3~ -KQ~ ~ CN 3. m L 1.0% fires: UO_\ KO C§XK no ..I..~.cc-u- ooooooooo u.. ..a..‘ u ..l\ ,u. .~.\v .m on mpasmm as» mmuzomc cows: cowumpau_mu mucmwcm> new mmmcw>m ms“ :_ cmu:_u:w mw mmczmww mmmgp to mac apco .cmmmwn m=_mn we zuwpwncnoca one mmoauocucw gown: mcowumwcm> o: sogm mmofi .flmmH .ouoa :_ =m>wm mmczm_m on» a 14 mucoamm mmpuwcosu=< uwmcmch mumum :0 women Fwoczou xamwmm Pacowumz .muomm acmuwou< "mucaom E. m; 3 o.~ : E Z 3 ms to no to 29:0 .2 mm 8. Z 3 j 3 Z Z 3 E 3 .3 3 2323 25 £3 25 mm 8. mm. to so m... no 2 a; .3 3 i: 3 2323 9:8 58 mm mm. o; .2 m; o; no as 9o 3 3. Two no 233 :< mm 3 m8 ~.m m8 2 3 2 3 E 3 .3 3 888m 25 mm a. ma ~.m ma 3 3 2 ma 2 3 1: m.m 2323 25 .EE 95 mm. mm.) M: S 2 on 3 3 Z 3 9m 9m o.m .2323 8.8 58 w H.N m.o~ a.mfi ~.- m.- m.- ~.m~ ~.c~ m.mH a.mfi a.- ¢.~fi apmc< 8a m=_cau=u 3 n: 8.8 “.2 I; 53 98 8mm 9: 98 98 gm mcwwfiwmwfiafimcncakes” a 85:2, 82a: 22 22 E: 23 £2 :2 23 a2: :2: 35 as; 28.82 to 8.: mcomuummcmucm sang: um macmuvoo< mpovsm> Lopez 03H .H m4m

ogo m.e m.¢ m.¢ ~.o m.m m.m m.m m.m o.¢ o.¢ Pmcmwm umugmmmcmwo m.cm c.0w a.mm a.mw H.Hm a.mm m.om m.mw a.mm w.mm mcw>wco cmaocasm quoh mm ms mm mm mm em mu us an an mcw>Pgo cmaosaEH mo vcwx mmzawczomh mcw>wgo zupzmu mo mmmwucougmq .m ”Sm: I‘-’ (-1 ' v .T" ‘A n ‘1. 9"»..1. 63””‘r "' ‘i. U . :q‘ I, g. ‘ lg. 9 ‘ ‘ 3 "-aw,‘ ‘5 Q. ‘1]. “'5? '1 ‘2. .. “ a): l hr“. 1 17 data on the methods of learning to drive, mileage driven yearly, educa- tional level, and other information. The authors concluded: 1. Violations and accidents are closely correlated and records of violations give a fair index of potential accidents. 2. About one person in four has an accident record and the same ratio holds true for violations. 3. A higher percentage of accidents involve men under 30 than other age groups. 4. The hypothesis that accidents are distributed equally among age groups is rejected. Lynn (23) indicated that there is evidence to support the fact that accidents do not occur completely by chance but rather according to a demographic pattern. For example, in 1970, males between the age of 20 and 24 comprised 10.1% of all males in Virginia and 5% of the total population of the state. However, this demographic group comprised 20.7% of all male accident fatalities during the same year. Lynn gave other examples and concluded that certain demographic variables can be used to predict accident patterns. Waller and Koch (27) evaluated accident data and types of driving exposure in terms of sociodemographic variables. They found that more highly educated individuals drive more than those with less education but there was no significant difference in the accident rate. Plez and Schuman (28) found that among young drivers, those who were alienated from the educational system (dropouts, poor'grades) had more accidents and incurred more violations. Baker (29) found that more males and more young drivers were at fault in fatal accidents and that lower economic class drivers were also an olf‘ in i I) V Turf: l u ul- s-‘ uu'u , .. .123 F . t... . .1- - ‘vv ‘0‘ 1w q n. J Ca .1 .: u... n. a... ..V‘:[ U I. ‘ .1" .‘ 18 more often at fault. McGuire (59) compiled data on 2,799 drivers in an attempt to iden- tify personal characteristics of drivers that possessed a causal rela- tionship to accident-producing behavior. The study reported a multiple correlation between characteristics of drivers and accidents, with the highest observed simple R values corresponding to age (.38), occupation (.38), and income (.30), among 29 other tested variables. All of these studies indicate that there are dirver characteris- tics related to accidents which if known in advance, could be used in accident prediction models. However, many of these characteristics can only be determined through interviews with drivers involved in acci- dents or stopped at a survey station. Unless the distribution of these characteristics can be determined, it is not possible to determine the relative hazardness of intersections. 2.4 Traffic Conflicts Traffic conflict analysis has been proposed as a method of addres- sing these concerns. Because traffic conflicts are measured on-site, variations due to different driver characteristics or changing volumes can be captured in the data. The adoption of this technique, however, has been limited by the lack of research indicating that these conflicts are strongly related to accidents. The traffic conflict technique was developed by Perkins and Harris (30) to be used as a dynamic tool in evaluating accident potential and operational deficiencies of intersections. Using this technique, it was.hypothesized that intersections could be evaluated with no accident data. Instead, the safety evaluation would be based on the observed corlflict frequencies at the intersection.. l‘) a a... .- V'Iv‘ III a O O-A'. I1. “ 19 Since this initial paper on the traffic conflict technique, re- searchers have directed their studies in two main areas: 1. Studies concerned with finding relationships between accidents and conflicts. 2. Studies using the technique to identify potential hazards and operational deficiencies at intersections, and to substantiate the value of operational traffic improvements by measuring a reduction in conflict frequency or rate. The Federal Highway Administration (FHWA), in cooperation with the State Highway Departments of Washington, Ohio and Virginia, evaluated the traffic conflict technique in an effort to find statistical rela- tionships between accidents and conflicts. Their objectives were also to field test the technique to determine if conflict data proVided in- formation advantageous to pinpointing the need for safety improvements at intersections. The studies concluded the following (31): 1. The hypothesis that conflicts and accidents are associated is accepted. 2. Safety deficiencies at intersections can be described more quickly and reliably by using the traffic conflict technique than by conventional methods. 3. The traffic conflict technique costs less than conventional methods for finding intersection deficiencies and operational problems, and is also useful in analyzing spot improvements (before/after studies). Paddock and Spence (32) developed prediction equations to estimate the number of accidents expected over a two year period using ccwrflicts (and additional parameters) as components of these equations. Equaltions (1) and (2) were developed from their data for signalized and Fllv Nd- 20 non-signalized intersections respectively: AP2Y = 1.1653 + 11.6345 (ADT) - .0503 (CPT) - .0321 (RROPP) + .0387 (OCP02) (1) + 0.285 (TTOPP) - .02255 (OPOPP) AP2Y = 0.36 + ADT (22.3568 + 17.773 (SPLIT) - 36.7045 (ADT)% - 1.6785 (SPLIT)2 (2) + 18.2544 (ADT) - .0264 (OPOPP) + .8385 (OPCON) where AP2Y = Accidents predicted in 2 years ADT = Average Daily Traffic in 10,000's SPLIT = Ratio of the sum of the cross volume to the approach volume CPT = Total of conflicts/10 opportunities OPOPP = Opposing conflict opportunities (opportunity is a term used to denote traffic volume) RROPP = Rear-end conflict opportunities OPCON = Opposing conflicts TTOPP = Total conflict opportunities OCP02 = Square of opposing conflicts/10 opportunities The regression equations had a prediction error for all points (611 locations, 220 signalized, and 391 nonsignalized intersections) of i 4.2 accidents/year at a 95% confidence level. Glennon (33) pointed out that these equations are based more on traffic volume counts (ADT, SPLIT, OPOPP, RROPP, AND TTOPP) than the conflict counts OPCON, OCP02 and CPT). In fact, in equation (1), the rnunber of accidents is negatively correlated with conflicts (CPT and OCP02). Thus, the use of these regression equations adds little to the expcnsure based models discussed earlier. 21 All studies of conflict rates did not reach the same conclusion. Cooper (37) and Glennon (33) found that conflicts are very much volume dependent and could not account for differences in accidents when cor- rected for exposure. They argued that studies which concluded strong association between accidents and conflicts are either brief and not statistically reliable, or have disregarded important parameters that may explain the association (such as site parameters). Not all conflicts are, in fact, near accidents. Drivers vary in their response to other vehicles, some braking when it would not be necessary to avoid a collision, for example. In an attempt to gain reliability, Spicer (34,35,36) classified conflicts by severity (routine, moderate, and severe) and attempted to find relationships between severe conflicts and intersection accidents. He concluded that using only severe conflicts improves the prediction equation. However, accidents at only six intersections were included in this analysis. Hayward (53) used Time Measured to Collision (TMTC) as another method of identifying conflicts. TMTC is defined as "the time required for two vehicle to collide if they continue at their present speeds on the same path." This too improves the reliability of prediction equa- tions be reducing the variation due to a driver related response in the equation. A second type of research used the Traffic Conflict Technique to identify potential hazards and operational deficiencies of highways and intersections. Findings from these studies were used to justify arui substantiate thevvalue of many types of operational traffic im- prxivements. This use seems to have had more success with the Traffic Conf‘l ict Technique. Among the reports indicating successful use of the Traffic Conflict up!“ ‘ .~ ‘ "' n O (\J -, ”.1. Qa- t-a - ‘I n “A... 5'-‘ 22 Technique in practice are: ‘1. Before and after improvement studies (38,40,41,42), 2. Construction zones (43), 3. Freeway acceleration lanes (44), Freeway lane drops (45), Freeway weave areas (46), 01014:- Intersection improvements (42,41,39,40,47,48), 7. Interchanges (49), Oversize loads (50), 9. Pedestrians (51), 10. Signing and traffic signals (38,39,48,43,40,52). The conclusion of all these studies was that traffic conflict fre- quencies or rates are sensitive to changes in geometry or traffic control measures. Thus, traffic conflicts appear to possess one of the proper- ties necessary for a valid surrogate measure - that they respond to changes in the independent variable being tested. The remaining ques- tion is whether the conflict response maps reliably into an accident response, and thus can be used to make an accurate prediction of future accidents. - nag... c¢“bu‘ u .1 O:{§3l. "4' H.» :II( ”All! 0,... .. :7f’;:1 a. | - .H~4 1"1‘ '~—l [\J (.k.’ CHAPTER 3 METHODOLOGY As the literature review indicated, many past studies have suggested a statistical relationship between a change in the number of conflicts and accidents. In addition, research (59) (22) has identified the existence of an overrepresentation of certain char- acteristics among drivers involved in accidents. Therefore, it is a reasonable hypothesis that there also exists a relationship between conflict involvement and driver characteristics. This hypothesis was tested by studying the characteristics of conflict involved drivers- and comparing these with the general driving public to see if they are significantly different. 3.1 Location The experiment was conducted in the Eastern province of Saudi Arabia in the city of Dammam. Dammam is about 18 kilometers from Dhah- ran, the home of the University of Petroleum and Minerals. This location was chosen based on the following considerations: 1. The University's interest in this type of research and their willingness to fund the data collection phase. 2. The cooperative attitude of the Dammam police agency. 3. The availability of facilities and students to assist in the data collection activities. 3.2 Choice of Intersections Due to the extensive construction and development program in Saudi Arat>ia, many of Dammam's streets were under construction. The task of 23 M" 'he I t “-‘ . t I\ . I o . ‘5- ‘. try.“ "“V Kr ::'biy, Q in :.r :. a, . 5 A . '. \- ‘.‘ H - u . 4 Q.‘e “ .' . ' u . . .‘\ .- . V.‘ . 3"} ..\\ 1“.‘ .~ “A ‘ .~ I . ..-!‘ "'51 ‘.a 1 .'_~ i ~ ‘ an! c S 75:. ._“ 24 locating suitable sites to undertake the experiment was a difficult one. One day was spent touring the city, and nineteen intersections suitable for the experiment were identified. Five of those were signalized and fourteen were controlled by stop signs. The city was then divided into three radial sections extending from the central business district. Two intersections were chosen randomly from each of the three sections, to serve as data collection sites. One of those was disqualified after the second visit and a similar inter- section in the same sector was substituted for that intersections. The intersections chosen had no obstructions to the traffic stream for at least 300 feet from the center. They all were properly signed, paved, and painted. The signalized intersections were lighted at night, and there were no observed driver distractions. The Average Daily Traffic (ADT) at these intersections ranged from 2000 to 26000. The two signalized intersections had ADT's of 21000 and 26000, with the unsignalized intersections carrying lower volumes. 3.3 Observer Training Special attention was given to the training of conflict observers for the data collection phase. Fifteen candidates were selected as pos- sible conflict observers, and all were required to attend four training sessions designed to provide them with an overall understanding of traf- fic law, traffic safety, and traffic conflict techniques. The sessions consisted of movies, slides, video tapes and lectures. At the end of the last session, the candidates were taken to different intersections arm! each was asked to conduct volume counting, interviewing, and conflict observations. Three of the fifteen participants were chosen as conflict observers, 01h 25 with the others assigned to volume counts or interviews. The choice of conflict observers was based on: 1. Their background in traffic engineering. (Two were seniors, one a graduate student, all majoring in Transportation.) 2. Their attendance and participation in all of the training sessions. 3. Their understanding of the definition of traffic conflicts, and their ability to distinguish conflicts in the field. 3.4 Experimental Design Each intersection was observed for three consecutive working days during which two different samples were collected. 1. Intersection p0pulati0n sample (subsequently referred to as sample 1). The first day of observation at each intersection was used to collect data to identify the characteristics of the general driver population using the intersection. Three interviewers were sta- tioned about 200 feet upstream from intersections on each approach, and for a period of twenty minutes, every tenth vehicle was stopped and the driver interviewed. The interviewers then relocated on a second approach, and the process was repeated until samples from all intersection approaches were collected. 2. Conflict involved drivers sample (subsequently referred to as sample 2). The second and third days were used to collect data on con- flict involved drivers. Conflict observers identified those vehicles that caused a particular conflict, and then signalled con- flict type and make and color of that vehicle to interviewers who (JD . a . .~ \" .1: vs — .1 'r up n. ‘9 1'. . -~ . ~0- I D'- I " '7. . . ‘2 u , . . 1 i 'b ‘A V O - h C ‘ l 26 stopped and interviewed the driver. Table 8 gives the number of samples collected at each intersection. 3.5 Schedule of Activities The data collection phase of this project took three weeks. The first week was devoted to training participants and selecting adequate sites, and the following two weeks were devoted to data collection. A total of six hours of three consecutive working days were re- quired to collect both the data for sample 1 and sample 2 at each inter- section. The first two hours of the first day were used to collect the general driving population sample of each intersection with a twenty minute period designated to collect data for each approach. The same two hour period of the following two days was used to collect data on conflict involved drivers. A minimum of seven participants for each observation period were needed to collect the required data. The participants included one conflict observer, one volume observer, three interviewers, and the two policemen. Data collection on conflict involved drivers was designed so that data on each traffic movement is collected one time (for twenty minutes) during the total four-hour period of two consecutive days. This was accomplished by first stationing the conflict observer on the south approach and the interviewer on the north approach. The conflict ob- server would record all conflicts on the south approach and identify for the interviewer the subset of those drivers traveling north. After twenty minutes, the interviewer was relocated on the west approach, iwher'e the subset of conflict involved drivers turning left from the Stnrtliapproach were interviewed. Finally the interviewer was moved to l a o ‘O a ,, au' . .-.a E 10 u u v o f 5.; p “ UV A "I l. J) 27 the east approach to obtain the last data set. In the second hour of observation, the conflict observer relocated to the north approach and the process was repeated. The east and west approach were then used in the data collection phase on the second day of data collection for sample 2. This accounted for all traffic movements with the use of only two police officers. Simultaneous collection of all movements was not pos- sible because the use of communications devices by individuals other than policemen is banned in Saudi Arabia. l' A . . A to vb- .— It“ uU Q =4: r: o . 5| ' 4. WV..- 4‘ 4‘} {Stung .. I. . "its .4 ' U CHAPTER 4 DATA ANALYSIS 4.1 Observer Reliability Observer reliability is essential to this experiment, since the samples are relatively small, and some analyses required the aggregation of data from different observers. To determine whether data from the various ob- servers could be combined, a test was designed to determine if observa- tions made by different observers were significantly different. Two observers simultaneously recorded conflicts on eleven different occasions. In seven cases observers one and two recorded simultaneous data, and in four cases the observers were numbers one and three (see Table 4). The Pearson's xztest for goodness of fit was used to determine whether the resulting conflict counts were observer dependent. This test is used to study data which can be grouped into multiple categories across at least two dimensions. The categorization can be either scaler or vector level (contingency tables or cross tabulations), and is used to determine whether the categories are independent or whether certain levels of categories tend to be associated. An r x c contingency table with columns E A, rows E B, and cell frequency n is used to test the independance of A andlB by calcu- lating the value of x2 (60, 61): X2 = Z ("13' 7 £13) all Eij rc cells where: 28 V . out kit 7 ‘1‘. .xv N» 29 "ij = frequency of observations in cell AiBj n. x n . Eij = expected frequency of A, Bj = -19—N——91-if the charac- teristics are independent ith row total, or frequency of Aj 3 ll i0 "oj = jth column total, frequency of Bj n : Z nio = Z noj The null hypothesis of independance is rejected at a specified level of significance (0), if the calculated x: is greater than the upper tail of a x? distribution with degrees of freedom = (r-1)(c-1). The null hypothesis for observer reliability was; Ho: there is :10 difference in conflict counts by type of conflict between observers. 'The null hypothesis was accepted at the a = .1 level of significance, 611d thus data from different observers could be combined for further analysis (Tables 5 and 6). 4. 2 Representation of Driver Characteristics Since data on the characteristics of conflict involved drivers (sample 2) and the data on characteristics of the general driving public (sample 1) were collected on different days, an analysis was conducted two 'test the assumption that "drivers passing through an intersection at a. Stoecific hour during working days possess similar characteristics to 'd?‘i\/ers passing through that intersection at the same hour of other WCJr‘king days." Intersection populations were sampled for each signalized inter- SeCtion on the first and third working day during the specified hours 0f data collection. The null hypothesis that there is no difference b‘i‘tWeen characteristics of the population during these days was tested. h \-\ol 1 5...) Qovu. o‘.‘ \ \‘§o\\ 3O 8N NN NN NN HN 3N NN NN NN NN NH NN NN NN NN NN NH HH NH NH N oH HNHoN essHoN N N a o N H a a N N N a N N N N H H H H H H NNHN Lo HNNNHN NNHNNNONHN N N a N N N N N N N H H N a N N o o o o o o =N:N-N o N o H o o H H o o H H o o o o o o o o H o HNNHN EOLN UHNNNNH NNoaN N N N H o N N N o N o o H H o o o o o o H H HNNH socc UHNNNLH NNOLN o o H o H o o o o o o o o o o H o o o o o o HNNH acct caaH HNNN N o N N o o o o N o N N H H N N H H H o o o HNNHL ENLN ccsH HNNNN H o o o o o o o o H o o o o o o o o o H H N =L=H HNNH NNHNoaao N N N N N N N N N N N N N N a N N a N N N a NNNNNU NNNN H N N N N N N N N N N H N N N N N - N N N N N «HUHNms onN N N o o N N N N N N H o N N N N o o N o o o NOHHUNLHN aeNN .caaH HNNHN H H N N N N N N H o H N o o N N N H N H o o coHHuaLHN NENN .=L=H HNNH H N H N H N H N N H N H N H N H N H N H N H .62 LN>LNNNN aaNH NHzoN Hmaz NHLoz HNNN HNmz NHLoz HNNN NNNON NHNON HNNN Hmaz NaNoaaa< HaHHNNON Hza N-Nv NEHH Nza N-NV asHH N28 N-NV aeHH Na: NHN .mHasNN NNN N :oNHUNNNmucu Nae NHN .NHaeNN NNN H :owuuomcmuca Noe vcm .NFNENN ecu H coNHUmNcmucH .NHNNNN HNNHN=NN NNNNNNNN .4 NHNNH TABLE 5. 31 Conflict observation, reliability test for observers 1 and 2. Total Observer Counts (Seven simultaneous counts) Total Conflict Type 1 2 rows Right Angle 21 (20.36) (19.64) 40 Rear End (49.87 (48.13) 98 Turning O (30.02) 9 (28.96) 59 Side Swipe (42.75) (41.25) 84 Total Columns 143 138 281 Hypothesis: 0 type of conflict. x2 calculated: at 3 d.f.:accept H0.C TABLE 6. T H : The frequencies counted by observers 1 and 2 are similar for each 1.16 < x2 distribution 7.81 Conflict observation, reliability test for observers 1 and 3. Total Observer Counts (Four simultaneous counts) Total Conflict Type 1 3 rows fight Angle 6(26. 20) (25.80) 52 :ear End 4.(55 42) 56 (54.58) 110 .1:ir711n93(22'17) 21 (21.83) 44 Side Swipe 30 (29. 22) 28 (28.78) 58 Totai Columns 133 131 264 ———HH___ Hypothesis: *1C,: The frequencies counted by observers 1 and 3 are similar for each type of conflict. x: calculated = .181 < 7.81 at 3 d.fJ accept H0. "H. ' 1' » u ‘\ .7 . TIM 32 A xz-test was run to test for differences between population char- acteristics on the first day, and population characteristics on the third day of data collected for the same intersection. The sample sizes were (n; = 99, n; = 27) for population samples on the first and third day respectively. Results of this test are summarized in Table 8, which indicates that there is no significant difference between drivers of the first and third day when age, nationality, trip purpose, occupation, and income are tested. The null hypothesis for these characteristics is not rejected at a = .1. When the driver characteristics "miles driven per year" and "number of accidents in the last two years" were tested, they were found to be significantly different at the a = .1 level. About 58% of the drivers ()n the first day drove < 13000 miles per year compared to 30% of the cirivers on the third day, and three times as many drivers have had two (or: more accidents in the past two years on the third day than on the fLirst. This implies that there is more variance in these characteris- ‘ti<:s than in the others tested, or that a different group of drivers use the road on different days. 4.N3 Characteristics of Hazardous Drivers Nineteen characteristics of both conflict involved drivers, and the general driving public were collected in the field (see Appendix A). The average driver is 29 years old, with an income of about $1100 U.S. a. "nonth. Almost sixteen percent of the drivers did not possess a valid dr‘i ver's license. One of every three drivers had been involved in at 1east one accident in the last two years, and one of every two drivers rui<1 been involved in at least one accident since he started driving. 37 Xty percent of the drivers had completed a driving training course. 33 TABLE 7. Chi-square test of population assumption. x Characteristics of drivers a =T.1 xc Age 2.706 .122 Nationality 2.706 2.015 Trip purpose 4.605 4.278 Occupation 4.606 1.717 Miles driven/year 2.706 5.562* Number of accidents in past 2 years 2.605 9;§49_ Income 4.605 2.271 * H0 is rejected for underlined characteristic. HO: There is no difference between characteristics of drivers on different days of observation. TVXBLE 8. Sample sizes for studied intersection, number of drivers interviewed in each sample for each intersection. Sample 1 Sample 2 Double (Intersection (Conflict conflict population involved involved Intersection sample) driver sample) drivers 1 Signalized 56 66 -- 2 Signalized 70 64 -- 3 Unsignalized 19 41 34 4 Unsignalized 20 _ 23 8 5 Unsignalized 24 51 28 6 Unsignalized 23 51 10 (Total 7 l\ll intersections 212 296 -- S i gnal i zed 126 130 -- ) lJtisignalized 86 166 80 \ 'L‘vJ' ..Q‘ 1“ pi n‘u I rar- .u-u V 8 n ‘r' or o H,- a 'L’JF‘ - v' ‘ - . :q'tr "d' 1 l u I I C u . A. ‘- .“ "Ubu I. «We 5- ‘U1 34 About 40% of all drivers were professionals, followed by students (36%) and all other occupations (24%). Two of every three drivers were Saudi Arabian nationals. Determination of characteristics of conflict involved drivers In order to identify the characteristics of conflict involved drivers, it was hypothesized that conflict involved drivers possess dif- ferent characteristics from that of the general driving public. The testing of this hypothesis was done in two steps: 1. By comparing the intersection population sample to that of conflict involved drivers. 2. Statistically testing the significance of the characteristics that appeared to be different from step 1. 'The following paragraphs will discuss the findings of step 1, while section 4.4 will discuss the findings of step 2. The characteristics of conflict involved drivers (sample 2) were ccnnpared to those of the general driving public (sample 1) at both sig- ruilized and non-signalized intersections (Table 14), to determine whether haxzardous drivers can be identified by these characteristics. Driver characteristics overrepresented in the conflict sample at bcrth signalized and non-signalized intersections include the age groups < 25, the age group > 46, Middle Eastern (but non-Saudi) nationality, Occupation listed as a salesman, craftsman, student, or professional dr‘T‘ver, has completed 9 years of schooling, drives an average of 11,000 mi Tes per year, has had two or more accidents in the past two years, and has attended a local driver training school. On the other hand, safe drivers (those underrepresented in the con- Fl'T<:t samples) are of the age group (26-45), of Saudi or Western 35 nationality, have a professional occupation, college education, have had no accidents in the past two years, an income of $1800 U.S. a month, and had driving training outside the Kingdom. Several plots (1-5) were used to show the representations of dif- ferent age groups in conflict involvement, in order to determine whether a certain age group contributes more to conflicts than other age groups. The graphs show that drivers of age group (5_25 years of age) are overrepresented in conflict involvement, and that drivers of age group (> 45 years of age) were slightly overrepresented in conflict involvement. These observations support the hypothesis that there is a positive relationship between conflicts and accidents, as young drivers and drivers with an accident history were significantly more likely to be: rwapresented in the conflict sample than in the sample of the general toublic (see Fig. 6). 4 .4- Test for Driver Characteristics that Explain Conflict Involvement Since most characteristics of the general driving public are not significantly different across test sites and samples, these data were Dcnoled, and x2 tests conducted to determine whether the differences in characteristics identified in the previous section are statistically Significant. The frequency with which characteristics occur in the general popu- 1ation were compared to the frequency with which these same characteris- tics occur in drivers involved in conflicts. This comparison was con- duCted for data from all intersections, data from signalized intersec- tiOns only, and data from unsignalized intersections only. The sample size for each intersection is given in Table 9, and the 36 ‘ 6 4 9... Nu mermv 4 m2 1 1 1 1 umuzmmmcamccm>o Ncm>wco to pcmucma 1‘ no N» no 8 02 1 1 umucmmmcamccmucs Ncm>wco to pcwucma om A omnmv. mvnfic ovuom mmuHm omuom mm v Age Representation of driver age in conflict involvement. Fig. 1. 37 6 A.» 0 more 4.4 Md 1 1 1 12‘ cmuemmmcamccm>o Ncm>wco No ucmocma 4 9... A... no 8 02 1 1 umpcmchamccmuc: NLN>NLQ mo ucmocma cm A mvnflv oeuom mmuflm omnmm mm v Age Representation of driver age in turning conflict sample. F519. 2. 38 1! 6 A... 2. mu 80 4 2 1 1.. 1.. 1 cmpcmmmcamccmso Ncm>wco No Hcmucma A 2 4 no mo 0 2 1 1 emucmmmcamccmvcz Ncm>wco No ucmucma om A omaoe mequ oesom mmuHm omuom mN v Age Representation of driver age in right angle conflict sample. Fig. 3. 39 ‘ ll ‘1 No A» mu no ,0 n4 ma .1 1 1.. 12+ umucmmwcamcgm>o Ncm>Ngo mo ucwucma 4 no 4 6 mo mum/H 1 1 umpcmmmcamctmccz Ncm>Hco to ucmucma cm A omuov mvafiv oenom mmufim omamm mm v Age Representation of driver age in side swipe conflict sample. Fig. 4. 4O 1‘ 6 4 W. W 80 4 2 1 1. 1. l. umucmmmcamccm>o Ncm>wco No ucmucma ‘ 11 9. m. .b no my A2 11 umpcmmmcamccmvca NLm>HLo to pcmucma cm A menav oqnom mmuHm omnom mm v Age Representation of driver age in rear end conflict sample. . 5. 1'9 F 41 Population Sample Conflict involved drivers with 2 or more accidents in last 2 years DI 'U C) > I— U, OQ’L >Lt‘U :20.) .— >5 1323“ 40 u— H F050 2&5 ,— o 30 US: 44'!- “-‘r- 0319 20 «Hm: 588 u)..— 10 L-t-U GLO QD< L“ 3 3 N 1—4 N I I I <1' (‘4 to '43 V N 0") A Age Fig. 6. Age of conflict involved drivers with 2 or more accidents. 42 test results are shown in Tables 9, 10 and 11. These results indicate that there are significant differences in the characteristics of conflict involved drivers and the general driving public. When all intersection data was combined, the null hypothesis of no difference was rejected (at a = .1) for only nationality and driving schools (Table 9). When using only signalized intersection data, the null hypothesis was rejected (at a = .1) for age, nationality, occupa- tion, driving schools, miles driven per year and income (Table 10). These results suggest that there are differences between the general driving public and conflict involved drivers at signalized intersections. Some of these characteristics may be useful in predicting accidents at signalized intersections. When using only unsignalized intersection data, the null hypothesis of no difference was rejected (at a = .1) only for driving schools (Table 11). The fact that the null hypothesis is not rejected for nearly all tested characteristics at unsignalized intersection is due (in part) to the fact that nearly all drivers are conflict involved drivers. This is true because a violation of a traffic control device is considered a conflict, and nearly all drivers failed to observe the stop signs at these intersections. (An average of 80% of all conflicts observed at unsignalized intersections were right angle conflicts, which include control device violations.) In an effort to more clearly identify conflict involvement at un- signalized intersections, the test was repeated with Sample 2 including only those drivers involved in at least two conflicts. Using these data, the null hypothesis was rejected (at a = .1) for age, occupation, 43 educational level, and number of accidents in the past two years (Table 12). These results indicate that including violations as a conflict simply tends to mask real differences between hazardous drivers and the general driving public. At signalized intersections, like unsignalized intersections, those drivers involved in conflicts have significantly different characteristics than the general driving public. The age of drivers was found to be significantly different at a = .1, at both types of intersections. Thus, age appears to be the variable with the greatest explanatory power among the nineteen charac- teristics tested. This result is consistent with findings by Lynn (23) in which he concluded that age is the variable which best explains differences in accident involvement, and McGuire (59) who reported that age had the highest simple R value (.38) among all characteristics correlated with accident involvement. Drivers of age groups (< 18, 19-25, and > 46) were overrepresented in conflict involvement at both intersection types. The highest in- crease in representation found was for age group < 18 among all age groups. The age group (26-45) was underrepresented in conflict involve- ment at both types of intersections. Drivers with a middle eastern nationality (but non-Saudi), those with two or more accidents in the past two years, students with 9 years of education or less, and drivers with an occupation as salesman, craftsman, or professional driver were also overrepresented at both intersection types. On the other hand, drivers with a professional occupation, those from a high income group, and drivers with no previous accident history were underrepresented at both intersection types. 44 TABLE 9. Chi-square test for driver characteristics in Sample 1 vs. Sample 2 using all intersections. X2 T 2 Characteristics of Drivers at a = .1 X Age 6.251 4.605 Nationality 7.779 12.013 Occupation 10.645 8.200 Educational level 7.779 6.076 Driving schools 4.605 11.938 Miles driven/year 4.605 1.993 Number of accidents in past 2 years 6.251 1.112 Income 4.605 3.217 Drivers license availability 4.605 .649 0 combined. TABLE 10. Sample 2 (signalized intersections only). H : There is no difference between characteristics of population drivers and conflict involved drivers at all intersections Chi-square test for driver characteristics in Sample 1 vs. 2 XT X2 Characteristics of Drivers at a = .1 Age 6.251 7.802 Nationality 6.251 12.463 Occupation 9.236 18.109 Educational level 7.779 3.490 Driving schools 4.605 6.174 Miles driven/year 4.605 5.568 Number of accidents in past 2 years 4.605 .523 Income 4.605 5.450 Drivers license availability 2.706 2.203 H : There is no difference between characteristics of intersection population drivers and conflict involved drivers at signalized intersections. TABLE 11. 45 Chi-square test for driver characteristics in Sample 1 vs. Sample 2 (unsignalized intersections only). X2 T 2 Characteristics of Drivers at a = .1 Xc Age 6.251 .750 Nationality 6.251 .384 Educational level 7.779 1.405 Driving schools 4.605 12.636 Miles driven/year 4.605 1.590 Number of accidents in past 2 years 4.605 1.619 Income 4.605 1.997 H : 0 TABLE 12. There is no difference between characteristics of intersection population drivers and conflict involved drivers at unsignalized intersections. Chi-square test for driver characteristics in Sample 1 vs. Sample 2 (unsignalized intersections, drivers with two conflicts only). 2 XT X2 Characteristics of Drivers at a = .1 c Age 4.605 6.212 Nationality 4.605 1.562 Occupation 6.251 6.088 Educational level 6.251 18.359 Miles driven/year 4.605 .264 Number of accidents in past 2 years 4.605 12.530 Income 4.605 2.207 H : 0 There is no difference between characteristics of intersection population drivers and drivers involved in two conflicts at unsignalized intersections. 46 TABLE 13. Percentage increase or decrease in representations of con- flict involved drivers over intersection population sample. percent over or under in representation Unsignalized Signalized (drivers with 2 Characteristic Intersection conflicts only) Age: < 18 + 23% + 95% 19-25 + 28% + 1% 26-45 - 26% - 21% > 46 +150% N T * Nationality: Saudi + 1% - 6% Middle Eastern + 70% + 39% Far Eastern + 38% - 18% Others — 73% N T Occupation: Salesman + 53% + 43% Teacher - 17% + 43% Professional - 49% Craftsman + 40% + 46% Student + 94% + 50% Professional Driver + 82% N T Educational level: None + 18% - 47% Preparatory + 17% - 47% Intermediate + 23% + 28% High School - 9% + 74% College - 34% - 61% Driving Schools: No - 5% Dallah + 34% N.T. Others - 51% Miles Driven/Year: < 9000 miles - 33% + 2% 9000-13000 miles + 55% + 12% > 13,000 miles + 6% - 7% Number of Accidents Past 2 Years: None - 3% - 23% One + 4% - 7% __Two orgmore + 31% +150% Monthly Income (S): < 900 + 45% — 20% 900-1800 - 10% + 26% > 1800 - 29% - 13% * N.T. Not tested because of low frequencies of occurrences. 47 4.5 Evidence of Patterns Among_Conflict Involved Drivers . Data from conflict involved drivers were separated by intersection for each of the signalized intersections, and the distribution of these characteristics were compared with the characteristics found when data from all unsignalized intersections for drivers with two conflicts were combined. Pearson's goodness of fit test was used to test the null hypothesis of no difference among conflict involved driver's characteristics. The results are given in Table 14 and 15, which indicate that conflict in- volved drivers possess similar characteristics at all intersections. It may be concluded that a 'pattern' does exist among conflict involved drivers, and that this pattern may be useful in predicting future involvement in conflicts, and ultimately in accidents. It is suspected that the reason certain characteristics (occupa- tion, educational level, and income) were accepted in the first test (Table 14) and rejected in the second (Table 15) was due to the large number of young conflict involved drivers (high school students) with- out driver's license traveling through some of the 'back streets' away from police detection. 4.6 Conflicts by Type The literature (32,34,35) contains several articles that suggest a better correlation exists between accidents and specific types of conflicts than between accidents and total conflicts. Thus, an analy- sis was performed to determine whether specific conflict types are prevalent among conflict involved drivers. The null hypothesis of no difference between characteristics of drivers involved in a specific conflict type and the general driver population was tested. The result 48 TABLE 14. Chi-square test for conflict involved 1 vs. Intersection 2 signalized). drivers (Intersection X2 T X2 Characteristic a = .1 c Age 4.605 2.453 Nationality 4.605 5.161 Occupation 7.779 4.959 Educational level 7.779 .766 Driving schools 4.605 2.039 Miles driven/year 4.605 1.810 Number of accidents in last 2 years 4.605 4.107 Income 4.605 3.286 Driver's license availability 2.706 1.431 H : There is no difference between conflict involved driver 0 characteristics. TABLE 15. Chi-square test for conflict involved drivers (Intersection 1 vs. Intersection 2 vs. all unsignalized intersections with drivers of 2 conflicts). X2 T X2 Characteristics 0 = .1 c Age 7.779 4.586 Nationality 7.779 7.192 Occupation 13.362 24.656 Educational level 13.362 23.807 Driving schools 7.779 2.381 Miles driven/year 7.779 7.733 Number of accidents in past 2 years 7.779 13.290 Income 7.779 5.710 Driver's license availability 4.605 32.599 H : There is no difference between conflicted involved driver 0 characteristics. ‘P In Or on a 4 Ir‘ (d .er I 0 "'nl , . 'lu.v 3.. r 5.. ‘ "zr‘ ’ A 'i’li 49 of this analysis is provided in Table 18. Since no single characteristic is significantly overrepresented in drivers involved in all types of conflicts, predicting type of conflict using driver characteristics did not appear feasible. In contrast to the literature, total conflicts were more highly related to characteris- tics than were conflict types. One possible explanation for these results would be that the variance between days or locations in the dependent variable (conflict type) was large, and thus combining these data masked any true relation- ship. To determine whether this is true, a series of chi-square tests were conducted to see if the variance by day or location was signifi- cant. The specific tests conducted were: ‘ 1. HO: conflict type frequencies are similar at each unsigna- lized intersection. 2. H0: conflict type frequencies are similar for each day of conflict observation at unsignalized intersections. 3. H0: conflict type frequencies are similar at each signalized intersection. 4. HO: conflict type frequencies are similar for each day of conflict observation at signalized intersections. In tests 1 and 2 frequencies of conflict type were found to be similar on all days at all unsignalized intersections. Thus prediction of conflict type at these intersections cannot be improved by analyzing each data set separately. In tests 3 and 4 frequencies of conflict types at signalized intersections were found to differ significantly with the observation period and the intersection, indicating that separating the data may be helpful in determining the relationship between driver characteristics and conflict types. 50 TABLE 16. Frequencies of conflicts by type for unsignalized intersections (drivers involved in two conflicts only). Conflict Type Frequency Percent of Total Right angle 48 60 Turning 11 13.7 Rear End 9 11.2 Side Swipe 12 15.0 ES 136'" TABLE 17. Frequencies of conflicts by type for signalized intersections Conflict Type Frequency Percent of Total Right angle 14 11.2 Turning 23 18.4 Rear End 40 32.0 Side Swipe 48 38.4 125 I 100 51 TABLE 18. Characteristics of conflict involved drivers that explain types of conflict. Characteristics Found Significant at a = .1 Con§$gct Signalized Unsignalized All Intersections Right angle Nationality Income Nationality, number of accidents last 2 years Turning Age None None Rear end Occupation, None Educational level educational level Side Swipe Occupation, None Number of miles accidents last 2 years, miles driven per eyar 52 A test to determine whether subsets of driver characteristics are related to the distribution of conflict involvement by type for each of the signalized intersections was conducted. Each characteristic that explained conflict involvement (as shown in Table 10) was tested using a multiplier to maintain the proportionality of the test. _ 11. n . n. OJ J n = total conflict frequency observed "j = total column frequency of characteristic subset noj = new cell frequency of that subset Several subsets of characteristics were found to have a significant relationship to type of conflict. The significant variables are drivers of age group < 20, and with a Saudi nationality. As in previous tests, age and nationality proved to be important variables in explaining the variation in conflicts. 4.7 The Model One of the principal objectives of this study was to determine whether models could be developed to predict the conflict rate at inter- sections, based on driver characteristics. The fact that a 'pattern' of conflict involved driver characteristics, defined in (4.5), exists indicates that some characteristics of conflict involved drivers are consistently related to the conflict rate. It is, therefore, hypothe- sized that the overrepresentation of these characteristics can be used in the model to predict conflicts. Multiple linear regression was used in this study to relate 53 dependent variables to one or more independent variables by means of a linear relationship. One of the methods used in multiple regression is step wise regression where an independent variable is added or removed from the equation to achieve the highest coefficient of determination with the fewest number of independent variables. The model tested was of the form: Y X X k = a000 + a11k 11k + °°° T a14k 14k where: % of conflict involvement (pattern) for characteristic X = i subset j intersection k ijk % ofigeneral driving public for characteristic i subset j intersection k Total conflict count Y = conflict rate/ = ( for each approach ) (1000) k 1000 vehicles Corresponding VOTUme (20 min.) n n = number of observations corresponding volume = 20 minute volume for that approach The driver characteristic most clearly related to conflicts, age, was used as the first independent variable in the model tested. The multiple linear regression models developed as a result are Y = 508.9 - 242.9 X k 12k (1) X13 corresponds to age group (19-25) Y = 925.2 - 95.8 X - 349.5 X k - 149.3 X 13 (2) 11 12 X11 corresponds to age group 5.18 and > 46 years of age 54 X13 corresponds to age group 26-45 In all cases, the 't' test indicated that the coefficients were not significantly different from zero. The coefficient of determination R2 was .32 (32% of variations ex- plained by regression) with an F significance of .244 for the first model, and R2 = .58 for an F significance of .558 for the second model. All coefficients of the independent variables were negative with the lowest coefficient (95.8) corresponding to age group (< 18 and > 46) and the highest coefficient corresponds to age group (26-45). Because the sum of the percentages of all age groups amounts to 100%, the lowest negative coefficient represent that subset of the age that makes the greatest contribution to the conflict rate. Thus, a well defined and statistically reliable linear relationship between conflict rates and age does not exist in this data even though the x2 test indicated that there was a relationship between these variables. The number of data points was too small to test non-linear regression analysis, or to stratify the data into multiple regions for separate linear analysis. To illustrate the empirical relationship between these variables, a plot of the conflict index, which is defined as the ratio of the per- centage of conflict involved drivers of a certain characteristic to the percentage of the same characteristic of the general public, and driver's age is shown in Fig. 7. The graph suggests two linear relationships, with the conflict index decreasing with age for ages 16-40 and then increasing with age after age 40. More data would be required to test the statistical reliability of each of these relationships, but this explains why the x2 test was significant and the linear regression not 55 strong. 4 The conflict index plotted against age for both drivers who had been involved in at least one accident in the past two years, and drivers with no accident in this time period is shown in Fig. 8. Drivers with no accident history exhibited the same general pattern as all drivers. However, drivers with one or more accidents in the past two years exhibited a similar relationship for the under 40 age group, but the conflict index did not increase with age beyond this point. 56 S ml e v .1 P: D .T. C 3| 1|. ..T n O C 1' 1| A 1098765432109876543 . 221111111111 xaucH HUHHNNON om A omume mvnflv ovumm mmnam omnom mNuHN omumfi Conflict index by age of all conflict involved drivers. Fig. 7. 4.0 3.8 3.6 3.4 3.2 3.0 2.8 g 2.6 E 2.4 .3 2.2 E 2.0 3; 1.8 .3 1.6 1.4 1.2 1.0 .8 .6 .4 .2 Fig. 8. 57 —--Conflict Involved Drivers with No Accident History “‘ Conflict Involved Drivers with One or More Accidents ,in the Past Two Years k“-)S-oo.X- - X L!) 0 L0 C? Q' d' I I H go H m 0") <2- 16-20 21-25 26-30 46-50 0 L0 A Conflict index by age of conflict involved drivers with and without an accident history. CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS 5.1 Conclusions Based on the data collected and analyzed in this study, the follow- ing conclusions were reached: . 1. The most hazardous driver is in the age group 25, has not completed more than 9 years of schooling, has been involved in two or more accidents in the past two years, works as a salesman, craftsman, professional driver, or a student. His income is of a lower income group, and he drives an average of 11,000 miles per year. 2. Drivers with prior involvement in accidents were involved in ”more conflicts than drivers with no accident history. 3. Age was found to be the single most explanatory variable of conflict involvement. Age and occupation form the best multiple re- gression model among all tested characteristics. 4. Uniformity of conflict involved driver characteristics was found to improve when intersection type is considered. There was no significant difference between the characteristics of the general driving public and conflict involved driver for combined data from all intersections. However, differences were identified when the data was segregated by type of intersection control. 5. Characteristics of conflict involved drivers were found to be similar at all intersections, and the potential of using this pattern is promising in predicting future conflict occurrence and ultimately future accidents. 6. No single driver characteristic has a distribution matching t:he distribution of type of conflict occurrence. However, combinations 58 59 of individual characteristics (young Saudi drivers) correlates well with the occurrence of conflict types at signalized intersections. 7. Drivers licensed by local driving schools were involved in more conflicts than drivers without driving training or with training from schools outside the Kingdom. 8. The multiple regression model results indicate that a simple linear relationship between conflict rate and driver's age does not exist. However, when the relationship was presented vectorially two different relationships were observed; one for the age group 16-40 and a second one for the age group 40 and over. More data is needed to construct and test these relationships in a model. 5.2 Discussion and Recommendations The traffic conflict technique was determined to be a promising technique for identifying characteristics of hazardous drivers at urban intersections in Saudi Arabia. The relationship between the charac- teristics of hazardous drivers and conflicts was found to be analogous to that between driver characteristics and accidents. This strengthens the arguments for using conflicts as a surrogate safety measure when- ever accident history is not developed or not available, such as in the case of Saudi Arabia. The similarities in the observed characteristics of conflict in— volved drivers at different locations provides the basis for the iden- tification of a driver profile (or pattern) necessary for predicting future conflict rates. *Based on the data collected in this study, it appears that the prediction process should be carried out separately fflor signalized and unsignalized intersections. Young drivers were found to be overrepresented in conflict 6D involvement as well as drivers with one or more accidents in the past two years. One of every two drivers interviewed 18 years of age or younger, was involved in one or more accidents in the past two years. This group of drivers is considered a high risk group, and issuance of special driving permits, which is allowed for drivers below 18 years of age (the legal driving age), should be reviewed. These young drivers tended to traverse back streets away from police detection. It is recommended that local police introduce more patrol surveillance to cover remote roads, and better enforcement action be promptly and strictly applied to traffic violators to reduce the risk of accidents. Local driving schools have not been effective in minimizing the risk of involvement in conflicts. These schools should be evaluated and upgraded to achieve greater effectiveness in driver training. The establishment of an accident location system should be initiated in Saudi Arabia to provide a data base for research, improve- ment, and development programs. BIBLIOGRAPHY 10. 11. 12. l3. 14. 15. BIBLIOGRAPHY Russan, Sabey 8., "Accident and Traffic Conflicts at Junctions,: TRRL Report LR 514. Waller, P. F., and G. G. Koch, "Characteristics of North Carolina Drivers," Highway Safety Research Center, Univeristy of North Carolina, Chapel Hill, N.C. (December 1971). Smead, R. J., "The Statistics of Road Accidents," 3rd Int. Sympo- sium on Urban Traffic Safety, October, Budapest, 1972. Jorgenson, N. O., "The Statistical Detection of Accident Black Spots," 11th Int. Study Weak in Traffic Engineering and Safety, Brussels, 1972. Hall, P. A., "The Identification of High Accident Locations," 11th Int. Study Week in Traffic Engineering and Safety, Brussels, 1972. Mattewson, J., and Brenner, R., Indexes of Motor Vehicle Accident Liklihood, Highway Research Board Bulletin 161, 1957. Breunning, S. M., and Bone, A. J., Interchange Accident Exposure, Highway Research Board, Bulletin 240, 44-52, 1959. Thorson, 0., "Traffic Accidents and Road Layout," The Technical University of Denmark, Copenhagen, Denmark, 1967. Smith, w. L., Probability Study of Accident Locations in Kansas City, Missouri, Traffic Engineering 40 (7), 42-49, 1970. Gorssman, L., Accident Exposure Index, Highway Research Board Proc. 33! 129’ 19, 54. Sutri, V. H., "Accident Exposure for At-Grade Intersections," Traffic Engineering, December 1965. Gwynn, D. H., "Relationship of Accident Rates and Accident Involve- ments with Hourly Volumes," Traffic Quarterly, July 1967. Feuchtinger (1949) Road Junctions and Trans. Die Bautechnik (Germany). Raff, M. S. (1953) Interstate Highway Accident Study, HRB Bulletin 74, 18-45. McDonald, J. w., Relation Between Number of Accidents and Traffic Volumes at Divided Highway Intersections, HRB Bulletin 74, 7-17, 1953. 61 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. x29. 30. /31. 1,32. 62 Surti, V. H., Accident Exposure and Intersection Safety for At- Grade, Unsignalized Intersections, HRB 286, 81-95. Chapman, R. A., Traffic Collision Exposure, New Zealand Roading Symp., 184-198, 1967. Tanner, J. C., 1958, Accidents at Rural Three-Way Junctions, J. of Instn. Highway Engineers, 2(11), 56-67. Colgate, M. and Tanner, J. C. (1967), Accidents at Rural Three-Way Junctions, Road Research Laboratory Report, LR-87, Crowthorne, England. Webb,(i.M., 1955, The Relation Between Accidents and Traffic Volumes at Signalized Intersections, Proc. Inst. of Traffic Engr., 149-160. Thrope, J. D., 1964, Calculating Relative Involvement Rates in Accidents Without Determining Exposure ARRB Aust. Road Res. 2(1), 26-36. Seibrecht, E. B., Schumacher, C. F., Lauer, A. R., "Accident Char- acteristics of Drivers at Various Age Levels," Iowa State College, Traffic Safety, December 1959. Lynn, C. W., "Psychological and Sociodemographic Characteristics of Accident Involved Drivers," Highway Safety, Virginia. Betz, M. H., and Bauman, R. 0., "Driver Characteristics at Inter- sections," HRR, 19S. Faulkner, C. R., 1968, "Accident Debris and Reported Accidents at Roundabouts," TRRL Report LR 202, Crowthorne, England. Spicer, B. R., "A Traffic Conflict Study at an Intersection on the Andoversford By-Pass," Transport and Road Research Laboratory, TRRL Report LR 520, 1972. Waller, P. F., Koch, G. 6., "Characteristics of North Carolina Drivers," Highway Safety Research Center, University of North Carolina, Chapel Hill, N.C., December 1971. Pelz, D., Schuman, 5., "Drinking, Hostility, and Alienation in the Driving of Young Men," A paper delivered to the Third Annual Alco- holism Conference, NIAAA, Washington, June 1973. Baker, W. T., "Evaluating the Traffic Conflict Technique," FHWA. Perkins, S. R., Harris, J. 1., "Traffic Conflict Characteristics - Accident Potential at Intersections," GMR, December 1967. Baker, W. T., "Evaluating the Traffic Conflict Technique," FHWA. Paddock, R. D., Spence, D. E., "The Conflict Technique: An Acci- dent Prediction Method," Ohio DOT, August 1973. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. ‘49. 63 Glennon, J. C., "Critique of the Traffic Conflicts Technique," TRR #630. 1977. Spicer, B. R., "A Traffic Conflict Study at an Intersection on the Andoversford By-Pass,“ Transport and Road Research Laboratory, TRRL Report LR 520, 1972. Spicer, B. R., "A Study of Traffic Conflict at the Intersection Transport and Road Research Laboratory," TRRL, Department of the Environment, Crowthorne, Berkshire, 1973. Spicer, B. R., "A Pilot Study of TC at a Dual Carriage-Way Inter- section," Road Research Laboratory, TRRL Report LR 410, 1971. Cooper, P. J., "Predicting Intersection Accidents," Ministry of Transport, Canada Road and Motor Vehicle Traffic Safety Branch, September, 1973. Pugh, David, "Relevance of Conflict Studies to Safety Improvements at Intersections," Washington DOT, 1977. Pugh, D. E., Haplin, T. J., "Traffic Conflict in Washington State," Washington State DOT, April 18, 1975. Al-Ashini, Nasa, "Alternative Methods of Examining Correlation of Conflicts with Accidents," Michigan DOT Traffic Engineering, October, 1976. Parker, M. R., Jr., "Right-Turn-on-red," Virginia Highway at Transportation Research Council, September, 1975. Blunden, W. R., Munro, R. D., "Report on the Study of Traffic Con- flicts and Accident Exposure," Australian DOT, June 1976. Seymour, W. M., Deen, R. C., Havens, J. H., "Traffic Control for Maintenance of High-Speed Highways," Research Report 380, Kentucky DOT, December 1973. Cima, B. T., "An Evaluation of Freeway Merging Safety as Influenced by Ramp Metering Control," TRR 30, 1977. Pigman, J. 6., Agent, K. R., "Raised Pavement Markings as a Traffic Control Measure at Lane Drop," Public Roads, Vol. 39, No. 1, June 1975. Pigman, J. G., Seymour, W. M., Agent, K. R., Cornell, D. L., "An Operational Analysis of the I-69, I-65 and I-71 Route Junction at Louisville," Kentucky DOT, April 1972. Zegeer, C. V., "Effectiveness of Green-Extension of Systems at High Speed Intersections," Kentucky DOT, Divisions of Highways, May 1977. Clyde,WL N., Paper presented at Traffic Conflicts Seminar, March 20-21, 1974, Michigan DOT. SO. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 64 Parker, M. R., Jr., "Use of TCT to Assess the Hazards of Trans- porting Oversize Loads," Virginia Highway and Transportation Research Council, December 1977. Guttinger, V. A., "Conflict Observation Technique of the Work Done and On-Going Research in Holland," Holland 1977. Paddock, R. D., "The Conflict Technique Procedures Manual,” Ohio DOT, July 1975. Blanz, W. D., and Migletz, D. J., "Application and Traffic Con- flict Analysis at Intersections," Midwest Research Institute, October 1979. Clayton, M. E., Deen, R. C., "Evaluation of Urban Intersections Using TC Measures," Kentuck DOT, August 1971. Hayward, J. C., "New-Miss Determination Throng Use of a Scale of Danger," Pennsylvania Transportation and Traffic Safety Center, HRB Record #384. Lightburn, A., and Howarth, C., A Study of Observer Variability and Reliability in the Detection and Grading of Traffic Conflict, 2nd International Traffic Conflict Workshop. "Motor Vehicle Accidents in Relation to Geometric and Traffic Features of Highway Intersections," Stanford Research Institute. National Safety Council, "Accident Facts," 1970-1979. McGuire, F. L., "The Understanding and Prediction of Accident- Producing Behavior," North Carolina Symposium on Highway Safety, Chapel Hill, N.C., 1969. Nie, Hull, Jenkins, Steinbrenner, and Bent, "Statistical Package for Social Sciences," McGraw-Hill, 1970. APPENDIX A Driver Interview Form for Both Sample 1 and 2 A-l 55 mt" w)‘ 4” r- .l l l' . | . “4".H4h4, Counmmtlu '"I""“""" 44444441 ”and! . u‘ only 4444' 80444444445: Hadron ‘by We“! he on on you? “4’5 I I JJ’Q- 1 I Jay-1p l a... Doc 4. your 44.44.44.444” “444“ Ion-6.444“ (Spocuyl "b,oblg‘J' 4 i 4.44 4 ”.44 4 I 3.....44 4 1.1354.- 1 do) “to on you got”? 44.4444 Moo. Shopping 0444:" 42444:.le 4 “aw“ b an- lg'k I Solemn/Inn: I “I. to m melon 05‘ ’91- I l Tanner/Proton”: 4,4» I 4» I .4».- 1 l Prolonlooal/‘hchnlcol/Mmrut g4— J-b ’ *9 Mb 1 l Craftmn/nochantc/Poctory lotto: A .35 lfifi- I ,u I l cur4c44/ucncu4u/oulc- 4444444.: 4'35 4 1 "who: .4... 4 1 I444 Iron! 4.1—— 4 ) omor 'Jchuarde-Hb 1 I‘ll... 4 Din-,1- I IJ'O C 4.3-Io! 1.3;. In: 4- pl Mind in" m tum—flu- 4444444 34:44:31 mu.)- m tyuohdh» 4 I... I 4 1 II. you can “to an no ‘ 440 Mm Uh“ 25.5.4.4 mas-1_— 4 4,544.3“...5. 3.4. 4 4;. 4-4 4 1 .45. 4-4 4 4a,..- 3,44 In: my ”an did you an”? A 1-2 gun )-5 you-- nor. 4 ' “*J‘JHVL ‘ “fibh 4 A Jb‘aH- ‘ 4m! A 4405‘ 3.45- me tn- “ on you «In? Pun-90¢ a: mu I44. 0444." 'Mowuowbfi be my ”on 440'. you 11'. 1. 8M. om? 'oanlquI-Jlfi C I '0 I I-‘alga I “85‘ I My“ I dun-I love. you has n «4444.. about I. cum "one. on." «mum ~- I was! mfl'JeI-MMAV ' ' .0 any in...“ you «tn 1. a day. can. at you? 4wwumdhoowp; 4 4.3.4. 4 3.5 I Lab. 4 Dana-i“ In. .0447 munc- a“ you no. «c444. I044. 044. 1540 Mon :44 . cu ’m "" mv...u.n,.. 434.4,“, 4 4,... 4 4-4.4 4.4.44 4 a.m.,; an my mime. “at“ In you: an: «4444-. «9031044440. I944. 044. M More than in 4 and.» 4 H 4 4..- 4 4...: 4 41' m 10! actual? . In You In In 4 44.4.1.4.» Jo no you have children IIIUJJEUH o mmmvtun‘clzi‘y? ‘ '0‘...” ‘ ’fim ‘ .zytium 4qb¢y¢na 1 HmJ-n 4 Ha— ! lab—b I. you have m“ beau? Vary an All. r4944: Duh 0 mum-m» 4 4404-"...444 an to you linen, no. "ooo-’ooo ‘ ,‘."I'fi.fi' Lou than “.3000 It)”. - «on lion than 60“ “final—5,4...» 4 I... 4 H 4 45449.11 in you In. 0 I.“ Antiv- «no; no..." In In La“: 44. v 444 I no. “JV-”a- APPENDIX B Chi-Square Test of Population Assumption 66 Na u w .toommmm mo mwmwuo ~ Ivan ohmmn ozocumau ac umxwuo n rpm: ¢m~mu 5.99H nogm sud — cm H Nona — mono mono — moc— hh ” mu — @oh H muhn flown — no¢m hm ~ w ~-"'-"' mom 4<>oh > :cm «omuzcmz arccpucoonaocuo «Jazabmmcau ... . Nm\wnxmo u uhqo zanh¢u¢uv zo-<>zummo oz~wm~r we «mean: . n u¢~ hzaou ccccuccvcau91§i umq «a aaaaaaacaaaacoc uzo hmwaa. wank ugummam uu—«zzo—hmuac muubmnzuhuu¢c uz otttatt‘tfitbttbtttttttttttfi: I. .m >_H44:ou_¢z «a _ 4 4 q.m 4 b m m c v o s n a a o i o o t v » s o t t . t o wz m4“ m . Nox~du~= u mp4: zanpnmc Hzczu mum 68 HH 4 manHHHammmo wszmH: mo .rHQMHHL do Hmuaumo ~ rHHx HHHHN.: « mmaanm H HHWHH H.- m.HH H< OH nH m~ H an 4 z: Hco H a. . E H H . :.mH H o.m W .:m m HZHa HH _ H H HH H .n a.mn H .m m m.:m m.o~ H ..mH H an H H» H .~ » .3 H 4.5. . .:o H m.m: H ¢.ma H a. HN H JWHH H cm h oH * 94 H .H ..u H.~ H .ua Ho» H + .ca Hco :o« «amazou: >Hmo mumtaz :0 34¢ norm mtox xxo: ma s o o no t t o Hummam manm H2439 mum 69 H u mchH<>zmmmo oZHmmH: mo amoraz .soaouxa Lo muumouo H :HH: GHHHH.H n ua<=cm H10 2<¢ o.ccH o.Hm e.ah Hc m o.~H H ¢.wm H «c.H2ua:Hm.Hu¢¢u :n H o H mm H .m H o.¢ H ~.cH H H =.HH H m.e~ H . - m.- H ¢.m~ H w.o~ H mu4u.:u H Hum ch no: «ammzoua. >«ozzmH Hum 3oz H hzaou '0 uaacuacuwcciaiuccaca > a.§acccuccca . chhm H < H a c < H m m c c u c a a a a a a a a a a a a a a a a c uzo qummzm H «zmexmo n wHH¢o H2 m Haw omuzcu: >402: < \O a Cl a: HHHHfiHHHHHHO—IHHHHuHHHHHv-O pr) )0! i a a a a a i r c a c c a c c < H a m < H m w o a u a a a - mzo HHHLmam . chmemo u LH<¢ 7oHHH¢a Hz« < .oc .¢ .mu a O C ¢¢a iI-C c1. .4. 'xu ‘I 'I c c ‘I C gd. ,4 .0. 'I ‘I C C V3 71 .I «N: v-‘d' 3< \O 0 \C’ 0 GI- -‘ "'3 «a b‘" ~r .- flHHF‘H—Ib’t—‘Hl—fip. 1! 0' I C '0 I I U? I anon I O‘HG'H w | OOOde. 2 I (HOMO I FIGHT Q u #6! I “"WH '11 I I Z'l)- I I ' u-IHI-IHI—IHI-II-OU-Ih-IH 0| I N I I >- I NF *0“: I GIG-cm 4 It." 0 o o I e o o O C! I hhifl I comm Z I flmc I seen 3 I I m I HHHHHHHI—NHH—HH I v-v-r- I I-UUU I 0 0 20.0.l I '4 I") D I . OJJI- I K a: 00°C I >- )- ¢UI-I I I u-o r-o 2 1'. O J O . a O' c O '0 '0 H 1-0 V A 011 -1--------1-------- 7 c 9 21. 39 78.6 72 H u monb<>¢ummo oszmHz no mumzaz .zccuuqm no mwummuc m :HH: mmm:n.m u u¢ H pom sou .zcr «cmuzcu: > * a s c a y a c a I c c a ¢ a I a a a a u a n ’ u « I a c c a >m m¢¢u> c~p hm<4 mhzucHuu< mo mumzaz «Ho b < A D 1 4 h m w O x U a I c a « c c I s c a a a 1 a R c a uzo uaHumam . ns\«n\ma u u~¢o 20HhH¢o Hzmummo cszmH: mo mumznz ozcommmu mo muumouo N . IHH: HmohNoN u u¢ H Hum Aou no: «cmuzcus > a c i u a c a a a a a a I ¢ c . a a c c a I c a c a a c . c m ocomo>azhzo:.uxcozH ch < A D m < H m m o a U 5 c a a c a c i a u a t a c a ¢ I c was uaHmmzw H Nn\NH\No u wHH¢o H2416 APPENDIX C Chi-Square Test for Driver Characteristics In Sampe 1 vs. Sample 2 Using a1] Intersections 74 .zocuuxm Ho mumxouc n .zHHa mmwam.¢ u uz .acanacnucauacsacccccatitcaaa... .m um< Ho .Hco H241" Huo 75 a H .zooumcm mo mumgouc c eucHH hem Gui-.mauata.’ < A 3 o < h m m o a u H usxm n monH4>¢ummo qummHz kc cumin: :HHa chHong n U¢H szou a c a a a c a a a « a c a c t a a ¢ c >bH4H¢o Hz¢ummo rpm: hemho.o oz~mm~x mo cumin. u u¢<=om uzu :<. H usmexma u ur— . c a a c a . é a c a a I a a a c c . HH>HH onH.u=am no t c c c a c c c . a a c « a a . c . oaH Hz. HHHum 3m hmuac uJ—u mu~hm~¢upu~¢c ~2¢mmco oz—mmnz we «mean: .zoouHmu Ho muumouo m :HHz m~snm.HH u u H.9HH ~.om e.H. HquH Hem - mom H «Hm H zzzHou p--"--'- -""'-- ' H c.. H ..H H H m.m H o.cH H a.mH H a... H m.H. H mm H mm H a. H .. Huu---wu-Hr-----n-Hu H «.5. H H.5H H H a... H HJN. H a... H m.H. H «gun H 5mm H cmH H mm H .N H a.mm H ¢.aH H H H... H m..n H H.m¢ H ..H. H m.mn H mHm » mnH W mm w H H.~ H.H H Hum HcH HHHcH uH HHH Ho. Ho. 2cm ¢=«m can asH Hzacu 0 a a a a i a a a a c ¢ . y a c c a c a . i a a v m mdoorom uz~>~¢c < 4 D m < h m m c z u c a c c c c c a u . . c x~¢c ~zIH2 coonHA mmu H mnH H Hm H on H--------H--------H- H c.HH H m.~ H H m.o~ H o.HH H a.mm H ~.Hm H comm H mmmeuHocm mm H cm H mm H .m H a.mH H H.HH H H .nm H a.mn H HJHH H H..m H o.m¢ H ¢>qu ocaav HwH H mm H n: H .H H.~ H.H H um Ho» HqHaH UH _ MHH on How 2:: mz . a o a c . a . a . c . ¢ . . a c c a a c . ¢ . a u . u c . m >on mmuHMHOHHx HHc «AH—adhmmcmu licuccaccucac.aa.l or» uzo uaHmmam H mmmech u mHH¢o H2aummc qummHz.mc mumzaz .zoouuzm no mwmaeuo N thz Nncwm.n u m¢<=am qu 3<¢ a.c=H 0.mm m.H¢ HH Hznou I I.c«acaviaccacha#acac¢tciaHacc; Q m¢ 93H Hmaa mhzuaHuu< mo muazzz NHc d J D a < H m m o m U y a a . a a a a c c a a c a c . a . ozH uzo waHmmzm H mmmexuc u UHH¢o Hz«:c Hno 81 mm H monH<>¢ummo qummHz ma «mean: u .zocuumm no muwmouc :HH1 «mmH~.n u um<3cw Hzo 34x :.o¢H c.cm c... 4JIHzoz.uzoqu ch 4 A : z 4 b m m o z u . c a c . c . c . a a i c . a a a . c3» uzo uuHumnm H mnmeHNH u HHHmo Hz¢mmmc oszmHz mo zumxaz .zcauuzu mo wuumouc m :HHz nmm¢m. u u¢<3am qu n<¢ =.eeH c.5m ¢.~¢ HH¢o HaozhH: Hm H nH H nH H m H o.om H m.an H . H a.ma H H.mm H . c.c: H a.mr H c.~¢ H HH m.¢u>Hzo zHHn Hm: H can H HmH H .H H.. HoH H Hum HcH H«H¢H u; uHH Hum Hou . 2e. mt . a a a a a c c a a g c a a a c c c a a a . c c . . a a . ¢ I »m HHHHHc< uozmuHH m.¢u>H¢o ch r < 4 D a < b m m o c u . c c c 9 . . . c i c a a a u a a ozH wzc HHHumam H ommequ u uHeo chHH¢o H2H szou «cicccccascaecicaa« a c a . ¢ a c . cap wac u wbHzo Hz<=u Hue 84 \ H n mchH¢>¢mmmo .zeoHuHH Hc mummuua n :HHz Ha~¢¢.mH H.HHH .H.Hm a.mc mm“ HnH mmH H ¢.N H ~.¢ H H o.q H m.oH H e.HH H «.NN H ¢.HH H Hm H a H Hm H H ¢.n H H.« H H H.H H c.m H H.o H m.mm H «.Hs H H. H HH H H H H H.HH H H.o H H H.n~ H m.mH H ¢.xH H ¢.nm H c.0n H H: H an H HH H H--------H--------H- H m.~m H e.Hm H H a.¢c H H.¢o H n.¢: H ~.Hn H ¢.ce H qu H :c H cm H H.~ H.H H HHHOH H4 HHH so: Hx¢m cz~ asz Hme H> .ccccaaCcuucaCCCccacyc. z @szmHt mo «mean: u u¢<=em HIu 34¢ «hep zadou .m mauzho 44< .e azuHmbHHH¢o Hzmuwco oszmHz mo mumtaz .xocuwam uc muummmo m :HHn HHmaH.mH u u¢H¢c .moam es H an HH H .m N-.H-"-"~--""--H- H H.H ‘ H m.n H H H.¢H H «.H H o.cH H H.0m H n.nm H qucaHm Hm H mH H m H .m H H.m H m.n H . H H.HH H ~.H H . HHw H H.am H m.oc H z tacnuauccaacicaccaccaac«iii... onH<¢2uuo cc <43mH¢o H2¢ummo :HH: momm¢.n moon tomc. wmd mflu .H ¢o «A H com H H m. mm H coma H H ho¢m H meme H H ww H cm H H"---"'H--o"-'-'~' H New H «can H H flood H m.¢N H H {.90 H momn H H Hm H Hm H H"”"-'-H""-"'H' H «can H coda H H Noam H NoNN H H Hort H mon H H mm H cm H H coda H c.@ H H moNN H nomH H H comm H Note H H 6N n MN H H cum H cos H ~ ficed H eomn H H moem H non¢ H H ¢N H cm . H How Hofl H “A MJH 1t oszmHz.Ho Humxaz w «HOH :zdou ¢¢ow<¢m4 onHH¢c HzH szou I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I >m m400?0m GZH>H¢O oHc h < J 3 a < h w m o a U I I I I I I I I I I I I I I I I I I cab UZC UJHumDm . Nm\NH\Nc u UHn¢c HzIHx coonHA mNH H hm H H@ H .n H mgew H «.4. H H e.¢e H m.mH H n.9m H m.H¢ H m.mn H ammuHuHccm mm H mm H cm H .m _ H H.~H H H.HH H H a.»& H H.9n H ~.mm H mqnc H 0.0m H ¢>IHz cacmv as H Hm H me H .H H.m H." H mm Ho» JHHCH m4 MHH. um 450 300 a=H¢a mmuHuH 4H2 «Ho mow—U IIIIIIIIIIIIIIIIII 2 ozH uze HHHumam H ‘qwa\mo u uHH¢o Hz¢ummo :HHz Ennmmo qummH: mo «mean: u w¢ IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII H: m¢Hdc Hzm H < A 3 : III < h m n e u monh<>xummc qummH: to «moan: :HH: HHenIIm u umcaom qu 3<¢ o.ocH H.m¢ ooom H I I I I I I I I I I I I I I I I I I I I I I .zom I :4 :HZOIIHHoqu oHa o a U I I I I I I I I I I I I I I I I I I on» uzo UHHmmam H utwaxuo n whH¢o Hz 46 UCOO I a: H OI- ¢Uhl U K O > O H a I- D D Q I I I- I- I-I III 3 J 019 an: Luau or: 90" ‘22 FH- HH :2: scs mn- NF- cam ma~ O O (‘J ('4 IUHBER 0F MISSING OBSERVATIONS = l o APPENDIX E Chi-Square Test for Driver Characteristics in Sample 1 vs. Sampie 2 Using Unsignaiized Intersection Data 92 ozcouumm mo mumzowc n ooc¢H «mm 3:: :HHn encMHI a.mo Hcen mcH mm H QIN H mIH H H o.m H HI: H H QImm H IIII H H m H a H H a.e~ H mIHH H H nIHn H HIMn H H H.ce H wan H H mm H mo H H momm H r.¢H H H m.on H m.HI H H Iocm H moan H H mo H mm H H m.nH H H.H H H mIcm H cocH H H Homm H monn H H on H HH H ~"-"l'- H"""-'H -- H.~ H.H . H H u; HJH H mt IIIIIIIIIIIIIIIIIIIII < 4 D m < H m w o z u IIIIIIIII 03H H chmewo u UHH¢o Hzarm n um<=om qu 34¢ HH¢o HzH¢o H H u mchH¢>¢ummo cszmHz mo mumtnz m IHH: nmnnmIn n um<2cw qu 3<¢ HIcsH HImm HIIn H< HcH Hmu H mmH H. mm H z: 3400 H o H H «I: H H mIHH H IINH H IIuH H nInm H HIcn H ¢u>H¢o Imozm In H mH H HH H In H HIIH H oIoH H H mIm~ H HImw H HIwm H Homw H mIHn H HzmcaHm He H N: H IN H II H eoc H cIN H H HIm H m.m H oIm H HImm H ”Inn H z IIIIIIIIIIIIIIIIIIIIIII onHIQ=uuo Io memo IIIIIIIIIIIIIIIIII 02H uzc uaHmoam H mmmexwo u MHIG onH-IAH hav‘ssns I . \ 95 m Izocumxm ms mwummuo I IIIIIII < J D c < H m m :HH: memHIIH u u¢¢OH<¢ IIIIIIIIIIIIIIIIIIIIII Ju>m4 :oHHIunou mo emu IIIIIIIIIIIIIIIIII cJH uzo uHHmmam H mcx~Hsm= u wHHmo Hz¢uwmo oszmHz m5 aumzzz Izcoummm Ho wwmmoma N zHH: HemmoINH u u¢H Hzaou IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII >m mJOOIum ozH>H¢= oHc H 4 A D m < H m m o m u I I I I I I I I I I I I I I I I I I ozH uzc UJHucam H mmwaxuc u uHH¢o HzIHz coonHA HoH H Hp H on H on H H.HH H H.h H H m.cH H o.o~ H n.mH H moom H H.mn H mmmNHlHoom m: H cm H mH H .N H m.m~ H H.mH H H ¢.a¢ H ~.e¢ H ~.H¢ H conw H Nomn H ¢»st ooomv mcH H ho H an H .H H.“ H.H H Hum Hop 4 IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII >m >H¢c wzuHuaoaHx HHc h<43mH¢o H2 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I Hm mm cm» HmHH «HzmoHuuc to guess: «He H < 4 D u < H m m c m u I I I I I I I I I I I I I I I I I I onH mzo uJHmmzm H ammexmo u uHH¢o Hzxummo qummHz Lo mumzzz me u :HHz eHmmmIH u u¢4:Hzoquzoqu mHo H<43mH¢a.Hz¢ummc qummHt kc «umzaz .zoouucm m: mmmuuo H :HH: Imnwn. n um<2cm qu 1<¢ czcouumu uc umcomc H zHHz c u m¢«:cm qu ouHuuzzcu H.cHH IIIm H.mn 4H¢o HaozHHz. om H an H mH H .N . H m.Hm H ¢.H~ H H a.mH H H.0H H m.mH H H.mm H ¢.en H H4 mI¢u>H¢o :HH: moH H HmH H mm H .H H.N H.H . H Hum HoH 4 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I He HHHHHm<4H<>< uuzuuHJ w.¢->Hmo mHa H<...:H.HH¢o H=¢mw :hHa mcmHNIm m.pw «.ms an H mom I.HH H H.Hn n.H¢ H a.mm «.mm H m.¢~ an “ onH a.¢H H o.o~ a.mn H “.mn ~.Hm H m.mp cm w an e.m H mom H.n~ H n.HH m.~c H H.pm 1H w cu .m . H.H u; H4 axqm ozm az ’I..ICIOI I I u I I I I I no H. pmnmmmmqh—n—IHHH—thu l :hqo ZSHHHmc H24rc 102 h cicadum; Ho mHHzéma oorqH new ION Aqhsh 43H IIIIIIIIIII >w P a J 3 L a h m m D & H.ww\ u,.—4 w n monhq>xwmmo oszmH: mo «umzaz w zhHa «HwomIH u uzHao Hzczo NIH 103 In u mstHH>Hummo qummH=.Ho mumzal ezoauHLu LG mmuaaua m Ith mcxmogw H szzum H10 fidm comma Homw womh AdhCh an“ H NH- H {AHA H 2131—00 H mvh H nonH H H m.H~ H ..mH H . Hon H nINn H ccmfl H hzuoahm en H cm H cm H .0 H H.n . H Mom H . H H.HH H a.H H won H comm H c.nm H ZdzmhuH¢o HZIIQ mum um I I I I I I I I I I I I I I I I .zsauuxm Ho mmuxuua n thH: hmzmm.¢H u wzaaan HIU any song ¢.H~ a.mh aqpop me ma «Hm zzndou m.--“"0.d-'a-".'H- H H. H v.¢H H . H 9.5 H H.0H H H.6H H m.mH H «.5: H mougdeu H: H a H Hc H .v H nomH H a.mH H H 50m H momm a N.Hn H a.mn H :.uc H Hoczum I¢Hr Hm H cm H a; H .n R--'.O'-'-~--“'-‘.H- H ~.m H a.mH H H a.an H h.Hw H : m.¢w H mus» H h.mo H kpaHauzmquH as . H on H m« H .m H .3: H H. ..H H H m.sH H comm H auHm H h.cH H n.n= H nmua ¢o auwa :u H «H H as H .H H.N H.H H Hua HcH 4H Hggou I I I I I I I I I I I.I I I.I AU>H4 onpHmo H2¢rm K) I I 105 :hHJ manom- .zocmLHH Ho mHHHoHo H u HH<3Hm qu 24¢ H.HHH ¢.H~ a.mH HHHoH «mm . cm H NHN H zxaHou H H.HH. H ~.Hn H H H.o¢ H 0.“: H H.m¢ H H.o~ H H.¢H H HHqu HHHnHA nNH H mm H Hm H .n H m.n H .nH H. H H.H~ H a.HH H a.mH H o.a~ H H.9H H . mmmNH.Hoom Ha H oH H am H .w . H H.HH H ¢.u~ H H H.H¢ H ~.wn H H.an H m.H~ H H.HH H x>aHz HHHmv mHH H mm H n; H .H H.~ H.H H Hug HcH HHHoH H4 HHH HuH Hag er H44» 52¢ asz HmHH Hum HHH . H ngcu m> I I I I I 2».>~st war» I IIII IIIIIIIIIIIII )- It 2 Cull « H a a a H a m o H u .. H waanHHr u HHaa stH “HHHHHxHHnuac muHHmHa «Hag. Hmuaa HHHH HHuHaazu ¢u>HHo Hzaza 106 . H M ”ZOHH<>zuwcc ozsauqxu Ho mHHzcua.m Ih—J thwmoNH 3.0HH m.~w a.mp a.“ cm . HHN M ooh mob H H m.hm H cch H a.mu m wowc H HIHm H m: H mm H MN H H «Hm H mohd H m mama H NI¢N H honw H Homw H momh a ma m DA “ an M H Nona H Hohc H H a.am H m.ec H nouu m sown H cohh a hhH m o: H hnH H How Hon H Adhéb m4 mam 3o; H£ 92h Hw«4 mh=MDH004 kc yum a J D t q p m n a 1 U I I I I I I I I I H Nu\mH\mo HHquchhmuae muHhmHmHHUQmazu mu>Hzo H2¢mwmo ngmmH: no xumxaz zHH: ¢¢~o~.~ u Hz H anew 241H uzoz H 2H H m H .n H ~.mH H m.pn H H m.Hm H ..H: H . w H.¢ H ¢.H~ H uHhoaocon .n H w: H .N H 0.5 H o.v H H n.wn H m.u H - H u.mH H m.ct H coon Zqu omHJ H om H H» H .H - H.~ H.H H wu Ho» 44 HHH Hum goo Hz IIIIIIIIIIIIIIIIIII .m.m.>4;prcz.uzcu2H wHa IIIIIIIIIIIIIIIIII HHm: u mpaa zmeauzu. Hmmzc “AH; HHH4/49HHmuac muHhmHumpuumrzu mu>Hmo HzH H 5.0H H mooH H =.=m H coon H meA mHuv mm H HH H HH H oH H.~ HoH H hum hop HHHQH H Hum 400 33m H hum 30: H hzzau H> IIIIIIIIIIIIIIIIII.IIIIIIIII. ._ mm: H: JDnQHmmoxu III-IIIIIIIIIIIIII H masmenu u uhH¢o Hz«zo H:u 109 .zocuwmu mo mmuxeuu N :th nmcmHom u ux I H u a a a a a u u ¢ s a a g u ; a u a a a a a a u a a a u a HH HHHHHonHHz «a H a 4 3 1 ¢ p m w o H u a n a I v a u a c a ¢ a a H c a H a H mmmexna u whon Hz~t3 NZQHHO fl. {my 110 .xocumzu mo mmuzcua c c.ooH «H m ‘ n.n~ on oocH mH Ho¢H nH n.nw Jabs» 3cm uu.cc«aaaauhuaaaaa m . 4 4 D n < H m w c g u n aza-<>zummc :hH:.omwmmo: qummH=.uo Human: u uz<=cm qu me «.me e.om 4H¢a .Hoxa H mH H NH H .m H ¢.m . m.c H H m.aH H ¢.¢H H H a.mn H H.Hc H quosz H H H HH H .m H‘--'----H--‘--'-'~- o H a.» H H.n H H H.¢H H w.o H H «.mo H a.mn H z C fi C a C I i £ a C 9 l onHHHsuuo :c unauacnu. a a u ¢ a u c H a H «xmexna u ubao zmeqmzuH hmuac .mHHL HHHH?¢OHhmu:: mqumqup04¢4zu mu>H¢5 HZHIw mum 111 H u monh<>¢wmmo mszmHt.mo aunts: cicauwxu HO mamawuc ¢ IPHS Nomoho fl um<33w H20 34¢ coaOH Goat Non thDh mNH H no H mm H ZZDJQU H HocH H coNH H H o.c~ H w.e~ H moww .H fleet H «.60 H U29? mm ‘H nH H OH H on Ml--l|-'l~-'-3‘--- H- H mom H aoh H u nohH H NomH H nocH H comm H @050 H mouddou HN H HH H 6H H o¢ H Moan H «cud H H 00cm. H homH H Noam u 0.6m H cocm H Jfiozum IQHI cm H nH H mH H on H HoOH H coNH H H moow H No¢N H noNN H mace H New H UPdHoutzuth mm H nH H CH H 0N . H Hood H mo: H H moaN H thH H . monH H Nccm H mam: H >¢Oh<¢uH onHquaa. ma « H a a < H a m o H u . . . . . . . . . . . . . . . . . . . HHHHHHHQ u HHHQ onHauau. HmHaa uHHu HHHHHzaHHmHaz muHHmHaHHuqx«zo ¢u>H¢a Hzazo ¢-o 112 ¢Hsouu¢m m: muuxuuo N ooaoH onH a a i h a u a u a i a H < H 3 m 4 H m m o “.0 H ma\n :HHa momma-m ~.mc m.om 4 mmH¢a cu: ICC-CICCCCCICICCCI H\no u ub¢o onbH¢o HznHz cooan Hm H mm H mm H .n H H.nH H m.HH H H a.mm H ~.mm H o.¢~ H H.nn H m.o¢ H omme-Hoo¢ an H HH H mH H .m H ~.m H m.cH H H a.mH H moxm H a.mm H h.mn H n.Ho H x>IHx oocmv Hn w NH H mH H .H H.“ HoH H Hum HQH HquH H boa Hou acH H Hum 30¢ H Hznou H> a.uvc‘uab-c...anuuuuua«uacfiuauu a ><= ¢ zu>ch mzuHH:CHHx HHa «43¢.«hmmczu cncaaaac.uac¢¢«uu. H mmmexmo u HHH¢o Hszo oJH 114 .zcauuzm mu muuguua m :HH: mcHaH.e u Hz<=cm qu nag ccauH Nome mIcm HIHCH an H In H on H 212400 H ace H moo H H e.m H oInH H o.HH H cone H e cm H mac: mo onH mH H In :H I: HI on H Io. «HI HI ;H.H HI H H.m~ H «.mH H m.HN H n. mm H mocn H . uzo mm H oH H cH IHI .N H a on H «Ion IHI m m.oo * m HH H No?» none Hocn H wzoz om H mm H H: HI .H HIw HIH IH IHoa HoH H I I I I I I I.I I I I I I I I I I I I I I I I I I I I I I L maug» cIH Hm<4 mHzHaHuuc Ho mumzaz «Ha < 4 D m < H m w c a u I I I I I I I I I I I I I I I I I I H mwanxnc u uHHzo Hz¢wmmO Ith homnmon come {can 4JIHZOZINSOUZH I I I I I I I I I I I I H\n¢ u “#43 ZGthuxuo ch I I I I I I mHa I I I I I I hmuac maHm UHHaronHmuae muHHmquHu«zHao HzIIa wHV (%9 can": DRIVER CHARACTERISTICS uucsrrcxvaxpr ) : 03/15/52 (CREATION DATE GUEST I I I 019 IIIIIII FILE l->-I 'J- I! C .l I I: I a. i < I h- I (I) I 6‘) I C F 0 LIBILITY I I I I .98 Iii. * 116 .4 mm 3% N o 30— HIP- I33 m .— HHH IflONO‘HHHH—O O"! 0| NI INVVHI 0450' IIOOOQI o ‘0 “*6 C I I I O I-II-I 0| I-IC I I T 0‘ I 0’ I m > I “HHHO‘H HHHHHHPH I-l-r- O I-I I—UUUI 0.! 2.11;; I H D 6340-! 00:20! 10h! UITH DRIVER'S 019 cmnmI Inc‘s: 47.3 Q I O] a-. I-oI-IHI-n-II-IHI-IH 5 63.3 2 UITHOUT DRIVER 8 Id IO?! M 7.6 3.9 -I¢--&---1------ 63 48.8 66 51.2 COLUMN TOTAL P‘H NUMBER OF MISSING OBSERVATIONS 117 .zacwmmm a: mum...........c : .. rbnr :Namm.: u mats—am H15 3<¢ o.a—H ‘ u.nn m.on r.nr 44p¢h nHm a. :9 cm ; 2234:: .. :.H H H.Hm .H :.:H H . . m.nr w u.mm M m.c% H 0.36 H mslmN H? H mm H mm H o~ H .m H H.H. H m.m H c.n. m n ”.mn H H.HH H 5.x: : . H.3H w incm H anew w H.HH H - mmumH HH . 19 H Hm H :m y .« ”"--"--F-'-'-"'M"'--'--C' m a.» H ~.m H .m. H H H.m~ .H H.HH H H.c. u :.=H H c.~: H m.mu H H.c~ H m:. HHuv mt m cm h H 4 dd » .H H.n H.m H.H H Ham HJH quq h. u. PHI—nu ‘3'— :cm H boa :cw ..... haze: d> .Q$.O.‘10‘9‘OVO16,.99V.996§QQOOQ$. . mew He 10§It$¥$$IVfi$+ . t t t t v ? t $ t >hH4aroabq 1a a _ NC t§§sQI$ "'fi Mung-JD tfiiiitfi... I H mr\m«\mo fl wbmummo wszmHz ma ammzaz muumouo ¢ :th $05.... u u¢Hmo mama a oahm H Hm H aw H mm H .N H H.m H a.mH H H.~H H H H.m~ H H.9H H w.mm H H “.ma H H.Hn H ~.un H mmuxHo 444 H u H mm H mm H .H H.“ H.m H.H H hum h.uh H bug 409 w.m.¢.n H was 30¢ H. H2309 H> «IIIIIIaccacct¢i¢a¢a.—iat 29HhmHmHo oszmH; Ho Humzaz muumwua c thH= hmflcw.mH n um ..HHHHHHHHquHHtcaH-«itHHaHiaaac 4m>u4 onhH¢o H241m en: 121 . N n wZHHthHHmmmHHG min—.3 Hz ..H—J mucraz .tocmwau ..HQ mmmwlén. .H 1.75. «NammJH n ma. 42cm :5 2H&c ea: Pfi.Jq.—»45.WWZ¢C s‘$wot$¢$ooovv9099 Huc\mH\nc u mspcc onhcmmcH Hmmzc mam..— HHHHHHHHHHHlome. munbmmwmhuqvqro )wsHHHc H.339 mu: 122 $.03me mm uc muurcw: : :sz wmnms.~ u utfiacm MIC 34: o.HHH d.nn m.om :.«r thc» gnu - ”w H :m 4 Ho s 21:4:u sI'IIIIII .Illl'nl'l. .III‘I'II r'. H a.ma H ~.od H a.mfi H H a.s: H 5.:m H m.c: m . d.n+ M M.Mw H :.mn H .m.m. n 1>|Hv acorda ma H Hm H mm H Hm H .H H'"---'-'H'-o.u-'-'.H'-"-"-..-' h c.~ H H.r H n.~ w H «.mm H a.mm H h.~m r m.mm m Hum» H :.mH H m.HH m memH-Hch To. a. va H K“ .r KM .. 0N H--------H--------Huu------Hu H ~.HH H H.m H H.¢. w H H.H: H H.HH H H.cu r . e.:H H m.=m H =.m. H m.=. w z>-H: ===av no H r» H NH H 0“ H .H w--------H--------H--------Hs------- HHc n.w HoN H.“ w #91 50h 41......P .H. #3.... .-...Q 2:» M be: :cm H 2.50,. H> ‘§.‘§4.¢.§$.¢?¢.QI‘I§-$IQ§$4$." \ >40 4 twawno m1MHH304Hy HHC $H.mc :Hazw on: 123 .rcauu.vu ma n. can. ca M c. o muuxmu: : rt; -.r.m m.cw .H. :c 40-"-".-N---'-"' H m.:a H $.m w n.- » ¢.ma H m.rm H ~.op H mm m m H o.o H 2.0 H m.mm H s.mm H . 2.8 H .1.... A. ..M. .4 H"'-'-'-H-----'-' « n.9a H c.1a a :.em H ¢.e¢ H ~.Hn H e.«% H H: H on H'---‘ -'-H"---'-- H.& H.~ ~.A$¢$tt§$+fi§os$tfi :- .Io- nt 22h quJ wk. :« 4 2 m < b. m m c M o . ¢ q . n:\mu\nc qudeOHhvm3? 'H «Humewoma n umEZEJ. H.,... :5... 3.3. 43:; .. .Hm - :35..de * m.: w . a.mH m - H m.:m » macs «c :2... m a. M. .p H 0.4 m w «.mH n .‘ o 2 u.: w- m. M., ':-'-'-P' . H :. No H .. “1: m. H m.hr H wzoz u N: n .H H........»........ «do H.H H #01 3» w hum cu r p21 16¢ . .552. as. v o t .Y i # t $ $ $ 3 .v 1 0. L04 v.9 ~.....:t:z NH: v 9 .9 .Q t b. .9 1 tr 0 Q Q ..v t 0 q:— 6. Ha. zuuhfiuzu: .pvm: 3 man“. mgnhmzfinpofiuqru unwynwé Hzefio 124 HH n .nzo.~...«>~.w.n:¢ m.:Hmmwz mu zumrzz .chmma... mo muuvéwc a :HH: ranmwco u uni...“ :5 2:... ago." m...4.H o.Nm. ....m.. .757..— nmd av me am 3.5.7.7. H--quu--H--Mumr--H--MHm---H- H ...mm m c...m H 6;...” .n. ‘ m.mu I H203 H. a.c~ H oi; ... mic? ...=.. CI» om w Hm H .H H m H .m .o--"0--‘P’-""-'H"I¢-""F- H H..,... 4 H.,.n H N... H H 3.3N H e.h~ H 7.3m H m.H~ H cnmm H anH m HJQ~ H - Hzo m: H 5 H .2 u; H .o .PI.I-"-ul--H'Il---”ul--Hu--lv-"-I'm.I! H HJH H Hg. H 3.3 H . a. in”... H {g H 0.5.. 0.3.. w mfmm H s.m«. H. 5.7m .u. 9.92 n... H cm H Mm H mm H .H H"'--'I.-H.'-'I---I-H"'I.nl---H'Iu..-"lulv ”do H.» H.“ H... H ban. bob. 41.2... w .57. .8... re» . H boa 3c: M H.239 . H) s .v 6. 6. O $~ ‘ . O. 0 O. a. ‘ O O. ‘ ‘ Q 0 . .‘. .9 $ ‘ ‘ Q .0 $ ‘ § ‘ a 02:32: ”.0sz mhtwawcud no "Mia-Z“ mac q.amm2c cszmHz mu aunts: .roou man me muwzcwc 4 :HH? HNHHH.m u Ha<:cm Hzc 24m o.aad a.mm No~m :.u .4Q53F HHH no Hm Hm 22:.29 H--J-----H--------H--------H- c..m. H .. .. - w w.¢H H H. MH H H. m.H H . H.+. m m.¢m H c.4n w u.c~ w gags 2azH mac: on w an H m H s H .m ""-"-'-H-"I-H"H'.II-"'II.¢P»- . «....n 4.. 3...: H ...m: H. H c.Hm H H.ne H 9.2w H N.m.. .,. o.e: .4 rorm w m}: n... caccloccm H... .H. Mm « Gr w. an. ... .N H n.n4 H :.o. H :. NH H - H m.n. H ¢.=m H o. H ..H. wnsu H e.a: h H.H. r maom zHIH mmug H .H. H mm H cm H“ .H F-,-I"--- H'---"-'M""-",...I."Iu-"lu- ¢H° H.” How 5.. w hcu bch HapaH p Hon Hcv zc~ .. #00 33... H .352: H: $0$t$t§$t§¢ttv++ M¢¥+§¢§$V$t .u .. .. .233. ...:AZH 3c <4:m1.~mm¢ao sooto>oo.Q‘&to$¢-H1Q$ H m...\mH\rH. u mug-(P. 2C.».-......nfl.: FEE-c ..H—Hm ”......HqTHHCHbmmfc mgHHWHHCbUHnHEfi. ..wHHuc Hakim. mu: 126 .roommmm mo fi$§ d J : n.aafl mzm o.ma NH m u mchHHHHmyzc eszmH: ma ammzaz mmmrcm: m szx omamm.mm n wrascm at: :Hzo bactbhz H a. H H H u H .N I"--"-'*-'----"sc"'---"O.' H :.H~ H ~.HH H H.H~ n w m.$o H :.mm m 4.N0 H . H HHHN H H.me H H.mH H H4 m.am>H1c :HH: H ....u d . ~® H v0 .... ..H H--------H--------H--------H-------- HHc H.H H.~ H.H H Hog HaH 5 ham aim H hug :02 a m Haze: H: 90?HO$QQ¢.I$’9‘..Wl.l$t§?9$fi HHHHHHHHHHHH uuzmaHH H.«H>HH= HHc mama—U Ott9§00i$btfififitHQ$Q . mc\mu\nc " Mud: ZCHHch wzH¢a uz mm>b ZOHbummmmh‘H hZH coach QZHAJ «#100 QNGK >m mud G . m o z o H h 4 a a o < h m m a a c Q o t 0 o 9 Q o t t o s o o t o n t . Nomesna a HHHo onHHHao. Hmuaa .HHHH. HeHszOHHmHau moHHmHHHHHH¢H¢o Hz HHHH onHonaHHzH HzH . ..mou HzHHHoaHzoo H~a¢ HH , HHHHHonHHH ~a a ~o\m«\ne. a whqo ZoHpH¢o Hzczu 130 eoeed moo ~.o mam. ecu» coo: J m:d Md 0 m: «h z ------uH--------H--muuc--Huu-u----H--------Hu m H. H H H. H H .H m H.H H H.m H :.H~ ”umH H mumH m H.m~ Hnmm H Mum“ m zamHHHH uHaoH: HH m H H N w W H m HH H .H . Hum. . E. . He... H ... .... . ..H H H.HH w H.H M H.H m H.m m: H .Hm H Ha=¢m. HHH H m H H H H» H .H HHHHH HH - . HHH m wma wow sou Hz» HHHH ozH «Hum HzpzaaH za quHa H Hog sou H Hzaoo H~ou titt9t000.......03....QtittttbeOOz... uaaHH HHHH chHommcmHzH H zH on uznaa mbzoo owae >o >bHa2 H ..uom quaamahzoo a~d¢ >0 . onhdaaooo o z c H h c 4 a a c b m m o a c t t o o o t t o I w 0 t t o t t o t .. HHHHHxne u HHHH onHHHao. Hausa HHHH uxHszoHHmmac moHHchuHonHHo «HHHHH HszH mum 132 . HH _H.H H.H :. H.~H HHHHH HHm zzHHH HH H HH H m H. H H H.» WH H.H H. H. H H.Hn H .HHH H H mm ..H..HH .HHm H Ham H . m.m H H.H w .H.H H H H.Hm w HHHHH H H m H H H Hm .H . H.H H .H ".H .:H H H. HN H w H.,: H ... .... mm ...mn ...HHH H H. H.. H H.HH H.H H H. H H.HH H HHH :HHHH HHHHH HH m H H H H m H mm :H .H HHHHH HH HHH :m ”Hm +mw HHH HHH HHHH Ham «Hum HszHHH 2H HzHHH m HwnHuHH HHHH t..‘.....t.0..i...’.... ...Q... HHHHH HHHH zHHHHHmmuwzH HzH . .HHH HzHHHHHHzHH HHH HH zHHHHHHHHH HH 9 no 20Hhcasocbmmc¢o HH...HQ¢H.HHQHHHH.0.H H mo\ma\no u wh¢o 26Hh¢m¢oo Hausa .mAH... wdH¢ZZOHhmH=u muHhchmhodmazo ¢m>Hco Hz uu>h chhowm¢um¢m0m quawuwrzoo. 0N0¢ > Jm>ma 20HhHH onHaoaoH mu m o 2 OH H c A a o a b m m o a u H H H H H H H H H H H H H H H H H H H NQ\MH3ne u whao 20Hh¢wmne hmmau .uaHm wcHazonhmmac mlomHuwHocquo ¢u>H¢a H2439 k. NIH 135 H.HHH H.HH a.mH m.H ~.HH HHHHH HHN HH HH HHH H HHHHHH w H.~H H.H lm H.~H H.H~ H H H.HH H H.HH H H H.HH H H.HH H ..HH H H.~H H H H.HH H HHH w HH HH mm m HH w .H H H.H H H.H H.~ H H.H .H H ..H H .HHH H HHHH H HHHH H mm H H H H. HHH... H NH H ..H H HHHH HH H H.H~ w H.H w m.Hm H ;m HWH wmc H HH H H H. H H.H H .H HHHHH HH HHH w Hon Hmo :HH me HHHH HHH «Hum HszHHH zH HHHHH H HoH :HH H HzHHo HHHH t O O O O I t O O b O O t I I Q 0 O i D Q 0 t 0 0 OZ HHHHH HHHH .zHHHommmHHzH HHHH HH HHHH zHHHHH HHHHHHHHHH «50H: aeonHA @@ONHIHoao «>3H: oaamv «Ha Hack uZHJHMMHzoo m o z o H H c 4 a o c H m m G a o H H H H H H H H H H H H H H H H H H . H No\mH\no a wH¢a 29Hhcm¢oH Hmuac NHHm w¢H¢zonhmu=o mlochmh041¢zo ¢u>H¢c Hzuco uquu HE 137 H.HHH n. «H o.oH «.o o.m H H.Hm HH onHowmmquH HzH ..aom quHHoaHzoo =~a¢ Ho . mmcm> ozH Hm<4 mHzmaHou< mu uuozsz ~Ha . u a. ,. 6 av.— t .d a n. 4 >.m" g: 0 .t s H. Q .H_- t .v o .H_. o H. o .H o .H o H ~cmesne . upH¢o Hz¢au. HHIH 138 a. ed H.HH ”mew won Hom coo m.¢~ cgqm 4 ma>b 20HmommmmHzH HzH ..mom quawombzoo Nwa >0 mm o H hm.<4 mHzMOHoo< no «mozaz «a H H «.4 2 o d Hm mo 0 t t o t . Q O Q I t t 0 t t o o o o _ H ~HHHHHHH u mHHa onH¢u¢oH Hausa .HHHH uxHszoHHmuaa moHHchuHoHcH¢a Hz.:u NHnH [—13 139 m a.aed cgeu . m.nd Hom Mom h.dm . 4h 20Hhowm¢uH2H HZH ..«ou quHHomhzoo amdu >9 . ocomu>azhzozuwzoqu oHa o z o H b c 4 a n 4 H m m o a a t 0 0 o o o t o I I o 0 o 0 o t o o . . ~nmexna a “pH: zeHHH¢a H2119. mHIH VHIH 140 I . I II||.I .. H.HHH H.H o.H m.o N. mu a. can HHHoH HHH HH on zzaHoo m ...... M NH H H.HH H H. H H H.HH H H.HH H H. mH H m.HH H m.=~ H H. .mm H cage z _ m¢>H onHommmquH HzH , .aom quHHoaHzoo HHaa >o .m.» .HHtzos.m:oozH 9H: m c z c H,H a A a m c H m m o a o .0 t t t 5 t o s w t t t t I o t s o H ~o\md\ne n mbHao Hz~¢c ~24Iu HIx cc u mzo~h<>¢ummo OZummuz mo cumtaz .xocuuzm mo wu¢9uo a rpm: mmnnr.n u U¢0 . >b~4H¢H Hsze «Ix 145 H u mchh<>¢ummo aszmHz mo xumzzz ozoouumm mo ummouo H :hH: ¢mohm.m u u¢0 m¢ oak hm<4 mhzuoHuu« no mumzaz «Ha h < A D m < b m O z u a ¢ ¢ a a c c u ‘ . ; ¢ I c ¢ c § c H mcmexnc u HHH¢o Hz¢quo oszmH: mo munzaz .zoouqu Ho muumwuc m :HHz honm2.w u u¢arhzcz.u;:qu mHo H¢o HZuumgc cszmHz no xmmzzz H .2ouwmx: mg uuaaug H LHHz mnmmH.¢ u Hxm onhaanuuo :c hdJDxchmmoxu ¢¢c¢oo¢Ha£¢aaaCHHii H unxmuxmm u whqa onpH¢o Hz¢umao oszmH: mo ¢uztaz .zocuuxu mu uu¢muo H rpm: mmnm¢.u u H¢«Dow HIU 34¢ ... .zcomu¢m kc uu¢$u5 H 2H»: hnhmo.¢ u u¢<33m qu awhqm¢¢ou u.ooH m.nm ¢.mh 41 _ Ju>u4 zuHhauaou me P 4 H 3 fl 4 H m w 0 x u a u a c a O'CCOC§OC H ma\mH\nm n whqo chH¢H¢uH hwubc uaHu m¢Hazonhwunc mqumHuuhuqzazu ¢u>H¢o Hz¢Hmmo qummH: ma mumzaz .zoauuzm.uo mucouo H zHHz euamH.n u HH<=am qu zHa .zcouuau mo mucous H :HH: HHNHH.~ u u¢<=am qu oHHuuzxau H.HHH m.oH H.Hm HHHoH How H me H «Hm H zzagou H a. H a.mn H , H H.HH H a.mc H H.HH H H.HH H m.mm H Hoczum :on "A HHH w mH w Hm w .n . .H H.HH H H.HH H n. m H «umm H ”new H Hoczum :HH: v meH H mm H HHH m .H H.» He H Hum HOH HUH.chHH¢o Hz¢ummo @ZHmmHz mo ¢umz=2 .xoouu¢u mo uu¢ouo H1HH3.ioco~ c n u¢m m¢ 03h hm<4 mbzucHuu< mo ¢umzaz NHc h < A a m < h m m 6 ¢ 0 I I I I I I I I I I I I I I I I I I H ~¢\m H\no u uhH¢o HzIHt coonHA QNH H mm H Hm H In H Hom H oIcH H H ncnw H mohH H . HIGH H a.mw H Honk H @mmNHIHcom mm H ¢H H on H IN H coo H moan H H nocH H «Ian H mocn H onH * noon w ¢>IHz ecomv ¢m H H nc oH HI: an H hum ho» AH¢a m¢uhutOJH¥ HHc H J 2 m < h m m O x u I I I I I I I I I I I I I I I I I I H Namexno u uhH¢a HznHz ooanHA mm I Hm H Hm H .n H--------H--------H- I h.m H m.HH H H m.a~ H m.mH H u.hH H m.nn H h.ww H mmmmHnHuom an H cH H cm H .m M"'-"-'H'-'l'."'~- . p moc H mom“ H w 6 ed H Foam H «.ow p m.mH H m.wa H m>nHz ocumv um H h H ma H .H H--------H--------H-------- Hag Hoe .Hu H boa HaH JQHc» um H H9; 409 .9Iu Hzm uon H Ho; 30: H H4339 cwcm I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I H >43 q zu>Hao mzHHHzQJHx . HH: 4 4 D 3 < h w m Q m U I I I I I I I I I I I I I I I I I I H maxmmxna u quo onHIHmu. menc uaHu mquzonHmuao muHHmHumHuH¢o HZ Luv-0° ¢nwa (#10 DES (no. .NC (MIC * O Llnfi 11“!!! Non. ommontom HHthIQ$H ¢OHOFIOhm Hb2 thmm k m xczzm OHM m UJQ¢H¢<> hum flINm >bHJHQ no umwou HHrmromnH .ZOHh¢H>UC Ohm amonmoannmm om J no mHm>J‘z< ecmHho a UJthJzt HcHa Ion cwctnz amHm zo oucUH7u HmHqu 156 Inc. ”we. ;. .. . . . . Hauunwmmctcno noonmo ocNHHo nzHa wwwMNnc h QHQNO HnH :ncem mac H>2HHUHHm<4u uuz (Hum u m aomxu OHM m wam u--------uu anH ----ua---u -----.u---n.uuuuu-uuuu zaHh‘3cu um» :— mu4m¢H¢¢> -----uu------uuu-uuuu Hum on >HH4HmflH¢<> no unwou HIHHmoan onh« :H mu Ohm HooQIInonH cosugoHchoh co 04<30Hmu¢ oHHIH. . :4 cm a mwpmsfio« ccm omwoooH omHHooHchmn onHHooHchn .H v.0 Hmm.uacu¢ mHHHn. u¢¢aom x wuz< uHquuHm m uxqacw znu: mu¢ no mHm>4 neao¢.mcH o>wo oobm annoommN umzommuc z quozuauo 00...... c uu mmmm O z o.H w m u c a u x wmanx¢o uJQHhJDzIIIIIIIIIIIIIIIIIIIIIII uzczcz H Nc\nH\co n ubdo 7oHHHao Hz<1o 157 uuz luck KVU‘ 0.0.0.... .nnIHNInH. 44Hb¢ 44¢ mmm. mIonnmo. N Nthmomcm IchN.m~o HHz «Hum u m mocxu OHM a u4o IU'I-"-I| I--'----"-'--'-"---I wuo—P‘DOU “IF 2“ WUJO‘HU‘> I"|I|"-"IIII--II-"'| Hun coon >HH4HQ no muuou mo owaponnH n onHwo arm .5an $33.32.. .N 3323.. a 1. cm ._ 3523‘ 3:8 imaging... .r. 22323.. 93%. .333 a m 231 3593 no :3 no 32:5, .3 3332‘ 1.32.. I 515:: nrHc Ion «wars? muHm 2o ouaquu HmHu4m UH quozuuuo I I I I o H m m u x 9 u z w J a H H J 3 t I I I I I I I I I I I I I I I I I I I I I I I aon> mman H mmanxco u upao roHH¢u¢u. uzIzoz mnuqHu 2ch mu..cua ~m\nH\co umHH¢o H7 .¢Io 158 “31106 3'an «mom 0 O O uuzduHquon 04" (1‘00! NG‘C‘ ‘LG’N .1.-mo I I I II!!! n 44<¢u>o C C nHo mNHmII cmHhoI mcmNm. thmhI CHO. nHuono ncHo n Inasmol IncwH. mueom. aowHHo no». nvuoHoH Hch m mormmcn mhthI mthno ammon- Icm. cowoooH wao H uuZIIu u>ozu¢ co cwhzu ow>oxu¢ ouzwpzu a UJthm ux¢scm x u¢ auHm u A o 4 H b x < x x 3 m quz .ou4m quozuauo I I I I I I I z c H w m u x o u r u 4 a H H J 2 z I I I I I I I I I I I I I I I I I I I I I I I H menH\co u whdo roHH mmam umanx¢c u¢HH¢o szru