INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy subm itted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UM I a com plete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize m aterials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UM I directly to order. UMI University Microfilms International A Bell & Howell Information C o m p a n y 3 0 0 North Z e e b R oad . Ann Arbor. Ml 4 8 1 0 6 - 1 3 4 6 U S A 3 13 /7 6 1 -4 7 0 0 80 0 /5 2 1 -0 6 0 0 Order Num ber 9223224 Evaluation and rating of driver education programs in M ichigan based on driver records of accidents and convictions Mallick, Sayeedur Rahman, Ph.D. Michigan State University, 1992 UMI 300 N. Zeeb Rd. Ann Arbor, MI 48106 EVALUATION AND EATING OF DRIVER EDUCATION PROGRAMS IN MICHIGAN BASED ON DRIVER RECORDS OF ACCIDENTS AND CONVICTIONS BY SAYEEDUR RAHMAN MALLICK A Dissertation Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Civil and Environmental Engineering 1991 ABSTRACT EVALUATION AND RATING OF DRIVER EDUCATION PROGRAMS IN MICHIGAN BASED ON DRIVER RECORDS OF ACCIDENTS AND CONVICTIONS BY SAYEEDUR RAHMAN MALLICK The primary objective of this study is to evaluate driver education programs based on convictions and, specific accident criteria such as type of accident and accidents under different weather and light conditions. There are two major reasons for undertaking this study. First, there has been a growing concern about the effectiveness of such driver education programs. Second, prior research on the effectiveness of various driver education programs has not been conclusive. In general, prior research studies have lacked a valid measure of a ccid en t exposure. In the present work, an indirect accident exposure measure, in addition to traditional methods (ie, accident and conviction frequency per student for each school and program) was utilized in the analysis of the effectiveness of driver education programs. This indirect accident exposure method is based on the assumption that the accident exposure of any group of drivers is proportional to the innocent victim ( a driver who is not responsible in a multi-vehicle accident) involvements in multi-vehicle accident by that group of drivers. Various statistical techniques were utilized to test various hypotheses for comparing different criterion variables and to determine the relationship between types of program and performance variables. A general rating score was then determined for all schools and programs on the basis of both frequency and severity of accidents. Analyses indicated that students from competency program in commercial schools had significantly higher accident and conviction rates than the range, traditional, and competency programs in public schools. Whereas, no significant difference was found among all four programs when an indirect accident exposure measure was utilized in the analysis. Based on the rating score, developed on the combined criterion of accident frequency and severity, the range program was found to have the best performance. No significant relationship was found between types of programs and a set of their performance variables. Thus, as an overall conclusion there is no significant difference among public school based programs on various criterion variables; but the commercial school program had a significantly higher accident and conviction rate than public school programs. ACKNOWLEDGEMENTS I would like to express my great intellectual debt to my academic adviser Dr. William C. Taylor, for his guidance and advice during the course of my studies and this research, without which this research could not have been accomplished. I am also very grateful to Dr. Richard W. Lyles and Dr. Gerald L. Ockert for their valuable assistance and guidance throughout the completion of the project that constituted the basis of my research. My gratitude is also extended to Dr. Thomas Maleck and Dr. Cress for serving as members on the guidance committee and for reviewing this dissertation. I would like to express my appreciation to the Department of Civil and Environmental Engineering at Michigan State University for arranging financial support for my studies and this research. I wish to express my thanks to my sisters Sheema, Kauser and brothers Anis, Tanweer and Amir for providing their support and encouragement. My parents have provided continuous encouragement and moral support throughout my life and my appreciation for them cannot be expressed with words. And finally, I thank God for everything. TABLE OF CONTENTS Page LIST OF TABLES .............. ...................... vii LIST OF FIGURES .................................... ix CHAPTER 1. INTRODUCTION ........................... 1 CHAPTER 2. LITERATURE REVIEW ....................... 3 2.1 2.2 2.3 2.4 Characteristics of Young Drivers .... Young Drivers and Fatal accidents .... Driver Training and Education ....... Performance of Drivers Licensed with Driver Education Programs ........... 2.5 Conclusions ......................... CHAPTER 3. METHODOLOGY ............................ 3.1 3.2 3.3 3.4 3.5 3.6 3.7 CHAPTER 4. 3 5 6 9 17 23 Data Base ......................... Data Variables ...................... Induced Exposure Method ............. Hypotheses Testing .................. Development of a Scoring System ..... Model Development ................... Discriminant Analysis 23 25 26 30 32 34 36 DATA ANALYSIS........................... 38 4.1 Data Statistics ..................... 4.2 Accident and Conviction R a t e ........ 4.3 Quasi-Induced Exposure Method ....... 4.3.1 Weather Conditions ............ 4.3.2 Light Conditions .............. 4.3.3 Types of Accidents ............ 4.4 Result from Hypotheses Testing ...... 4.5 Rating S c o r e ........................ 4.6 Regression Models.................... 4.7 Discriminant Analysis ............... 38 40 51 61 63 65 68 84 90 104 CHAPTER 5. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS.. 5.1 S u mmary............................. 5.2 Conclusions ......................... 5.3 Recommendations ..................... REFERENCES ........................................ 108 108 110 115 116 APPENDICES ......................................... 121 Appendix A. A Fortran Program ................. 121 Appendix B. List of variables ................. 123 Appendix C. Values of different criterion variables for various programs ...... 125 Appendix D. Histograms of variables............. 145 Appendix E. ANOVAs tables 151 Appendix F. List of schools in ranking order.... 184 vi LIST OF TABLES Table 2.1 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 Page Number of crashes, by program types for different level of driver experience ......... 15 Accident rate for higher and lower ranked public schools using the range program ............... 43 Accident rate for higher and lower ranked public schools using the traditional program ......... 45 Accident rate for higher and lower ranked public schools using the competency (pub.) program 46 Accident rate for higher and lower ranked commer­ cial schools using the competency (comm.) program 46 Weighted average accident rate for various programs ...................................... 47 Weighted average single-vehicle accident rate for various programs .............................. 49 Weighted average conviction rate for various programs ...................................... 49 IR value for higher and lower ranked public schools using the range programs .............. 55 IR value for higher and lower ranked public schools using the traditional program ......... 56 IR value for higher and lower ranked public schools using the competency program .......... 56 IR value for higher and lower ranked commercial schools using the competency program .......... 57 Average relative involvement rato (IR) for various programs .............................. 58 IR value for various programs for different geographical areas ............................ 60 IR value for higher ranked schools under various programs for different weather conditions .... 62 vii IR value for lower ranked schools under various programs for different weather conditions .... 62 IR value for higher ranked schools under various programs for different light conditions ...... 64 IR value for lower ranked schools under various programs for different light conditions ...... 67 IR value for higher ranked schools under various programs for different types of accidents .... 67 IR value for lower ranked schools under various programs for different types of accidents .... 67 ANOVAs and "t" tests results ................. 69 Rating of schools by accident frequency and severity criterion ............................ 86 Rating of programs by accident frequency and severity criterion ............................ 86 Rating of schools by conviction frequency and seriousness criterion ........................ 88 Rating of programs by conviction frequency and seriousness criterion ......................... 88 Classification summary from discriminant analysis for year 1988 data ................... 106 Classification summary from discriminant analysis for year 1989 data ................... 106 vi i i LIST OF FIGURES Figure 4.1 4.1 Page Histogram of accident rate for the range program ..................................... 42 Histogram of IR values for the competency (pub.) program ............................... 52 ix CHAPTER 1. INTRODUCTION During the past half century, driver education programs have steadily become a standard curriculum in high school classrooms across the United States. From a shaky or unpromising start in a solitary school in Gilbert, Minnesota, in 1923, such programs have flourished to become the primary source of driver education today. Despite this growth and popularity, driver education programs have come under attack during recent years. Critics have charged that such programs are both inefficient and ineffective. They point to the fact that accident statistics among young drivers remain high despite the widespread use of driving education programs. Lastly, it is argued that the public is concerned about spending tax dollars on a program which has not been proven to be cost-effective. Not surprisingly, proponents of driver education programs advocate that public instruction is crucial since it reaches young people right when they attain legal licensing age, and thus, right when they are most highly motivated to learn. Further, the proponents note that it makes intuitive sense that driver education is helpful in reducing the number of accidents and injuries since such courses teach proper driving maneuvers and the rules of the road. Finally, they argue that the potential 2 consequences which result from a poorly trained driver being on the road are so serious that it is essential that society maintains driver education programs to ensure that quality instruction is provided. In light of this controversy, the appropriate objectives of this study are 1) to develop and calibrate a model for estimating the differences in the accident potential of various driver education programs, 2) to evaluate driver education programs based on accidents and convictions and on more specific accident criterion such as types of accident and accidents under various light and weather conditions and 3) to develop a scoring system to rate driver education programs and schools within each program based on the frequency and severity of accidents and convictions. In the next chapter, various articles about different types of driver education programs and their accident and conviction reduction performance on highways are reviewed. The methodology required to conduct this study is explained in chapter 3. Results of the data analyses are presented in chapter 4. The final chapter includes the summary and conclusion of this study. Recommendations made, based on this study, are also presented in the final chapter. CHAPTER 2. LITERATURE REVIEW The main purpose of a driver education program is to teach young drivers the minimum skill necessary to drive safely, with the ultimate objective of reducing the number of accidents and driving offenses. However, this objective is apparently not being acheived. Accident statistics for young drivers remain apallingly high despite the wide spread institution of driving education courses. The review of literature concentrated on the following areas of young drivers: (a) characteristics of young drivers, (b) young drivers and fatal accident records, (d) types of driver education programs, and (d) performance of drivers licensed with different types of driver education programs and without formal driver education. 2.1 Characteristics of Young Drivers Young drivers are most often characterized as inexperienced, aggressive, temperamental and competitive with greater self-confidence and higher risk taking behavior than older drivers. The mental and sensory abilities required for safe driving; such as visual acuity, visual field, night vision, perception and recognition power, and decision and reaction times are quite sound when compared to other age groups. 4 In spite of these physical advantages, the young drivers were found to have been over-involved in traffic accidents in many studies. Many research projects have been conducted to study the contribution of human factors in accidents. In one such study, McFarland found that youthfulness, temperament, and inexperience were major causes for higher accident frequencies of young drivers (1). In another study Beamish and Malfetti investigated psychological characteristics of young male drivers with and without violations using a variety of personality tests (2). The test result showed a statistically significant difference on variables - emotional stability, conformity, objectivity and mood, between the two different groups. In his study on minor driver performance, Ockert reported the finding of Gallagher and Moore who investigated the causes of accidents of young drivers by studying the relationship between a broad range of personal variables and accidents (3). The factors which were best predictors of accident history were the amount and quality of exposure. In addition to exposure, variables with a significant relationship to accident history were personal and emotional adjustment and dynamic personality traits. In summary, the higher accident and violation rates of young drivers appear to be the result of (i) lack of 5 experience at a new activity with a frequent lack cf awareness that certain actions may produce serious consequences, and (ii) psychological characteristics associated with a greater risk-taking behavior. 2.2: Young Drivers and Fatal Accidents Young drivers were also found to be over-involved in fatal accidents by many researchers. According to statistics provided in "Accident Facts" published by the National Safety Council for 1976, young drivers of age 15 to 24, have the highest fatality rate of 46.2 fatalities per 100,000 population (4). Another study based on data obtained from the Fatal Accident Reporting System (FARS) for the year 1978 reported that male drivers encounter the greatest rate of involvement in fatal accidents at the age of 18 (5). The greatest rate of involvement in fatal accidents by female drivers occurred at the age of 16. Based on data from the Fatal Accident File (FAF), as maintained by NHTSA, for the years 1973 to 1974, Wuerdeman et al. in their study titled "Drivers in Fatal Crashes With or Without Driver Training" indicated that of all the reported fatal accidents involving drivers through age 30, 51.4% had not completed a driver training program (6). Slightly fewer, 48.6% had completed driver training. According to Highway Statistics for the year 1985, publlished by the U.S. Department of Transportation, Federal Highway Administration, young drivers of age 15 6 through 24 were involved in approximately 46% of total fatal traffic accidents, even though they comprised only 18.36% of the total driver population nation-wide (7). This disproportinate trend has been consistent for several years. 2.3 Driver Training and Education The terms driver training and driver education refer to the task of producing drivers who are able to drive safely on the roadway. The training part refers to actual driving by the candidate - i.e. time spent in the automobile to get familiar with vehicle handling and to practice driving according to the traffic regulations that drivers have to obey. The education part refers to the the instruction that is given to the student driver out of the automobile, regarding traffic regulations, legal and moral responsibilities, as well as special theoretical maneuvers of the automobile. The objective of such instruction is to teach students the minimum skills necessary to drive and to enable students to make sound decisions under various driving circumstances. The ultimate purpose of all of this is, of course, to decrease the number of accidents and driving offences committed by young motorists. There are several methods of driver training and education in use, ranging from simple classroom teaching, to costly simulators. There are four types of driver education programs which are as follows (8): (1) the two-phase competency or traditional program, which involves classroom and behind the wheel (on­ street ) instruction; (2) the three-phase range program, which consists of a two-phase program with range training added; (3) the three-phase simulator program, which combines a two phase program with simulator training; and (4) the four-phase program, which combines classroom, behind-the-wheel, simulator, and range training. An explanation of various elements of these four driver education program may give further insight into these programs. A simulator, as the name suggests, imitates on-road driving. It is an immobile unit, closely resembling the inside of an automobile and is replete with safety belt, gear shift lever, steering wheel, gauges, speedometer etc. At the front of the unit is a screen which depicts various driving scenarios a person may encounter when on the road. The student sits in the unit and operates the simulator as if it were a moving vehicle. The instructor who also sits beside the student, can immediately correct the student's action. The adequacy of simulator training varies widely. In most commonly used simulators, there is no interaction between the actions of the student and the simulator, e.g. change of the scene on the screen acordingly to steering wheel or brake application input (9). More sophisticated simulators with interaction between the actions of the driver and the simulated road environment are available, but due to their very high cost, are rarely used in driver training courses. The theory behind the use of the simulator is that once a student begins to drive,they will be able to transfer the skills learned on the machine to the operation of a real car. The main advantage of simulators is that it allows students to be trained without the risk of an accident. On the other end of the spectrum is the "behind-thewheel" program which takes the student out onto the public roadways to put into practice those skills which have been taught in the classroom. Typically, the instructor sits next to the student and has access to a dual brake pedal. This allows the instructor to slow down and stop the vehicle when necessary. The behind-the-wheel program has been in use since the beginning of driving instruction, and is an integral part of most courses. Another driver education teaching tool is known as an off-street driving range. Usually driving ranges contain elements of the roadway system like intersections, curves, and merging lanes, painted street markings, signs, and curbs very similiar to those actually used in real life. A range consists of a large driving area constructed seperate and apart from any public roadway. This type of range training provides students with an opportunity to drive an automobile without being exposed to the dangers 9 inherent in on-road, in-traffic driving. 2.4 Performance of Drivers Licensed with Different Types of Driver Education Programs: The majority of the early studies on driver education programs have mainly focused on comparing different types of training. Comparison between trained and untrained drivers was not made in many studies. However, a comparison in performance between trained and untrained young drivers was made in the Dekalb study, which was a very comprehensive study conducted in Dekalb county schools, in Georgia (10). The primary objective of this study was to determine the crash reduction potential of a competency based driver training program known as Safe Performance Curriculum (SPC). The experiment design called for the random assignment of 18,000 volunteer high schools students in Dekalb county schools, Georgia, to one of the following* (1) Safe Performance Curriculum (SPC) - 70-hour course including classroom, simulation, range, and on-street training; (2) Pre-Driver Licensing (PDL) - a modified curriculum containing only the minimum training required to obtain a license; and (3) Control - no formal driving education in the secondary school. The student driving records were monitored for a period of 2 to 4 years to assess measures of intermediate and ultimate performance. Comparative analyses of SPC vs. 10 PDL vs. Control groups were then made in terms of crashes and violations. The results of this study showed that students who had completed the SPC or PDL driver education courses had 13% fewer accidents and 16% fewer violations during the first 6 months of driving than those students who had been placed in the control group. This difference was found to be statistically significant. However, this initial difference between these groups was marginal during the next year, and completly diminished over the two year observation period. These findings led the author to conclude that students receiving the SPC or the PDL instruction performed no better than students with no formal driver education. In addition, this study determined that there was no statistically significant difference in performance of those students who had received the lengthy SPC driver education instruction and those who had received PDL driver education instruction. In light of these facts, the researchers in the Dekalb county project concluded that the Safe Performance Curriculum was not an effective accident reduction countermeasure and there were no significant differences among the three experimental groups in either accident or violation rates over a two year observation period. Lund et al. reexamined the Dekalb study regarding the variables of licensure, crashes and violations (11). 11 By fitting a log-linear model to each variable, these researchers estimated the relative hazard (likelihood) of students becoming licensed, or having their first crash or violation, at each month following their sixteenth birthday. Conclusions reported were different than the original Dekalb report. The more recent study found that students assigned to the Safe Performance Curriculum program were at significantly greater risk of crashing and of receiving violations than were the comparable control group of students. This report indicated that even during the first six months of licensing eligibility, there was no evidence that the Safe Performance Curriculum or the Pre-Driver Licensing programs reduced the per capita likelihood of crashes or violations. There have been criticisms of the Dekalb study. First, while the students who participated in the project were randomly assigned, the initial group of 18,000 people consisted of only those individuals who had volunteered to be a part of the experiment. In his dissertation, Ockert reported that about 18 percent of the control group students had completed a formal commercial driver training program (3). Thus, their may have been an initial self selection bias and the students who participated may not have been an accurate cross section of high school students in Dekalb County, Georgia. Second, although the 18,000 students were originally divided evenly among three programs, a number of people dropped out of their assigned 12 program or did not go on to become licensed. Again this self-selection factor may have skewed the data. Third, it is possible that driver education graduates reduced their mobility as a result of a cautious approach to driving learned in the courses. Finally, there has been an opportunity to track these students's performance for only a few years. It may turn out that the SPC contains latent benefits which will not be fully evident for a number of years. In addition to these problems, the study was also criticized for not considering any kind of accident exposure measures such as vehicle-miles of travel, in its analysis. The higher accident frequency per student in one program may indicate that students in that program are worse drivers than students in other program, or it may simply be that students in this particular driver education program drive more miles than students in the other programs. Therefore, direct comparision, based on crash and conviction records without recognizing exposure, may be misleading. Proponents of driver education programs were disappointed in the Dekalb study as it did not fulfill their expectations. This is not surprising as several other studies on driver education program comparisions reported mixed results. In one such study in Michigan in 1977, the researcher asserted that students who had been exposed to only behind-the-wheel and classroom instruction 13 programs had a higher average incidence of violations and accidents than those students who had also been exposed to simulator training (12) . This finding was later supported by an author of another study conducted in Texas (13). In this study, 4,759 matched pairs of drivers from the same school environment licensed with or without public school driver education programs, were evaluated using accident and conviction records. Drivers without public school driver education were defined as having received "other training" in vehicle operation, where as the counterpart driver had 3-phase simulator training. All the drivers had a minimum of 18 months driving experience. No accident exposure measure was considered in this study. The conclusion was that those drivers who completed the 3-phase simulation program experienced fewer accidents and moving violations than the drivers having completed other training. However, two other studies, one conducted in Illinois and the other in California contradicted this finding (14). These two studies reported that there were no statistically significant differences in the number of crashes and convictions experienced by those students who had been enrolled in a simulator-enhanced program as compared to those students who had been enrolled in a behind-the-wheel program. In another study in Virginia, the Virginia Department of Highways & Transportation compared the performance of 14 different types of programs - two phase, three phase ( simulator and range) and four phase programs, using accident and conviction records for three driver experience levels (less than one year of driving experience, 1 to 2 years and 2 to 3 years) (15). The comparisions among programs showed that young people who received their training in the two phase program generally accumulated fewer convictions per 100 students than did their counterparts who received training which included a simulator, a range or both. However, the data concerning the number of crashes per 100 students for each different type of program did not show a consistent pattern for the three different levels of driver experience. That is, the program which had the fewest number of accidents during one level of driving experience, had the highest number of accidents during another period of driving experience, as shown in table 2.1. Therefore no definite conclusions could be reached regarding the crash reduction potential of any particular program. One of the reasons stated in the study for the absence of any pattern was that the number of crashes is too small and too volatile. An important factor which might have a potential effect on the accident frequency per student, is the identification of the guilty and innocent driver in the computation of an accident rate. It is not evident that this factor or any kind of accident exposure measure was considered in this study. 15 Table 2.1: Crashes per 100 drivers, by program types for different level of driver experience (15). Types of program Level of experience < 1 years 1-2 years 2-3 years 2-phase 5.9 11.8 10.3 3-phase simulator 6.3 11.2 12.3 3-phase range 6.9 10.4 12.9 4-phase 5.5 13.5 8.8 16 Due to the high cost of simulator programs, many researchers directed their interest towards the range program. In 1977, the California State Department of Motor Vehicles made a study of range versus nonrange (two phase) driver training in the San Juan area using accident and conviction frequencies of young drivers (16). The study indicated that range students had fewer total accidents per student than non-range students in the year following the begining of training. However, a North Carolina Highway Safety Research Center Study indicated that there was no significant difference between the performance levels of the range students and those students who had received other types of driving instruction (17). Thus one can see that there is no consistent evidence that enhanced programs using simulators and/or ranges are more effective than programs consisting strictly of classroom instruction and behind-the-wheel training. Research was also conducted to determine the performance of public and commercial schools. One study conducted by the Washington Division of Motor Vehicles in 1969 found that commercial driving schools are more effective in teaching safe driving habits than are public high schools (18). The researchers argued that this was particularly true for men, because they found that the male commercial school students had significantly fewer accidents and violations than their public school counterparts. However, a 1973 California report indicated 17 that there was no difference in the accident rate observed between publicly and commercially trained students (19). A Virginia study found that students graduating from commercial driving schools in Virginia had a significantly greater incidence of accident involvement and a higher rate of convictions for motor vehicle offenses than students who received their driver training at a public school (15). In a comparision between public and commercial schools in Ohio, a total of 59,496 driver records of public schools trainees were compared to a total of 37,642 driver records of commercial trainees in terms of the number of accidents and convictions accumulated over a time period of 6, 12, 18, 24, and 30 months immediately following the issuance of a license (19). The analysis of data led to the finding (at a very high level of confidence) that public school graduates had a greater percentage of records without accidents and violations, a lower accident involvement rate and a lower conviction rate than their commercially trained peers for all time intervals. 3.4 Conclusions: Despite the obvious benefits gained from previous studies, there are some limitations and deficiencies in each study. The overall general deficiencies and limitations found in the reported studies are as follows: 18 1. Conclusions drawn from previous studies (either large scale or small-scale) are not free from various external contaminating factors (e.g. lack of control for exposure, enforcement irregularities, etc.). 2. Students receiving different driver education programs may have been psychologically, physiologically or socio-economically different from each other, as well as from those students who did not receive driver education. 3. Since the analysis periods are short, the data on conviction and specifically on accident records are too small to allow for a conclusive result. 4. Accident exposure measures were not considered in any of the above mentioned studies. Therefore, direct comparisons between different groups of young people based on their accident and conviction records v/hils ignoring their average miles driven or other exposure measures which could be an essential factor in their crash and conviction statistics, may be misleading. 5. An additional concern is the accuracy in accident and conviction records, which might significantly affect the accident and conviction rate - the criterion variables used for evaluating driving education programs. Although in theory there is standardization in the reporting of accidents, in practice, this is not always the case. For instance, two-car accidents involving no personal injuries and no major vehicle damage may be underreported in large urban area. Similarly, there is almost certainly a sizeable variance among jurisdictions in regards to the type of driving behavior which will prompt a citation for a traffic violation. Sometimes drivers are not correctly reported as the guilty or innocent drivers in multi-vehicle accidents, and this will bias the result. Moreover, it is not clear that previous studies considered the concept of innocent driver and guilty driver in computing the accident rate. 6 . Finally, studies on the effectiveness of existing driver education programs are not conclusive. This study was undertaken to overcome some of these concerns. The major deficiency in previous studies, that of not considering an exposure measure in their analysis, will be overcome by using an accident exposure measure in this study. Vehicle-miles of travel on a specific highway under specific condition is generally considered to be the most appropriate measure of exposure. However, it is quite cumbersome to collect these data. This problem has led to the use of some kind of indirect exposure measure method, among them is the quasi-induced accident exposure method. 20 This method is based on the assumption that accident exposure by any group of drivers is proportional to the innocent victim involvements in multi-vehicle accidents by that group of drivers (20). This method will be further elaborated in the following chapter. Like many other studies, this study has also some limitations which are mentioned below: 1. This study concentrates on a comparison of performances of various driver education programs. There is no control group in this study - that is performance of various drivers with and without driver training are not compared. This is because 16 and 17 year old drivers can not obtain a driver license in Michigan without receiving driver training. 2. There are various external contaminating factors which might affect the results. Such factors include: Geographical location of schools: Drivers from schools in different geographical areas may encounter different driving conditions which might affect their road performance as well as the performance of the school and program. However, this factor is partially reduced by considering schools from three major geographical areas - Detroit metropolitan area, other urban areas and rural areas. However, there is still room for variation in driver exposure in each area. The reporting of accident and conviction incidents also varies from place to place which might also affect 21 the results. Drivers with different socio-economic background: Students receiving different driver education programs may have been psychologically/ physiologically or socio­ economically different from each other, as well as from those 18 year old students who did not receive driver education. This factor may also bias the results. 3. The analysis period for this study is limited to two years because driver school codes were not captured prior to 1987. However, some of the important relationships that remain unknown following this review of the literature will be addressed by testing the major hypotheses in this study: 1. There is no difference in the mean accident rate (number of accidents/student) among various driver ir\n nrnrrramc 2. There is no difference in the mean single-vehicle accident rate (number of single-vehicle accidents/ student) among various driver education programs. 3. There isno difference in the mean conviction rate (number of convictions/student) among various driver education programs. 4. There isno involvement programs. difference in the mean relative ratio (IR) among various driver education There is no difference in the mean relative involvement ratio among various driver education programs under different weather and light conditions. There is no difference in the mean relative involvement ratio for different types of accidents, among various driver education programs. There is no difference in the mean relative involvement ratio among different types of schools for different types of accidents and under different weather and light conditions. CHAPTER 3. METHODOLOGY The general purpose of this study was to determine the crash reduction performance of various driver education programs in Michigan. The performance was evaluated on the basis of accident and conviction records of drivers who received their driver training under each of the programs tested. A scoring system was developed, and a model was calibrated to predict the crash performance under each of the different driver education programs, and individual schools within these programs. To accomplish these objectives, the following methodology was followed: 3.1 Data Base; A new data base was created by extracting required information from three existing data files, and merging them together in one file. These three data files are the Highway Accident Master file (accident file), Driver Accident and Conviction Record file (driver license file), and Driver Education Program and School Information file (school file). The Highway Accident Master file contains information about each accident that occurred within the State of Michigan, including the accident location (time and place), road and weather condition, type and severity of the 23 24 accident and data regarding drivers and vehicles involved in the accident. This file also contains the identification of the driver most at fault and the driver considered to be the "innocent victim" which are coded as the first and second driver respectively as determined by the investigating officer. The Driver Accident and Conviction Record file contains accident and conviction records of each driver, as well as driver parameters such as age, sex, residence etc. For 16, 17, and 18 years old drivers this file also contains a code number for the driving school where they completed their driver education program. The lay out of this file is different from the accident file. A fortan prgram (shown in appendix A) was written to change the format of the driver accident file to make it compatible with the accident file. The Driver Education Program & School Information file contains information regarding each driving school such as school code, school location, number of students graduated each year and types of program offered. The accident file and the driver accident file share a common element (the accident report number). Using this common variable it was possible to merge the accident and license file into one file. The driving school file was merged with the license file on the basis of a common variable - the school code. 25 3.2 Data Variables From the three original data files a new data base was created that included the data necessary to accomplish the objectives of this research. The variables that were selected from the three original data files and included in the new data file were as follows: — the environmental characteristics at the time of the accident: that is time, date, weather and light condition; — general characteristics of the accident such as accident type, type of violation and accident report number; — conviction, type of offense code, date of conviction; — the characteristics of the driver: such as age, sex, date of birth, driver license number, original and issue license date, driving school; d2ri.vi.n5 school infcriPiStion t such ss typs of school (public or commercial), location of school, number of students, type of program offered. Once the entire new data file was built, a subset of this data file was created which included only those schools which graduated a minimum of one hundred students each year for the two year analysis period (1988 and 1989). 26 3.3 Induced Exposure Method; The traditional method of determining whether a certain driving school or a particular driver education program has a better performance than other driving schools or driving programs is to compare accident rates. These rates are traditionally based on the number of accidents per driver in each school or program. This method is not a very accurate representation of true risk, since it is based on the implicit assumption that drivers in all the schools or programs have an equal exposure to accident situations. To overcome this problem, an alternative method was utilized in this research (20, 21). This method uses an indirect measure of accident exposure, generally called the quasi-induced exposure method. This method is based on the assumption that the accident exposure by any group of drivers is proportional to the innocent victim involvements in multi-vehicle accidents by that group of drivers. An innocent victim in an accident is defined as the driver not responsible for the accident. Those drivers who are involved in multi-vehicle accidents and are atfault, or responsible for accidents, are defined as driver-1 , whereas drivers who are not at-fault, or not responsible for the accident, are defined as driver-2 . Driver-2 is used as a measure of accident exposure. This measure of accident exposure provides a tool to study 27 the relative accident propensity of different groups of drivers from different driving schools and programs under different driving conditions. To obtain a measure similiar to an accident rate for groups of drivers from different driver education programs, a ratio that indicates the involvement of drivers of that specific group to their respective exposure measure, was used. This ratio is defined as the ratio of the percentage of accidents where driver-1 comes from a given driver education scenerio to the percentage of accidents where driver-2 comes from the same scenerio. This ratio is called the relative accident involvement ratio (IR), which is a measure of the relative freguency of accident involvement for the various groups of drivers from different driver education programs. A value of 1.00 for this relative involvement ratio denotes equality between the accident involvement and accident exposure for drivers from a given scenerio. Similarly, when the ratio is less than 1.00, this means the driver group is less likely to be responsible for an accident, which constitutes under-involvement. When the ratio is greater than 1.00 the driver is more likely to be responsible for an accident which constitutes over­ involvement. However, there were certain limitations or issues related with this technique as identified by many previous studies (20, 22). These are discussed as follows: 28 1) How one-vehicle accidents are considered in this concept of exposure. Studies have shown that if the characteristics (proportion) of the at-fault drivers are the same for both types of accidents (i.e. single-vehicle & multi-vehicle accidents), the involvement ratio is uneffected (20). However, there is no compelling reason that the characteristics of at-fault drivers in one and multi-vehicle crashes should be the same. If the characteristics of at-fault drivers in one-and multi­ vehicle crashes are different, these two types of accidents should be analyzed separately. Therefore, in this study one- and multi-vehicle accidents are analyzed separately. 2) Another important issue related with this concept of exposure measure is that innocent drivers (driver-2 ) should constitute a random sample of all drivers. That is driver-1 's choose their innocent victim at random from all drivers present on the system. It implies that subsets of driver-1 's should choose driver-2 's in the same proportion - that is, the row proportion should be identical. Therefore, in this study only those schools under each program were selected which had a high number of accidents. The relative involvement ratio for each school within each program was computed. An average value of the relative involvement ratio was determined for all schools within each program. This average value represents the 29 relative involvement ratio for each driver education program. In addition to this approach of computing IR value for different driver education programs (i.e. using the average of IR value for all schools under each program), the IR value for each program was also computed separately for three different geographical areas. These areas are the Detroit metropolitan area, other urban areas and rural areas. The purpose behind this was to reduce the differences in driver exposure due to different geographi­ cal areas. The performance of drivers from different schools or programs was determined under different driving conditions, to find out if drivers from a particular school or program have specific problems under certain driving conditions. The variables considered for defining various driving scenarios six’s ss follows? 1) light condition: day, dawn or dusk, and night, 2) weather condition: clear, raining, and snowing, and 3) types of accidents. For each school, the relative involvement ratio was obtained for each level of the above mentioned variables. The relative involvement ratio for each program was then determined by computing the average value of this ratio for all schools under each program. In addition to the analyses using the accident 30 exposure, other traditional methods were also used to analyze the data. These methods involved the calculation of the value of certain criterion variables for each selected school and then for each program. The criterion variables used were: 1 ) total number of accidents per student; 2 ) number of single vehicle accidents per student; 3) number of convictions per student; These criterion variables in addition to the IR value were used for hypothesis testing as discussed in the next section. 3.4 Hypotheses testing; After the values of all criterion variables were determined, hypotheses were tested using the "t" test and analysis of variance models (ANOVA). ANOVA procedures n«v»Tn 1 ^ A. W a w a 1 ifc*l A j> M A W W A n w i a m m l A r» a w » a 1 « AW> r* i M A M M M U 1 IWt V44W simulataneous testing of more than two samples, where as the "t" procedure can test only two samples at a time. First, ANOVA procedure was utilized to examine the differences in the mean criterion variable among different groups. If any difference among different groups was found statistically significant, then the "t" test was used to find out exactly which two groups differed. The Statistical Analysis System performing the tests (23). (SAS) software was used for 31 The major hypotheses which were tested are as follows: 1. There isno difference in the mean accident rate (number of accidents/student) among various driver education programs. 2. There is no difference in the mean single-vehicle accident rate (number of single-vehicle accidents/ student) among various 3. There isno difference driver education programs. in the mean conviction rate (number of convictions/student) among various driver education programs. 4. There is no difference in the mean relative involvement ratio among various driver education programs. 5. There is no difference in the mean relative involvement ratio among various driver education programs under different weather and light conditions. 6 . There is no difference in the mean relative involvement ratio for different types of accidents, among various driver education programs. 7. There is no difference in the mean relative involvement ratio among different types of schools for different types of accidents under different weather and light conditions. 32 3.5 Development of a Scoring System; Comparisons among different schools and programs based on the criterion variables, classified certain schools as being either higher or lower ranked under certain conditions. However, it may be misleading to draw a conclusion regarding the over-all performance of various driving schools and programs using total accident rate as a criterion variable. This is because students from different schools may experience accidents of different types and degree of severity. It may not be appropriate to rate a school which had a fewer number of accidents with a high degree of severity better than another school which had a higher number of accidents but with lesser degree of severity. To determine the rating of different schools and programs based on frequency and severity of accidents, an attempt was made to define a common base for all types of accidents. Previous studies have determined the average dollar value of accidents by degree of severity (such as fatal, injury and property damage) but these averages did not account for the various types of accidents (24, 25). To determine the average dollar value by type of accident, three steps were followed: 1 ) the percentage of fatal, injury, and property damage accidents in each type of accident were determined from the state-wide accident data for the year 1988 (26), then 33 2 ) these percentages were multiplied by the respective average dollar value of fatal, injury and PDO accidents; and finally 3) these three products were summed . For the purpose of developing a scoring system based on accident frequency and severity, a weighting score is required for each type of accident. And this weighting score is taken to be equivalent to the dollar value for each type of accident. The score for each school was then obtained by summing the product of the frequency of each type of accident and their respective weights. Each school and program was rated on the basis of the score/student for each school. The same concept was utilized for developing the score for different types of convictions. Points associated with different types of conviction was used as weight for their respective type of convictions. The conviction rating score for each school was obtained by summing the product of frequency of each type of conviction and their respective weight. Each school and program was also rated based on the conviction score/student for each school. In order to determine the consistency of various schools in their performance on different criterion variables over the two year period, the following steps were followed: 1. For each of the two years, schools were ranked based 34 on each of the three criterion variables - IR value, number of accidents/student and score/student. 2. In each year's ranking, schools were classified into two groups for each criterion variable. The first group constitutes those schools whose criterion variable value is lower than the mean value. The second group includes those schools whose criterion variable is higher than the mean value. The first and second groups are called higher and lower ranked groups according to a specific criterion variable. 3. Those schools which appear in both years higher or lower ranked group under each criterion variable, were identified as consistent schools on a given criterion variable. Schools which were consistent across different criterion variables were also identified. These rating scores, in addition to other criterion variables, were further used in developing models, as discusssd in ths nsxt section. 3.6 Model Development: One of the prime objectives of this study, was the development of a model to estimate the performance of various driving schools and driving programs. The independent variables used in developing the model included the number of students, number of accidents, relative accident involvement ratio, accidents/student, convictions/student; and dummy variables for the type of 35 program (range, competency public, competency commercial, and traditional. The simulator and the 4-phase programs were not considered due to small sample size.), and geographical location of school (Detroit metropolitian area, other urban area and rural area). The dependent variable (year 1989 score/student) for each school was regressed against the year 1988 values of the independent variables for the same school. The purpose was to use the information from the first year accident and conviction records for each school to calibrate a model which predicts the score/student for the following year. Models were calibrated for the following cases: 1. For each combination of type of program and geographical location of school ( such as range-urban, range-rural, etc.) 2. For each type of program irrespective of geographical iccation, 3. For each geographical location of school irrespective of type of program. 4. An overall model for all types of programs and geographical locations. In each of the above cases, two types of regression models were calibrated. The first one was the simple linear regression model between the dependent variable (score/ student) for the year 1989 and the independent variable 36 (score/student) for the year 1988. This was done to determine the consistency between year 88 and year 89. The second type of model was a multiple linear regression model, using all the possible independent variables. The tool of analysis was STEPWISE regression, which either allows each variable to enter into the model at each step if it makes a significant contribution to the model, or drops the variable out of the model at each step if it does not make a significant contribution to the model. The level of significance selected for variables to enter into the model or stay in the model was 0.1. SAS software were utilized for performing the stepwise regression analysis. 3.7 Discriminant Analysis: Discriminant analysis was performed to determine if it was possible to classify schools into different driver cduCwiwicn pr’c^r’ctms or* discriiuinwinti functions developed from a set of performance predictor variables. A comparision between the actual classification of schools under each program and the predicted classification would determine how successfully schools can be discriminated into different programs based on these performance variables. A small difference between criterion groups with respect to predictor variables results in more error in classification in discriminant analysis. Any relationship between qualitative criterion 37 groups and quantitative predictor variables could be identified based on this analysis (27). For performing discriminant analysis, discriminant functions were developed from the first year data, using types of program as a classification variable and IR value, score/student, accidents/student and convictions/student for each school as a set of predictor variables. Using these discriminant functions, the classification of schools into various programs was predicted. The accuracy of classification was examined by comparing the actual with the predicted classification of schools into different programs. The comparision between the actual and the predicted classification of schools into different programs was also made for the second year, using the same discriminant functions which were developed from the first year data. Based on these comparisons, it can be inferred whether a relationship between types of program and a set of performance variables exists. This will further indicate whether there is any difference among the programs based on this set of performance variables. CHAPTER 4. DATA ANALYSIS The newly created data file, which contains information regarding accidents, convictions, and driver education programs for a two year analysis period (1988 and 1989) was used to evaluate relationships. Analyses were conducted to assess differences in accident characteristics when different driver education programs and driving schools were compared. Additionally, a scoring system was developed to rate various schools and programs. Models were calibrated for predicting program and school performance under varrious conditions. This chapter reports the examination and assessment of the data. 4.1 Data Statistics: For years 1988 and 1989, the total number of 16, 17 and 18 year old drivers in the state of Michigan was 222,647. About 41 percent of these total drivers come from the Detroit metropolitan area counties. The male and female drivers were 49.6 and 50.4 percent respectively. Female drivers were found to have a better accident record than male drivers in this age group. Of 75.6 percent of the drivers with no accident records, there were 52.4 and 47.6 percent female and male drivers respectively. For those drivers who had at least one accident, the male and female drivers were 55 and 45 38 39 percent respectively. There were only 4 percent of the drivers who had two or more accident records. The most common type of accident which drivers in the given population experienced was the rear-end accident. Rear-end accidents constituted about 27 percent of the total accidents. The percentage of other types of accidents is as follows: fixed object hit = 13.6% angle straight = 11.5% head-on-left turn = 7.1% rear-end drive = 7. 1 % angle turn = 6 .6 % Female drivers had an even better record in convictions than they had in accidents. Of 68 percent of the total drivers who had no conviction record, there were 57 percent female and 43 percent male. For drivers with a minimum of one conviction record, female and male drivers were 35 and 65 percent respectively. Only 12 percent of the drivers had two or more convictions on their record. The most common type of conviction reported in the data set was related to speed violations, which constituted 48 percent of total convictions. Convictions involving violations of basic traffic laws were about 18 percent. There were a total of 517 public and 75 commercial schools in the data set. About 85 percent of total driver records in the age group studied had the code number for schools where the drivers obtained their driver training. 40 To perforin a detailed analysis of the data, the data set was reduced to contain data from only those schools which had at least one hundred graduates each year for the two year analysis period. This data set included 255 public schools and 35 commercial schools and includes about 69% of the total number of students. For the selected subset, the distribution of schools under various training programs was as follows: (i) 124 public schools using the range program (ii) 63 public schools using the traditional program (iii) 57 public schools using the competency program (iv) 35 commercial schools using the competency program (v) 5 public schools using the simulator program (vi) 6 public schools 4.2 using the 4-phase program. Accident and Conviction rate; The total number of state wide accidents that involved 16, 17 and 18 years old drivers for years 1988 and 1989,was 54,500. The accident rate, that is accidents per student, for all selected schools stratified by program, is shown in appendix C. The frequency distribution of accidents/student for all programs is shown in appendix D, whereas the frequency distribution curve for the range program is shown in figure 4.1. The frequency distribution of accidents/student for all programs were found to be slightly skewed to the the right, indicating relatively fewer schools with higher accident rate. The range program had the least skewness among all programs (fig. 4.1). About 50% of schools (which are to the left of the mean) fall between the mean (.220 ) and the mean-l*std.dev. (.179) and 43% of the schools (which are to the right of mean) fall between the mean (.220 ) and mean+i*std. dev. (.261). The competency program had the maximum skewness in its distribution (appendix D). About 70% of schools whose accident rate is less than the overall mean accident rate, lie between the mean (.226) and the mean-l*std.dev. (.186) and about 61% of schools whose accident rate is greater than the overall accident rate fall between the mean (.226) and the mean+1 * std.dev.(.266). Thus, due to the uniformity in accident rate values at both ends of the frequency distribution curve for all programs, there is no obvious cut-point for separating higher and lower ranked schools from the rest of the schools. A sample of about 20% of all schools with the lowest and highest accident rates under each program are shown in tables 4.1 to 4.4. Table 4.1 shows the accident rate for a sample of twenty four from the top and twenty three from the bottom schools using the range program, in order of increasing accident rate. These top and bottom schools are called "higher" and "lower" ranked schools respectively in the following analysis. The same table shows that the accident rate for the lower ranked schools 42 RANGE PROGRAM FREQUENCY BAR CHART FREQUENCY ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** 0.150 0.175 0.200 0.225 0.250 0.275 + 20 + 15 + 10 + + ACCRT Figure 4.1: Histogram of accident rate for the Range program. ***** ***** ***** ***** ***** ***** ***** ***** 0.300 43 Table 4.1: Accident rate for higher and lower ranked public schools using the range program. S.N. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. Code for Accident higher ranked rate schools 740 623 171 785 543 736 462 795 741 042 420 547 701 256 020 706 276 746 750 760 738 535 365 555 0.130 0.136 0.143 0.152 0.153 0.154 0.157 0.158 0.159 0.160 0.165 0.167 0.169 0.169 0.170 0.173 0.174 0.175 0.175 0.176 0.177 0.177 0.178 0.179 S.N. Code for lower ranked schools Accident rate 1. 2. 377 252 622 267 347 258 081 303 348 366 616 492 559 560 394 323 541 549 059 471 301 332 544 0.260 0.263 0.265 0.266 0.266 0.266 0.267 0.267 0.267 0.268 0.270 0.273 0.276 0.278 0.280 0.289 0.290 0.298 0.301 0.302 0.302 0.308 0.313 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 44 is approximately 2.2 times higher than that of the higher ranked schools. The accident rate for higher and lower ranked schools under the traditional program are shown in table 4.2. The accident rate for lower ranked schools is slightly more than 2.5 times the accident rate for higher ranked schools in this program. The ratio of the accident rate between higher and lower ranked public schools using the competency program are approximately the same as the range program, as shown in table 4.3. For the competency program in commercial schools, the accident rate for the lower ranked eight schools are 67 percent higher than the accident rate for the higher ranked seven schools in the same program as shown in table 4.4. The same table indicates that there is less variation in the accident rate among commercial schools as compared to the variation among public schools. Table 4.5 shows the average value of accident rate for different programs which is weighted by number of students in each program. The competency program in commercial schools had the highest accident rate among all programs. A detailed comparison among the various schools and programs using statistical methods is discussed in the hypotheses testing section. Single-vehicle accident rate (number of single vehicle accidents/student) was also computed for all schools and is shown in appendix C. A large difference (up to 13 45 Table 4.2: Accident rate for higher and lower ranked public schools using the traditional program. S.N. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. Code for Accident higher ranked rate schools 763 675 260 009 198 635 439 770 188 103 772 486 0.120 0.154 0.155 0.156 0.163 0.166 0.173 0.176 0.178 0.180 0.187 0.189 S.N. Code for lower ranked schools Accidenl rate 1. 2. 163 114 483 153 506 545 707 416 350 450 455 0.273 0.278 0.279 0.282 0.283 0.283 0.288 0.304 0.309 0.314 0.332 3. 4. 5. 6. 7. 8. 9. 10 . 11. 46 Table 4.3: Accident rate for higher and lower ranked public schools using the competency program. S.N. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10 . 11. 12. Code for Accident higher ranked rate schools 790 226 429 777 136 618 570 769 442 670 128 425 0.135 0.143 0.168 0.170 0.173 0.174 0.181 0.184 0.184 0.186 0.188 0.189 S.N. Code for lower ranked schools Accident rate 1. 2. 495 412 507 431 633 134 629 194 538 269 413 0.267 0.270 0.272 0.286 0.290 0.293 0.298 0.305 0.313 0.315 0.325 3. 4. 5. 6. 7. 8. 9. 10. 11. Table 4.4: Accident rate for higher and lower ranked comm­ ercial schools using the competency program. S.N. 1. 2. 3. 4. 5. 6. 7. Accident Code for higher ranked rate schools A77 980 OSS A56 975 984 A3 5 0.217 0.218 0.222 0.226 0.228 0.236 0.238 S.N. 1. 2. 3. 4. 5. 6. 7. 8. Code for lower ranked schools 966 A45 A48 A65 A82 A24 A62 959 Accident rate 0.296 0.302 0.307 0.307 0.310 0.325 0.341 0.353 47 Table 4.5: Weighted average accident rate for various programs. S.N. Type of program 1. Range 2. Number of schools in each program Weighted average accident rate 124 0.219 Competency (Public) 57 0.227 3. Competency (Commercial) 35 0.281 4. Traditional 63 0.228 5. Simulator 5 0.265 6. Four-phase 6 0.243 48 times) in the single-vehicle accident rate exists between various schools as can be observed in appendix C. However, the average (weighted by number of students) value of the single-vehicle accident rate for all programs lies in a close range of .048 to .063 accidents /student, as shown in table 4.6. There were about 80,000 conviction records for 16, 17 & 18 years old drivers in the state of Michigan for the two year period of 1988 and 1989. The conviction rate (number of convictions/student) for all schools under all programs is also shown in appendix C. The conviction rate for lower ranked schools was approximately 2.7 times higher under the traditional and competency (public) program and 3.5 times higher under the range and competency (commercial) than the higher ranked schools in their respective programs. Conviction rates of commercial higher and lower ranked school were respectively 1.75 times and 2.0 times higher than public higher and lower ranked schools under any program (appendix C). The average conviction rate (weighted by number of students) for all programs are shown in table 4.7. Similiar to the case for accident rates, the competency program in commercial schools had the highest conviction rate, which is approximately 1.6 times higher than the rate in public schools under any program (excluding 4-phase and simulator programs which had very few schools). This review of the data indicates that: 49 Table 4.6: Weighted average accident rate for single-vehicle accidents for various programs. S.N. Type of program 1. Range 2. Number of schools in each program Weighted average accident rate 124 0.048 Competency (Public) 57 0.054 3. Competency (Commercial) 35 0.050 4. Traditional 63 0.061 5. Simulator 5 0.063 6. Four-phase 6 0.048 Table 4.7: Weighted average conviction rate for various programs. S.N. Type of program Number of schools in each program Weighted average accident rate 124 0.284 Competency (Public) 57 0.262 3. Competency (Commercial) 35 0.448 4. Traditional 63 0.274 5. Simulator 5 0.378 6. Four-phase 6 0.344 1. Range 2. 50 1) There are relatively small differences in the average number of accidents per student and the average number of convictions per student across various programs used in the public schools. The commercial schools have a higher rate in both measures than that for public schools. The 16, 17 and 18 years old drivers were about 11 % more involved in accidents than the drivers of all age group in the state of Michigan for the two year period (1988 and 1989) (26). Of the 24.4% of the 16, 17 and 18 years old drivers with accidents there were about 17% of the drivers who had more than one accident as compared to 14% of the total driver population who had at least one accident, and 14% of drivers with accidents, who had two or more accidents. Similarly 16, 17 and 18 years old drivers are 12.5% more involved in convictions than the total driver *oonulation * (2G\ . « « Of the 32.5% of 16 to 18 acre group drivers with a minimum of one conviction there were 36.5% drivers who had more than one conviction. About 20% of all drivers had convictions and 35% of drivers with convictions, had two or more convictions in their record. 2) There are large differences in the rates among various schools within each of the driver education program categories. The data above describes the performance of schools 51 and programs based on the traditional method of determining the accident and conviction rate; that is accidents or convictions per student. However, the main thrust of this study was to modify the analysis by introducing an accident exposure measure to assess the performance of driver education programs and schools. The application of one such method called, the quasi-induced exposure measure, is discussed in the following sections. 4.3 Quasi-induced accident exposure method: In this method a criterion variable called the relative accident involvement ratio (IR), as explained in the previous chapter, was computed for each school under each program. Relative accident involvement ratio (IR) for all schools under consideration are shown in appendix C. The frequency distribution of IR values for different program is shown in appendix D. The IR distribution for all the four programs were found to be skewed to the right. In each case schools with a low value of IR were predominant and there were relatively few schools with higher values. The competency (public) program had the maximum skewness in its distribution as can be seen in figure 4.2. About 99 percent of all schools whose IR value is less than the overall mean value, lie between the mean (1.75) and mean-l*std.dev. (1.0) and about 52 percent of all schools whose IR value is greater than the overall mean value, 52 COMPETENCY (PUB.) PROGRAM FREQUENCY BAR CHART FREQUENCY *** *** 15 + *** 10 + 5 + *** *** ■kick k kk k kk 1.00 •kkk k kk kkk kk k k kk k kk kk k k kk kk k kk k kk k kkk k kk kkk kk k kkk kkk kk k kk k kk k k kk k kk k kk .25 1.50 1 .7 5 kkk kkk kk k kk k kk k kk k kkk kk k k kk k kk k kk *** *** *** kkk kkk *** *** *** k kk kkk k kk k kk k kk kkk kkk 2.00 2.25 2.75 3.50 4.00 4.25 IR Figure 4 .2 : Histogram of IR values for the competency (pub.) program. 53 fall between the mean (1.75) and mean + l*std.dev. (2.50). The mean and the standard deviation statistics are shown at the end of this section in table 4.12. The competency program in commercial schools had the least skewness in its distribution (appendix D). About 78% of all schools (whose IR value is less than the mean) fall between the mean (1.63) and the mean-l*std.dev. (1.0) and about 75% of the higher scores lie between the mean and mean + 1 * std. dev. (2.24). Thus due the to higher uniformity in IR values in the beginning of the distribution curve, there is no obvious cut-point for separating higher ranked schools. However, due to the skewness in the distribution curve, there is a more obvious cut-point of lower ranked schools. Like the previous section 20 percent of the schools with lowest and highest IR values under each program are shown in table 4.8-4.11. Those schools with lowest and highest IR value are called "higher ranked" and "lower ranked" schools respectively according to this criterion. An overall review of these tables reflects that except for very few schools, drivers were over involved in multi-vehicle accidents. In other words, the 16, 17 and 18 year old drivers were responsible for more multi-vehicle accidents than would be expected based on their exposure. The IR value is fairly consistent across higher ranked schools (according to this criterion) for all four programs i.e. range, competency (pub.), traditional and 54 competency (comm.) as shown in table 4.8 to 4.11. This was also found in the frequency distribution of IR values for each program. Under the traditional program (table 4.9) the IR value for higher ranked schools varies in a very close range of 0.9 to 1.15. The IR value for lower ranked schools was approximately 2.20 times higher than the IR value for their respective higher ranked schools under the range and traditional program. The IR value for lower ranked schools under the competency program varies in a wide range (1.9 to 4.25) as can be seen in table 4.10. The IR value for lower ranked schools under the competency (pub.) program are approximately 2.7 times higher than the IR value for their respective higher ranked schools. That is, drivers from lower ranked schools under these programs were respectively 2.2 to 2.7 times more over­ involved or responsible for multi-vehicle accidents than their counterpart drivers in higher ranked schools under the same eroaram. - - - w The IR value for lower ranked commercial school groups were 1.8 times greater than their respective higher ranked schools, as indicated in table 4.11. There is more variation in the IR value for public schools under any program as compared to the variation in commercial schools. The average values of IR for all programs ( table 4.12 ) shows that drivers under the competency program (public schools) were approximately 10 percent more over-involved in multi-vehicle accidents than drivers in the range, 55 Table 4.8: Relative involvement ratio (IR) for higher and lower ranked public schools using the range program. S.N. Code for higher IR ranked school ratio 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 623 365 062 170 548 547 597 674 733 750 738 624 553 760 430 535 549 508 255 037 736 680 S.N. 0.700 0.889 1. 2. 1.000 3. 4. 5. 6. 7. 8. 9. 10 . 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 1.023 1.042 1.054 1.125 1.156 1.159 1.184 1.191 1.218 1.219 1.231 1.235 1.250 1.254 1.273 1.278 1.286 1.286 1.300 Code for lower ranked school 059 314 529 307 323 651 701 785 492 183 556 420 081 349 270 165 706 166 795 331 391 256 043 IR ratio 1.931 1.938 1.968 2.000 2.000 2.000 2.000 2.000 2.045 2.149 2.159 2.250 2.308 2.308 2.318 2.321 2.333 2.348 2.538 2.556 2.667 3.091 4.000 56 Table 4.9: Relative involvement ratio (IR) for higher and lower ranked public schools using the traditional program. S.N. Code for higher IR ranked school ratio 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 450 198 114 385 506 725 482 457 635 483 486 753 295 0.935 0.941 0.976 1.000 1.015 1.034 1.053 1.107 1.121 1.125 1.137 1.147 1.158 S.N. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10 . 11. 12 . 13. Code for lower ranked school 439 103 147 705 153 187 776 217 009 763 772 035 675 IR ratio 1.929 2.000 2.000 2.109 2.118 2.118 2.150 2.214 2.286 2.375 2.545 3.000 3.800 Table 4.10: Relative involvement ratio (IR) for higher and lower ranked public schools using the competency program. S.N. Code for higher IR ranked school ratio 1. 2= 3. 4. 5. 6. 7. 8. 9. 10 . 11. 053 641 650 478 670 629 615 029 182 036 425 0.800 1 .000 1.094 1.111 1.120 1.156 1.160 1.190 1.294 1.300 1.300 S.N. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Code for lower ranked school 781 669 570 554 778 136 128 469 186 618 226 IR ratio 1.938 2.000 2.100 2.136 2.565 2.700 2.824 3.500 4.091 4.167 4.250 57 Table 4.11: Relative involvement ratio (IR) for higher and lower ranked commercial schools using the competency program. S.N. Code for higher IR ranked school ratio 1. 2. 3. 4. 5. 6. 7. A04 975 A05 984 980 A45 999 S.N. 1.147 1.179 1. 2. 1.222 3. 4. 5. 6. 7. 1.263 1.330 1.362 1.384 Code for lower ranked school A10 977 A 86 A08 A56 A3 5 A83 IR ratio 1.820 1.875 1.889 1.912 1.933 2.077 2.400 58 Table 4.12: Average relative involvement ratio (IR) for various programs. S.N. Types of program Number of schools in each program IR ratio Std. Dev. 124 1.650 0.444 1. Range 2. Competency (Public) 57 1.750 0.732 3. Competency (Commercial) 35 1.637 0.256 4. Traditional 63 1.591 0.513 5. Simulator 5 1.883 0.640 6. Four-phase 6 1.604 0.238 59 traditional and competency (comm.) program. The IR value was also computed for 16 to 18 year old age group of drivers for various programs under the three different geographical areas of schools. These three geographical areas are the Detroit metropolitan area, other urban areas and rural areas. It is assumed that drivers from different geographical area have different driving exposure. That is driving conditions or traffic patterns in Detroit metropolitan area may be different than the driving conditions in rural areas. It may be further assumed that driver exposure may be more uniform within the same geographical boundry. The accident exposure of drivers with different driver education training is assumed to be similiar within the same geographical area. The IR value of young drivers under various programs for the three geographical areas are shown in table 4.13. The same table 4.13 shows that the IR value is consistent for various programs in the Detroit metropolitan and other urban areas. In rural areas the IR value for the competency program indicates that drivers from commercial schools were 12 percent more over-involved than drivers from the public school programs. One of the possible reasons for this higher IR value may be that there are very few commercial schools in rural areas. There are small differences in the IR value (a maximum of 10% in competency (comm.) program) across different geographical 60 Table 4.13: Relative involvement ratio for various programs under different geographical areas. Type of program Detroit metropolitan area Urban area Rural area Average value Range 1.513 1.662 1.581 1.583 Traditional 1.532 1.572 1.457 1.513 Competency (pub.) Competency (comm.) 1.652 1.643 1.595 1.621 1.584 1.630 1.742 1.663 61 areas within each program. The drivers from public school programs in rural areas are approximately 7 percent less over-involved in multi­ vehicle accidents than drivers from the same programs in other urban areas as well as in the Detroit metropolitan area (except for the range program). The comparisions between the average IR value for various programs computed in table 4.12 (i.e. average of all schools IR value under each program) and in table 4.13 (i.e. average of IR value for all three geographical areas) shows a maximum of a 7.5% difference which in the competency (pub.) program. The above discussion indicates that there are small differences in IR values among various programs and schools due to different geographical areas. To determine the performance of schools and programs under different driving conditions, a sample of higher and lower ranked schools based on the IR value under each program was used for further analyses. 4.4 Weather Condition; The IR value for each higher and lower ranked school under each program, for the three different weather conditions - clear, raining, and snowing condition is given in appendix C. The average IR values of higher and lower ranked schools under each program for these three different weather conditions are shown in table 4.14 and 4.15. Table 4.14 shows that all average IR values of even 62 Table 4.14: Average relative involvement ratio (IR) for higher ranked schools under various programs, for different weather conditions. S.N. Type of program f of schools per program Average IR ratio for Clear Rainy Snowy weather weather weather 1. Traditional 13 1.117 1.115 1.218 2. Range 22 1.211 1.149 1.317 3. Competency (Public) 4. Competency (Commercial) 11 1.194 1.192 1.268 7 1.145 1.371 1.261 Table 4.15: Average relative involvement ratio (IR) for lower ranked schools under various programs, for different weather conditions. S.N Type of program # of schools per program Average Clear weather IR ratio Rainy weather for Snowy weather 1 . Traditional 13 2.408 2.439 2.654 2 . Range 22 2.072 2.545 2.386 3. Competency (Public) 4. Competency (Commercial) 11 2.917 2.574 3.545 7 1.894 1.964 3.147 63 the higher ranked schools under each program and each weather condition, are greater than 1.0. The average IR for each weather condition was consistent across each program for both higher and lower ranked schools. The IR value under snowy weather condition was 10 and 25 percent higher than the IR value under clear weather conditions for the higher and lower ranked schools respectively for all programs. Table 4.15 shows high IR values for lower ranked schools under snowy condition, indicating a very high over-involvment of drivers from these schools across all programs on snowy days. The average IR value of lower ranked schools averaged two times higher than the IR of their respective higher ranked schools under the same program and same weather condition. These data indicate that none of the driver education programs prepare students to drive under adverse weather conditions. The traditional program results in the lowest and ----..... second . . . .lowest TR value under snowv .4 and. rainv 4 weather conditions for higher and lower ranked schools respectively. 4.5 Light Conditions: Appendix C shows the IR values for higher and lower ranked schools under each program and three different light conditions - day, dawn/dusk, and night time. The average IR for higher ranked schools under each program were consistent under each light condition as shown in table 4.16. There is a clear-cut pattern of higher night 64 Table 4.16: Average relative involvement ratio (IR) for higher ranked schools under each program, for different light conditions. S.N. Type of program number of schools under each program Average IR ratio for light condition under day dawn/dusk night time time time 1. Traditional 13 1.107 0.831 1.275 2. Range 22 1.197 1.423 1.314 3. Competency (Public) Competency (Commercial) 11 1.254 1.366 1.375 7 1.343 1.276 1.433 4. Table 4.17: Average relative involvement ratio (IR) for lower ranked schools under each program, for different light conditions. S.N. Type of number of schools under Average IR ratio for light condition under day dawn/dusk night time time time 1. Traditional 13 2.399 1.805 2.771 2. Range 22 2.499 2.085 2.568 3. Competency (Public) Competency (Commercial) 11 2.745 2.136 2.864 7 2.172 2.370 2.418 4. 65 time IR values over day time IR values across all programs for higher ranked schools. The same pattern of higher night time IR values was found for lower ranked schools, as shown in table 4.17. Under day and night conditions, the average IR values of lower ranked schools was almost two times greater than their respective higher ranked schools. These data indicate that young drivers are about 15% more likely to be involved in a night accident than a day accident (after correcting for exposure). The ratio is even higher among the bad schools, with very high values of the IR exhibited by all public school programs. This ratio means that young drivers from these schools are involved in between 2 and 3 accidents as the guilty party for each accident in which they are the innocent victim. The commercial school program results in the highest and lowest IR value for higher and lower ranked schools respectively for night time accidents. The lowest IR value for night time accidents for lower ranked schools might reflect the fact that most of the commercial schools are in the urban areas where street lighting mitigates the difference between day and night driving. 4.6 Types of Accidents: The four accident types which constitute the highest percentage in total accidents were considered in determining the performance of higher and lower ranked 66 schools under different programs. These four types of accidents were rear-end, angle turn, angle-straight, and head-on-left turn. Fixed object accidents could not be considered because there is no innocent victim in one vehicle accidents. The IR value of the higher and lower ranked schools under each program by accident types is shown in appendix C. Table 4.18 shows the average IR values for higher ranked schools for different types of accidents under each program. It can be observed from this table that the average IR for angle-turn accidents under the traditional program was less than 1.0. All other average IR values were between 1.01 to 1.5. For lower ranked schools, the IR values for all types of accidents ranged between between 1.9 and 3.2, as evident from table 4.19. Tables 4.18 & 4.19 illustrate that the IR of lower ranked schools for different types of accidents under different types of programs averages two times higher than their counterpart higher ranked schools average IR value. An interesting result of this analysis is that the traditional program had a lower IR value for all four accident types than the range program. Thus, at least for those four common accident types, the additional driving experience gained on the driving range did not result in a lower accident experience (after counting for exposure). In this section, the quasi-induced accident exposure measure method was used to determine a criterion variable 67 Table 4.18: Average relative involvement ratio (IR) for higher ranked schools under each program, for different types of accidents. S.N. Type number of of schools under program each program Average IR ratio for angle- rear- angle head-on strait end turn left turn accd accd accd accident 1. Traditional 13 1.012 1.159 0.949 1.025 2. Range 22 1.477 1.189 1.352 1.500 3. Competency (Public) Competency (Commercial) 11 1.285 1.363 1.198 1.437 7 1.308 1.366 1.412 1.288 4. Table 4.19: Average relative involvement ratio (IR) for lower ranked schools under each program, for different types of accidents. Type S.N. number of schools under program each program Of Average IR ratio for angle- rear- angle head-on strait end turn left turn accd accd accd accident 1. Traditional 13 2.415 1.980 2.442 2.135 2. Range 22 2.953 2.339 2.625 2.278 3. Competency (Public) Competency (Commercial) 11 2.962 2.270 3.187 2.583 7 2.066 2.198 1.981 2.369 4. 68 - IR, for different schools and programs under various driving conditions. This exposure measure criterion variable, in addition to other traditional criterion variables was used for hypotheses testing in the following section. 4.7 Hypotheses Testing; Based on the hypotheses defined in the previous chapter, the hypotheses were tested using the ANOVA and student "t" test. These tests were performed by comparing the mean value of criterion variables for two or more than two groups by the use of "t" and ANOVA procedures respec­ tively. The ANOVA procedure was first applied to determine the significant difference in criterion variables among different groups. If this difference was found to be statistically significant, then the "t" test was used to determine exactly which two groups differ. All these hypotheses were tested at 95 percent level of confidence. Some of the results of ANOVA and "t" tests are shown in table 4.20, however the complete results are shown in appendix E. The result from the test for each hypothesis is discussed below: Hypothesis 1: HO: There was no difference in the mean accident rate among various driver education programs. 69 Table 4.20: The ANOVA and "t" results from hypotheses testing. Hypothesis number 1 a) 1 b) 2 3 3 4 5 5 5 a) b) a) b) c) 6 7 8 a) 8 b) 8 c) 9 a) 9 b) 9 c) 9 d) 10 10 10 11 11 11 12 12 12 12 13 13 13 14 14 14 15 15 15 15 16 16 17 17 18 18 a) b) c) a) b) c) a) b) c) d) a) b) c) a) b) c) a) b) c) d) a) b) a) b) a) b) Comparing groups Dependent variable F /"t" value accds/stud R, C, T & P accds/stud R, C, & T accds/stud ("t") 2 Ph. & 3 Ph • convs/stud R» C, T & P convs/stud R, c, & T convs/stud ("t") 2 Ph. & 3 Ph • sngl/stud ("t") P & T R & T sngl/stud ("t") C & T sngl/stud ("t") & P IR C, T R, IR 2 Ph. & 3 Ph • C, T & P IR (Detroit area) R, & IR (Urban area) C, T P R/ C, T & P IR (Rural area) R/ GLCl GLC2 & GLC3 IR (Range) GLC1 GLC2 & GLC3 IR (Comp., Pub.) GLCl GLC2 & GLC3 IR (Trad.) GLCl GLC2 & GLC3 IR (Comp., Comm.) IRCLR (Hig. grp.) R1 , Cl, T1 & PI R1 , Cl, T1 & PI IRRAN " IRSNW " R1 , Cl, T1 & PI IRDAY " R1 , Cl, T1 & PI IRDWN " R1 , Cl, T1 & PI R1 , Cl, T1 & PI IRNGT " R1 , Cl, T1 & PI IRRER " R 1 . Cl . T1 & PI IRATR " IRAST " R1 f Cl, T1 & PI R1 , Cl, T1 & PI IRHLT " R2 , C2, T2 & P2 IRCLR (Lwr. grp.) R2 , C2, T2 & P2 IRRAN " IRSNW " R2 , C2, T2 & P2 R2 , C2, T2 & P2 IRDAY " R2 , C2, T2 & P2 IRDWN " R2 , C2, T2 & P2 IRNGT " R2 , C2, T2 & P2 IRRER " R2 , C2, T2 & P2 IRATR " R2 , C2, T2 & P2 IRAST " R2 , C2, T2 & P2 IRHLT " R1 & R2 IRDAY R1 & R2 IRNGT IRDAY Cl & C2 Cl & C2 IRNGT IRDAY PI & P2 IRNGT PI & P2 14.41 0.44 0.92 52.11 1.02 2.23 2.24 2.76 1.15 1.06 0.11 0.71 1.54 0.48 2.42 0.83 0.05 3.17 0.49 0.69 0.07 1.86 1.51 0.20 1.25 0.55 0.99 0.78 2.01 0.48 1.08 0.49 1.21 0.24 0.35 0.49 1.08 0.13 24.58 9.91 17.91 30.91 13.29 4.89 PR > F/ "t" 0.0001 0.6458 0.3588 0.0001 0.1510 0.0263 0.0271 0.0061 0.2577 0.3651 0.7560 0.5500 0.2090 0.6950 0.0935 0.4398 0.9510 0.0654 0.6910 0.6910 0.9770 0.1483 0.2283 0.8960 0.3024 0.6485 0.4041 0.5122 0.1238 0.6967 0.3680 0.6871 0.3194 0.8714 0.7909 0.6939 0.3682 0.9440 0.0001 0.0030 0.0004 0.0001 0.0034 0.0472 Tabl2 4.20 (Continued on next page) 70 Table 4.20: Continued. Hypo- Comparing thesis groups number 19 a) 19 b) 20 a) 20 b) 20 c) 21 a) 21 b) 21 c) 22 a) 22 b) 22 c) 23 a) 23 b) 23 c) 24 a) 24 b) 24 c) 24 d) 25 a) 25 b) 25 c) 25 d) 26 a) 26 b) 26 c) 26 d) 27 a) 27 b' 27 c) 27 d) Tl Tl Rl Rl Rl Cl Cl Cl PI PI PI Tl Tl Tl Rl Rl Rl Rl Cl Cl Cl Cl PI PI PI PI Tl Tl Tl Tl & & & & & & & & & & & & & & & & & & & & & & & & & & & T2 T2 R2 R2 R2 C2 C2 C2 P2 P2 P2 T2 T2 T2 R2 R2 R2 R2 C2 C2 C2 C2 P2 P2 P2 P2 T2 fc T2 & T2 & T2 Dependent variable IRDAY IRNGT IRCLR IRRAN IRSNW IRCLR IRRAN IRSNW IRCLR IRRAN IRSNW IRCLR IRRAN IRSNW IRRER IRAST IRATR IRHLT IRRER IRAST IRATR IRHLT IRRER IRAST IRATR IRHLT IRRER IRAST IRATR IRHLT F /"t" value 50.77 110.91 98.25 21.99 5.88 9.20 11.99 12.13 61.04 3.88 5.75 13.90 17.69 10.91 25.88 5.88 5.63 4.09 4.89 25.08 12.76 2.85 15.35 15.08 3.49 2.90 26.48 29 ,27 23.36 16.41 PR > F/ "t" 0.0001 0.0001 0.0001 0.0001 0.0204 0.0066 0.0025 0.0028 0.0001 0.0732 0.0356 0.0011 0.0003 0.0034 0.0001 0.0204 0.0234 0.0504 0.0441 0.0001 0.0034 0.0810 0.0020 0.0022 0.0914 0.1140 0.0001 0,0001 0.0002 0.0009 T = traditional, R = range, C = competency (pub.), P = compe­ tency (comm.), Tl, Rl, Cl, PI = higher ranked schools using traditional, range, competency (pub.& comm.) program, T2, R2, C2, P2 = lower ranked schools using traditional, range, competency (pub.& comm.) program, GLCl=Detroit, GLC2 & GLC3 = urban & rural areas, IRCLR, IRRAN & IRSNW = IR value under clear, rainy and snowy weather conditions, IRDAY, IRNGT & IRDWN = IR value under day, night & dawn/dusk time light conditions, IRRER, IRAST, IRATR & IRHLT = IR value for rear-end, angle-straight, angle-turn & head-onleft turn accident respectively. 71 The results from this test (F=14.4) showed that this hypothesis (hypothesis 1(a) in table 4.20) can be rejected at the 95 percent level of confidence. However, no statistically significant difference (F=0.44) was found among all the three public school driving programs (hypothesis 1(b)) as can be seen in table 4.20. The mean accident rate for the competency program in commercial schools was found to be significantly different (higher) than the mean accident rates for public school. The average number of accidents per student for the commercial school competency program was 0.281, compared to 0.219, 0.227 and 0.228 for the public school range, competency and traditional program respectively. Hypothesis 2: HO: There was no difference in the mean accident rate between the three phase program (range) and the two p h a a e pjTCCjjTciiu ( u i a u i t i O n a l a n u CGIupctcnCy} xii p u b l i C schools. The results from this test ("t" = 0.92) showed that this hypothesis can not be rejected at the 95 percent level of confidence. And it can be concluded that there was no difference in the mean accident rate between the two and the three phase program in public schools. 72 Hypothesis 3: HO: There was no difference in the mean conviction rate among various driver education programs. The test for this hypothesis (F=52.1) (hypothesis 3(a) in table 4.20) led to the rejection of the hypothesis. There was no statistically significant difference (F=1.02) in the mean conviction rate among various programs in public schools (hypothesis 3 (b)). The mean conviction rate for the competency (commercial schools) program was signi­ ficantly higher than the rest of the programs. The number of convictions per student for the commercial (competency) schools was 0.448, compared to 0.284, 0.262 and 0.274 for the range, competency (public) and traditional program respectively. Hypothesis 4; UA tl V • I ^ T.r-»e* n w a « 4W M ^ ^ W< b W 4 4 W W i n f m oan PA nT ri p f i n n W A W y*af a ^ M between the three phase program (range) and the two phase program (traditional and competency) in public schools. The test for this hypothesis led to the rejection of the hypothesis. The mean conviction rate for the three phase program was significantly higher than the two phase program in public schools. Hypothesis 5; HO: There was no difference in the mean single vehicle accident rate among various driver education programs. The result of this test showed that this hypothesis can be rejected. Further tests showed that the mean single vehicle accident rate for the traditional program was significantly different (greater) than the mean single vehicle accident rate for the range and competency (commercial) programs. The average number of single vehicle accidents per student from the traditional program was 0.061 compared to 0.048, 0.054 and 0.50 for the range, competency (public) and competency (commercial) program respectively. Hypothesis 6: HO: There was no difference in the mean IR value among various driver education programs. The test showed that this hypothesis can not be rejected and concluded that there was no difference in the mean IR value among various education programs. Hypothesis 7: HO: There was no difference in the mean IR value between the three phase program ( range ) and the two phase program ( traditional and competency ) in public 74 schools. The test showed that this hypothesis can not be rejected and concluded that there was no difference in the mean IR value between the two and three phase program in public schools. Hypothesis 8; HO: There was no difference in the mean IR value among various driver education programs in each of the three geographical areas - (a) Detroit metropolitan area, (b) urban areas and (c) rural areas. The test showed that this hypothesis can not be rejected and concluded that there was no difference in the mean IR value among various education programs in each of the three geographical areas. Hypothesis 9: HO: There was no difference in the mean IR value for each driver education program across the three different geographical areas - Detroit metropolitan area, urban areas and rural areas. The test showed that this hypothesis can not be rejected and concluded that there was no difference mean IR value for each of the four driver education in the 75 programs across the three different geographical areas. Hypothesis 10; HO: There was no difference in the mean IR value among various driver education programs for higher ranked schools under (a) clear, (b) rainy and (c) snowy weather conditions. The result of this test indicates that there is no significant difference in the mean IR value among various driver education programs at the higher ranked schools under clear, rainy and snowy weather conditions. Hypothesis 11: HO: There was no difference in the mean IR value among various driver education programs for higher ranked schools under (a) day, (b) dawn/dusk and (c) night time light conditions. The test result shows that there was no significant difference in the mean IR value among various driver education programs under day, dawn/dusk and night time light conditions. Hypothesis 12: HO: There was no difference in the mean IR value among different driver education programs for higher ranked 76 schools for the following types of accidents (a) rearend (b) angle turn (c) angle straight and (d) headon-left turn. The test result shows that there was no significant difference in the mean IR value among various driver education programs at the higher ranked schools for rearend, angle-turn, angle straight and head-on-left turn accidents. Hypothesis 13; HO: There was no difference in the mean IR value among various driver education programs for lower ranked schools under (a) clear, (b) raining and (c) snowing weather conditions. The results of this test indicate that there is no significant difference in the mean IR value among various driver education programs for lower ranked schools under clear, rainy and snowy weather conditions. Hypothesis 14; HO: There was no difference in the mean IR value among various driver education programs for lower ranked schools under (a) day, (b) dawn/dusk, and (c) night light conditions. 77 Based on the test results, it can be concluded that there was no significant difference in the mean IR value among various programs under all the three light conditions. Hypothesis 15; HO: There was no difference in the mean IR value among different driver education programs for lower ranked schools for the following types of accidents (a) rear -end (b) angle turn (c) angle straight and (d) headon-left turn. The test results showed that the above null hypothesis can not be rejected, as no significant differences in the mean IR value was found for the four different types of accidents among the various programs. Hypothesis 16: HO: There was no difference in the mean IR value between higher and lower ranked schools under the range program for the three different light conditions (a) day (b) dawn/dusk and (c) night. The results from this test showed that there was a significant difference in the mean IR value for day, dawn/ dusk and night time light conditions, and no difference in the dawn/dusk light condition, between higher and lower 78 ranked schools under the range program. Hypothesis 17; HO: There was no difference in the mean IR value between higher and lower ranked schools under the competancy (public) program, for three different light conditions - (a) day (b) dawn/dusk and (c) night time. The results from this test showed that there was a significant difference in the mean IR value for day and night light conditions and no significant difference in the case of dawn/dusk light conditions, between higher and lower ranked schools under the competency (public) program. Hypothesis 18: HO: There was no difference in the mean IR value between higher and lower ranked schools under the competancy (commercial) program for three different light conditions - (a) day (b) dawn/dusk and (c) night. The results from this test showed that there was a significant difference in the mean IR value for day, dawn/ dusk and night time light conditions between higher and lower ranked schools under the competancy (comm.) program. Hypothesis 19: HO: There was no difference in the mean IR value between 79 higher and lower ranked schools under the traditional program for the three different light conditions (a) day (b) dawn/dusk and (c) night. The result from this test showed that there was a significant difference in the mean IR value for day, night and dawn/dusk time light conditions between higher and lower ranked schools under the traditional program. Hypothesis 20: HO: There was no difference in the mean IR value between higher and lower ranked schools under the range program for three different weather conditions (a) clear (b) raining and (c) snowing conditions. The results from this test showed that there was a significant difference in the mean IR value for all the three weather conditions between higher and lower ranked schools under the range program. Hypothesis 21; HO: There was no difference in the mean IR value between higher and lower ranked schools under the competency (public) program for the three different weather conditions - (a) clear (b) raining and (c) snowing conditions. 80 The results from this test showed that there was a significant difference in the mean IR value for all the three weather conditions between higher and lower ranked schools under the competency (public) program. Hypothesis 22: HO: There was no difference in the mean IR value between higher and lower ranked schools under the competency (commercial) program, for the three different weather conditions - (a) clear (b) raining and (c) snowing conditions. The test showed that there was no difference in the mean IR value for raining conditions but there was a significant difference in the clear and snowing weather conditions between higher and lower ranked schools under the competency (commercial) program. Hypothesis 23: HO: There was no difference in the mean IR value between higher and lower schools under the traditional program for the three different weather conditions (a) clear (b) raining and (c) snowing conditions. The test showed that there was a significant difference in the mean IR value for all the three weather conditions between higher and lower ranked schools under the 81 traditional program. Hypothesis 24; HO: There was no difference in the mean IR value between higher and lower ranked schools under the range program for four different types of accidents (a) rear-end (b) angle-straight (c) angle turn and (d) head-on-left turn accident. The results showed that there was a significant difference in the mean IR value for all the four types of accident between higher and lower ranked schools under the range program. Hypothesis 25: HO: There was no difference in the mean IR value between higher and lower ranked schools under the competency fmibliel/ orooram for four different tvoes of * -» «» accidents - (a) rear-end (b) angle-straight (c) angle-turn and (d) head-on-left turn accident. The results showed that there was a significant difference in the mean IR value for rear-end, anglestraight and angle turn accidents between higher and lower ranked schools under the competency (public) program. There was no significant difference IR values for head-on-left turn accidents for the same groups. 82 Hypothesis 26: HO: There was no difference in the mean IR value between higher and lower ranked schools under the competency (commercial) program, for four different types of accidents - (a) rear-end (b) angle-straight (c) angle turn and (d) head-on-left turn accident. The resuls showed that there was a significant difference in the mean IR value for rear-end and anglestraight accidents and no difference in the case of angleturn accidents and head-on-left turn accidents for the same groups of schools under the competancy (comm.) program. Hypothesis 27: HO: There was no difference in the mean IR value between higher and lower ranked schools under the traditional program for four different types of accidents (a) rear-end (b) angle straight (c) angle turn and (d) head-on-left turn. The results showed that there was a significant difference in the mean IR value for rear-end, angle turn, angle-straight and head-on-left turn accidents between higher and lower ranked schools under the traditional program. Results from these tests of hypotheses are concluded 83 as follows: The competency program in commercial schools had significantly higher accident and conviction rates than the range, traditional, and competency programs in public schools. There was no statistically significant difference in the mean accident rate between the 3-phase range program and the 2 -phase traditional and competency programs in public schools. However, the range program had a significantly higher conviction rate than the twophase traditional and competency programs in public schools. There was no statistically significant difference in the mean IR value among all four programs including two-phase and three-phase programs. There was no difference in the performance of drivers from different driving education programs due to different geographical areas (i.e. different driver exposure), as no statis­ tically significant difference was found in the mean IR value (i) among all the programs in each geographical area and (ii) for each program under three different geographical areas. There was no statistically significant difference in the mean IR value among all four programs under different weather and light conditions for both sample of higher and lower ranked (according to IR criterion) schools. However, a significant difference was found in the mean IR value, for different weather and light conditions, between higher 84 and lower ranked schools under each program. No significant difference was found in the mean IR value for all the four accident types among all programs for both higher and lower ranked schools. These results indicate that, when corrected for exposure, there is no difference in the accident patterns for young drivers based on the type of driver training program they attend. This conclusion extends to type of accidents, weather conditions and light conditions as well as the total accident experience. However, there are significant differences in performance across various schools within any program type. This may mean that it is the instructor, rather than the mode of instruction, that determines driver performance. 4.8 Rating Score; The above analysis described the performance of schools and programs based on the frequency ofaccidents. It may not be appropriate to rate a school better which had a fewer number of accidents with a high degree of severity than another school which had a higher number of accidents but with lesser degree of severity. To determinethe performance of schools and programs onthe basis of accident frequency and severity, a rating score was derived for each school. This score is the sum of the products of the frequency and weights for each type of accident. The weight was taken as the equivalent dollar 85 value of each type of accident, determined by summing the product of the percentage of fatal, injury, and property damage in accidents by the corresponding average dollar values of fatal, injury, and PDO accidents. The weights obtained for each type of accident is shown in appendix F. The score per student for each school is also shown in appendix F. Based on ascending values of score/student for each school, fifteen higher and fifteen lower ranked schools are shown in table 4.21. The score for each program is shown in table 4.22. These values were obtained by taking the average score/student for all schools under each program. According to this measure, the range program had the best performance in terms of a criterion which combines the number and severity of accidents. Nine of the fifteen high ranked schools come from those using the range program. In contrast, none of the commercial schools ranked in this group. Four commercial schools were ranked in the lowest 15 schools, where there were also four schools from the range program. The public (competency) and traditional program had about equal representation in both the highest and lowest rated schools. Similarly, using frequency and offense points associated with each type of conviction as weights, a conviction rating score for all schools was obtained. The ranking of schools, based on this score, is shown in 86 Table 4.21: Rating of schools by accident frequency and severity criterion. Rank School Code 763 790 740 623 785 795 462 226 260 736 042 420 198 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 020 035 Types of Program T C R R R R R C T R R R T R T Score per student Rank 6.20 6.67 7.07 7.10 7.46 7.56 7.76 7.84 7.98 8.01 8.22 8.22 8.38 8.42 8.71 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 School Types Code of Program A62 471 059 545 616 506 450 A48 416 538 A24 544 413 959 044 P R R T R T T P T C P R C P S Score per studenl 16.94 16.97 17.00 17.02 17.13 17.14 17.32 17.36 17.74 17.76 17.79 18.50 18.70 21.41 33.01 T = traditional, R = range, C = competency (pub.), P = competency (comm.), S = simulator and F = 4-phase Table 4.22: Rating of programs by accident frequency and severity criterion. Rank Program Score/student 1 Range 11.790 2 Traditional 12.206 3 Competency (Public) 12.500 4 Four-phase 12.791 5. Competency (Commercial) 14.563 6. Simulation 17.111 87 appendix F. Fifteen higher and lower ranked schools are shown in table 4.23. The score for each program is shown in table 4.24 which indicates that the competency (pub.) program had the best performance in average conviction points per student. Based on points, the public school competency and traditional program had three and six schools in the highest fifteen and no school in the lowest fifteen schools. Conversly, the commercial schools had no representation in the highest fifteen schools while ten of the fifteen lowest rated schools come from this group. The consistency of schools in their performance on various criterion variables - IR value, accidents/student and score/student, were also investigated for the two year period. All schools were classified into two groups for each year according to each criterion variable. The first group constitutes those schools whose criterion variable value is lower than the average value. The second group includes those schools whose criterion variable value is higher than the average value. The first and the second groups are termed as higher and lower ranked groups according to a particular criterion variable. The number of schools which appeared in each group for both years by each criterion variable is shown below: (1) The number ofschools which appeared in both year above average ranked group according to IR 88 Table 4.23: Rating of schools by conviction frequency and seriousness criterion. Rank School Code 182 256 457 442 187 530 411 367 421 446 042 169 177 675 188 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Types Score of per Program student c R T C T R T T R C R R R T T 0.216 0.287 0.307 0.317 0.328 0.334 0.335 0.335 0.338 0.339 0.347 0.347 0.370 0.377 0.381 Rank School Types Score Code of per Program student 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 966 039 A48 316 951 A10 A09 A62 509 317 A 88 975 A04 959 044 P S P R P P P P R R P P P P S 0.943 0.969 0.983 0.991 1.028 1.039 1.052 1.084 1.113 1.134 1.374 1.401 1.454 1.478 1.641 T = traditional, R = range, C = competency (pub. ), P = competency (comm.), S = simulator and F = 4-■phase Table 4.24: Rating of programs by conviction frequency and seriousness criterion. Rank Program Score/student 1 Competency (Public) 0.538 2 Traditional 0.565 3 Range 0.575 4 Four-phase 0.714 5. Competency (Commercial) 0.886 6. Simulation 0.915 89 criterion = 130 The number of schools which appeared in both year below average ranked group according to IR criterion = 41 Thus 130 and 41 of a total of 290 schools (i.e. 45 and 14 percent of the schools) were consistent for two conse­ cutive years in acheiving above or below average rank respectively on a criterion of IR value. These schools are shown in appendix F. (2) The number of schools which appeared in both year above average ranked group according to accidents per student criterion = 100 The number of schools which appeared in both year below average ranked group according to accidents per student criterion = 98 According to accidents/student criterion 34 and 33 percent of total schools were consistent. These schools are also shown in appendix F. (3) The number of schools which appeared in both year above average ranked group according to score per per student criterion = 90 The number of schools which appeared in both year below average ranked group according to score per student criterion According to score/student criterion 31 and 30 = 88 90 percent of the total schools were consistent. These schools are also shown in appendix F. Based on all the three criterion variables, a very poor consistency was found among schools in both the groups. Only fourteen and five schools were found to be above average or below average for two consecutive years according to all three criterion variables. 4.9 Model Development: Models were developed to estimate the performance of various driving schools and programs in terms of the rating score. The dependent variable (1989 score per student) (SCR/STU)2 for each school was regressed against the 1988 values of various independent variables for the same schools. The independent variables used in developing models were; number of students in each school for year 1989 (NSTU)2 , number of accidents (NACC)1, number of convictions (NCNV),, relative accident involvement JL ratio (IR)1 , accidents per student (ACC/STU)^ convictions per student (CNV/STU)^ year 1988 rating score per student (SCR/STU)^, dummy variables for the four types of program (PRGM) - range, traditional , competency in public schools and competency in commercial schools and for the three different types of geographical location of schools (GLOC) - Detroit metropolitan area, urban area and rural area. The variable GLOC was considered because the driving 91 pattern differs from place to place. Drivers in the Detroit metropolitan area encounter a higher density of traffic as compared to drivers in the rural or small urban area. Exposure of different driving conditions also affects the driving performance. In order to take into account or reduce the variation in different driving patterns due to different driving locations, models were developed separately for each type of geographical location, as well as for different types of programs as described below. Models were calibrated for the following cases: 1. For each combination of type of program and geographical location of school. 2. For each type of program irrespective of geographical location. 3. For each geographical location of school irrespective of type of program. A •I « i ^ 1 1 U fkX T> /omm \ R — 2 2 = *J - . n*7 W . J-. n“ • - 0.18 and e.* / CPT5 / CfTTT\ “ V W W . . / W * W / ^ p = .09 (ii) (SCR/STU)2 = 1.07-2.19* (SCR/ STU^+SS. 3* (ACC/STU) + 1.29* (IR^ R2 = 0.57 and p = .0002 The simple regression model, like the above cases, shows an inconsistent relationship between the two years data. The multiple regression model shows a moderate 98 correlation. The dependent variable is positively related with accident rate and the exposure variable (IR) and negatively related with score/student for the year 1988. The simple regression model shows a positive relationship in score/student between the two years, which indicates a multi-collinearity problem between accident rate and score per student. When the variable score/student was dropped out of the model, the explanatory power of the model 2 reduces to 40% (R =.40). Competency (i) (commercial) (SCR/STU)2 = R 2 = (ii) (SCR/STU)2 = R2 = and Urban area: (Sample size = 15) 1.87 + 0.522*(SCR/STU) 0.13 and p = .08 1.10 + 16.81*(CNV/STUD)^ 0.35 and p = .011 The model in this case shows a relationship between score/student and conviction rate. This leads to the interpretation that schools with a high conviction rate tend to have a bad performance in the year after. Competency (commercial) and Rural area: (Sample size = No model was calibrated due to the very small sample. 4) 99 It can be observed that most of these models were both insignificant at a high level of confidence and very poor in explaining the variance (R ) of the dependent variable (score per student) except for the models for range and competency programs in the Detroit metropolitan area. In . . . these cases, the models were significant with R 2 values varying from .44 to .57. However, there are some concerns regarding the sign of parameters. The conclusion from this set of models is that there is no consistent relationship between the score/student for 1989 and the selected set of independent variables for 1988. Schools with a bad performance in 1988 could have either a better, the same or a worse performance in the following year. Case 2: Models for different geographical location Detroit metropolitan area; (Sample size = m • rscR/sTin„ • = 94) i.6i + o.65* (s c r /s t u ), ■ ^ R2 = 0.40 and p = .001 (ii) same as the above. The regression model for schools in the Detroit metropolitan area, irrespective of their types of program, 2 show a moderate relationship (R =.40) m score per student between the year 1988 and 1989. This relationship indicates that schools with low/high score per student 100 tends to have the same pattern in the second year. Urban area: (i) (Sample size = 96) (SCR/STU)2 = 2.99 + 0.24*(SCR/STU) R (ii) 2 = (SCR/STU)2 = R2 = 0.06 and p = .01 3.01+ 6.15*(CNV/STU)1 0.13 and p = .002 The regression model calibrated in this category had a low correlation. Rural area: (i) (Sample size = 89) (SCR/STU)2 = 3.07 + 0.17*(SCR/STU)x R (ii) 2 = 0.06 and p = .01 (SCR/STU)2 = 3.22 + 6.12*(CNV/STU) ± - 0.17*(IR)1 R2 = 0.10 and p The multiple regression model for = .009 schools in rural areas show a weak correlation. Except for the Detroit metropolitan area, models by areas explain only about 10% of the total variance. Even the model for the Detroit metropolitan area was only moderately successful as its R 2 value is 0.40. 101 Case 3: Models for different types of program Range: (Sample size = (i) 124) (SCR/STU)2 = 2.26 + 0.40*(SCR/STU)± R2 = 0.13 and p = .0001 (ii) same as the above. The only significant variable which entered the regression model for schools using the range program was score per student for the year 1988. The model shows a poor consistency in the overall performance of schools for the two year period. Competency (public!: (Sample size = (i) 57) (SCR/STU)2 = 2.93 + 0.24*(SCR/STU) R 2 = (ii) (SCR/STU)2 = R2 = 0.09 and p = .02 3.7 - 0.005*(STU)2+0.05*(NACC) 0.14 and p = .01 The regression model for all schools under the competency program show the same independent variables with the same parameter sign as in the case of the competency program in the Detroit metropolitan area. 2 However, the explanatory power is quite low (R =.11) in 102 2 this case as compared to the explanatory power (R =.44) in the Detroit case. This shows a wide variation in the performance of public schools under competency program in areas other than the Detroit metropolitan area. Competency (commercial): (Sample size = (i) 35) (SCR/STU)2 = 1.53 + 0.427*(SCR/STU) R2 = 0.09 and p = .12 No significant model was obtained . Traditional: (Sample size = (i) 63) (SCR/STU)2 = 2.63 + 0.27*(SCR/STU)^ R2 = fiil fSCR/STU)„ = C R 2 = 0.07 and p = .02 3.26 - 6.4*(NCNV),-0.255*(IR), a. 0.16 and m p = .005 The regression model for schools under the traditional program shows a negative relationship with the number of convictions and IR with a very low R 2 value of .16. All the models for different types of programs were both statistically insignificant and had a very low R value. 2 103 Case 4; Over all model for all types of program and locations (Sample size = (i) 290) (SCR/STU)2 = 2.48 + 0.37*(SCR/STU) R2 = 0.13 and p = .001 (ii) (SCR/STU)2 = 2.22 + 8 .5*(ACC/STU)±+ 3 .72*(CNV/STU) R2 = 0.17 and p = .0001 Although the overall model is statistically significant, (at 99% level of confidence) it has a very low R 2 value of .17. The significant variables which entered into the model are accident and conviction rate. These variables had a positive relationship with score/student for year 1989 which indicates schools with high or low accident and conviction rate tends to have the same pattern in the second year. Based on these analyses it can be concluded that except for the models for the range and competency programs in the Detroit metropolitan area, all the models were statistically not significant (at 90% level of confidence) and had very low R 2 values, and thus can not be used for prediction purposes. There is more consistency in the performance of schools in the Detroit metropolitan area as compared to schools in other areas. The regression models for the range and competency programs in the Detroit metropolitan areas were statistically significant, but had only a moderate R value. These models can not be used 104 confidently for prediction purpose. 4.10 Discriminant analysis: Discriminant analysis was performed to determine if it was possible to classify schools into different driver education programs, based on a discriminant function derived from a set of predictor variables. A comparision between the actual classification of schools under each program and the predicted classification would determine how successfully schools can be discriminated into different programs based on these performance variables. A small difference between criterion groups with respect to predictor variables results in more error in classification in discriminant analysis. Any relationship between types of programs and a set of performance predictor variables could be identified based on this analysis. Discriminating functions were developed, using type of program as a classification variable, and IR, score per student, accidents/student and convictions/student for the year 1988 as the four predictor variables . Based on discriminant functions, the classification of schools into different programs was predicted as shown in table 4.25. The same table shows that predictions of schools under the range program are correct in 33 out of 124 schools (26.6%); predictions of commercial schools using the competency program are correct in 28 out of 35 schools 105 (80%). The higher correct classification for competency (comm.) program indicates that commercial schools were very successfully discriminated based on a set of predictor variables. The first discriminant function (which separates commmercial schools from the rest of the schools) was highly significant. Whereas the other two discriminant functions were not significant even at 80% level of confidence. This results in a high prediction error. In total, the correct classification of schools are 111 out of 279, for an overall percentage of 40% correct classification. Based on the developed discriminating function and using another set of predictor variables (IR, accidents/ student, score/student and convictions/student) for the year 1989, the predicted classification of schools into the different programs is shown in table 4.26. It can be seen from table 4.26, that, except for schools under the competency programs, the percentage of correct classification for all other types of program decreased. For the commercial schools the percentage of correct classification is 82%, an increase of 2% from the 1988 set of data. The overall percentage of correct classification for all schools is 35%. Using three categories of the classification variable - 1 ) range, 2 ) traditional and competency program combined and 3) competency program in commercial schools, a new discriminating function was developed. And the resulting 106 Table 4.25: Classification summary for year 1988 data. From Traditional Tradi­ tional Classified into CompeCompe­ tency tency (pub.) (comm.) Range Total 24 38.10% 18 28.57% 12.70% 13 20.63% Competency (pub.) 17 29.82 26 45.61 3 5.26 11 57 19.30 100.00 Competency (comm.) 1 2.86 1 2.86 28 80.00 5 14.29 100.00 35 28.23 35 28.23 21 16.94 33 26.61 100.00 77 27.60 80 28.67 60 21.51 62 279 22.22 100.00 Range Total Percent 8 63 100 .00 % 35 124 Table 4.26: Classification summary for year 1989 data. From Traditional Competency (pub.) Competency (comm.) Range Total Percent Tradi­ tional Classified into CompeCompe­ tency tency (pub.) (comm.) Range Total 17 26.98% 34.92% 13 20.63% 17.45% 14 24.56 24 42.11 9 15.79 10 57 17.54 100.00 2 1 2.86 29 82.86 3 8.57 100.00 27 21.77 22 29 23.39 5.71 46 37.10 79 28.32 22 74 26.52 17.74 13 26.16 11 53 19.00 63 100 .00 % 35 124 100.00 279 100.00 107 classification shows a 50% accuracy in overall classification as compared to 40% correct classification in the case of a classification variable having four categories. This indicates that there is more difference among these three types of program with respect to the predictor variables. Classification of schools were also predicted using each of the four predictor variables separately as well as a combination of two and three predictor variables together. Discriminant functions were also developed using a stepwise procedure to select the predictor variables for the model. This caused only two variables (IR, accident rate) to enter the model. All these classifications show a higher percentage of error of classification as compared to the classification using all four predictor variables together. The discriminant function analysis shows a 60 to 65 percent error of classification. Whereas, the high percentage of correct classification for commercial schools supports the earlier findings that the competency program in commercial schools had a significantly higher accident and conviction rate. The high percentage of error results from a small difference among types of program based on these predictor variables. This indicates that there is not a good relationship between types of program and these set of performance variables. CHAPTER 5 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS 5.1 Summary; Based on the fact that the frequency of accidents among young drivers remains high despite the wide spread use of driver education programs, such programs have come under attack. The critics charge driver education programs are inefficient and cost-ineffective. Prior research on the effectiveness of various driver education programs was not conclusive. Many researchers cast doubt upon the earlier studies because they lacked a valid measure of accident exposure. In light of this concern, the present study was conducted utilizing an indirect accident exposure measure in the analysis of the effectiveness of different driver education programs. This indirect accident exposure method, called the quasi­ induced accident exposure method, is based on the assumption that accident exposure by any group of drivers is proportional to the innocent victim involvements in multi-vehicle accident by that group of drivers. The criterion variable used in this method, is called the relative accident involvement ratio (IR). The IR is a measure of the relative frequency of accident involvement for drivers from different driver education programs. 108 109 A new data base was created by extracting information from three existing data files - The Highway Accident Master File, Driver Accident and Conviction Records File, and Driver Education Program and Information File. In addition to the relative involvement ratio (IR), other traditional criterion variables such as accident frequency per student and conviction frequency per student were computed for each school and program. In order to take into account the difference in driver exposure from different geographical areas the IR value was also computed for various programs under three different geographical areas - Detroit metropolitan area, other urban areas and rural areas. To determine the performance of drivers from different schools or different programs under different driving conditions, the relative involvement ratio for a sample of higher and lower ranked schools ( according to the IR criterion) under each program was determined for different weather and light conditions. Hypotheses were constructed to determine if there were statistically significant differences in the mean relative involvement ratio, and the mean rate of accidents and convictions, among various driver education programs and schools. Analyses were also performed comparing the performance of various programs and schools under different weather and light conditions. To determine the rating of different schools and 110 programs, on the basis of both frequency and severity of accidents, a rating score was determined for all schools and programs. This score is a summation of the product of the frequency and weight of each type of accident, where the weight for each type of accident equaled the average dollar value of fatal, injury and property damage accidents. The consistency of schools in their performance on various criterion variables was also investigated over a period of two years. Regression models were calibrated to predict the crash performance of individual schools under (i) four different programs (range, competency (public), competency (commercial), traditional) (ii) three different geographical locations of schools (Detroit metropolitan area, urban area, and rural area), and (iii) combinations of each program and geographical location. Finally, discriminant analysis was performed to determine how successfully schools can be discriminated into different programs based on a discriminant function derived from a set of predictor variables. This analysis would further determine whether a relationship between types of program and a set of predictor variables exists. 5.2 Conclusions: Based on the analyses presented in the previous chapters, the following conclusions were drawn: The 16, 17 and 18 year old drivers were about 11 and Ill 12.5 percent more involved in accidents and convictions respectively than the drivers of all age group state-wide for the two year period (1988 and 1989). These numbers are based on the frequency of accidents and convictions, not adjusted for exposure. The most common type of accident and conviction were rear-end accidents and speed related violations. The competency program in commercial schools had significantly higher accident and conviction rates than the range, traditional and competency programs in public schools. There was no statistically significant difference in the mean accident rate between the range program (3phase) and traditional and competency programs (2-phase) in public schools. However, the range program (3-phase) had a significantly higher conviction rate than 2 -phase competency and traditional programs in public schools. The mean single-vehicle accident rate was significantly higher for atudcnts enrolled in uhe traditional program than the students enrolled in the range program. The average IR value indicates that drivers from all programs were over-involved in multi-vehicle accidents. There was no statistically significant difference in the mean relative involvement ratio (IR) among the four programs including two-phase and three-phase program, when the induced exposure measure of accidents was utilized in the analysis. There was no difference in the performance of drivers from different driving education programs due 112 to different geographical areas (i.e. different driving enviornment), as no statistically significant difference was found in the mean IR value (i) among all programs in each geographical area and (ii) for each program under three different geographical areas. No statistically significant difference in the mean relative involvement ratio was found among different programs for samples of both higher and lower ranked schools under clear, rainy and snowy weather conditions. The traditional program results in the lowest and second lowest IR value for higher and lower ranked schools respectively under rainy and snowy conditions. However, the IR value was very high in snowy weather across all programs which indicates that none of the driver education program prepares students to drive under adverse weather conditions. Under all three light conditions - day, night, and J n * tw / J uawit/uu M a 0 m 4 0mmm 4 4 2 m 0mn 1. ; iiw 01^1111 xwaitu *9 «! £ £ a w . m m 0ma 0m ui.i.1ci c u wc o * • 0 mmm0 m wcic £ 0 m. « m #9 xwuin* «! m xii ^ 0 * . unc mean IR value among all programs for both higher and lower ranked schools. The competency program in commercial schools results in the highest and lowest IR value for higher and lower ranked schools respectively under night time accidents. Moreover, the data indicates that young drivers are about 20 % more likely to be over­ involved in a night accident than a day accident (after correcting for exposure). The ratio is even higher among the lower ranked schools with high values of the IR shown 113 by all public school programs. These ratios show that young drivers from these schools are involved in between 2 and 3 accidents as the guilty party for each accident in which they are the innocent victim. No statistically significant difference was found in the mean IR value among the four programs for all four accident types - angle-straight, rear-end, angle turn and head-on-left turn accidents. However, an interesting finding was that the traditional program had a lower IR value for all four accident types than the range program. This indicates that at least for the four common accident types, the additional driving range experience did not result in a lower accident experience (after counting for exposure). Based on a scoring system developed on the combined criterion of frequency and severity of accidents, the range program was found to have the best performance. rrU ^ i r* 4- 9 ^ W-. c* ww**ww.*. ^w-. rsv w ■ » « / * » « w-.«.w-*ww /m WWW-- 9 w . *♦“ t.7/-n w.,w year period indicates that schools in the higher ranked groups are more consistent than schools in the lower ranked groups. In the higher ranked group, schools were more consistent in their performance using the IR criterion variable than on any other criterion variable. Whereas in the lower ranked group, schools were most consistent based on the accident rate criterion variable. All the regression models, except for the range and competency (comm.) programs in the Detroit metropolitan 114 area, were either statistically not significant or had very poor explanatory power. This shows that there is more consistency in the performance of schools in the Detroit metropolitan area as compared to schools in other areas. The models for the range and competency (comm.) programs in the Detroit metropolitan area were statistically significant but still did not have a high explanatory power, so even these models can not be used for predicting school performance. Using the discriminant analysis, it was found that only 40% of total schools can be correctly classified into their programs based on the four predictor variables - IR, accidents/student, score/student and convictions/student. This shows a small differences among the programs with respect to the predictor performance variables. Thus, it can be concluded that there was not a good relationship between types of program and their performances. As an overall conclusion, there is no evidence of significant difference among public school driver education programs based on the performance predictor variables used in this study. The commercial school program did have a significantly higher accident and conviction frequency per student than the public school programs. However, when corrected for experience, this difference was no longer statistically significant. 115 5.3 Recommendations: From the above conclusions, the following recommendations can be made: First, the certification requirements which are imposed on commercial driving schools should be scrutinized to determine whether they are effective in ensuring quality driver education. Second, school districts which currently use the two-phase programs should not seek to enhance their programs by investing in simulators or driving ranges. Schools districts which currently use three-phase or fourphase programs should, in light of the maintenance costs, consider implementing a two-phase program instead. Third, public and commercial schools should enhance their curriculum in order to prepare students to drive under adverse weather conditions, and the public schools should provide better training for night driving. The above conclusions were based on only a two year analysis period. This period is very short for a valid computation of driver performance. In order to draw a a more reliable conclusion, a longer evaluation period is recommended. A follow-up-study is recommended with a longer evaluation period to verify the findings of this study. REFERENCES 1. McFarland, R. A. Psychological and Psychiatric Aspects of Highway Safety. Journal of the American Medical Association, vol. 163, no. 4, pp. 233-237, 1957. 2. Beamish, J. J. and Malfetti, J. L. A Psychological Comparision of Violator and Non-violator Automobile Drivers in the 16 to 19 Year Age Group". Traffic Safety Research Review, vol. 6 , no. 1, pp. 12-15, 1962. 3. Ockert, G. L. Assessment of the Impact of the 1983 Minor's Restricted Driver's License Law Change in Texas. A Ph. D. Dissertation , Texas A & M University, May 1983. 4. National Safety Council. Accident Facts. Chicago, Illinois. 1984. 5. Karpf, R. S. and Williams, A. F. Teenage Drivers and Motor Vehicle Deaths. Accident Analysis and Prevention, vol. 15, no. 1, pp. 55-63, 1983. 6 . Wuerdeman, H., Belew, W. W. et al. Drivers in Fatal 116 117 Crashes With or Wihout Driver Training. National Highway Traffic Safety Administration, U.S. Department of Transportation, 1976, Washington, D. C. 7. Highway Statistics for 1985. Federal Highway Administration, U. S. Department of Transportation, Washington, D.C., 1986. 8 . American Automobile Association. Teaching Driver and Traffic Safety Education. New York: McGraw-Hill Book Company, 1965. 9. Shettel, H. H., and Schumacher, S. P. Driver Training Simulators, Ranges and Modified Cars. American institutes for Research, Pittsburgh, July, 1971. 10. Stock, J. R., Weaver, J. K., et al. Evaluation of Safe Performance Secondary School Driver Education Curriculum Demonstration Project. National Highway Traffic Safety Administration, 1983, U.S. Department of Transportation, Washington, D. C. 11. Lund, A. K., Williams, A. F. and Zador, P. High School Driver Education: Further Evaluation of the Dekalb County Study. Accident Analysis and Prevention, vol. 18, no. 4, pp. 349-357, 1986. 118 12. O'Leary, P. J. Report on the effectiveness of current Driver Education Program. Michigan Department of Education, September, 1972. 13. Vernon, R. J. and Phillips, M. B. A Study of Public School Driver Education in Texas. The Texas Transportation Institute, 1972. 14. Margaret, H. J. California Driver Training Evaluation study: Summary of Final Report, p. 18, 1973. 15. Kevin, A. 0., and Stokes, B. Charles. Driver Education in Virginia: An Analysis of Performance Report Data. Virginia Highway and Transportation Research Council, 1986. 16. Drever. Dell and Janke. Mary. The Effects of Range Versus Nonrange Driver Training on the Accident and Conviction Frequencies for Young Drivers. Accident Analysis and Prevention, vol. 11, pp. 179-198, 1979. 17. Forest, M. Council, et al. Effects of Range Training: Comparision of Road Test Scores for Driver Education Students, University of North Carolina Highway Safety Research Center, 1975. 119 18. Rodell, M. A Comparision of Public and Private Driver Training Courses. Washington Division of Motor Vehicles, 1969. 19. Barner, B. Monroe. Ohio Youthful Driver Record Comparisions Commercial vs. High School Trainees Licensed 1984 and 1985. Report to Office of the Governor's Highway Safety, October, 1987. 20. Haight, F. A. Induced Exposure. Accident Analysis and Prevention, vol. 5 1973, pp. 111-126. 21. Taylor, W. C., et al. Validation of the Innocent Victim Concept, Report to Office of Highway Safety Planning, Michigan Department of State Police, by Michigan State University, March 1986. 22. Richard, W. Lyles., et al. Quasi-Induced Exposure Revisted. Accepted paper for Accident Analysis and Prevention, 1991. 23. SAS Institute Inc. SAS users guide: Statistics, Carry, North Carolina. 24. Rollins, J. B. and McFarland, William, F. Cost of Motor Vehicle Accident and Injuries. Transportation Research Record 1068, pp. 32-41, 1986. 120 25. The Economic Cost to Society of Motor Vehicle Accident. Report DOT HS 806342, NHTSA, U.S., Department of Transportation, 1983, pp. vi-l-vi-8 . 26. Michigan Traffic Accident Facts, 1988. Prepared by the Michigan Department of State Police. 27. Stopher, Peter, R. and Meybur, Armin, H. Survey Sampling and Multivariate Analysis for Social Scientists and Engineers, Lexington Books, Massachusetts, 1980. APPENDIX A A Fortran Program for Changing the Layout of Accidents and Convictions Records File 121 APPENDIX A A Fortran Program for Changing the Layout of Accidents and Convictions Records File C C + 10 11 31 12 13 PROGRAM FOR CHANGING THE LAY OUT OF DRIVER'S ACCIDENTS & CONVICTIONS RECORD FILE INTEGER COUN, BIRTH, ORIG, ACCDT, ACRN, VEH, INJ, KILL, CONDAT, FRM CHARACTER*1 X, SEX CHARACTER*2 VETY CHARACTER*3 SCH, OFF CHARACTER*5 SPD, CODD CHARACTER*13 LIC N = 0 J = 1 K = 1 OPEN (UNIT = 5) OPEN (UNIT = 7) OPEN (UNIT = 8 ) DO 44 I = 1, 999999 READ ( 5, 11, END = 99) FORMAT (Al) IF (X.EQ.'B')THEN N = N+l IF (N.GE.2)THEN WRITE (7, 31) LIC, ORIG, SEX, BIRTH, COUN, SCH FORMAT ( A13, 17, Al, 17, 12, A3) WRITE (8 , 31) LIC, ORIG, SEX, BIRTH, COUN, SCH ENDIF IF (N.EQ.l)THEN IF (K.EQ.1)THEN WRITE (7, 31) LIC, ORIG, SEX, BIRTH, COUN, SCH ENDIF IF (J.EQ.0)THEN WRITE (8 , 31) LIC, ORIG, SEX, BIRTH, COUN, SCH ENDIF ENDIF BACKSPACE (5) READ (5, 12) X, LIC, ORIG, SEX, BIRTH, COUN, SCH FORMAT (Al, A13, 18X, 17, IX, Al, 17, 12, A3) J = 0 K = 0 GO TO 10 ENDIF IF (X. EQ. 'M') THEN K = K + 1 BACKSPACE (5) READ (5, 13) X, CONDAT, OFF, SPD, VETY FORMAT (Al, 13X, 17, 7X, A3, A5, IX, A2) WRITE (7, 32) LIC, ORIG, SEX, BIRTH, COUN, SCH, + CONDAT, OFF, SPD, VETY 122 32 FORMAT (A13, 17, Al, 17, 12, A3, 2X, 17, A3, A5, A2) N = 0 GO TO 10 ENDIF IF (X .EQ. 'S') THEN J = J + 1 BACKSPACE (5) READ (5, 14) X, ACCDT, VEH, INJ, KILL, CODD, ACRN, FRM 14 FORMAT (Al, 13X, 17, 312, 3X, A5, 16, 17) WRITE (8 , 33) LIC, ORIG, SEX, BIRTH, COUN, SCH, ACCD, + VEH, INJ, KILL, CODD, ACRN, FRM 33 FORMAT (A13, 17, Al, 17, 12, A3, 2X, 17, 312, A5, + 16, 17) N = 0 GO TO 10 ENDIF 44 CONTINUE 99 WRITE (7, 31) LIC, ORIG, SEX, BIRTH, COUN, SCH WRITE (8 , 31) LIC, ORIG, SEX, BIRTH, COUN, SCH STOP END APPENDIX B List of variables used in the study 123 APPENDIX B List of variables used in the study The following list indicates various variables used in different chapters in the study. PRG Various driver education programs. C Competency program in public schools. F Four-phased program in public schools. P Competency program in commercial schools. R Range program. S Simulation program in public schools. T Traditional program. Cl & C2 Competency (pub.) program in higher and lower ranked schools (according to IR criterion). PI & P2 Competency (comm.) program in higher and lower ranked schools (according to IR criterion). R1 & R2 Range program in higher and lower ranked schools (according to IR criterion) respectively. T1 & T2 Traditional program in higher and lower ranked schools (according to IR criterion) respectively. GLC1 Detroit metropolitan area. GLC2 Urban area. GLC3 Rural area. ACCRT # of accidents per student. CNVRT # of convictions per student. 124 SNGRT # of single-vehicle accidents per student. IR Relative accident involvement ratio. This ratio is defined as ratio of percentage of the at-fault drivers from a given driver education program scenario to the percentage of the innocent drivers from the same scenario. IRCLR IR value under clear weather conditions. IRRAN IR value under rainy weather conditions. IRSNW IR value under snowy weather conditions. IRDAY IR value under day time light conditions. IRNGT IR value under night time light conditions. IRDWN IR value under dwn/dusk time light conditions. IRAST IR value for angle-straight accidents. IRRER IR value for rear-end accidents. IRATR IR value for angle-turn accidents. IRHLT IR value for head-on-left turn accidents. APPENDIX C Values of various Criterion Variables for various Schools 125 APPENDIX C Values of Various Criterion Variables OBS SCH PRG ACCRT CNVRT SNGRT 1 2 029 036 053 076 128 134 13 6 182 1 86 194 226 259 269 392 393 410 4 12 4 13 415 417 4 22 425 429 431 441 442 444 469 478 495 5 07 525 531 537 538 539 546 554 570 6 15 6 18 629 633 638 6 41 C C C C C C C C C C C C C C C C C C C 0.218 0.252 0.224 0.228 0.188 0.294 0.174 0.219 0.216 0.305 0.143 0.215 0.315 0.217 0.259 0.217 0.270 0.325 0.213 0.262 0.194 0.188 0.169 0.287 0.223 0.185 0.196 0.198 0.257 0.267 0.273 0.247 0.222 0.238 0.314 0.189 0.260 0.206 0.181 0.229 0.175 0.299 0.290 0.245 0.204 0.232 0.252 0.310 0.278 0.236 0.300 0.270 0.103 0.273 0.225 0.221 0.235 0.313 0.249 0.240 0.213 0.282 0.341 0.167 0.333 0.226 0.261 0.212 0.295 0.278 0.144 0.170 0.315 0. 311 0.282 0.323 0.332 0.241 0.246 0.385 0.185 0.387 0.309 0.243 0.222 0.188 0.203 0.302 0.263 0.175 0.096 0.109 0.085 0.083 0.064 0.082 0.074 0.077 0.041 0.149 0.046 0.035 0.125 0.075 0.057 0.049 0.056 0.072 0.051 0.052 0.032 0.043 0.014 0.060 0.065 0.090 0.061 0.091 0 . 098 0.124 0.051 0.041 0.042 0.064 0.091 0.046 0.069 0.050 0.083 0.078 0.048 0.056 0.078 0.064 0.102 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 c c c c c c c c c c c c c c c c c c c c c c c c c c 126 OBS SCH PRG ACCRT CNVRT SNGRT 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 650 669 670 687 714 769 777 778 780 781 784 790 049 075 334 355 527 528 A04 A05 A08 A09 A10 A21 A24 A3 5 A39 A45 A48 A56 A60 A62 A63 A65 A77 A82 A83 A 86 A 88 OSS 951 959 965 966 973 C C C C C C C C C C C C F F F F F F P P P P P P P P P P P P P P P P P P P P P P P P P P P 0.247 0.238 0.187 0.192 0.246 0.184 0.171 0.215 0.364 0.213 0.184 0.079 0.098 0.050 0.089 0.055 0.027 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 0.201 0.201 0.207 0.135 0.218 0.251 0.302 0.223 0.266 0.201 0.279 0.246 0.282 0.293 0.282 0.275 0.326 0.238 0.265 0.302 0.307 0.226 0.284 0.341 0.271 0.308 0.218 0.310 0.240 0.260 0.243 0.222 0.280 0.353 0.249 0.296 0.287 0.201 0.229 0.213 0.236 0.327 0.328 0.265 0.302 0.224 0.282 0.340 0.459 0.302 0.454 0.277 0.740 0.410 0.436 0.504 0.502 0.382 0.431 0.352 0.344 0.304 0.448 0.308 0.388 0.590 0.392 0.396 0.239 0.477 0.336 0.448 0.646 0.399 0.510 0.604 0.326 0.495 0.423 0.021 0.034 0.026 0.045 0.033 0.004 0.075 0.077 0.074 0.044 0.038 0.027 0.071 0.066 0.066 0.057 0.056 0.059 0.061 0.058 0.049 0.078 0.056 0.068 0.034 0.098 0.099 0.063 0.051 0.069 0.086 0.065 0.044 0.066 0.053 0.108 0.040 0.073 0.062 127 OBS SCH PRG ACCRT CNVRT SNGRT 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 974 975 977 980 981 984 992 999 014 015 017 020 037 042 043 052 057 059 062 070 077 081 088 152 165 166 169 170 171 172 177 178 180 183 184 206 252 254 255 256 258 267 270 276 301 P P P P P P P P R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R 0.238 0.228 0.275 0.219 0.287 0.237 0.291 0.265 0.239 0.235 0.242 0.170 0.246 0.160 0.184 0.211 0.227 0.302 0.233 0.227 0.239 0.266 0.197 0.236 0.225 0.185 0.193 0.192 0.143 0.194 0.232 0.200 0.191 0.242 0.220 0.256 0.263 0.243 0.230 0.170 0.266 0.266 0.186 0.175 0.302 0.332 0.814 0.343 0.498 0.406 0.401 0.405 0.380 0.280 0.372 0.287 0.236 0.220 0.147 0.260 0.330 0.245 0.318 0.277 0.242 0.213 0.377 0.228 0.219 0.202 0.193 0.185 0.254 0.221 0.229 0.172 0.200 0.201 0.232 0.233 0.300 0.327 0.384 0.265 0.149 0.258 0.266 0.244 0.252 0.387 0.059 0.041 0.036 0.039 0.053 0.075 0.058 0.053 0.062 0.074 0.083 0.054 0.134 0.057 0.061 0.077 0.073 0.049 0.063 0.112 0.049 0.131 0.083 0.085 0.051 0.037 0.041 0.034 0.016 0.022 0.054 0.037 0.050 0.051 0.063 0.083 0.088 0.053 0.065 0.035 0.090 0.101 0.095 0.052 0.087 128 OBS SCH PRG ACCRT CNVRT SNGRT 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 303 307 314 316 317 320 321 323 326 331 332 340 342 343 344 345 347 348 349 365 366 377 391 394 407 419 420 421 428 430 462 468 471 492 503 508 509 518 529 530 532 535 541 543 544 R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R 0.267 0.216 0.196 0.254 0.259 0.230 0.216 0.289 0.256 0.210 0.308 0.253 0.236 0.223 0.217 0.241 0.266 0.268 0.207 0.178 0.268 0.261 0.259 0.280 0.246 0.212 0.165 0.219 0.191 0.223 0.157 0.229 0.302 0.274 0.244 0.257 0.246 0.185 0.233 0.230 0.226 0.177 0.291 0.153 0.313 0.261 0.205 0.297 0.483 0.529 0.320 0.272 0.357 0.242 0.210 0.324 0.303 0.335 0.370 0.305 0.344 0.367 0.259 0.310 0.192 0.310 0.322 0.279 0.331 0.313 0.225 0.275 0.165 0.239 0.200 0.223 0.296 0.412 0.377 0.321 0.338 0.538 0.203 0.369 0.166 0.232 0.315 0.337 0.267 0.382 0.088 0.089 0.036 0.036 0.048 0.053 0.083 0.092 0.084 0.083 0.132 0.051 0.041 0.042 0.078 0.040 0.110 0.089 0.065 0.105 0.094 0.097 0.093 0.081 0.071 0.043 0.027 0.039 0.032 0.035 0.066 0.076 0.104 0.120 0.064 0.063 0.059 0.033 0.029 0.056 0.031 0.054 0.076 0.034 0.031 129 OBS SCH PRG ACCRT CNVRT SNGRT 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 547 548 549 551 553 555 556 559 560 597 598 599 601 603 616 622 623 624 627 651 674 680 684 700 701 706 722 723 733 735 736 738 740 741 746 750 760 766 773 785 789 795 039 044 253 R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R S S S 0.167 0.196 0.299 0.204 0.223 0.179 0.201 0.277 0.279 0.233 0.240 0.240 0.248 0.216 0.270 0.266 0.136 0.204 0.251 0.245 0.184 0.201 0.197 0.220 0.169 0.173 0.203 0.207 0.194 0.182 0.154 0.177 0.131 0.160 0.175 0.175 0.177 0.201 0.197 0.152 0.206 0.158 0.256 0.376 0.201 0.223 0.282 0.241 0.255 0.307 0.257 0.210 0.290 0.273 0.317 0.398 0.383 0.303 0.356 0.307 0.271 0.252 0.244 0.224 0.401 0.203 0.181 0.250 0.331 0.202 0.187 0.293 0.297 0.284 0.323 0.322 0.306 0.324 0.283 0.347 0.333 0.263 0.370 0.229 0.272 0.250 0.259 0.417 0.453 0.383 0.027 0.032 0.060 0.032 0.032 0.031 0.026 0.044 0.059 0.061 0.039 0.051 0.038 0.053 0.066 0.055 0.033 0.040 0.052 0.087 0.065 0.059 0.088 0.078 0.060 0.049 0.042 0.040 0.030 0.029 0.034 0.024 0.021 0.028 0.029 0.046 0.025 0.031 0.036 0.041 0.044 0.066 0.068 0.137 0.042 212 213 214 215 216 217 218 219 220 221 222 223 224 225 130 OBS SCH PRG ACCRT CNVRT SNGRT 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 710 754 009 035 041 103 114 147 153 154 163 167 187 188 198 208 217 260 289 295 339 350 367 385 387 395 406 408 409 411 416 434 439 450 455 457 482 483 486 490 S S T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T 0.262 0.213 0.156 0.197 0.237 0.180 0.278 0.215 0.282 0.216 0.273 0.210 0.214 0.179 0.162 0.231 0.238 0.156 0.261 0.232 0.266 0.309 0.193 0.221 0.265 0.255 0.189 0.239 0.195 0.243 0.305 0.234 0. i73 0.314 0.332 0.224 0.266 0.279 0.189 0.235 0.329 0.236 0.174 0.243 0.278 0.373 0.295 0.360 0.369 0.318 0.257 0.231 0.185 0.171 0.188 0.317 0.321 0.227 0.272 0.258 0.254 0.373 0.159 0.292 0.398 0.293 0.207 0.278 0.248 0.189 0.342 0.266 0.246 0.364 0.254 0.157 0.293 0.333 0.223 0.294 0.083 0.031 0.080 0.126 0.053 0.082 0.080 0.066 0.141 0.056 0.114 0.075 0.078 0.042 0.042 0.078 0.101 0.084 0.075 0.090 0.078 0.089 0.070 0.080 0.075 0.089 0.037 0.034 0.053 0.038 0.093 0.069 0. 056 0.119 0.098 0.134 0.074 0.072 0.054 0.105 131 OBS SCH PRG ACCRT CNVRT SNGRT 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 494 506 526 536 545 583 591 600 617 635 675 705 707 712 715 719 720 725 753 757 763 770 772 776 782 T T T T T T T T T T T T T T T T T T T T T T T T T 0.244 0.283 0.243 0.197 0.283 0.232 0.208 0.234 0.209 0.166 0.154 0.227 0.288 0.196 0.218 0.215 0.235 0.214 0.205 0.218 0.120 0.177 0.187 0.194 0.197 0.299 0.323 0.263 0.315 0.233 0.253 0.279 0.330 0.180 0.178 0.167 0.354 0.222 0.235 0.288 0.330 0.312 0.245 0.226 0.319 0.287 0.314 0.255 0.261 0.217 0.122 0.051 0.040 0.050 0.087 0.124 0.068 0.064 0.108 0.057 0.066 0.042 0.082 0.051 0.056 0.057 0.060 0.034 0.018 0.062 0.022 0.031 0.058 0.041 0.043 132 SCH 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 029 036 053 076 128 134 136 182 186 194 226 259 269 392 393 410 412 413 415 417 422 425 429 431 441 442 444 469 478 495 507 MR W 531 537 538 539 546 554 570 615 618 629 633 638 641 PRG C C C C C C C C C C C C C C C C C C C C C C C C C C C c c c c r w* c c c c c c c c c c c c c IR 1.190 1.300 0.800 1.750 2.824 1.617 2.700 1.294 4.091 1.824 4.250 1.818 1.400 1.867 1.486 1.743 1.529 1.569 1.368 1.649 1.675 1.300 1.545 1.462 1.619 1.364 1.375 3.500 1.111 1.400 1.333 -1 vww COO 1.316 1.667 1.585 1.429 1.371 2.136 2.100 1.160 4.167 1.156 1.680 1.500 1.000 133 OBS SCH PRG IR 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 650 669 670 687 714 769 777 778 780 781 784 790 049 075 334 355 527 528 A04 A05 A08 A09 A10 A21 A24 A3 5 A39 A45 A48 A56 A60 A62 A63 A65 A77 A82 A83 A86 A88 OSS 951 959 965 966 973 C C C C C C C C C C C C F F F F F F P P P P P P P P P P P P P P P P P P P P P P P P P P P 1.094 2.000 1.120 1.455 1.765 1.895 1.423 2.565 1.750 1.938 1.716 1.563 1.500 1.840 1.688 1.257 1.467 1.870 1.147 1.222 1.912 1.498 1.820 1.698 1.697 2.077 1.580 1.362 1.653 1.933 1.469 1.553 1.560 1.652 1.625 1.604 2.400 1.889 1.600 1.525 1.667 1.639 1.500 1.477 1.674 134 OBS SCH PRG IR 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 974 975 977 980 981 984 992 999 014 015 017 020 037 042 043 052 057 059 062 070 077 081 088 152 165 166 169 170 171 172 177 178 180 183 184 206 252 254 255 256 258 267 270 276 301 P P P P P P P P R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R 1.606 1.179 1.875 1.330 1.797 1.263 1.721 1.384 1.423 1.900 1.773 1.714 1.286 1.824 3.700 1.357 1.879 1.931 1.000 1.488 1.343 2.308 1.500 1.346 2.321 2.348 1.370 1.023 1.545 1.424 1.891 1.722 1.774 2.149 1.320 1.889 1.375 1.552 1.278 3.091 1.667 1.350 2.318 1.353 1.441 135 OBS SCH PRG 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 303 307 314 316 317 320 321 323 326 331 332 340 342 343 344 345 347 348 349 365 366 377 391 394 407 419 420 421 428 430 462 468 471 492 503 508 509 518 529 530 532 535 541 543 544 R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R IR 1.741 2.000 1.938 1.554 1.667 1.647 1.706 2.000 1.762 2.556 1.500 1.889 1.481 1.750 1.316 1.580 1.458 1.909 2.308 0.889 1.373 1.538 2.667 1.800 1.444 1.444 2.250 1.922 1.923 1.231 1.333 1.438 1.923 2.045 1.520 1.273 1.684 1.450 1.968 1.419 1.885 1.250 1.590 1.778 1.600 136 OBS SCH PRG IR 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 547 548 549 551 553 555 556 559 560 597 598 599 601 603 616 622 623 624 627 651 674 680 684 700 701 706 722 723 733 735 736 738 740 741 746 750 760 766 773 785 789 795 039 044 253 R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R R S S S 1.054 1.042 1.254 1.681 1.218 1.320 2.159 1.340 1.823 1.125 1.644 1.548 1.386 1.905 1.844 1.848 0.700 1.196 1.833 2.000 1.156 1.300 1.235 1.313 2.000 2.333 1.314 1.524 1.159 1.917 1.286 1.191 1.424 1.458 1.585 1.184 1.219 1.600 1.691 2.000 1.647 2.538 1.500 1.789 1.440 212 213 214 215 216 217 218 219 220 221 222 223 224 225 137 OBS SCH PRG IR 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 710 754 009 035 041 103 114 147 153 154 163 167 187 188 198 208 217 260 289 295 339 350 367 385 387 395 406 408 409 411 416 434 439 450 455 457 482 483 486 490 S S T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T 3.000 1.684 2.286 3.000 1.511 2.000 0.976 2.000 2.111 1.516 1.261 1.238 2.118 1.515 0.941 1.579 2.214 1.222 1.293 1.158 1.391 1.907 1.667 1.000 1.500 1.174 1.571 1.600 1.538 1.441 1.769 1.360 1.929 0.935 1.407 1.107 1.053 1.125 1.137 1.375 138 OBS SCH PRG 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 494 506 526 536 545 583 591 600 617 635 675 705 707 712 715 719 720 725 753 757 763 770 772 776 782 T T T T T T T T T T T T T T T T T T T T T T T T IR 1.833 1.015 1.810 1.367 1.292 1.455 1.615 1.444 1.750 1.121 3.800 2.109 1.714 1.727 1.574 1.333 1.492 1.034 1.147 1.824 2.375 1.313 2.545 2.150 1.684 139 OBS SCH PRG IRCLR IRRAN IRSNW IRDAY 1 2 029 036 053 182 425 478 615 629 641 650 670 128 136 186 226 469 554 570 618 669 778 781 A04 A05 A45 975 980 984 999 A08 A10 A3 5 A56 A83 A 86 977 Cl Cl Cl Cl Cl Cl Cl Cl Cl Cl Cl C2 C2 C2 C2 C2 C2 C2 C2 C2 C2 C2 PI PI PI PI PI PI PI P2 P2 P2 P2 P2 P2 P2 1.105 1.071 1.250 1.333 0.950 1.500 1.364 1.385 0.733 1.300 1.143 3.250 1.571 2.333 1.786 2.083 1.500 3.000 4.250 1.333 1.500 N.A. 0.667 0.833 1.400 0.500 3.000 1.750 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 1.000 1.333 1.000 1.286 1.167 1.000 1.200 0.800 2.000 1.000 1.500 2.600 1.500 2.429 1.875 2.000 1.000 0.750 N.A. 4.000 2.500 1.500 1.667 6.000 3.667 5.000 6.500 6.000 2.000 2.200 8.000 2.500 2.118 2.250 1.180 1.154 1.428 1.375 1.582 1.263 1.615 3.667 1.400 1.667 1.667 1.545 3.333 3.500 3.000 1.500 0.833 1.000 1.536 1.000 1.077 1.220 0.984 1.204 1.909 2.091 2.013 1.875 1.886 1.583 1.900 2.000 2.000 2.000 1.800 1.000 1.000 1.363 1.500 1.625 4.000 1.714 1.636 2.200 2.333 N.A. 7.000 1.000 1.250 0.913 1.222 1.455 1.455 1.417 0.800 1.375 1.167 3.500 1.750 2.400 2.087 1.722 2.200 3.222 5.750 2.429 2.500 2.643 0.960 1.407 1.000 1.545 1.390 1.651 1.447 3.375 2.042 2.047 2.125 1.851 1.923 1.842 140 OBS IRNGT IRDWN IRAST IRRER IRATR IRHLT 1 2 0.769 1.667 0.750 1.333 1.714 1.429 1.286 1.444 1.600 1.286 0.750 3.000 3.000 1.667 2.083 3.750 3.000 4.667 1.667 3.000 3.000 2.667 1.750 1.176 2.600 1.333 1.273 0.625 1.273 2.400 2.143 1.824 1.364 1.600 4.000 3.600 N.A. 2.000 1.000 1.800 1.500 1.500 0.500 N.A. 0.800 1.285 1.667 0.714 1.833 1.500 0.533 1.333 1.000 2.000 0.600 2.000 1.000 0.500 1.200 1.200 1.333 1.500 N.A. 3.000 0.500 2.000 1.333 1.143 1.667 1.334 3.000 4.000 3.000 3.000 1.667 4.000 5.000 2.500 2.750 1.667 1.571 1.000 1.000 1.000 N.A. 2.333 1.750 0.250 0.500 2.500 N.A. 4.000 N.A. 1.750 5.000 2.000 2.000 • • 1.500 • 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 1.000 2.000 2.000 1.500 1.000 0.667 1.000 2.000 0.500 3.000 2.000 5.000 3.500 1.000 2.000 2.000 0.500 1.000 3.000 1.000 2.000 2.000 1.000 1.333 2.167 1.333 1.000 1.222 1.083 0.667 1.500 4.000 0.500 5.333 1.239 1.278 0.923 2.333 2.500 1.800 2.000 2.000 5.000 1.583 2.000 2.000 2.500 2.250 1.200 1.167 1.857 N.A. 1.700 1.000 2.429 6.000 1.333 • 1.889 1.500 1.235 0.778 1.667 1.519 1.148 1.720 2.800 2.077 1.847 2.364 1.645 2.750 1.909 • • • • • 1.000 2.000 N.A. 3.000 N.A. 2.000 6.000 1.000 2.000 1.000 0.600 1.600 2.000 0.667 0.957 2.250 1.500 1.500 1.393 0.692 1.333 2.000 • 2.000 2.308 2.000 2.000 2.330 1.250 1.000 2.000 2.333 4.000 5.500 0.750 1.000 141 OBS SCH PRG 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 037 062 170 255 365 430 508 535 547 548 549 553 597 623 624 674 680 733 736 738 750 760 043 059 081 165 166 183 256 270 307 314 323 331 349 391 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R1 R2 R2 R2 R2 R2 R2 R2 R2 R2 66 67 <«•r* DO 69 70 71 72 ▼a^ R2 R2 R2 R2 IRCLR 1 . 0 0 0 1.406 1.222 1.545 1 .0 0 0 0.571 1.071 1.073 1.500 1.214 0.905 1.333 1.412 1.158 1.515 1.333 1.500 1.889 1.129 0.886 0.960 1.032 2.167 2.375 2.333 1.853 2.077 2.375 2.077 2.714 2.316 IRRAN IRSNW IRDAY 1.333 1.455 1.333 0.333 1.333 0.800 1.500 1.875 0.667 3.000 1.500 1.342 1.500 1.250 0.912 0.900 1.333 1.262 0.714 1.308 0.957 1.113 1.188 1.083 1.405 1.563 • 1.500 1.333 0.857 1.100 0.950 1 .0 0 0 0.800 0.667 1.444 1.545 • 0.800 1.667 0.333 1 .0 0 0 0.444 1 .0 0 0 2.667 1.333 1.308 0.500 1.200 1 .0 0 0 • 0.013 3.000 2.438 2.600 3.667 1.500 1.429 1.250 2.200 1.333 2.200 1.667 7.000 1.857 1 .0 0 0 1 .0 0 0 1.333 1.941 2.115 2.500 1 .0 0 0 1.214 1.200 2.000 2.211 0.750 2.000 2.000 X. 009 aaa 1.333 1 .0 0 0 1.100 2.500 1.333 4 .000 a 1 . 0 0 0 • • 1 .0 0 0 • 1.125 7.000 2.000 2.333 2.750 1.061 1.294 1.158 1.700 7.500 2.000 2.190 2.700 2.625 1.947 2.444 2.091 I. 760 2.375 1.600 3.000 2.577 142 OBS IRNGT 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 0.500 1.375 0.833 66 67 68 69 70 71 72 IRDWN IRAST IRRER IRATR IRHLT 0.500 2.000 1.750 0.857 1.824 1.091 2.000 2.000 1.500 0.750 1.000 2.000 2.000 2.000 1.429 0.625 1.500 3.000 0.500 0.500 1.000 1.000 2.000 1.000 2.000 3.000 1.167 1.571 1.778 1.429 1.375 0.667 1.333 2.167 1.167 1.750 1.444 0.941 0.857 4.000 2.333 9.000 1.708 1.273 5.000 4.000 1.333 3.000 1.375 1.833 2.833 2.286 1.500 1.000 1.000 1.100 1.000 1.000 0.800 1.125 1.050 0.500 • 1.000 1.000 2.667 1.167 1.556 2.250 0.818 0.857 1.269 1.200 0.800 3.000 0.900 3.000 1.000 1.000 0.000 2.000 2.500 • 5.000 0.692 0.556 2.500 1.786 1.000 • 0.692 1.000 6.000 1.333 • 0.250 1.667 0.667 1.250 0.500 1.333 0.714 0.900 1.571 1.364 1.313 1.000 1.000 1.200 1.333 • 0.667 1.056 1.333 5.500 1.750 0.750 0.250 1.800 0.500 1.333 1.000 1.000 2.000 1.000 0.778 4.000 3.000 1.000 • 4.000 0.667 5.000 • 1.500 1.750 2.000 2.000 • 2.375 1.333 • 3.500 1.000 6.000 3.500 7.000 2.000 1.800 1.500 2.400 1.714 5.000 2.500 2.000 2.000 6.000 2.333 2.833 1.500 5.000 2.000 2.000 • 0.800 1.000 1.500 0.500 0.667 1.333 1.000 1.200 1.000 1.000 2.000 1.000 1.000 1.000 • • 1.500 • • 2.000 4.000 1.167 5.000 9.000 3.500 5.000 2.167 1.200 • 5.000 1.667 2.750 0.833 1.667 1.500 • 1.200 • 1.200 5.000 2.000 2.000 2.000 1.667 • 1.000 2.000 3.000 1.750 143 OBS SCH PRG IRCLR IRRAN IRSNW IRDAY 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 420 492 529 556 651 701 706 785 795 114 198 295 385 450 457 482 483 486 506 635 725 753 009 035 103 147 153 187 217 439 675 705 763 772 776 R2 R2 R2 R2 R2 R2 R2 R2 R2 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T1 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 2.143 1.705 2.556 1.941 1.618 1.682 1.733 2.000 1.826 1.200 1.222 1.150 1.231 0.783 1.211 0.981 1.094 1.222 1.235 0.951 1.083 1.167 5.333 2.571 3.500 1.300 1.636 2.000 1.875 4.333 2.143 1.933 1.968 1.467 1.250 2.500 2.087 2.000 3.667 2.273 3.000 2.833 3.000 1.500 1.500 1.000 0.667 1.000 1.429 0.875 1.000 0.944 1.000 0.333 1.667 2.000 1.600 2.000 3.500 • 4.500 2.750 2.000 1.500 1.000 2.250 3.500 2.273 1.500 2.500 • 2.143 1.500 3.500 7.000 1.200 2.000 1.500 • 1.000 0.714 1.500 1.667 1.444 1.000 2.000 0.929 1.500 0.750 1.333 1.000 1.000 1.500 1.000 1.500 1.500 3.500 5.000 2.000 2.000 2.000 2.000 2.500 6.000 4.000 4.000 1.803 2.556 2.000 2.296 1.727 2.043 2.654 1.905 1.250 1.222 0.947 1.000 1.074 1.167 1.038 0.932 1.000 0.957 1.024 1.286 1.500 3.250 2.714 3.667 1.600 2.000 2.750 1.778 2.600 1.938 2.700 2.483 1.615 2.100 144 OBS IRNGT 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 0.750 2.188 1.667 4.333 1.235 2.000 1.615 1.571 2.250 1.400 0.833 1.200 1.333 0.833 1.700 1.471 1.059 1.667 1.000 1.556 1.125 1.400 3.000 3.500 3.000 3.000 2.500 2.500 3.000 2.000 3.250 2.800 2.667 2.143 2.667 IRDWN • 2.000 2.000 2.000 2.000 1.667 2.333 1.429 1.500 • • t • • 0.400 0.833 1.750 1.000 0.667 0.500 • 0.667 • 1.000 • • 3.000 1.333 1.500 • • i. 000 3.000 • • IRAST • 1.900 3.000 1.500 0.000 1.800 1.500 3.667 4.000 0.667 0.250 1.500 1.000 1.143 1.250 0.866 1.250 0.500 1.400 1.000 1.333 1.000 1.833 2.000 1.500 3.000 1.667 3.000 2.500 2.000 2.334 4.000 4.000 1.500 2.000 IRRER IRATR 2.500 2.103 1.800 2.333 2.235 1.643 2.000 2.000 2.778 1.000 1.250 1.125 1.167 1.385 1.000 1.143 1.462 1.333 0.800 1.182 0.800 1.428 2.250 1.625 1.000 1.556 2.000 1.500 1.333 2.000 1.500 2.000 2.667 0.500 1.000 1.000 2.000 0.500 2.000 0.250 0.429 1.000 1.000 0.714 • 1.833 2.000 1.200 • 3.000 2.428 1.400 1.714 1.833 2.500 • 1.000 2.600 1.500 • • * 2.500 • • 2.500 4.000 2.000 • 2.000 IRHLT • 2.250 4.000 • 0.714 2.000 1.400 5.500 2.333 1.000 • 0.600 0.666 2.000 1.111 1.000 1.000 1.500 0.625 • 0.750 2.500 2.000 1.333 • • 3.000 • 3.000 1. 500 1.250 2.500 • APPENDIX D Histograms of IR and ACCRT Variables for various Programs 145 APPENDIX D Histograms of IR and ACCRT Variables for various Programs RANGE PROGRAM FREQUENCY BAR CHART FREQUENCY 30 + 20 + 10 + ** ** ** 8 0 ** 0 0 4c* ** *4r * 4: 4r4r 4c 4c * * * * * * 4c 4c 4c 4r ** 4r 4r 4c 4c 4c 4c 4c 4c * * 4c 4c 4c 4c ** * 4r 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c* 4c 4c 4c 4r 4c 4c * * 4 c* 4c 4c 4c 4c * * 4c 4c 4c 4c 4c 4c ** 2 5 5 0 4c 4c 7 5 xiv 0 0 ** ** ** ** ** ** ** ** ** 2 2 3 3 2 5 0 7 5 0 0 5 5 0 0 0 146 TRADITIONAL PROGRAM FREQUENCY BAR CHART FREQUENCY *** *** I 15 + kkk k kk •kkk kkk k kk k kk k kk kkk kkk kkk kkk kkk kkk k kk *** *** *** kkk kkk 10 + k kk kkk kkk *** *** 5 + kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk kkk k kk kkk kkk kkk kk k kkk kkk kkk kkk kkk k kk kkk *** kkk kkk k kk 0.90 1.00 1.25 1.50 1.75 2.00 2.25 2.50 3.00 3.75 IR 147 COMPETENCY (COMM.) PROGRAM FREQUENCY BAR CHART REQUENCY 13 + | 12 + | 11 + | 10 + | 9 + | 8 + | 7 + | 6 + | 5 + | 4 + | 3 + | 2 + | 1 + 1 ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** 1.25 1.50 1.75 2.00 IR ***** ***** 2.25 2.50 148 TRADITIONAL PROGRAM FREQUENCY BAR CHART FREQUENCY ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** 0.12 0.16 0.20 0.24 0.28 0.32 20 + 15 + 10 + 5 + ACCRT 149 COMPETENCY (PUB.) PROGRAM FREQUENCY BAR CHART FREQUENCY ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** 0.15 0.18 0.21 0.24 0.27 15 -- 10 •- 5 •- ACCRT ***** ***** ***** ***** ***** ***** ***** ***** 0.30 150 COMPETENCY (COMM.) PROGRAM FREQUENCY BAR CHART FREQUENCY ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ***** 0.225 0.250 0.275 0.300 12 + | 11 + | 10 + | 9 + | 8 + | 7 + | 6 + | 5 + | 4 + | 3 + | 2 + | 1 1 + 1 ACCRT ***** ***** ***** ***** ***** ***** 0.325 0.350 APPENDIX E ANOVA Tables used in Hypotheses Testing 151 APPENDIX E ANOVA Tables Used in Hypotheses Testing ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 2 7 9 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: ACCRT SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.07440236 0.02480079 ERROR 275 0.47316988 0.00172062 CORRECTED TOTAL 278 0.54757224 MODEL F = 14 .41 PR > F = 0.0001 R-SQUARE C.V. ROOT MSE ACCRT MEAN 0.135877 18.0800 0.04148033 0.22942652 DF ANOVA SS 3 0.07440236 SOURCE PRG F VALUE PR > F 14.41 0.0001 152 ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 2 7 9 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: CNVRT SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.88006532 0.29335511 ERROR 275 1.54801961 0.00562916 CORRECTED TOTAL 278 2.42808493 MODEL F = 52.11 PR :■ F = 0.0001 R-SQUARE C.V. ROOT MSE CNVRT MEAN 0.362452 25.5032 0.07502774 0.29418996 DF ANOVA SS 3 0.88006532 SOURCE PRG F VALUE PR > F 52.11 0.0001 153 ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 3 VALUES CRT NUMBER OF OBSERVATIONS IN DATA SET = 2 4 4 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: ACCRT SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 2 0.00157281 0.00078641 ERROR 241 0.43263714 0.00179517 CORRECTED TOTAL 243 MODEL F = 0.44 R-SQUARE C.V. 0.003622 SOURCE PRG 0.43420995 PR > F = 0.6458 ROOT MSE ACCRT MEAN 18.9736 0.04236950 0.22330738 DF ANOVA SS 2 0.00157281 F VALUE 0.44 PR > F 0.6458 154 ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 3 VALUES CRT NUMBER OF OBSERVATIONS IN DATA SET = 244 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: CNVRT SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 2 0.02620329 0.01310164 ERROR 241 1.04664492 0.00434292 CORRECTED TOTAL 243 MODEL F = 1.02 R-SQUARE C.V. 0.024424 24.1185 SOURCE PRG 1.07284821 PR > F = 0.1508 ROOT MSE CNVRT MEAN 0.06590087 0.27323770 DF ANOVA SS 2 0.02620329 F VALUE PR > F 1.02 0.1508 155 ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 279 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.82825570 0.27608523 ERROR 275 71.40220301 0.25964437 CORRECTED TOTAL 278 72.23045871 MODEL F = 1.06 R-SQUARE C.V. ROOT MSE IR MEAN 0.011467 30.8191 0.50955311 1.65336559 DF ANOVA SS 3 0.82825570 SOURCE PRG F = 0.3651 F VALUE 1.06 PR > F 0.3651 156 2 PHASE VS 3 PHASE ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 2 VALUES 2PH 3PH NUMBER OF OBSERVATIONS IN DATA SET = 2 4 4 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 1 0.02778580 0.02778580 ERROR 242 69.90796421 0.28887589 CORRECTED TOTAL 243 MODEL F = 0.10 R-SQUARE C.V. 0.000397 32.4049 SOURCE PRG 69.93575001 PR > F = 0.7567 ROOT MSE 0.53747175 IR MEAN 1.65861066 DF ANOVA SS F VALUE PR > F 1 0.02778580 0.10 0.7567 157 DETROIT ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 9 4 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.21040042 0.07013347 ERROR 90 8.94699941 0.09941110 CORRECTED TOTAL 93 9.15739983 SOURCE MODEL F = 0.71 R-SQUARE C.V. ROOT MSE TRI MEAN 0.022976 19.9170 0.31529527 1.58304255 DF ANOVA SS 3 0.21040042 SOURCE PRG PR > F = 0.5512 F VALUE 0.71 PR > F 0.5512 158 URBAN ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 9 6 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 1.35358119 0.45119373 ERROR 92 26.97341005 0.29318924 CORRECTED TOTAL 95 28.32699124 PR > F = 0.2097 MODEL F = 1.54 R-SQUARE C.V. ROOT MSE TRI MEAN 0.047784 31.5016 0.54146952 1.71886458 DF ANOVA SS 3 1.35358119 SOURCE PRG F VALUE 1.54 PR > F 0.2097 159 RURAL ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 89 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: TRI DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.56651798 0.18883933 ERROR 85 33.30166901 0.39178434 CORRECTED TOTAL 88 33.86818699 SOURCE PR > F = 0.6957 MODEL F = 0.48 R-SQUARE C.V. ROOT MSE TRI MEAN 0.016727 37.7750 0.62592679 1.65698876 DF ANOVA SS 3 0.56651798 SOURCE PRG F VALUE 0.48 PR > F 0.6957 160 Conpetency program (public) ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS GLC 3 VALUES 12 3 NUMBER OF OBSERVATIONS IN DATA SET = 57 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 2 0.90186847 0.45093424 ERROR 54 29.17102725 0.54020421 CORRECTED TOTAL 56 MODEL F = 0.83 R-SQUARE C.V. 0.029989 41.9416 SOURCE GLC DF 2 30.07289572 PR > F = 0.4395 ROOT MSE 0.73498586 TRI MEAN 1.75240351 ANGVA SS F VALUE 0.90186847 0.83 PR > F 0.4395 161 Range ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS GLC 3 VALUES 12 3 NUMBER OF OBSERVATIONS IN DATA SET = 1 2 4 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 2 0.88049027 0.44024514 ERROR 121 22.03672615 0.18212170 CORRECTED TOTAL 123 MODEL F = 2.42 R-SQUARE C.V. 0.038420 25.8937 SOURCE GLC 22.91721642 PR > F = 0.0935 ROOT MSE 0.42675720 DF ANOVA SS 2 0.88049027 TRI MEAN 1.64811290 F VALUE 2.42 PR > F 0.0935 162 Traditional Program ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS GLC 3 VALUES 12 3 NUMBER OF OBSERVATIONS IN DATA SET = 63 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 2 0.02705935 0.01352968 ERROR 60 16.14383192 0.26906387 CORRECTED TOTAL 62 MODEL F = 0.05 R-SQUARE C.V. 0.001673 32.5332 SOURCE GLC 16.17089127 PR > F = 0.9510 ROOT MSE 0.51871366 TRI MEAN 1.59441270 DF ANOVA SS F VALUE PR > F 2 0.02705935 0.05 0.9510 163 Competency Program (comm.) ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS GLC 3 VALUES 12 3 NUMBER OF OBSERVATIONS IN DATA SET = 3 5 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 2 0.37070645 0.18535323 ERROR 32 1.87049315 0.05845291 CORRECTED TOTAL 34 2.24119960 MODEL F = 3.17 R-SQUARE C.V. ROOT MSE TRI MEAN 0.165405 14.9536 0.24177037 1.61680000 SOURCE GLC DF 2 PR > F = 0.0654 ANOVA SS F VALUE PR > F 0.37070645 3.17 0.0654 164 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 3 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRCLR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.08191564 0.02730521 ERROR 49 2.73270308 0.05576945 CORRECTED TOTAL 52 2.81461872 MODEL F = 0.49 R-SQUARE C.V. ROOT MSE IRCLR MEAN 0.029104 20.0777 0.23615556 1.17620755 DF ANOVA SS 3 0.08191564 SOURCE PRG PR > F = 0.6911 F VALUE 0.49 PR > F 0.6911 165 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 53 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRRAN SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.27981229 0.09327076 ERROR 48 6.48353978 0.13507375 CORRECTED TOTAL 51 6.76335208 PR > F = 0.5623 MODEL F = 0.69 R-SQUARE C.V. ROOT MSE IRRAN MEAN 0.0413/2 30.8893 0.36752380 1.18980769 DF ANOVA SS 3 0.27981229 SOURCE PRG F VALUE 0.69 PR > F 0.5623 166 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 3 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRSNW SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.08104876 0.02701625 ERROR 45 18.19395724 0.40431016 CORRECTED TOTAL 48 18.27500600 MODEL F = 0.07 R-SQUARE C.V. ROOT MSE IRSNW MEAN 0.004435 49.8765 0.63585388 1.27485714 DF ANOVA SS 3 0.08104876 SOURCE PRG PR > F = 0.9772 F VALUE PR > F 0.07 0.9772 167 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 3 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRDAY SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.28511733 0.09503911 ERROR 49 2.49951286 0.05101047 CORRECTED TOTAL 52 2.78463019 MODEL F = 1.86 R-SQUARE C.V. ROOT MSE IRDAY MEAN 0.102390 18.7177 0.22585497 1.20664151 DF ANOVA SS 3 0.28511733 SOURCE PRG PR > F = 0.1481 F VALUE PR > F 1.86 0.1481 168 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 3 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRDWN DF SUM OF SQUARES MEAN SQUARE MODEL 3 1.94968783 0.64989594 ERROR 42 18.18953732 0.43308422 CORRECTED TOTAL 45 20.13922515 SOURCE MODEL F = 1.50 R-SQUARE C.V. ROOT MSE IRDWN MEAN 0.096810 50.9916 0.65809135 1.29058696 DF ANOVA SS 3 1.94968783 SOURCE PRG PR > F = 0.2283 F VALUE 1.50 PR > F 0.2283 169 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 3 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRNGT SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.13481934 0.04493978 ERROR 49 11.08815273 0.22628883 CORRECTED TOTAL 52 11.22297208 MODEL F = 0.20 R-SQUARE C.V. ROOT MSE IRNGT MEAN 0.012013 36.2538 0.47569826 1.31213208 DF ANOVA SS 3 0.13481934 SOURCE PRG PR > F = 0.8968 F VALUE 0.20 PR > F 0.8968 170 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 53 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRAST SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 1.77590894 0.59196965 ERROR 49 29.21960430 0.59631846 CORRECTED TOTAL 52 30.99551325 MODEL F = 0.99 R-SQUARE C.V. ROOT MSE IRAST MEAN 0.057296 59.3332 0.77221659 1.30149057 DF ANOVA SS 3 1.77590894 SOURCE PRG PR > F = 0.4041 F VALUE 0.99 PR > F 0.4041 171 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 3 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRRER SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.41913600 0.13971200 ERROR 49 5.48418087 0.11192206 CORRECTED TOTAL 52 5.90331687 MODEL F = 1.25 R-SQUARE C.V. ROOT MSE IRRER MEAN 0.071000 26.9452 0.33454754 1.24158491 DF ANOVA SS 3 0.41913600 SOURCE PRG PR > F = 0.3024 F VALUE 1.25 PR > F 0.3024 172 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRATR DF SUM OF SQUARES MEAN SQUARE MODEL 3 1.43706578 0.47902193 ERROR 44 38.07857014 0.86542205 CORRECTED TOTAL 47 39.51563592 SOURCE MODEL F = 0.55 R-SQUARE C.V. U .U J O J O i /O .U i O J SOURCE PRG PR > F = 0.6485 ROOT MSE U .S J U ^ O U D J IRATR MEAN 1.22379167 DF ANOVA SS F VALUE PR > F 3 1.43706578 0.55 0.6485 173 Higher Ranked schools ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 53 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRHLT SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 1.56997688 0.52332563 ERROR 40 26.84317703 0.67107943 CORRECTED TOTAL 43 28.41315391 MODEL F = 0.78 R-SQUARE C.V. ROOT MSE IRHLT MEAN 0.055255 60.8142 0.81919438 1.34704545 DF ANOVA SS 3 1.56997688 SOURCE PRG PR > F = 0.5122 F VALUE 0.78 PR > F 0.5122 174 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRCLR SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 6.69943375 2.23314458 ERROR 50 55.42763596 1.10855272 CORRECTED TOTAL 53 62.12706970 MODEL F = 2.01 R-SQUARE C.V. ROOT MSE IRCLR MEAN 0.107834 45.7390 1.05287830 2.30192593 DF ANOVA SS 3 6.69943375 SOURCE PRG PR > F = 0.1238 F VALUE 2.01 PR > F 0.1238 175 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRRAN SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 2.03510032 0.67836677 ERROR 49 69.04816889 1.40914630 CORRECTED TOTAL 52 71.08326921 MODEL F = 0.48 R-SQUARE C.V. ROOT MSE IRRAN MEAN 0.028630 48.4427 1.18707468 2.45047170 DF ANOVA SS 3 2.03510032 SOURCE PRG PR > F = 0.6967 F VALUE 0.48 PR > F 0.6967 176 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 4 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRSNW SOURCE DF SUM OF SQUARES MEAN SQUARE MODEL 3 9.98792466 3.32930822 ERROR 43 132.67578019 3.08548326 CORRECTED TOTAL 46 142.66370485 MODEL F = 1.08 R-SQUARE C.V. ROOT MSE IRSNW MEAN 0.070010 62.0989 1.75655437 2.82863830 DF ANOVA SS 3 9.98792466 SOURCE PRG PR > F = 0.3681 F VALUE 1.08 PR > F 0.3681 177 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 5 4 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRDAY DF SUM OF SQUARES MEAN SQUARE MODEL 3 1.53250338 0.51083446 ERROR 50 51.79341899 1.03586838 CORRECTED TOTAL 53 53.32592237 SOURCE PR > F = 0.6887 MODEL F = 0.49 R-SQUARE C.V. ROOT MSE IRDAY MEAN 0.028738 40.9855 1.01777619 2.48325926 DF ANOVA SS 3 1.53250338 SOURCE PRG F VALUE PR > F 0.49 0.6887 178 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRDWN DF SUM OF SQUARES MEAN SQUARE MODEL 3 6.24510541 2.08170180 ERROR 39 67.18209250 1.72261776 CORRECTED TOTAL 42 73.42719791 SOURCE PR > F = 0.3194 MODEL F = 1.21 R-SQUARE C.V. ROOT MSE IRDWN MEAN 0.085052 59.2264 1.31248534 2.21604651 DF ANOVA SS 3 6.24510541 SOURCE PRG F VALUE 1.21 PR > F 0.3194 179 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRNGT DF SUM OF SQUARES MEAN SQUARE MODEL 3 1.21616046 0.40538682 ERROR 50 86.18359768 1.72367195 CORRECTED TOTAL 53 87.39975815 SOURCE PR > F = 0.8714 MODEL F = 0.24 R-SQUARE C.V. ROOT MSE IRNGT MEAN 0.013915 49.3903 1.31288688 2.65818519 DF ANOVA SS 3 1.21616046 SOURCE PRG F VALUE 0.24 PR > F 0.8714 180 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRAST DF SUM OF SQUARES MEAN SQUARE MODEL 3 5.94237829 1.98079276 ERROR 47 86.49017587 1.84021651 CORRECTED TOTAL 50 92.43255416 SOURCE MODEL F = 1.08 R-SQUARE C.V. 0.064289 50.3305 SOURCE PRG PR > F = 0.3682 ROOT MSE DF ANOVA SS 3 5.94237829 IRAST MEAN F VALUE 1.08 PR > F 0.3682 181 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRRER DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.98237360 0.32745787 ERROR 46 43.32012048 0.94174175 CORRECTED TOTAL 49 44.30249408 SOURCE PR > F = 0.7909 MODEL F = 0.35 R-SQUARE C.V. ROOT MSE IRRER MEAN 0.022174 43.5508 0.97043379 2.22828000 DF ANOVA SS 3 0.98237360 SOURCE PRG F VALUE 0.35 PR > F 0.7909 182 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRATR DF SUM OF SQUARES MEAN SQUARE MODEL 3 3.75458787 1.25152929 ERROR 32 82.29124635 2.57160145 CORRECTED TOTAL 35 86.04583422 SOURCE PR > F = 0.6939 MODEL F = 0.49 R-SQUARE C.V. ROOT MSE IRATR MEAN 0.043635 63.0052 1.60362135 2.54522222 DF ANOVA SS 3 3.75458787 SOURCE PRG F VALUE 0.49 PR > F 0.6939 183 LOWER RANKED SCHOOLS ANALYSIS OF VARIANCE PROCEDURE CLASS LEVEL INFORMATION CLASS LEVELS PRG 4 VALUES C P R T NUMBER OF OBSERVATIONS IN DATA SET = 54 ANALYSIS OF VARIANCE PROCEDURE DEPENDENT VARIABLE: IRHLT DF SUM OF SQUARES MEAN SQUARE MODEL 3 0.73417465 0.24472488 ERROR 35 68.28326965 1.95095056 CORRECTED TOTAL 38 69.01744431 SOURCE PR > F = 0.9444 MODEL F = 0.13 R-SQUARE C.V. ROOT MSE IRHLT MEAN 0.010638 60.4056 1.39676432 2.31230769 DF ANOVA SS 3 0.73417465 SOURCE PRG F VALUE 0.13 PR > F 0.9444 APPENDIX F Determination of Weights for Different Types of Accident. List of Schools in Ranking Order. List of Consistent Schools on various Criterion Variables. 184 APPENDIX F Determination of Weight for Different Types of Accidents: The weight for each type of accident is taken as equal to the average dollar value of each type of accident. To determine the average dollar value by type of accident, three steps were followed: 1. The percentage of fatal, injury and property damage (PDO) accidents in each type of accident were determined from the state-wide accident data for year 1988 as shown below: Types of accidents Pedestrian Bicyle Hit train over-turned Fixed object Other object Parking Backing Animal Head-on Angle Rear-end Side swipe (meeting) Side swipe (passing) Driveway % of fatal accidents 4.91 0.94 6.0 1.13 0.62 0.05 0.05 0 0.008 2.70 0.49 0.11 0.0 0.0 0.12 % of injury accidents 91.33 84.50 41.35 56.82 30.33 11.98 8.3 5.79 2.34 37.62 38.85 21.65 20.65 12.46 28. 68 % of PDO accidents 3.76 14.56 52.65 42.05 69.05 87.97 91.65 94.21 97.65 59.68 60.66 78.24 79.35 87.54 71.17 2. These percentages were multiplied by the respective average dollar value of fatal, injury and PDO accidents. The average dollar value taken for fatal, injury and PDO accidents are as follows (25): i) The cost of a fatal accident is $ 32,500. 185 ii) The cost of an injury accident is $ 6,100. iii) The cost of a PDO accident taken is $ 1,150. 3. The weight of each accident type is equal to the summation of these three products. Weight of an accident type = % of fatal * cost of fatal + % of injury * cost of injury + % of PDO * cost of PDO The weight computed for each type of accident is as follows: Types of accidents Pedestrian Bicyle Hit train over-turned Fixed object Other object Parking Backing Animal Head-on Angle Rear-end Side swipe (meeting) Side swipe (passing) Driveway Weight 72,1012 56,2744 50,7783 43,1685 28,4570 17.5540 15,6200 14,3700 12,6800 38,6000 32,4560 23,5660 21,7300 17,6182 20,5880 186 List of Schools in Ranking Order ACCIDENT RATING SCORE RANK 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 SCH TYP 763 790 740 623 785 795 462 226 260 736 042 420 198 1 2 020 635 256 171 543 675 009 442 701 741 276 770 555 429 188 746 769 518 777 738 535 547 760 570 439 365 103 539 169 781 618 422 4 4 4 4 4 2 1 4 4 4 1 4 1 4 4 4 1 1 2 4 4 4 1 4 2 1 4 2 4 2 4 4 4 4 2 1 4 1 2 4 2 2 2 PRG T C R R R R R C T R R R T R T R R R T T C R R R T R C T R C R C R R R R C T R T C R C C C SCR/STU 6.20 6.67 7.07 7.10 7.46 7.56 7.76 7.84 7.98 7.98 8.22 8.22 8.38 8.42 8.71 8.72 8.74 8.75 8.79 8.80 8.89 9.03 9.03 9.06 9.06 9.11 9.14 9.15 9.17 9.19 9.24 9.25 9.30 9.35 9.37 9.38 9.40 9.40 9.50 9.57 9.58 9.66 9.67 9.67 9.73 187 ACCIDENT RATING SCORE RANK SCH TYP 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 772 712 270 735 409 706 674 750 180 088 444 166 128 136 556 406 035 259 253 687 528 722 349 548 773 428 367 469 A77 776 766 425 723 314 172 782 670 733 684 186 617 551 486 043 624 1 1 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 4 4 1 4 4 4 4 4 2 4 2 2 4 1 1 2 5 2 6 4 4 4 4 4 1 2 3 1 4 2 4 4 4 1 2 4 4 2 1 4 1 4 4 PRG T T R R T R R R R R C R C C R T T C S C F R R R R R T C P T R C R R R T C R R C T R T R R SCR/STU 9.74 9.78 9.81 9.86 9.92 9.95 9.99 10.01 10.02 10.03 10.04 10.05 10.08 10.19 10.28 10.33 10.34 10.35 10.39 10.40 10.41 10.42 10.44 10.47 10.47 10.48 10.49 10.49 10.50 10.51 10.52 10.54 10.55 10.59 10.63 10.69 10.70 10.71 10.73 10.76 10.76 10.83 10.86 10.87 10.94 188 ACCIDENT RATING SCORE 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 SCH TYP 052 980 415 680 029 154 536 187 778 392 182 715 753 789 049 784 757 A83 147 217 385 167 441 457 419 307 591 603 178 331 725 170 A56 532 430 355 410 975 295 343 421 OSS 057 342 780 4 3 2 4 2 1 1 1 2 2 2 1 1 4 6 2 1 3 1 1 1 1 2 1 4 4 1 4 4 4 1 4 3 4 4 6 2 3 1 4 4 3 4 4 2 PRG R P C R C T T T C C C T T R F C T P T T T T C T R R T R R R T R P R R F C P T R R P R R C SCR/STU 10.95 10.95 10.99 11.11 11.14 11.15 11.15 11.16 11.16 11.19 11.23 11.27 11.31 11.34 11.35 11.38 11.43 11.45 11.49 11.51 11.52 11.56 11.59 11.63 11.64 11.64 11.66 11.67 11.69 11. 69 11.72 11.73 11.87 11.88 11.88 11.91 11.91 11.94 11.96 11.96 11.99 12.01 12.03 12.03 12.04 189 ACCIDENT RATING SCORE SCH 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 070 344 184 053 530 597 490 705 641 754 700 720 320 554 531 468 177 255 525 434 077 076 321 017 719 345 529 553 075 503 537 A05 615 599 408 062 165 509 984 A3 5 650 208 152 015 583 TYP 4 4 4 2 4 4 1 1 2 5 4 1 4 2 2 4 4 4 2 1 4 2 4 4 1 4 4 4 6 4 2 3 2 4 1 4 4 4 3 3 2 1 4 4 1 SCR/STU PRG R R R C R R T T C S R T R C C R R R C T R C R R T R R R F R C P C R T R R R P P C T R R T 12.09 12.12 12.14 12.18 12.21 12.21 12.29 12.30 12.37 12.37 12.39 12.39 12.40 12.42 12.42 12.44 12.45 12.55 12.59 12.61 12.65 12.66 12.67 12.70 12.73 12.76 12.77 12.79 12.79 12.79 12.80 12.82 12.83 12.87 12.89 12.93 12.94 13.01 13.02 13.04 13.04 13.05 13.09 13.11 13.11 190 ACCIDENT RATING SCORE SCH 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 377 041 494 526 974 627 183 714 965 014 598 638 600 669 A 88 393 081 206 037 417 387 391 999 254 407 348 395 A 86 411 A63 651 114 495 316 977 527 601 508 323 252 317 622 710 366 A39 TYP 4 1 1 1 3 4 4 2 3 4 4 2 1 2 3 2 4 4 4 2 1 4 3 4 4 4 1 3 1 3 A 1 2 4 3 6 4 4 4 4 4 4 5 4 3 PRG SCR/STU R T T T P R R C P R R C T C P C R R R C T R P R R R T P T P R T c R P F R R R R R R S R P 13.14 13.20 13.20 13.24 13.25 13.27 13.29 13.33 13.35 13.39 13.43 13.46 13.47 13.47 13.49 13.49 13.53 13.58 13.61 13.66 13.66 13.83 13.84 13.89 13.92 13.93 13.96 14.00 14.02 14.03 14.03 14.15 14.15 14.19 14.20 14.31 14.33 14.34 14.40 14.44 14.50 14.51 14.57 14.60 14.61 191 ACCIDENT RATING SCORE RANK 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 SCH 052 680 981 153 A56 525 036 326 340 039 546 478 347 289 258 267 482 992 303 339 492 163 559 560 483 412 507 A04 A21 394 431 707 301 334 541 194 A10 A08 633 951 TYP 4 4 3 1 3 2 2 4 4 5 2 2 4 1 4 4 1 3 4 1 4 1 4 4 1 2 2 3 3 4 2 1 4 6 4 2 3 3 2 3 PRG R R P T P C C R R S C C R T R R T P R T R T R R T C C P P R C T R F R C P P C P SCR/STU 14.67 14.70 14.70 14.72 14.73 14.75 14.75 14.79 14.80 14.81 14.82 14.84 14.85 14.87 14.88 14.90 14.92 14.93 14.95 14.96 14.99 15.01 15.05 15.09 15.11 15.14 15.17 15.36 15.41 15.54 15.57 15.61 15.66 15.70 15.74 15.84 15.90 15.96 16.05 16.11 192 ACCIDENT RATING SCORE RANK 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 SCH A45 332 A65 629 A82 549 134 455 350 269 A62 471 059 545 616 506 450 A48 416 538 A24 544 413 959 044 TYP 3 4 3 2 3 4 2 1 1 2 3 4 4 1 4 1 1 3 1 2 3 4 2 3 5 PRG P R P C P R C T T C P R R T R T T P T C P R C P S SCR/STU 16.17 16.22 16.42 16.46 16.48 16.55 16.85 16.85 16.89 16.93 16.94 16.97 17.00 17.02 17.13 17.14 17.32 17.36 17.74 17.76 17.79 18.50 18.70 21.41 33.01 193 CONVICTION RATING SCORE RANK 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 SCH PRG 182 256 457 442 187 530 411 367 421 444 042 169 177 675 188 365 539 415 680 618 629 429 623 670 635 166 152 687 556 706 009 641 617 307 178 518 790 422 180 410 165 077 198 769 419 C R T C T R T T R C R R R T T R C C R C C C R C T R K C R R T C T R R R C C R C R R T C R SCR/STU 0.2165 0.2870 0.3075 0.3170 0.3175 0.3285 0.3340 0.3355 0.3355 0.3380 0.3380 0.3470 0.3475 0.3700 0.3770 0.3805 0.3815 0.3845 0.3855 0.3930 0.3935 0.3960 0.3970 0.3990 0.4020 0.4020 0.4025 0.4030 0.4030 0.4035 0.4060 0.4075 0.4120 0.4140 0.4175 0.4180 0.4185 0.4195 0.4210 0.4310 0.4315 0.4315 0.4395 0.4400 0.4410 194 CONVICTION RATING SCORE RANK SCH PRG 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 430 547 615 167 406 782 029 R R C T T T C R C R T R C R R R R R T R T R R R T R C R C R R C T C T R R C T R C T R T C 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 020 669 674 260 773 194 627 701 331 462 088 707 184 763 171 037 555 035 549 226 684 777 551 276 128 712 393 409 172 532 259 545 624 570 486 070 439 537 SCR/STU 0.4410 0.4415 0.4435 0.4440 0.4440 0.4455 0.4460 0.4465 0.4465 0.4475 0.4500 0.4515 0.4530 0.4535 0.4545 0.4550 0.4550 0.4565 0.4620 0.4630 0.4640 0.4645 0.4650 0.4655 0.4655 0.4725 0.4735 0.4735 0.4740 0.4750 0.4765 0.4775 0.4775 0.4805 0.4815 0.4820 0.4825 0.4830 0.4835 0.4850 0.4855 0.4870 0.4875 0.4875 0.4905 195 CONVICTION RATING SCORE RANK SCH PRG 91 92 93 94 95 96 97 98 99 753 A77 785 714 057 183 420 425 428 776 560 270 754 295 391 543 795 526 255 326 741 789 043 559 772 348 455 733 781 760 076 736 062 738 548 725 715 638 392 740 525 408 186 531 267 T P R C R R R C R T R R S T R R R T R R R R R R T R T R C R C R R R R T T C C R C T C C R 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 SCR/STU 0.4915 0.4935 0.4935 0.4985 0.4990 0.4995 0.5010 0.5020 0.5050 0.5060 0.5065 0.5115 0.5165 0.5170 0.5170 0.5235 0.5240 0.5250 0.5255 0.5280 0.5290 0.5315 0.5340 0.5375 0.5400 0.5425 0.5425 0.5435 0.5455 0.5475 0.5520 0.5520 0.5525 0.5535 0.5555 0.5580 0.5585 0.5600 0.5645 0.5645 0.5650 0.5670 0.5695 0.5710 0.5740 196 CONVICTION RATING SCORE RANK SCH PRG 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 441 014 258 163 339 049 591 036 633 303 528 321 349 746 583 494 314 622 385 114 723 735 412 535 355 041 017 434 541 154 482 289 344 597 490 A56 784 170 395 332 720 553 965 974 770 C R R T T F T C C R F R R R T T R R T T R R C R F T R T R T T T R R T P C R T R T R P P T SCR/STU 0.5740 0.5750 0.5765 0.5785 0.5795 0.5810 0.5840 0.5845 0.5850 0.5865 0.5905 0.5930 0.5970 0.5980 0.6015 0.6020 0.6025 0.6035 0.6070 0.6085 0.6085 0.6100 0.6105 0.6115 0.6140 0.6145 0.6150 0.6150 0.6160 0.6165 0.6245 0.6285 0.6295 0.6295 0.6305 0.6310 0.6310 0.6315 0.6325 0.6345 0.6350 0.6370 0.6370 0.6375 0.6380 197 CONVICTION RATING SCORE RANK SCH PRG 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 503 075 366 601 136 431 206 340 554 A83 778 345 468 700 766 536 977 A45 342 153 417 394 722 506 053 134 600 750 495 320 469 507 413 147 A3 5 603 059 407 377 208 252 705 508 538 217 R F R R C C R R C P C R R R R T P P R T C R R T C C T R C R C C C T P R R R R T R T R C T 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 SCR/STU 0.6385 0.6390 0.6390 0.6395 0.6405 0.6420 0.6425 0.6435 0.6495 0.6510 0.6515 0.6525 0.6540 0.6540 0.6545 0.6580 0.6615 0.6635 0.6640 0.6675 0.6680 0.6710 0.6730 0.6745 0.6755 0.6770 0.6780 0.6785 0.6805 0.6835 0.6835 0.6880 0.6890 0.6900 0.6930 0.6935 0.6955 0.6970 0.7025 0.7105 0.7120 0.7130 0.7145 0.7155 0.7175 198 CONVICTION RATING SCORE RANK SCH PRG SCR/STU 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 052 710 269 416 483 253 A3 9 616 323 529 015 546 780 757 599 343 999 478 A65 081 450 301 544 A21 387 719 992 598 350 A63 A60 651 347 981 254 650 A05 OSS 980 103 R S C T T S P R R R R C C T R R P C P R T R R P T T P R T P P R R P R C P P P T 0.7185 0.7190 0.7195 0.7215 0.7290 0.7305 0.7320 0.7325 0.7360 0.7385 0.7455 0.7460 0.7490 0.7530 0.7580 0.7590 0.7610 0.7615 0.7670 0.7690 0.7755 0.7805 0.7810 0.7835 0.7860 0.7865 0.7870 0.7880 0.7905 0.7935 0.7970 0.8045 0.8075 0.8095 0.8125 0.8140 0.8180 0.8330 0.8370 0.8430 199 CONVICTION RATING SCORE RANK SCH PRG 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 492 984 A 86 A24 A08 527 973 471 A82 334 966 039 A48 316 951 A10 A09 A62 509 317 A 88 975 A04 959 044 R P P P P F P R P F P S P R P P P P R R P P P P S SCR/STU 0.8510 0.8615 0.8715 0.8770 0.8795 0.9160 0.9245 0.9300 0.9315 0.9410 0.9430 0.9690 0.9830 0.9905 1.0280 1.0395 1.0515 1.0840 1.1130 1.1340 1.3745 1.4010 1.4540 1.4785 1.6415 200 List of consistent schools on various criterion variables in higher ranked group over the two year period. S.N. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28 ; 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. Code of consistent schools in higher ranked group on criterion variable IR Acc/stu Scr/stu S.N. IR Acc/stu Scr/stu A04 A05 A09 A21 A3 9 A62 A65 A82 A 88 014 029 036 037 039 041 049 052 053 062 070 088 114 152 163 167 170 171 172 182 184 188 208 252 253 254 255 258 267 289 295 301 316 317 339 344 347 009 009 020 020 035 042 053 077 103 128 147 166 169 170 171 172 178 180 184 186 188 198 208 226 256 260 270 276 320 344 367 385 395 408 409 410 411 415 420 421 422 428 429 442 462 478 486 508 035 042 052 053 075 077 088 103 128 136 121 147 166 169 172 178 180 184 186 188 198 208 226 256 260 270 276 320 342 349 365 367 377 391 406 408 409 411 415 420 421 422 428 429 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66 . 67. 68 . 69. 70. 71. 72. 73. 74 = 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 355 366 385 387 393 395 406 407 415 417 419 429 430 431 434 441 442 450 455 462 478 482 483 486 490 495 506 508 518 527 530 531 536 539 545 546 547 548 549 553 555 559 591 597 598 599 509 518 525 526 529 530 532 535 543 547 553 554 555 556 570 617 618 625 624 635 670 674 675 701 712 722 723 725 733 735 736 738 740 741 746 750 754 757 760 763 766 769 772 773 776 777 439 442 462 486 508 509 518 525 526 529 530 532 535 539 547 554 570 618 623 624 635 674 675 701 706 712 719 733 736 738 740 741 750 760 763 770 772 778 780 781 784 785 790 795 Continued on next page 201 List of consistent schools on various criterion variables in higher ranked group over the two year period (continued). S.N. 93. 94. 95. 96. 97. 98. 99. 100 . 101 . 102 . 103. 104. 105. 106. 107. 108. 109. 110 . 111 . 112 . 113. 114. 115. 116. 117. 118. 119. 120 . 121 . 122 . 123. 124. 125. 126. 127. 128. 129. 130. Code of consistent schools in higher ranked group on criterion variable IR Acc/stu Scr/stu S.N. IR Acc/stu Scr/stu 600 601 615 623 624 635 638 641 650 670 674 680 684 700 720 722 723 725 733 736 738 740 741 750 753 754 757 760 770 789 790 959 965 966 975 980 984 999 778 780 781 784 785 789 790 795 202 List of consistent schools on various criterion variables in lower ranked group over the two year period. S.N. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10 . 11 . 12 . 13. 14. 15. 16. 17. 18. 19. 20 . 21 . 22 . 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. Code of consistent schools in lower ranked group on criterion variable Acc/stu Scr/stu S.N. IR Acc/stu Scr/st\ IR A83 A86 009 035 043 057 075 081 128 147 153 165 178 186 187 226 270 307 349 391 392 410 420 428 469 471 532 543 556 616 622 675 701 706 707 710 763 778 781 785 977 A04 A05 A09 A10 A21 A24 A3 9 A45 A48 A62 A63 A65 A82 A86 OSS 015 029 044 049 057 059 062 081 114 134 152 154 163 182 183 194 206 217 252 253 254 258 267 289 301 314 316 317 323 326 332 A04 A05 A09 A10 A21 A24 A3 5 A3 9 A45 A48 A62 A63 A65 A82 A86 OSS 015 044 057 059 062 134 152 154 163 183 194 206 252 253 258 289 301 314 316 317 321 326 332 334 339 340 350 366 387 394 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66 . 67. 68 . 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86 . 87. 88 . 89. 90. 91. 92. 334 339 340 345 349 350 366 387 394 407 412 413 416 417 434 441 450 455 469 471 482 483 492 527 538 541 544 545 546 549 559 560 597 598 599 601 615 627 629 633 651 669 700 705 710 951 407 412 413 416 417 434 455 471 482 483 492 527 538 541 544 545 546 549 559 560 598 599 600 601 615 616 627 629 633 641 651 669 684 700 705 710 951 959 966 973 977 999 Continued on next page 203 List of consistent schools on various criterion variables in lower ranked group over the two year period (continued). S.N. 93. 94. 95. 96. 97. 98. 99. Code of consistent schools m lower ranked group on criterion variable IR Acc/stu Scr/stu S.N. IR Acc/stu Scr/stu 959 966 973 977 981 992 999