1 Freeway Dens) ES Uninterrupted Flow | Introduction Levels of service on uninterrupted flow facilities as defined by the Highway Capacity Manual 2010 (HCM) are based on the density of traffic on the roadway ‘as measured in passenger cars per mile per lane. Uninterrupted flow facilities include freeways, multilane highways, and two lane highways; sections of the transportation network where travel conditions are determined by the geomet- rics of the roadway and the amount of traffic present. Intersections, both signal- ized and unsignalized, along with interchanges and other delay causing ‘components of the system are referred to as interrupted flow facilities While much research has been devoted to identifying the parameters which affect the operations at each type of facility, interrupted flow facilities have received most of the attention, This has occurred in part because the remediation of problems identifiable at an intersection are usually much less costly to solve than those associated with a section of freeway or multilane highway. The cost of retiming a signal or even the widening of an approach to an intersection usually cannot compare to the high cost of adding a lane to a freeway section. ‘This chapter presents a method whereby the investigator, using the proper equipment, can rapidly determine the Level of Service of a freeway segment. This knowledge can then be used as input to a transportation management scheme which functions in real time or over a longer period, Easily finding the level of service of a freeway segment frees the engineer from many of the assumptions required when estimating the quality of the traffic flow using the ICM methodol- ogy. As explained later, the HCM uses measured values to develop a baseline condi- tion for a roadway and uses this baseline condition to estimate the characteristics of the traffic flow under other loadings. A basic assumption is that certain charac teristics of the traffic flow and driver behavior will be consistent throughout the day and over time. This may or may not be the case, By actually measuring the density of the flow, the investigator need not make such an assumption. Basic freeway segments may be thought of as those sections of the freeway where traffic flow is relatively unaffected by interchanges and the maneuvers associated with them. The procedure presented here involves measurement of speed and volume, and computation of density using these two measured values 99 Heer EEEEEd He eeeeEEEoe eeeEquipment Procedure for basic freeway segments-a very simple method. It is contained herein to demonstrate that with minimal effort, real time levels of service can be deter- mined for uninterrupted flow facilities. It is not presented as a complete treatise ‘on the topic of freeways. The reader is referred to the HCM for more in depth, investigation of the topic. Equipment needed to ascertain densities can be very minimal or very extensive. ‘The choice lies with the investigator. The study procedure presented takes the minimalist approach and uses note taking materials, data collection forms, a stop watch or other such timing device and a measuring tape or wheel. Using the minimum equipment requires the use of two field personnel. If available, the engineer may wish to take advantage of speed measuring devices such as radar ot laser units or permanent in place speed detectors. Volumes may be recorded using portable clectro-mechanical or electronic traffic counting units or permanent traffic detectors. Video monitoring may be used as well if real times can be obtained from the images. 1. Choose a site and time to observe the traffic. ‘The location of the study should be an reeway traffic is high speed traffic so the observation point should be chosen with safety in mind. The important con- sideration here is the ability to sce every vehicle which passes and the ability to identify the ‘speed of randomly chosen vehicles. These points become somewhat mute if the investigator has portable traffic counting units with speed detecting capabilities available. The units will greatly reduce field time, however, will greatly increase the risk of injury, since nearly all such. devices must be placed across or in the lanes of moving traffic. The ability to observe the traffic is not the only requirement of the site. A point at which the influences of interrupted flow facilities, namely on and off ramps, are minimum is desirable. Most attractive is the section of the freeway lying midway between interchanges. It can be assumed that the influences of merging and diverging traffic flows are minimal at this loca- tion and thus, data collected here best represents the uninterrupted flow condition being assessed. _ When using permanently mounted video monitoring equipment the engi-- (Giovement is recorded/in real time) The time will be used to compute the speed measurements which are critical to the process. (CORE MOMFU NARA WAGER to demonstrate the process and the investigator is free to choose any hour of the day to conduct the study. The 100 Freeway Densityonly requirement is that the hour be during daylight hours. An hour during non-peak periods is best for demonstration purposes. Collect data. Following site selection, the investigator should lay out a(gpeed trap follow) ing recommendations shown in the spot speed study. Again, since freeway trafficis high speed traffic, proper safety precautions should be taken. The speed trap must allow for times to be measured for any lane and the mark- ings used to denote the start and end easily seen from the chosen observa- tion point. If possible, the observer should make use of existing markings in or on the edge of the roadway. By doing so, accuracy should be maintained land results obtained without the data collection activities interfering with the traffic stream Two data collection forms are included for use in this phase of the study, The speed form allows for the collection of 100 randomly select sspeeds, This number is consistent with the number of observations recom- mended by the HCM when measuring free flow speed on a facility and is sufficient to demonstrate the process. THESecond jformnisjused t6/collect the volume data. ‘At the start of the hour one observer starts the timer and immediately starts recording the volume. Since the number of heavy vehicles is needed for the density computation, each and every vehicle is recorded by type. ‘The types recorded are passenger vehicle, truck, bus, and recreational vehi- cle and follow the HCM definitions of each. While the volume data is being collected, the second observer will collect the speed data. Randomly selected vehicles at the rate of approximately ‘one every 30 seconds will be timed through the speed trap and the values recorded on the collection form. The observer should choose vehicles from all lanes observed in the approximate proportion to the total number of vehicles using each lane. Since speeds are being measured to identify the average speed of the traffic stream, it is recommended that vehicles travel- ing extremely slow or extremely fast be avoided. Check data before leaving the field. The data should be checked for completeness before leaving the field. Special emphasis should be given to making sure that the minimum num- ber of observations were completed, the observations are within the expected range of values, and that any field observations needed to describe where, when, and how the data was obtained are complete. Freeway Density 101Data Analysis 102 1 Freeway Density Reduce field data. ‘The speed and volume data is summarized. The observed speeds are com- puted from the recorded times and entered on the data collection sheet. ‘These speeds are summed and divided by the total number of observations, in this case 100, and the average speed recorded Similarly, the volume data is summarized in order to find the hourly vol- ume and peak rate of flow. The peak rate of flow computed by multiplying the traffic measured during the peak 15 minute period by 4 will be used with the average speed to compute the density. When summarizing the vol- ume data to find the peak rate of flow the investigator should identify the vehicle mix occurring during the time frame. This is found by suraming the number of each vehicle type and dividing by the total number of vehicles observed during the peak 15 minutes. Results should be expressed in per- centage of the total traffic stream by type. For example, a typical break- down might be: 83% passenger cars; 10% tucks; 5% buses; and 2% recreational vehicles. Coimpute passenger car equivalents. Actual computation of the density is the division of the peak rate of flow by the average speed. With the peak rate of flow in vehicles per hour and the density in passenger cars per mile per lane, the computation must include conversion of vehicles to equivalent passenger cars and recognition that multiple lanes were observed. The following equation taken from the HCM converts the rate of flow to,equivalent passenger cars per hour per lane. wp = V/(PHF*N*fHV+fp) where: V/PHF = the volume divided by the peak hour factor = the rate of flow N = the number of lanes observed JAY = a heavy vehicle adjustment factor Jp = a driver population factor ‘Since the rate of flow has been determined by field observation, V / PHF can be replaced in the equation by R, where R = the peak rate of flow for the observation period. The driver population factor can be assumed to be 1.0 if the observation period occurred during or near the peak hour for the facility. Also, if the route chosen to assess is mainly a typical freeway and not a route handling significant tourist volumes, then the driver population, adjustment factor of 1.0 can be used. The investigator is referred to the HCM for further discussion regarding the driver adjustment factor.Freeway Density Study ‘Speed Data Collection Form Location: Observers: = Date:_. Time: Direction of Travek, Posted Speed: Weather Conditions: Freeway Density 103Freeway Density Study Volume Data Collection Form Location: Observeris: Date: Timo: Direction of Travet Posted Speed: Weather Conditions: time | Cars | Trucks | Buses | RV's | Time | Cars Freeway Density 105‘The heavy vehicle adjustment factor is used to convert the volume from vehicles per hour to passenger cars per hour. This conversion is necessary 30 that the units of density match those used to determine the level of service. ‘The heavy vehicles are converted to passenger car equivalents using the percentage of heavy vehicles observed and the terrain of the roadway. A detailed explanation of heavy vehicle adjustment factors can be found in Garber and Hoel including the following formula which is used to compute the adjustment factor: fie = WL PT(ET-1)+PR(E-RI)] where: PI = percentage of trucks and buses expressed in decimal form. PR = percentage of recreational vehicles expressed in decimal form passenger car equivalent of trucks and buses passenger car equivalent of recreational vehicles EF ‘The following table is used to identify the passenger car equivalents when the grade of the roadway is unknown. When the grade is known, then a more rigorous HCM method as explained in Garber and Hoel should be used. Mountainous ETifor trucks and, buses TER for recreational vehicles Given a freeway in rolling terrain and the example breakdown of vehicle type shown above, the heavy vehicle adjustment factor is calculated as: PAV =1/ (1 + .15(2.5—1) + .02(2.0~1)] = 0.803 Once the heavy vehicle factor has been computed, the passenger cars per hour per lane (pcphpl) can be calculated. Assuming the observer found a rate of flow of 3200 vehicles per hour for a 3 lane freeway, then the passen- ger car equivalent is: vp = 3200 / {(3) (0.803) (1.00)] = 1328.4 pcphpl Compute freeway density. As stated carlier, density is simply the rate of flow divided by the speed expressed in passenger cars per mile per lane (pepmpl) or: D= pl Freeway Density 107where D = density, V = the average speed, and yp = the rate of flow. If the observed average speed was 61 miles per hour, then: D = 1328.4 pephpl / 61 mph = 21.88 pcpmpl This calculation is performed for the observed data and record the results. |. Determine Level of Service. The computed density is compared to the HCM limits for the various levels of service. The level of service for a roadway is the measurement of the quality of flow on the uninterrupted flow facility. The levels of service for a freeway range from Level of Service A through and including Level of Service F with A representing the highest quality of flow and F the lowest. ‘Numerically the levels of service are defined by the density. The ranges for each level are shown in the table. Levels of Service for Basic Freeway Sections oo Level of Service ‘Maximum Density (pepmp!) 4 varies Comparing the example density of 21.88 pepmpl to the tabular values reveals that for this example the projected level of service is C. If C is con- sidered the lowest acceptable level of service for a freeway, then it can be said that this freeway is operating within the design limits. However, for each level of service for a freeway, a minimum speed criteria must be satisfied and that minimum speed is based on the Free Flow Speed of the roadway: Eree flow speed is the mean speed when measured when the equivalent hourly flow rate is no greater than 1300 pephpl. Therefore, if the value found for vp is 1300 or less, then the average of the speeds col- lected represents the free flow speed. In many cases when the density indicates a level of service of mid C level or higher, the average of the observed speeds will be the free flow speed. As the density approaches the limits of level C, vp will usually be greater than 1300 and the average of the observed speeds can not be used. In cases such 108 Freeway Densityas this, the posted speed limit plus 5 mph may be viewed as a good estimate of the free flow speed. Having found the free flow speed, a check of the minimum speed require- ments for the level of service indicated by the density can be accomplished. The following table lists the minimum speed requirements for the each level of service for 5 different free flow speeds. If in the example, the free flow speed was found to be 60 miles per hour, then the 61.0 mph average speed observed would confirm that the level of service is C. Minimum Speeds for Levels of Service 60.0 70.0 60.0 70.0 60.0 68.0 58.0 40 51.0 53.0 varies varies 5. Summarize and interpret results. Determination of'the level of service and the exercise of collecting the data and calculating the results should be summarized in report form address- Ing the questions which injtiated the study. Tn addition, questions which may be answered by this process include but are not limited to: + At what level of service is the roadway operating? + Was the traffic flow fairly consistent throughout the observation period? + Can this method of observation and analysis be used to assess the traffic impact of future land use in the area? How? + What can be done to improve the level of service at this location? + Describe one way that this method can be used to provide “real time” information to better manage the transportation network. ‘+ Investigate the HCM method for the analysis of multilane highways and evaluate the procedure presented herein for its applicability to multilane highways. If it can be used, then what changes would be required? If not, then why not? Freeway Density 109110 12 Vehicle Occupancy Study = MEE Measuring Persons per Vehicle transportation engineering. The statistic which best relates the number of people being moved to the number of vehicles being used to move them is the vehicle occupancy rate. This measure of effectiveness is usually expressed in persons per vebicle or persons per passenger car and is useful in many areas of transportation engineering One of the obvious ways the vehicle occupancy rate can be used is to assess transportation management strategies. Identification of the change in vehicle ‘occupancy rate as the result of a change in the transportation system can be used to measure the effectiveness of the system modification. For example, a high occu- pancy vehicle (HOV) lane on a freeway can be studied for its effects by comparing the vehicle occupancy rate of vehicles entering the roadway before and after it has been opened. Similarly, the vehicle occupancy rate can be used to assess the effects on the traffic stream of transit services that utilize the HOV lane. ‘Another use of the vehicle occupancy rate concerns before-and-after studies of parking management schemes to identify the number and type of facilities being used. Also, management strategies such as car pooling or ride sharing are evaluated and fine tuned to reduce the amount of traffic on the roadway, thus improving air quality. ‘The great potential of ride sharing or car pooling for reducing congestion problems cannot be understated. Merely increasing the vehicle occupancy rate from 1.0 to 1. persons means reducing the number of vehicles using the road- ways by nearly 10%. It follows that air quality could be expected to improve nearly 10%; the delays encountered by vehicles throughout the system could be expected to decrease by nearly 10%; and many. traffic problems could be solved or improved simply because of fewer vehicles using the system. With so many potential benefits, and so many resources being used to build and manage the roadway network, it is troubling that relatively few resources are devoted to the marketing of ride sharing and car pooling, ‘The exercise contained in this chapter presents the steps to be taken to assess the vehicle occupancy rate. The step-by-step procedure is used to compute the T: safe and efficient movement of people and goods is the primary goal ofvehicle occupancy rate of passenger cars entering a freeway, but it also can be used to evaluate almost any type of transportation facility, including multilane high- ways and two-lane rural highways. Parking lots and garages serving office build- ings, office parks, institutions, and shopping areas can be evaluated as well. Equipment Note-taking materials and a clock or watch to keep time is all that is necessary, although other equipment can be used. Videotaping equipment is helpful, allowing replay of the activity to check occupant counts. Data Collection 1. Select a site, and identify best observation point. For this study, the presence of the observer can be obvious to the drivers. In fact, it may help by slowing the passage of vehicles, thus making occupant count easier. A single-lane freeway entrance or exit ramp is chosen—preferably with 4 grade—so that the observation point is slightly higher than the elevation of the vehicles to be observed. This makes an easier task of identifying the number of occupants in cach vebicle. Information describing the ramp and the location of the observation point should be entered on the collec- tion form. ‘ 2. Select the sample size and sample time. Often, the intent of a vehicle occupancy study is to obtain information about a specific group of transportation network users. This group may consist of all vehicles using a facility during a specific time or a using a spe- cific location within the facility, or both. If the effort focuses on commuters travelling to a certain industrial park, then the data should be obtained during peak hours for that park. Similarly, should a large traffic generator be of interest—such as a manufacturing plant or college—then the time of + shift or class session change may be the best time to collect the data, In such situations, all vehicles entering and/or exiting the location may be measured for a predetermined length of time, usually an hour. Examining the behavior of all motorists at various points along the freeway system during an extended peak period, is not economical, and it is unnec- essary if a valid sample size is selected. The procedure for determining a statistically-valid sample size when secking the vehicle occupancy rate, and for other transportation studies, is introduced in the appendix. For the study of the freeway ramp, the sample size should be 400 vehicles or one hour of data collection (but a minimum of 50 vehicles}, whichever is reached first. Vehicle Occupancy Study 1113. Prepare to collect data. The investigator can collect the data using one of two procedures that differ in the level of driver participation in the collection process. In the first, the observer views each vehicle as it passes and records the best estimate of the number of occupants. This can prove difficult at times, since small chil- dren can be obscured from view as the vehicle passes. Moreover, tinted windows on many vehicles make accurate occupant counts very difficult. In the second method, the motorist is informed through signage that they are part of the study and are asked to show how many people are in their vehicle. The second approach can yield very good results if the signage is presented well in advance of the observation point, allowing drivers enough time to respond. With so many vehicles having tinted windows and the difficulty in seeing small children in the rear of sport utility vebi- les and vans, this method should prove more accurate than the first. Signage should not interfere with the driving task. Long, relatively straight, interchange ramps are suited to this method. Two signs, each placed approximately 100 to 150 apart, are needed. ‘The first should read “Vehicle Occupancy Study” and the second “How many. in vehicle?” or words to that effect, The signs should adhere to signage guidelines found in the Manual on Uniform Traffic Control Devices. 4. Collect data. : At the start of the observation period, the clock is started and the number of occupants in each vehicle is recorded (page 115). Since this study is focusing on passenger cars, only the number of occupants are recorded for passenger vehicles; these include autos, motorcycles, light trucks, vans and sport utility vehicles. If video taping equipment is being used, then the camera should be set up next to the primary observer at approximately the same eye height so that a record of what the observer views is obtained. 5. Check the data before leaving the field. All data should be checked for accuracy and completeness before leaving the field. Signage should be removed as soon as data collection is complete. Data Analysis ’ 1. Compute the vehicle occupancy rate. ‘The average vehicle occupancy rate is found by summing the number of occupants and dividing by the number of vehicles. The standard deviation is computed using the common expression: ‘ 45 where Xj = the ith observation, and NV = the number of observations, 112 Vehicle Occupancy StudyConstruct frequency groups, and compute frequencies. The Vehicle Occupancy Rate Data Summary Form (page 117) allows a grouping of the observations so that a cumulative frequency curve, similar to that for spot speed studies, can be produced. The number of observations showing one occupant is entered in Column 2, Number Observed, and so on for the entire range of values until all observations are represented on the summary sheet, Next, the frequency for cach entry is computed by dividing the number of observations in Column 2 by the total number of observations and record- ing the answer in Column 3. The decimal value is converted by multiplying by 100 so that it is expressed in percent. Cumulative frequency is simply the total percent of the observations that were observed to be equal to, or less, than a certain value. Plot frequency and cumulative frequency curves. ‘The frequency curve is plotted using the number of occupants, Column 1, as the abscissa (X-axis) and the frequency, Column 3, as the ordinate (Y-axis). This plot should illustrate the randomness of the observations. If this plot is bell shaped, then the occupancy distribution is random. However, if it is skewed toward either the lower or higher values, then other useful conclusions can be drawn, ‘The cumulative frequency curve can be plotted using the number of occu- pants, Column 1, as the abscissa and the cumulative frequency, Column 4, as the ordinate, The plot is produced to illustrate the distribution of the observations and to graphically identify the relationship of the average and the median to the rest of the observations. Useful values that may be obtained directly from this plot include the 50th percentile or the median, the 85th percentile, and the 90th or 95th percentile. Interpret the results, and draw conclusions. ‘This step in the process requires that the observer focus on the reason the study was accomplished. If the purpose of the study was to ascertain ‘whether ride sharing is above a particular level, then few conclusions are needed. However, many inferences can be made about the driving public and its performance during the observation period. Included among questions that may he addressed through this procedure are: + What is the average vehicle occupancy rate for the sample? The 50th per- centile? The 90th percentile? + What do the values identified in response to the previous question imply about drivers’ perceived value of ride sharing? Vehicle Occupancy Study 113* Do the data show higher occupancy rates during specific periods of time? « What, if any, difficulties in the data collection process affected the quality of the observations obtained? How can these difficulties be overcome? + What effect does ignoring transit vehicles and heavy trucks have on the study? What would their inclusion add to the study? ++ Ifa goal for the area is to increase ride sharing, what suggestions can you make to achieve this goal? ‘A summary report that includes all data, computations, and conclusions should be prepared and presented. 114 Vehicle Occupancy StudyVehicle Occupancy Rate Data Collection Form Observer/s: Date: Start Time: Finish Time: Comments: Note: Form will accommodate 25 observations and a total per column. Fill Column 1 before proceeding to next column and so on. Vehicle Occupancy Study 115Vehicle Occupancy Rate Data Summary Form Location: Observeris: Date: Start Time: Finish Time: Comments: ‘Number Observed 2 ee Average = Standard deviation = Vehicle Occupancy Study 11713 Origin—Destination Study === EE Where and When Do They Travel? travel in the future so that development of the transportation network can best accommodate the public’s needs. An essential ingredient of the trans- portation planning recipe is the origin-destination study. It provides the planner with the existing travel patterns and characteristics of the population, so that future needs and desires can be realized. Decisions including development of transit routes, widening of highways, dnd extensions of rail lines use results of origin-destination studies as a major criterion for action. ‘The area of study and its purpose must be properly predefined for an origin~destination study.to be useful. Gathering the travel characteristics of the population is a complicated and labor-intensive task that demands pre-planning of all activities. Testing of data-gathering methods prior to their implementation in the field is a necessary step to’ avoid wasted efforts. Data needed to decide when and where to extend a rail transit line is very different from that needed to decide the location of a bus terminal. ‘The best methodology for acquiring the data differs as well. Many methods are used to obtain the travel patterns, and most involve a direct interaction between the investigator and the public. Some of the methods include home interviews, roadside interviews of drivers and passengers, terminal interviews of transit riders, and in-route interviews of passengers. Other meth- ods are telephone interviews, photographic techniques, license plate surveys, employee questionnaires, and direct-mail questionnaires. A key component to the success of any of these methods is to secure the cooperation of the sampled population, otherwise, data may not be valid. In addition to the method employed, the complexity of the origin-destination study will vary in direct proportion to the size of the area to be studied. A study of the travel patterns affecting a bus terminal may involve merely interviewing artiv- ing drivers and recording data about their trip. Useful data in this case should be casy to obtain and verify. On the other hand, a study aimed at an update of a land use plan for a metropolitan county means dealing with a very large proportion of the traveling public. All travel patterns into and out of the county, as well as patterns within the county, need identification and documentation. Teens! planners are required to predict when and where people will 19Equipment Data Collection ‘The exercise presented in this chapter is a small origin-destination study that will give the investigator a taste of what these studies are like, the complex- ity of their development, and the limitations of the data-collection metiiod. tis hoped that this will provide the investigator with an appreciation of the enor- mous task with which transportation planners are faced when planning for the future mobility needs of the public. ‘Whereas many other methods are available, the interview is employed here to facilitate easy implementation on a campus. Equipment needed includes writing instruments and clipboards, printed questionnaires, and note-taking materials. Alternately, video equipment may be used and the interviewee’s responses recorded for summary at a later date. Two or more investigators are required. 4. Define the scope of the study, and identify the target group. This origin-destination (OD) study will seek to identify the travel patterns of a small population. The investigator should choose a target population such as those arriving at a building or business during a specific time. For this exercise, the target population may be all students arriving at a specific building or restaurant during the peak hour. Others may wish to assess the lunch hour of their place of employment. Before any work is accomplished, the investigator needs approval from the appropriate authority to interview the subjects. Whatever location is chosen for the study, it is the purpose of the study to identify the travel pattems of the interviewees regarding how they arrived at the building or, if exiting, where they are going. Gathering of this data should provide an indication of what future travel patterns and transporta- tion system modifications may benefit these users. 2. Refine the method by which data will be obtained. ‘Although there are many different methods to obtain O-D data, itis essen- tial to refine the selected method to suit the application at hand. For this investigation the interview method is chosen, calling for the interviewer to ‘ask questions of users and record their responses. Simple as it may sound, there are many ways this can be accomplished. The interviewer might verbally interview each person, but this may develop an impatient queue of people, jeopardizing data reliability. ‘To facilitate fast processing of interviewees, the investigator can hand peo- ple questionnaires as they pass by. These can be filled out at the person's convenience and dropped at a central collection point. However, the 120 Origin-Destination Studyresponse rate will depend on whether the sampled population takes the time to provide the responses and make the effort to deposit them at the collection point. An alternate method is to require that the question- naires be immediately filled out and returned. By doing so the probability of a queue developing is decreased, since multiple respondents are answer- ing the questions at the same time. ‘This exercise can be carried out in classroom buildings where people can be ‘expected to stay at a certain location during the assessment period. The investigator can visit each classroom while classes are in session—or visit each department during a specified hour—and interview people or distrib- ute the questionnaires. This method is recommended for this exercise. Develop the questionnaire based on the information needed. Nearly all origin-destination studies try to answer the following questions regarding the travel characteristics of the sampled population, and this study does, as well. The questions are as follows and may be concerned with a single trip or many irips throughout the day or week. + When did they travel? Specifically, what time of day did they travel both to and from their origins and desfinations? * Where did they travel? Where did they begin and end each trip? + How did they travel? Which mode or modes of travel were used when making each of the trips? * Why did they travel? What was the purpose of each trip? Was the trip purpose to reach a specific destination or was there a stop on the way to another destination? + Who was traveling? What were the characteristics of the people being studied? For this exercise, the study will focus on profiling the typical traveler and the travel patterns that led them to the study location. “Who are you?” and “How did you get here?” are the primary questions of interest. The ques- tionnaire on page 123 will be used for this study. Collect the data. ‘The data-collection method used depends on the chosen location. Assum- ing that a classroom building is the location, then the interviewer visits all classrooms during one class period and has each occupant complete the questionnaire. Origin-Destination Study 121Data Analysis ‘The time of day when data is collected should be considered so that a suffi- cient sample may be obtained. The Appendix contains more information relating to the choice of sample size for this and other studies. Interviewers choosing to sample people entering and exiting a facility should use the same questionnaire and strive to keep the queues of waiting people (0 a minimum. One hour of data will be sufficient as long as the number queried is greater than 30. Check the data on each questionnaire. ‘The completeness of each respondent's questionnaire, should be checked before termination of collection efforts. The investigator should quickly check all parts of the questionnaire when received and make sure responses to all questions have been entered. Summarize the data. ‘The Travel Information and Data Summary Form on page 127 provides for the reduction of the data collected on the questionnaires. The form is simi- lar in layout to the questionnaire for easy interpretation of the results. Where applicable, spaces are provided for the mean and standard deviation of the responses. The mean is the average response and can be calculated for each question where the respondent is asked to fill in an answer. This is also applicable to questions where the respondent is asked to rate an activ- ity or experience on a quality level from low to high. For example, ques tions such as those asking the respondent to rate the quality of the roadway system used during a trip on a scale of 1 to 5 can be summarized using the mean. The mean is found by weighting each level of response, similar to the way a grade-point average is calculated. The standard devia- tion is computed in much the same way, using an assigned numerical value to each level of response. : Other questions, such as “Which mode of travel did you use?” lend them- selves to a simple summary of responses expressed in percentage of the sample. Interpretation of the responses is limited, unless the question allowed énly two responses similar to a “Yes-No” question. If so, then the percentage can be indicative of the majority opinion of the population as a whole. Interpret the results. To aid in the interpretation of the results, the investigator should consider graphical representation of the summary. Graphical summaries similar to the pie charts and bar graphs commonly found in newspapers such 2s USA TODAY, are very effective at presenting the findings of a survey. O-D study 122 Origin-Destination StudyTravel Information Form Please help us identity the transportation needs for our area by completing this form. Please print, or circle answers. Your help in this study is greatly appreciated. 1. How did you travel here? Auto. Bus Taxi_—-Bike Walked Motorcycle 2. If you traveled by automobile, did you uavel here with someone? Yes. No 3. Was transit an option for you? Yes No 4, Where did you tome from? Home — Work Shopping Other 5, How many miles did you travel to get here? 6. To the nearest 30 minutes, when did you artive here? 7. To the nearest 30 minutes, when will you leave here? 8, What was your total travel time from thé start to the finish of your trip? 9. Should car pooling be encouraged in this area? -Yes_——No 10, Should transit service be expanded? Yes. No ; 11, What is your: age? sex? M F Thanks again for your help with this important study. Origin—Destination Study 123Travel Information Form Data Summary Sheet Investigator. Date Survey Taken: Location: Duration: Comments: L. Mode of travel? Auto Bus Taxi Bike Walked Motorcycle 2. Carpool? Yes__-% ~=No__% 3. Transit available? —Yes_% +No__% 4. Origin, Home___% Work____% Shopping ___% Other__% 5. Travel distance?. Average j Standard deviation ; Range 6, Arrival time? Average ; Standard deviation ; Range 7. Departure time? Average ; Standard deviation 8, Travel time? Average ; Standard deviation 9. Encourage car pooling? Yes__% +No__% 10. Expand transit? —Yes__% No___% 11. Age: Average ; Standard deviation ; Range 12. Sex: M, % F % ‘Total Number of Respondents: Origin—Destination Study 125findings are presented to decision makers to aid them in their efforts to develop transportation strategies. Few of these decision makers are engineers, and they are likely to have limited time available for the engineer or plan ner to present the information, Since they usually demand that the presen- tation be made as quickly and clearly as possible, the use of graphics is very effective. When viewing the results of the effort, the investigator should identify whether or not the goals of the study have been satisfied. At least, an effort should be made to see that the basic questions of who, when, where, how, and why have been answered adequately to ensure that significant conclu sions can be drawn, Summarize the data, and draw conclusions. ‘Many questions can be answered using the information obtained, For this study, some of the questions whose answers can shed light on the travelers being sampled include: + How do the travel patterns of male and females differ? + Is there a difference in how different age groups responded? + Does the distance traveled appear to affect how a respondent views car pooling or transit? ‘* What other conclusions can be drawn from the sample? ‘The final report should seek to draw as much information as possible from the data and present it in such a manner that its conclusions are easily understood. Origin-Destination Study 127128 Trip Generation Study ———— and Use. and Trip Production impact of development on traffic becomes more and more important. ‘raffic impact studies which address the pre- and post-development traffic conditions are produced by the traffic engineer to identify deficiencies in the transportation network and to develop a range of solutions to these potential problems. The key to producing successful estimates of future traffic volumes rests with the engineer's ability to iddntify a reliable tip generation model. ‘A trip generation model is a mathematical model used to predict trip-making behavior based on land use. It uses up to three types of independent variables: character, intensity, and location of the land use. The effect of these variables on anticipated trips (usually measured in vehicle trips per hour) can be computed for a specific land use, using a trip generation model. The Institute of ‘Transportation Engineers (ITE) has compiled models for various types of land use in their publication “Trip Generation.” These models and rates represent the average values for different land use types found throughout the United States and are used daily by traffic engineers in forecasting future traffic loading. Some states and locales have modified the ITE rates to better reflect conditions occur ring in their geographical region and require traffic engineers to use these local- ized models to forecast traffic conditions, Whether the engineer uses the ITE or localized values to forecast traffic loading, good practice dictates that the values be verified as representative of traffic generation in the area under study prior to their actual use, An engineer should not use the ITE value of 10.6 vehicle trips-per-day-per-dwelling for a subdivision without first checking one or more subdivisions to see if that value is representative of the trip-making behavior in that area. By first verifying the values to be used ina traffic impact study, the quality of the estimates and the significance of the conclusions drawn in the study are of much greater significance. Several methods can be used to either verify or develop trip generation rates for a land use type. The logical starting point is the ITE Trip Generation Manual. ‘The manual contains hundreds of land use types and variations. It was produced from actual trip generation studies conducted by engineers throughout the A s land use changes throughout a community, the need to assess theEquipment Data Collection country: It is continuously being updated by studies submitted to ITE for analy- sis and inclusion in the trip generation database. Those who conduct trip gener- ation studies are encouraged to donate their data to ITE for inclusion in the manual. The method to be used in this exercise focuses on verifying ITE rates so that a traffic impact study can be conducted. Many transportation planning efforts demand that trip generation models be much more detailed. Information about the types of trips in an area, and detailed characteristics of the people making the trips, should be included. This exercise does not address such detail, and the investigator is directed to the many transportation planning references available that deal with this topic in depth. This chapter's exercise allows the engineer to verify or produce trip generation rates for use in a traffic impact study of a proposed development. Through application of the data-collection techniques presented here, the investigator should be able to verify the applica- Dility of the ITE rates to the situation at hand and modify the ITE values to best represent the trip-making behavior most likely to occur at the site. The essence of the data collection part of this study is the measurement of traffic volumes entering and exiting a facility during a specific time frame. Because of this, the investigator needs minimal equipment. Although video monitoring, or electronic or electromechanical counters may be used, the study requires just, basic note-taking materials. 1. Choose a location and time for the study. ‘Many traffic generators are available for study; however, this exercise requires the investigator to look at the trip-making of a subdivision, a small apartment complex, or a fast-food restaurant having one drive-in window. ‘The choice should be based on the number of entrance/exit points to the land use and the number of observers or the equipment available. The essence of the data collection for this study is the measurement of vehicles entering and exiting the area, so the number of observers needed varies in direct relationship to the number of access points. It is recom- mended that any location chosen have only one or two access points through which all traffic must pass. Important to the decision of which facility to examine is the time of day when the data will be collected. This directly relates to how the trip genera- tion rate being verified will be used. If the rate is to be used to generate the amount of traffic expected on the adjacent streets during the morning peak hous, then the data should be collected during the morning peak hour of the adjacent streets. Should the trip generation rate of the facility during evening peak hour be of importance, then the data should be gathered dur- ing the evening peak hour. For this study the evening peak hour between 4 and 6 Ra. will be studied, Trip Generation Study 1292. Decide which independent variable will be examined. The purpose of the study is to verify that the ITE trip generation rates are applicable to the region being examined. The independent variables associ- ated with the land use of that type of region are reviewed and one is selected. The following table includes three land use types and some of the independent variables included in the ITE trip generation manual. ‘Land-UseType. ~ Independent Variables Used ‘Single Family Detached Housing Dwelling Units Vehicles ‘Actes “Apartments : Dwelling Units Parking Spaces Fast-food restaurant—one drive-in window ‘Gross floor area in square feet. Seats Parking Spaces It is recommended that dwelling uniis be used if a subdivision or apartment complex are being studied and that gross floor area be used if the study involves a fast food restaurant. When performing a study such as this, engineers usually choose the independent variable that has proven to be the best predictor of traffic flove The ITE manual identifies these for each land-use type. 3. Measure the independent variable. To calculate the trip generation rate for the land use, the independent vari able chosen must be measured, If the number of dwelling units is chosen for the subdivision, then the number of houses is counted. For a fast-food restaurant, the proprietors may already know the gross floor arca of the building. If not, then the exterior dimensions should be measured and the gross floor area calculated in square feet. 4. Record traffic data Vehicles entering and exiting the facility are recorded during the chosen hour. It is imperative that counts be taken at all driveways and access points simultaneously. The type of vehicle should be counted; although it is not important to verification of the applicability of the ITE trip generation rates, counting by vehicle type gives insight to the vehicle mix using the facility. Observers should record entering and exiting vehicles separately. 130 Trip Generation StudyData Analysis : z “ITE Trip Generation Rates 446P Peak Hour Total. 5. Check the data before leaving the field. Standard practice dictates that all data recorded in the field is to be checked before leaving the ficld. Investigators should check to see that the start and finish times are the same for all observers and that the data recorded will satisfy the analysis needs of the study. 1. Calculate the generation rate of the studied land use. The trip generation rate is calculated by summing the total number of vehi- cles entering and exiting the facility during the study hour and dividing by the magnitude of the independent variable used. For example, a 200-house subdivision is examined and found to generate 221 vehicles per hour dur- ing the evening peak hour of the adjacent roadways. Dividing 221 by 200, the magnitude of the independent variable, the trip generation rate meas- ured was 1.05 vehicles per hour per unit, Obtain ITE rates. ‘The ITE rates for the land-use type should be obtained from the trip gener- ation manual. Particularly needed are the average rate, the standard devia- tion of that rate, and the number of studies used to develop the rate. For the three land use types mentioned here, the following table of values can be used. These values are taken from the ITE Trip Generation Manual, 6th Edition, 197, Average trip rate ‘Standard deviation Number of studies LOLs = 105 ber diving vnie | one drive-in window “Fast-food restaurant— >| thousand gross-square feet 3. Calculate the confidence interval of the ITE rates. ‘To compare the ITE rates to the observed values, an analysis the reliability of the ITE rates must be conducted. The confidence interval of a trip gener- ation rate is the range of values within which the actual mean of all the trip generation rates for that land use will fall. For example, the confidence Trip Generation Study 131132 Values for t Sasa [= [= [| =| interval for a particular apartment complex is the range of trip generation. rates within which the average of the trip generation rates of all apartment complexes will likely fall. It is stated in terms of the probability of the true mean or average falling within the range. A statement that there is a 90% probability that the average grade point average of the graduating class will be between 2.50 and 2.90 is a statement of the confidence interval. Calculation of the confidence interval is achieved by the following equation: ts oat Scectt Se Va Va where x is the average of the sample (in this case the average ITE rate); s is the standard deviation; 1 is the number of studies; 1 is the true mean of the population; and t is the statistic of the f-distribution. Values for ¢ vary with the number of studies conducted and the reliability desired in the confidence interval. However, as long as the number of stud“ ies conducted to produce the ITE rate being examined exceeds 30, then the following values can be used. ‘Therefore, calculation of the confidence interval within which we are 95% confident that the true generation rate for a subdivision lies is: (0.89 <# <1.13] 1,96 1.05 1.96 + 1.05 101— <#<101+ [ ( VI ) a ( 294 I Stated another way, we can expect 95 of every 100 subdivisions studied to produce a rate between 0.89 vehicle trips per dwelling unit and 1.13 vehicle trips per dwelling unit during the evening peak hour of the adjacent streets. Compare computed value to measured value. For the example subdivision it was found that 1.05 vehicle trips per dwelling unit were generated during the observation period. This falls within the 95% confidence interval and implies that the ITE rates can be used. Computing the 90% confidence interval, the range of values is Trip Generation Studydecreased to 0.91 t0 1.11. The observed value still falls within the range, and the implication js still that the ITE rate of 1.01 trips per dwelling unit can be used. The investigator should compare the values for the observed land use. What do the observations imply? Care should be taken during this step that the investigator does not con- clude more than the sampling can support. Using the confidence interval as an indicator of the applicability of the ITE rates, based on a single obser- vation, provides only an indication how the values compare. To address the original question as to whether the ITE rate for subdivisions are applicable to the local region, multiple observations of like land uses must be con ducted and a comparison made of the mean of the observed rates to the mean ITE rate. While requiring more field effort than is practical here, an instructor or manager may wish to assign the same land use to multiple groups, thereby facilitating the aforementioned comparison. This should be done to produce ten or more data points to be used in the next step. Calculate the difference between the mean trip generation rates, and compute the confidence interval of the difference. This procedure is a simple application of the following computation and interpretation of its results. If the confidence interval of the difference between the average of the observed rates and the ITE rates spans a range from less than zero to greater than zero, then there is effectively no differ- ence between the two, and use of the ITE values is acceptable. However, if the range lies entirely below zero or above zero, then a significant differ- ence does occur at that confidence level. : Values for x, n, and s are the average, the number of studies, and the stan- dard deviations of the ITE studies and the 10 or more local observations, respectively. The f values used previously can be used here as Jong as the number of studies used to produce the ITE rates is greater than 30, and 10 ‘or more local areas are measured. (a ‘To illustrate this step, assume that 10 observations of subdivisions yielded an average trip generation rate of 1.10 trips per hour, with a standard devia- tion of 0.95. The comparable ITE rate is 1.01 trips per hour, with a standard ‘Trip Generation Study 133,134 deviation of 1.05 based on 294 studies. Computing the 90% confidence interval: 0.95" | 1.05? (1.10 — 1.01) — 1.645 = Hy < (1.10 = 1.01) + 1.645 ‘This reduces to 0.09 = 0.50 or an interval of —0.41 to +0.59. Based on the calculation, we are 90% confident that the difference between the ITE rate and the observed rate lies between —0.41 and +0.59. This shows that the difference could be zero, and it cannot be proven that one rate is significantly different from the other. Therefore, either rate could be used to forecast subdivision traffic in this area. To be conservative in the forecast, the higher of the two rates likely would be used. Summarize findings in report form. ‘At this point, the findings should be summarized and presented in report format. Questions that should be addressed include: + How important is a thorough knowledge of statistics to the transporta- tion engineer? ‘+ What did the field observations reveal? ‘+ Should ITE rates be used'wherever possible? Why? + How should independent variables be chosen? * How should the time and length of study be determined? * Did the data reveal that the development of unique trip generation rates for this locale should be considered? + How can the verification process be improved? Trip Generation StudyAppend x of the population as a whole. In order to analyze such data and reach valid conclusions, the investigator needs to know some basic statistical methods. In this appendix the basics of statistical methods needed to conduct the investigations in each of the chapters are presented. The discussion is limited to the equations and procedures needed to conduct the investigations and is not meant to be a substitute for a complete study of the topics. This appendix should serve only as a refresher of knpwledge gained in more a thorough study of statistics. Looking through the chapters, itis evident that only two basic data types— measured values and measured proportions—need to be addressed. Measured values are speeds, voltimes, queue length, and the like. Measured proportions are the parts of the whole—expressed as percentages—attributable to some activity; for example, the right-turn on red found in the compliance study. To analyze these two data types, and obtain statistically significant results, calls for applica- tion of the following statistics and equations. Ti ‘engineering decisions are based on data that represents the behavior Mean, Standard Deviation, and Variance Most traffic engineering data are assumed to be randomly distributed. This means that if all values were measured, not just a small sample, a graphical plot ‘would yield a bell-shaped curve. Some values would be high and some would be Jow, but most would be clustered about a central point. In a purely random pop- ulation, the midpoint of this curve, or central point, would be the mean or aver- age. The median is the value in the middle of the population—that point where 50% of the values are greater and 50% are less. It has the same value as the mean ina purely random population. ‘The mean or average for a sample is computed as follows: a4 yah ” where X= the arithmetic mean %, = sum of all observations x = number of observations 135‘The standard deviation of a sample/represents how varied the sampled data {s, and it is the square root of the variance. The variability of a purely random distribution determines the shape of the bell-shaped curve by defining the range of the values and how clustered the values are about the center of the plot. Distributions with data clustered over a very short range will have a small stan- dard deviation, whereas those dispersed over a wide range will have a large standard deviation. If the population from which a sample is taken is assumed to be a random distribution, then 68.3% of the data will fall within a range of +1 standard deviation of the mean, 95.5% of the data will fall within +2 stan- dard deviations of the mean, and 99.7% of the data will fall within a range of +3 standard deviations of the mean. ‘The standard deviation of a sample is computed as follows: where s = standard deviation of the sample, X; = the ith observation and the other values are as previously defined Confidence Limits of the Mean 1 136 ‘Appendix When looking at the mean of a sample we are looking at an estimate of the true mean of the entire population from which the sample was taken. The reliability of the estimate can be computed, and a confidence limit can be developed. This computation identifies the range within which the true mean of the population sampled is likely to be found, The range varies with the degree of reliability sought, For example, a 95% confidence interval is a range within which we are 95% confident that the true mean of the population will fall, and it is broader than the range of a 90% confidence interval. Computation of a confidence interval, or limits of the mean, is accom- plished by applying a standard relationship. The standard relationship presented ' here assumes a sample size =30. For sample sizes <30, different values for the f statistic must be used. The reader is referred to Probability and Statistics for Engineers And Scientists, 8th Edition, by Walpole and Meyers, for a further discus sion regarding sample sizes <30 as well as the derivation of the confidence limit computation presented here. Calculation of the confidence limit is accomplished by application of the following: w=kt ‘where 4 = true mean of the population, f = the statistic of the f-distribution, and the other variables are as previously defined. While the standard statistic4 varies, depending on the sample size, it can be assumed to be constant when the sample size >30, Values for ( are shown in the following table, Values fort Computation examples of the confidence interval for measured values can be found in the Trip Generation Study presented in Chapter 14. Confidence Limits of a Proportion Much like the confidence limits of the mean, the confidence limit of a propor- tion can be computed. Many of the assumptions are the same, including the requirement that the sample size be =30. For sample sizes <30, the reader is referred to a statistical text for further guidance. Proportions are used in many traffic engineering endeavors, including a Traffic Control Compliance Study, as shown in Chapter 10. In studies where a proportion is needed, the reliability of the estimate of that proportion found during sampling is a necessity. The confi- dence interval for a proportion is computed as follows: where p = the true proportion, f the sampled proportion, § = 1 ~ f = the sam- ple size and z, = the normal statistic for a. confidence level. For sample sizes >30, the value for z, and { are equal, Therefore f can be substituted in the above equa- tion and the table values appearing in the previous table used in the calculation. For example, should it be found by sampling 80 drivers that 20 favor spending gas tax money on promoting car pooling, then: + 196, [2D p= .25 * 1.96, sgetit ‘The computation concludes that the investigator can be 95% confident that the true proportion of all drivers who favor this use of the gas tax is between 0.16 and 0.34, or 16-34%, 09 Difference Between Two Means When trying to differentiate between two different situations, such as a before- and-after study of traffic, it is important to identify whether or not differences appearing between two samples are statistically significant or if they are attrib- utable to random sampling error. Two methods can be employed to differentiate between the means, First, the investigator can simply compare confidence intervals. If the range of the confidence limits overlap when computed at the same confidence level, Appendix 137then there is no significant difference between the values. For example, if the ITE trip generation rate for a single family dwelling has a 95% confidence limit of 0.89 to 1.13 vehicle trips per hour, per dwelling, and a study of the same generation rate for local subdivisions reveals a 95% confidence limit of 1.00 to 1.23 trips per hour, per dwelling, then no statistically significant difference exists between the two rates Second, the difference between the two values can be examined. This proce- dure is a simple application of the following computation and interpretation of its results. If the confidence interval of the difference between the average of the observed rates of two samples spans a range from <0 to >0, then it can be said that there is effectively no difference between the two. If the two samples repre- sented values for before-and-after studies, then no statistical difference has been found between the two. However, if the computed range is entirely positive or entirely negative then a significant difference does occur at that confidence level. -%, ¢ mn Values for X, n, and s are the average, the number of observations, and the standard deviations, respectively, of the two samples. Testing the above equation for various values of 1 reveals that the t values used previously can be used here as long as m and 1, are >30. In most situations, the traffic engineer can reduce the minimum 7, to 10 and still safely use the f values shown. To illustrate this step, assume that 10 observations of subdivisions yielded an average trip generation rate of 1.10 trips per hour with a standard deviation of 0.95. The comparable ITE rate is 1.01 trips per hour with a standard deviation of 1.05 based on 294 studies. Computing the 90% confidence interval requires the following: (1.10 ~ 1.01) — 1.645 My < (1.10 — 1.01) + 1.645- ‘This reduces to 0.09 + 0.50 or an interval of -0.41 to +0.59. Based on the calculation, we are 90% confident that the difference between the ITE rate and, the observed rate lies between —0.41 and +0.59. This shows that the difference could be zero, and it cannot be proven that one rate is signif- icantly different from the other. Therefore, either rate could be used to forecast subdivision traffic in this area. To be conservative in the forecast, the highest of the two rates would likely be used. Difference Between Two Proportions Comparison of proportions can be accomplished in much the same fashion as the comparison of two means. The confidence limits of two proportions can be calculated and compared, or the following calculation can be done. As was 138 AppendixSample Size stated previously, the values for £ shown in the included table can be used for z, and the following calculation can be performed when the sample sizes >30. Computing the sample size required for a study is difficult when the standard deviation is unknown. Assumptions regarding the expected standard deviation of the sampled population can be obtained through the review of other similar studies. From this work produced by others, the investigator can obtain the val- ues needed to compute the minimum number of observations needed to obtain data within a chosen confidence interval. In addition to some insight into the anticipated standard deviation, the permitted acceptable error of the estimate must be chosen as well and input to the following equation: Ey where ¢ = the permitted error of the estirhate, and the other variables are as pre- viously defined. Suppose we wish to determine the number of spot speeds that must be observed to estimate the true’ mean of spot speeds on a roadway, with a 95% level of confidence that our estimate will have a standard error <1.0 mph. If the standard deviation is expected to be 5.0, then the number of observations required is: _ [uso “| 10 2 | = 96 observations A similar approach can be taken to estimate the required sample size when @ proportion is the outcome. When estimating the sample size needed for a pro- portion, the investigator must estimate the results of the study in advance, since the magnitude of the proportion is a determining factor in the sample-size com- putation. The computation is as follows Examination of this equation reveals that when values of p and q are 0.5 and 0.5, respectively, n will be the most conservative estimate of the sample size for the chosen confidence interval and error allowance, For example, to calculate the sample size needed to identify the percentage of drivers who come to a Appendix 139Goodness of Fit 140 Appendix complete stop before executing a right-turn-on-red maneuver, and to do so with 195% confidence that the estimate is within 10% of the actual proportion, then: (1,967)(0.5)(0.5) 0.10? 96 observations In traffic engineering, the observer must often hypothesize as to the mathematical ‘equation or statistical distribution which best represents the traffic behavior being studied. An example is the Poisson study (Chapter 7). The statistical test used to verify whether the expected values—those computed from a mathematical ‘equation—and the observed values are sufficiently similar to validate the equation or distribution is referred to as a Goodness-of-Fit test or a Chi-squared test. In a nutshell, the Goodness of Fit identifies significant differences between the expected values and the observed values by computing the statistic Chi- square, x?, and comparing the computed value to the standard x? value. The fol- lowing equation is used for the computation: fpreap yo peee where the symbols o, and ¢; represent the observed and expected frequencies respectively for the ith cell. The decision criteria should not be used if the expected frequency of a cell is <5. The observed and expected (computed using the Poisson equation) fre- quencies from the Poisson distribution study in Chapter 7 are shown below. ‘The Goodness-of-Fit test will be applied to this data to determine if the Poisson distribution properly reflects driver behavior on this approach. - Number of Vehicles Observed Frequency |... Computed Frequency Computing the x? using the stated equation yields the following: (6 = 22.2" | (16 = 13 (7The reader will notice that the values for 3, 4, 5, and 6 vehicles were com- bined because the number of observations in the cells was <5. If the Poisson distribution is a good estimator of the arrivals, then the tabular or theoretical Chi-squared will be greater than the computed Chi-squared. In order to identify the proper Chi-squared, the number of degrees of freedom must be found. By definition, the number of degrees of freedom, v, in a Chi-squared Goodness-of-fit test is equal to the number of cells minus the number of | quantities obtained from the observed data that are used to calculate the expected frequencies minus 1. For this example, the average arrival rate was the only quantity obtained from the observed data to compute the Poisson frequencies. Therefore, the degrees of freedom for this example is 5 - 2 = 3. From the table shown, ¥7595 = 7.815 and 3.424 < 7.815, so it can be concluded that the Poisson distribution provides a good fit for the distribution of arrival rates. Values for Chi-squared 6 12.592 As stated earlier, this appendix is presented as a refresher for the investiga- tor using this work, not as a substitute for a formal study of statistics. Readers are urged to pursue a rigorous study of statistical analysis as part of their formal transportation engineering education. Appendix 141142 References EEE A Policy on Geometric Design of Highways and Streets, American Association of State Highway and ‘Transportation Officials, Washington, D.<., 2004. Banks, James H. Inteductin to Transportation Engineering, 2nd Edition, McGraw HU, Boston, ‘Massachusetts, 2001. Box, Paul C. and Joseph C. Oppenlander. Manual of Traffic Engincering Suadie.tnstivute of ‘Transportation Engineers, Washington, D.C, 1994 ‘Chapra, Steven C. and Raymond P Canale. Nipmerical Methods for Engineers, Fifth dition ‘MeGraw Hil, Boston, Massachusetts, 2006, Garber, Nicholas J. and Lester A. Hoel. Tafic and Highway Engineering, Fourth Edition, CL Engincers, California, 2008. Highway Capacity Manual, 2010. Transportation Research Board, National Research Council, ‘Washington, D.C, 2011, Homburger W.S. J. H. Kell, and D. D.Peskins, Fundamentals of Taf Engincering, 13th Eton Institute of Transportation Studies, University of California, Berkeley. CA, 1992. Khisty .Jotin and B, Kent Hall. Transportation gineering: An Introduction, Third Editon. Prentice Hall, Buglewood Cliffs, New Jersey, 2002, ‘Mannering, Fred 1. and Walter P. Kilaresk Principles of Highway Engineering and Traffic Analysis, + Fowih Baiion J. Wiley and Sons, New York, 2008 May, A.D. Trafic Flow Fundamentals. Prentice Hall, Englewood Cliffs, New Jersey, 1990. ‘McShane, Willam R. Roger P Roess and Elena 8, Prassas. Trafic Engineering, Second Baton, Prentice Hal, New Jersey, 1998. Murthy, ASN. end R. Henry Moble. Transportatin Engineering Basics. American Society of Civ Engineess, New York, 2002, Pignataro, Lous 5. Trafic Engineering Theory and Practice. Prentice-Hall, Englewood Cliffs, New Jersey 1973. ‘Trafic Engineering Handbook, 6h Editon. tstiute of Transportation Engincers, Washington, D.C, 2008. Transportation Planning Handbook, 3rd Editon, Instiuute of Transportation Engineers, Washington, D.C, 2008.‘Trip Generation, Institute of Transportation Engineers, Washington D.C. 2008, Welpole Ronald F. and Raymond H. Myers, Probably and Statics for Baginers and Seni, (6th Eition. MacMillan Publishing. New York, 1997, een reget Norman J Ashford and Rober J Stammer i: ranspriation Engineering. Witey and Sons, New York, 1998 Weis Paul Hand Radnor J. Paquette Highuay Engineering, Fifth Edo. J. Wiley and Sons, New York, 1987. References 143‘animomya aneinortome uledrazeoundinng simundernisito | 2 9 GA. 258) $ or Oh chet we an a aie aysit “xowniauesdtninuod Wiinurduiiaty lates vowind ss aisrivat 126 mvanhsewnysie vasa svnsjany panne 10140oo CENGAGE © Learning Tolear more about Cengage esnieg st win cengage com For your course and ering solos, vs wew.cngagebran.comn Purchase any of our products a you local bookstore or at our preferred online store wwew.cengagebrain.com o rgon-ags 97% Sacase 2494-3 118978) | 14 os7asi-9 | ete 20000