TRANSPORTATION RESEARC RECORD 121 19 Characteristics of Pedestrian Accidents in Montreal Central Bsiness District PRIANKA N. SENEVIRATNE AND EVAN M. SUSTER Pedestrian trips constitte a sbstantial portion of the total daily intra-central bsiness district (CBD) jorneys. As a conseqence, conflicts between pedestrians and vehicles contine to be a major concern for transportation planners and engineers. CBDs are expected to expand in terms of persons employed and persons entering the area for bsiness and social activities. Despite crackdowns on jaywalking and extensive traffic management schemes, the nmber of pedestrian accidents in the Montreal CBD has remained virtally nchanged between 1985 and 1987. Analysis of available data pertaining to the Montreal CBD sggests that more pedestrians involved in accidents are in the 2- to 29-yr-ld category than in any other age grop considered in the stdy. Over 4 percent of the accidents occr dring the 12-6 p.m. period and 8 percent are within commercial land se areas. Very similar characteristics are evident in the accident statistics for the Calgary CBD. The fact that almost 4 percent of the CBD pedestrians in Calgary are in the above age category, and that they walk the farthest bt always choose the shortest path, partly explains the higher rate of involvement of this age grop in accidents. The data available in Montreal are sfficient for reporting bt not for planning. The research sggests that frther detailed stdies are reqired to determine movement patterns and identify the highly involved grops and their need to travel as pedestrians. Means of obtaining and analying sch information are sggested. In many cities, over 7 percent of daily intra-central bsiness district (CBD) jorneys are made on foot. As pointed ot by Bondada (1), this figre may indeed be higher, depending on the availability of pblic transit services, climatic conditions, and density of land development. The primary prposes of the jorneys are dependent on the time of day. Dring the mornings and the early evenings, the jorneys are primarily work-based as area residents and people arriving in the area by varios modes sally walk from transportation terminals and parking facilities to their places of work, and vice versa. On the other hand, dring the midday period, many jorneys are for recreational or social prposes, with people walking to and from shops, restarants, and recreational areas. Depending on the time of day and the trip prpose, the length of the jorneys have been fond by Seneviratne and Morrall (2) to range anywhere from a few hndred feet to over one mile. Ths, the level of exposre or the potential for conflicts between pedestrians and motor vehicles is higher in the CBDs than in other areas. Unfortnately, some of these conflicts end in fatalities and many case serios injries to pedestrians. Compared to 1.4 percent of the injred car occpants who die in a given accident, Moore and Older (3) have Department of Civil Engineering, Concordia University, 1455 de Maisonneve Blvd., West, Montreal, Qebec 3G 1M8, Canada. observed that 3.8 percent injred pedestrians are fatal cases. In spite of these alarming figres, the general attitde of transportation agencies toward pedestrian needs and the approaches to managing pedestrian safety have not changed sfficiently in the last two decades. The same cold be said of the research efforts. On the safety side, the main emphasis of research has been on risk and identification of haardos sites. For example, over a decade ago Jacobs and Wilson (4) proposed the classification of roadway sections according to the level of risk, defined as the ratio of accidents in 2 1 /2 years to 12 minte pedestrian flow. The ratio is designed to identify the relative safety of different pedestrian facilities sch as ebra crosswalks, signalied crosswalks, etc. owever, according to the definition, the accidents wold have occrred prior to the period in which pedestrian volme conts are taken. Ths, the appropriateness of the ratio for evalating relative risk is qestionable. The more recent work [Grayson (5) and Jonah and Engel (6)] is qite similar to that of Jacobs and Wilson (4). The primary focs has been on identifying target grops and comparing risk levels. For instance, Jonah and Engel ( 6) examined the relationship among pedestrian crossing freqencies, walking distances, time spent on the streets, and a ratio defined as relative risk. While this type of ratio is appropriate for evalating relative risk in large poplations and sefl for identifying target grops or haardos locations in these poplations, the practical insight provided at the micro-level seems negligible. In other words, in calclating risk, the athors assme that the level of exposre to conflict (i.e., the instances in which pedestrians actally cross the path of moving vehicles) is similar for every trip and every location. An nderstanding of the haards and the high risk grops or sites is an important reqirement for formlating safety management schemes. owever, nless the factors inflencing risk or risk-taking behavior are clearly evident from the risk analysis, this knowledge and information alone is nlikely to provide the insight needed to formlate effective schemes. For this reason, one needs to nderstand pedestrian travel patterns and circlation needs. In this article, we discss a stdy of pedestrian accident characteristics in the CBD of Montreal, Qebec, Canada. An attempt is made to identify certain trends or significant factors that have contribted to the occrrence of accidents over a 3-yr period and to determine whether sch information cold be sed to plan appropriate remedies. We also discss the database as well as the presently sed data acqisition techniqe. Based on the Montreal experience and pedestrian accident characteristics of two other Canadian cities, we sggest certain isses that we feel are important in planning for pedestrians.
2 CARACTERISTICS OF STUDY AREA AND METOD OF DATA COLLECTION Urban Montreal, with a poplation of approximately 2 million, is the second largest metropolis in Canada. It is a highly cosmopolitan region sbdivided into several mnicipalities. The City of Montreal is one of these mnicipalities with a CBD that occpies an area of 6 sqare miles (1,57 hectares). The nmber of pedestrian accidents in the Montreal CBD has remained high and virtally nchanged over the last 3 yrs. In 1986, for example, pedestrian accidents constitted 48 percent of all accidents and 52 percent of fatal accidents within the City of Montreal. If this trend is allowed to contine withot immediate action to nderstand and correct the potential conflict sitations, the City's Jong term plans (7) to revitalie the CBD and make it an attractive place to Jive and work cold be ndermined. The area otlined in Figre 1 was chosen for analysis. It consists of 5 intersections and is approximately 173 acres (7 hectares). Travel patterns and demographics of the stdy area are classified in Table 1 according to 1986 censs data. The physical characteristics of accident sites were recorded dring field visits; the accident information was extracted from records maintained by the City of Montreal. The primary sorce of information for the City is police accident report forms. It shold be noted that accidents occrring at midblock are assigned to the nearest intersection. Ths, the nm- TRANSPORTATION RESEARC RECORD 121 bers represent reported pedestrian accidents in the entire road network in the stdy area. The records contained the following information for each of the 23 accidents that occrred in the stdy area dring the analysis period from Janary 1, 1985 to December 31, 1987: 1. Age of pedestrian; 2. Time and day of accident; 3. Date of accident; 4. Direction of travel of vehicle; 5. Severity of accident (fatal or otherwise); and 6. The approach leg of the intersection on which the accident occrred. In terms of physical characteristics, we recorded the availability of pedestrian signal phases (i.e., walk/don't walk signs), the predominant land se in the vicinity of the intersection, the nmber of signal phases, delays at intersections, and featres that indce conflicting behavior. The Statistical Package for Social Sciences (SPSS) was sed to analye the data. Ths, characteristics of the pedestrian as well as times and accident site details were classified as described below. In order to examine the characteristics of pedestrian grops that are more ssceptible to accidents, the involved persons were first classified according to age. Five age grops, which FIGURE 1 Stdy area within the Montreal CBD.
Seneviratne and Shster 21 TABLE 1 CARACTERISTICS OF MONTREAL CBD AND STUDY AREA Greater Montreal Urban Area Poplation (1986) 1,752,582 CENTRAL DISTRICT CCITY OFMQNTREAW Nmber of people entering for work 135, for stdy 2, for leisre 25, for shopping 4, other 28, BlKilQ L!ND.6B STUDY Area Poplation (1986) 7 hectares 2,15 were expected to reflect different lifestyles, were chosen. The times of accident occrrence were divided into seven intervals. In addition, accidents were classified into for general types according to the direction of travel of the vehicle. These inclded a) " direct hits," or conflicts that occrred when a pedestrian crossing a street was hit by a vehicle moving straight throgh; b) "left-trn hits;" c) "right-trn hits;" and d) "reverse hits," or vehicles backing p from parking spots and driveways. Only two physical characteristics of the accident site (adjacent land se and the availability of pedestrian signals) seemed to provide even a minimm of insight. The intersections were, therefore, classified according to the most dominant land se types, defined as commercial, residential, edcational, recreational (open spaces), indstrial, parking space, or mixed. As for pedestrian signals, sites were groped according to the existence of exclsive pedestrian walk/don't walk phases as well as co_ncrrent pedestrian phases. Where pedestrians had to depend on the vehiclar indication, the site was classified as lacking in pedestrian signals. Accident statistics for the CBD of Calgary, Alberta and City of alifax, Nova Scotia were sed to examine the existence of any similarities in the pedestrian accident characteristics in CBD areas. The data were obtained from the traffic departments of the two cities. The manner in which the data were smmaried made it difficlt to isolate CBD accidents from otside-cbd accidents in alifax. owever, alifax has a relatively small CBD compared to Montreal or Calgary and it is not clearly distingishable from the neighboring residential areas. Therefore, we consider the data for the entire city of alifax. ANALYSIS OF DATA Age of Pe"estrian The crosstabs fnction of the SPSS was able to highlight the relative ssceptibility to accidents of the different age grops. From Figre 2, when data are aggregated over the 3-yr period, it is apparent that approximately 28 percent of the accidents in Montreal involved pedestrians in the 2-29 category. The 3 to 39 and 4 to 59 grops were also involved in a significant nmber (-24 percent each) of accidents. Conversely, the -14 category was the least involved grop within the stdy. As expected, the involvement of 2 to 6 yr olds in approximately three qarters of all pedestrian accidents is de to the fact that the majority of the people who are present in the stdy area for different prposes (see Table 1) fall into this age bracket. As seen from Figre 3, the low percentage of involvement of the to 14 category in accidents is likewise the reslt of their low presence in the CBD. owever, a closer look at the age profile is necessary to estimate the relative risk, except in the case of the over 6 grop who are clearly a high risk grop on the basis of the 5 percent representation in the pedestrian poplation in the stdy area. Time of Day It is apparent from Figre 4 that more than 22 percent of all accidents occr in the p.m. peak (3:-6: p.m.) and 2 percent occr in the p.m. off-peak (12noon-3: p.m.) periods. Of the 2 percent of accidents that occrred dring the p.m. off-peak, almost 4 percent were between 12 noon and 1: p.m. This cold be attribted to the large nmber of jorneys generated arond noon for lnch and other social and recreational prposes. Similarly, both pedestrian and vehiclar traffic flows dring the p.m. peak period are more prononced compared to the a.m. peak period. This leads to more interaction between the two modes and, therefore, explains the difference between the nmber of a.m. peak and p.m. peak accidents. Moreover, it is apparent from Figre 4 that 13 percent of the accidents that occrred dring the 3-yr period were dring the late p.m. (9 p.m. to 12 midnight) period while 17 percent occrred in the early a.m. (12 midnight to 7: a.m.) period.
JO UJ 2 p:; UJ a. r... p:; a. 1-14 YRS 15-19 YRS 2-29 YRS 3-39 YRS 4-59 YRS 6 + YRS FIGURE 2 Percent of pedestrian accidents by age grop (1985-1987). AGE OF PEDESTRIAN 4 El 1985 r9 1986 UJ ml. 1987 p:; UJ a. r... p:; a. 3 2 1-14 YRS 15-19 YRS 2-29 YRS 3-39 YRS 4-5 9 YRS 6 + YRS AGE OF PEDESTRIAN FIGURE 3 Percent of pedestrian accidents by age grop.
Seneviratne and Shster 23!/l r.::a P':!/l r.::a r.::a "'... 2 1 r.::a P': r.::a "' EARLY AM AM PEAK AM OFF-PEAK PM OFF-PEAK PM PEAK EVENING LATE PM FIGURE 4 Percent of pedestrian accidents by time of day (1985-1987). TIME OF DAY As evident from Figre 5, dring the late p.m. period, the majority of accidents involved an eqal nmber of people in the age grops of 2s, 3s and 4 to 5s. Whereas dring the early a.m. period, the 2 to 29 yr olds who are more likely to remain in the city for social and other engagements are involved in 35 percent of the accidents that occrred dring that period. as shown in Figre 8, seems to be different from one year to another. Althogh data collected between 1981 and 1985 for another prpose indicate that the early part of fall sally experiences the highest vehicle volmes, as well as the fact that pedestrian volmes wold be higher becase of stdents retrning to school, the existing data are inadeqate to make a definitive conclsion. Land Use It seems that every year at least 8 percent of the accidents occrred at intersections located in predominantly commercial srrondings, which make p approximately 8 percent of the land se in the stdy area. Figre 6 indicates that, on the average, 1 percent of the accidents occrred in the neighborhood of the two niversities, which are also major generators of pedestrian trips dring the day. The land area occpied by the niversities is less than 8 percent of the stdy area. owever, in the absence of details on pedestrian volmes or trip generation rates, one cannot make definitive conclsions regarding the relationship between land se and accidents. Time of Year Early fall (September to October) appears from Figre 7 to be the period dring which most accidents occrred. The portion of accidents that occrred dring the remaining months, Pedestrian Signals Traffic flow throgh a large portion of the 5 intersections in the stdy area is one-way, and 8 percent of the intersections had walk/don't walk phases. Frthermore, right trns (right and/or left when it is one-way) on red are not permitted. Althogh it is difficlt to jdge withot knowing the exact position of the accidents within the intersection, Figre 9 shows that the portion of accidents that occrred while a vehicle was trning are mch less compared to the direct hits, which averaged abot 7 percent over the 3-yr period. owever, the proportion of left-trn accidents dobled from 14 percent in 1985 to 28 percent in 1987. The average share of reverse accidents remained relatively nchanged at abot 5 percent, and right-trn accidents declined from 14 percent in 1985 to 5 percent in 1987. At first, the high percent of direct hits appeared to be the reslt of the absence of walk/don't walk signs or other warnings for pedestrians and drivers. owever, over 14 percent of the total accidents occrred at the for intersections that had pedestrian walk/don't walk signals. Ths, if the ratio of the nmber of accidents to the nmber
3 EARLY AM (/) 25 AM PEAK m AM OFF-PEAK PM OFF-PEAK D PM PEAK 2 Pi: (/) 15 r.. 1 Pi: p.; 5 a EVENING 6J LATE PM - 14 YRS 15-19 YRS 2-29 YRS 3-39 YRS 4-59 YRS 6 + YRS AGE OF PEDESTRIAN FIGURE 5 Percent of pedestrian accidents by age grops and time of day (1985-1987). 1 (/) Pi: (/) r.. 4 Pi: 121 1985 1986 8 m 1981 6 2 COMMERCIAL RECREATIONAL EDUCATIONAL VACANT LAND USE FIGURE 6 Percent of pedestrian accidents by land se type.
3 en r.l lj lj 2.:r:!>: en r.l r.l p, ri. 1 r.l lj!>: r.l p, JAN-FEB MAR-APR AY-JUNE JULY-AUG SEPT-OCT NOV-DEC MONTS FIGURE 7 Percent of pedestrian accidents by time of year (1985-1987). 3 en r.l 81 [] 1985 fs:i 1986 EJ 1987 JAN-FEB AR-APR AY-JUNE JULY-AUG SEPT-OCT NOV-DEC FIGURE 8 Percent of pedestrian accidents by time of year. /\GE OF PEDESTRIAN
26 TRANSPORTATION RESEARC RECORD 121 ti) E-- 6 I( E-- ti) 4 "" E-- 2 "" El 1985 ISi 1986 m 1987 DIRECT REVERSE LEFT TURN RIGT TURN TYPE OF CONFLICT FIGURE 9 Percent of pedestrian accidents by direction of travel of vehicle. of sites in each category is considered, the intersections with pedestrian signals seem more haardos. Nevertheless, more information abot pedestrian and traffic volmes is needed to confirm or reject the hypothesis. Location In 1985, accidents were clstered at a few intersections within the stdy area. For example, eight accidents occrred at the intersection of de la Montagne and Ste-Catherine, while seven occrred at Peel and Ste-Catherine. In contrast, most other intersections experienced less than three accidents. Over the next two years (1986 and 1987), when no changes to the physical characteristics were made, accidents were more evenly distribted, with the maximm nmber occrring at any one location in the sample (inclding the high accident intersections mentioned above) being only for. Incidentally, even thogh it spans only 25 percent of the accident sites, Ste-Catherine Street experienced 35 percent of the 23 accidents. The larger portion of accidents may be explained by the relatively higher level of vehiclar and pedestrian movement on this primary artery at all times of the day. Day of Week The nmber of accidents that occrred dring weekdays was only slightly greater than dring weekends. The exception was Monday and Friday, when the nmber of accidents was twice as great as Satrday or Snday, as shown in Figre 1. This indirectly indicates the difference in the level of exposre dring weekdays and weekends. Conflict In order to permit comparisons of safety levels at different locations, we defined two simple measres. First is conflict, which we define as the ratio of pedestrian crossings per nit time to vehicles crossing the path of pedestrians dring the same period of time. In other words, the ratio of the nmber of pedestrians crossing while vehicles are attempting to trn or proceeding straight throgh to the nmber of vehicles attempting to trn or proceeding straight throgh dring the time periods when pedestrians have the right-of-way. The second is an accident rate, defined as the ratio of he nmber of accidents dring a given time period to the prodct of the nmber of vehicles and pedestrian crossings (defined the same way as in exposre) dring the same interval. Partial pedestrian and vehicle conts were performed in Jly 1988 at two intersections in the stdy area. One was at the intersection of de la Montagne and Ste-Catherine where there were 12 accidents in the 3 years. The other was at the intersection of Gy and de Maisonneve where 8 accidents occrred dring the same period. From these data and expansion factors obtained from partial conts at adjacent locations, the expected conflicting volmes dring the time of accidents were derived for the two locations. In addition, we observed pedestrian and driver behavior. Dring the p.m. peak period, pedestrians ai the latter intersection were law abiding and crossed with the fight-of-way
Seneviratne and Shster 27 1 8 rl 6 i:i:: Ill rl rl P r.. 4 rl rl P 2 [] 1985 [:::I 1986 ml 1987 SUNDAY MONDAY TUESDAY WEDNESDAY TURSDAY FRIDAY SATURDAY DAY OF WEEK FIGURE 1 Percent of pedestrian accidents by day of week. even thogh the average delay compted from 1-sec arrival rates was abot 67 percent of red time. The exception was approximately 2 percent who crossed against traffic and away from the intersection. The conflict dring this interval was 7.2 x 1-2, bt only one accident had occrred at this time over the period of 3 years. Therefore, the accident rate dring this interval was fond to be 2.2 x 1-4 accidents per pedestrian per 1, vehicles. Conversely, noon hor observations at the intersection of de la Montagne and Ste-Catherine, which experienced similar pedestrian volmes to the former site and averaged waiting times of 5 percent of red time, showed over 13 percent crossing the traffic stream withot the right-of-way. owever, the exposre of 6.1 x 1-2 is lower than that of the former case where the north-soth approaches were two-way. This site also experienced only one accident dring the noon hor over the same period, bt the accident rate dring the noon hor was 1. x 1-4 accidents per pedestrian per 1, vehicles. DISCUSSION OF RESULTS Several common characteristics became evident from the comparison of the three databases. For example, in alifax, the largest percentage of accidents also occrred dring the months of September and October, whereas Calgary CBD experienced the second largest share of pedestrian accidents dring the same period (approximately 2 percent) averaged over the 3 years (1985-1987). The largest share was in the May-Jne period, althogh the difference is negligible as seen from Figre 11. Also, the age grops of most individals involved were similar among the three cities. The percentages of the 3 to 6 yr olds involved in Montreal, Calgary (see Figre 12) and alifax were approximately 47 percent, 41 percent, and 49 percent, respectively. Also, in Calgary, 33 percent (from 1985-1987) of the accidents occrred dring the p.m. peak period (see Figre 13) as opposed to 2 percent in Montreal. On the basis of this information and information on pedestrian travel patterns [Seneviratne and Morrall (2)], it can be reasonably conclded that the higher level of exposre is the primary factor inflencing 3 to 6 yr olds to be involved in a relatively larger share of accidents. For example, the walking distance distribtions for Calgary indicate that this age grop walks farther on the average (-995 ft) than any other age grop. Frthermore, when selecting rotes, the pedestrians in the same category were fond to select the qickest path, which is likely to reslt in more crossings of conflicting vehiclar traffic. What may also be inferred from this is that, since pedestrians wish to gain access by the qickest (shortest) path, satisfying this criterion while minimiing conflicts is likely to be more effective than simply redcing conflicts throgh median barriers, reroting pedestrians, and grade separation. Sch measres may redce exposre, bt add to the walking distance, inconvenience the elderly and the handicapped, and may sometimes shift the conflicts to a mch more complicated site. It also works against the force attempting to provide eqal priority and space for pedestrians. Observations at two sample sites (de la Montangne and St. Catherines and Gy and de Maisonneve) between 3: and
4 Ul E-i 3 rx: E-i Ul r..:l "" E-i 2 l'l Ji; 1 1986 rj 1987 El 1988 FIGURE 11 JAN-FEB MAR-APR MAY-JUN JUL-AUG SEP-OCT NOV-DEC Percent of pedestrian accidents in Calgary CBD by time of year. MONTS 4 E-i Ji; E-i U) ll "" CJ E-i 3 2 1 r..:l Ji; ll III. 1986 1987 El 1988-14 YRS 15-19 YRS 2-29 YRS 3-39 YRS 4-59 YRS 6 + YRS NOT STATED AGE OF PEDESTRIAN FIGURE 12 Percent of pedestrian accidents in Calgary CBD by age.
Seneviratne and Shster 29 4 WI 1986 en 1987 m 1988 3 r:x: en r.. 2 1 r:x: EARLY AM AM PEAK AM OFF-PEAKPM OFF-PEAK PM PEAK EVENING LATE PM NOT STATED FIGURE 13 Percent of pedestrian accidents in Calgary CBD by time of day. TIME OF DAY 6:p.m. showed close to 17 percent of the eastbond vehicles on St. Catherines Street rnning the red light. Moreover, 6 percent stopped for the red light across the pedestrian crosswalk (ahead of the stop line) and 14 percent failed to yield to pedestrians while trning on green. While this small sample is insfficient to make a final conclsion, it sggests that the large nmber of direct-hit and trning accidents in Montreal cold be as eqally attribtable to aggressive driving habits as to inattentive pedestrians. This observation also sggests that exposre shold be defined in terms of actal conflicts as opposed to general ratios made p of the total nmber of vehicles on the road, nmber of street crossings, or total nmber of pedestrians. Sch macro-level ratios cannot provide mch insight to planners and engineers. CONCLUSIONS It was fond that the 2 to 29 yr age grop is involved in a disproportionately larger share of accidents within the stdy area. Frthermore, most accidents seem to occr between 3: p.m. and 6: p.m. These two pieces of information imply a strong relationship between pedestrian volme and traffic volme. In other words, dring the p.m. peak period the stdy area pedestrian volme consists of a large portion of persons in the 2 to 29 yr age category, and traffic volme is at its highest dring this period. In order to develop proposals for effective management of pedestrian, flow and safety, it is essential that p-to-date information concerning pedestrian volmes; walking distances; and travel patterns by time of day, trip prpose, and nder different weather conditions, as well as corresponding vehiclar volme data are careflly considered. In the absence of sch information, it is difficlt, if not impossible, to make inferences or increase or knowledge abot the tre factors that lead to a high level of conflicts. As is indeed the case in many other cities, adeqate information of this natre is navailable in Montreal and, therefore, the ability of athorities to design effective contermeasres is limited. Several extensive srveys are sally needed to acqire specific information. For example, an attitdinal srvey and a volme srvey shold highlight some of the deficiencies in the pedestrian network that reqire adjstments. Sch srveys are expensive and time consming. Nevertheless, the benefits that accre from efficient adjstments have been fond (8) in several areas to far otweigh the costs. The forms of adjstments made in these instances range from the creation of traffic-free ones to pedestrian/transit malls as well as the establishment of time-dependent traffic management schemes to balance pedestrian and vehiclar space in existing areas. The task of information acqisition shold be a continos and ongoing process. owever, extensive srveys are not reqired for this information pdating. It can be achieved by systematic sampling at a minimal cost. Sch sampling can be sed to establish growth factors, trends, and alterations in ser needs. Frthermore, the presently sed accident data management methods are not the most efficient and ser friendly. The information contained in standard police accident report forms is primarily geared to assist in legal proceedings rather than to provide details for rban planners. Even the little information that is of se is obscred by crrent data maintenance procedres. An effective database management system to store the information in a format that cold be qerried according to ser needs can, nevertheless, be easily created sing any database package. Sch a system can provide more accrate
3 and helpfl statistics related to the existing information. Similarly, information pdating and ftre srvey data can be stored and effectively analyed. ACKNOWLEDGMENTS The athors wish to thank Jean Bertrand and Phillip Oneson of the City of Montreal for providing the accident data. This research was fnded by the Natral Sciences and Engineering Research Concil of Canada. REFERENCES 1. M. V. A. Bondada. Intra-CBD Secondary Travel Patterns of Downtown Workers. ASCE Transportation Division. Janary 1982, pp. 15-26. 2. P. N. Scncviratne and J. F. Morrall. Analysis of factors affecting the choice of rote of pedestrians. Transportation Planning and Technology. Vol. 1, 1985, pp. 147-159. TRANSPORTATION RESEARC RECORD 121 3. R. L. Moore and S. J. Older. Pedestrians and motor vehicles are compatible in today's world. Traffic Engineering, 1965. 4. G. D. Jacobs and D. G. Wilson. A Stdy of Pedestrian Risk in Crossing Bsy Roads in For Towns. Report LR 16. Road Research Laboratory, Crowthorne, England, 1967. 5. G. B. Grayson. Pedestrian risk in crossing roads: West London revisited. Traffic Engineering and Control, Vol. 28, No. 1, 1987, pp. 27-3. 6. B. A. Jonah and G. R. Engel. Measring the relative risk of pedestrian accidents. Accident Analysis and Prevention, Vol. 15, No. 3, 1983, pp. 193-26. 7. Planning Statement for the Central Area. osing and Urban Development Department, City of Montreal, Canada, March 1988. 8. C. A. yde and. Johnson. Attitdes t Envirnmental Changes in Barnsley Town Centre (and Comparison With Similar Reslts From the Liverpool Stdy. Working Paper 12. Institte of Transport Stdies, University of Leeds, England, September 1978. Pblication of this paper sponsored by Committee on Pedestrians.