Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities (2012)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

and Sustainability by Dannenberg, Frumkin, and Jackson, published by Island Press in August 2011. This book, aimed at students but appropriate for all involved in community design, presents in- depth diagnoses of problems related to the built environment and offers practical treatments. TRB’s National Cooperative Highway Research Program (NCHRP) is preparing a number of guides and related products, some jointly with the Transit Cooperative Research Program (TCRP), on NMT safety. Included to date are NCHRP Report 500, Vol. 18: A Guide for Reducing Collisions Involving Bicycles, NCHRP Report 562/TCRP Report 112: Improving Pedestrian Safety at Unsignalized Crossings, and NCHRP Report 674: Crossing Solutions at Roundabouts and Channelized Turn Lanes for Pedestrians with Vision Disabilities. CASE STUDIES Special Mini-Studies in Montgomery County, Maryland Situation. The Handbook authors encountered, during chapter development, several areas of interest that seemed particularly poorly quantified. Some of these were relationships that the literature dis- cusses in qualitative terms but apparently without hard numbers to support the logic. Opportunities were therefore taken to conduct or obtain simple counts paired with careful observation to address the topics in question. The actions and situations covered, all illustrated in real life within Montgomery County, Maryland, include sidewalk reconstruction and expansion, sidewalk indirectness effects, a downtown mid-block crossing installation, volume variability in response to count timing and other factors, provision of a pedestrian connection to a transit stop, and hiker-biker off-road trail traffic mix relative to use of a parallel on-road bike route. Montgomery County is a generally upper-income sub- urb of Washington, DC, but the count locations are mostly located in older “down-county” transition areas between the higher and more modest incomes. Actions/Analysis. The actions and situations examined are described under the applicable paragraph headings: “Results” if a specific change is the subject, and “More” if a static condition was observed to develop the information. The analyses have consisted of taking brief-duration counts or obtaining pre- existing full-day intersection counts, and pairing the count information with descriptive analysis. The limitations of using mostly single-day counts of mainly short duration are partially counterbalanced by careful attention to detail and consistency. The results are presented without any tests of statistical significance. They are organized in order of their cross-referencing from the main body of this chapter. Results—Sidewalk Improvements. A State Highway 547 improvement project in Garrett Park, Maryland, was taken advantage of to obtain a small-sample count, before-and-after sidewalk improve- ments, under static land use and highway/transit network conditions. A morning 3-hour peak-period count taken in late January 2002 represented the “before” situation. The pedestrian and cross-street vehicular count elements were replicated under nearly identical conditions 4 years later, to provide a post-improvements “after” count.91 16-410 91 Pedestrian and partial vehicular counts were made from 6:45 AM to 9:45 AM at the intersection of MD 547 (Strathmore Ave.) and Montrose Ave. in Garrett Park, MD, on Friday, January 25, 2002 (in clear, cool, dry weather), and on Friday, January 27, 2006 (in clear, cold, dry weather). Walkways were fully clear of snow in both instances. An infant and the school patrols in the before condition, and a meter-reader and a bicyclist in the after condition, are omitted from the count presentation. These and other short-duration counts were staged and taken by the lead chapter author. The pedestrian crossing improvements referred to were, one block west, a 24-hour pedestrian-activated crossing signal added at a MD 547 crosswalk long controlled during school hours by an adult crossing guard, and two blocks east, a pedestrian-presence-activated in-pavement-lights crosswalk installation.

The before condition was a two-lane state highway with a rural cross-section and a poorly main- tained 4-foot-or-narrower asphalt sidewalk on one side for five blocks and none for an additional block. In the after condition MD 547 had been reconstructed with a two-lane curbed and drained urban section and ADA-compliant mostly 5-foot concrete sidewalks on both sides for five blocks and one side for the last block, connecting to a pre-existing side path and hiker-biker trail. In the time between the counts, pedestrian crossing improvements were made one block west and two blocks east. The improved section of MD 547 is entirely within the single-family-housing historic town of Garrett Park, laid out as an isolated railroad suburb with an irregular street grid reflect- ing 19th century “garden-city” influences, but now surrounded by post-World-War-II suburban tracts with limited access points into the town. MD 547 divides the community and carries an aver- age weekday traffic volume on the order of 20,000 vehicles. The 3-hour adult (and teenager) pedestrian count crossing the intersection parallel to MD 547 increased from five to 21 persons. Of these, one was accompanying a schoolchild in the before con- dition and five were doing so in the after condition. The comparable child pedestrian count, not including grade school patrols assigned to protect schoolchildren walking parallel to MD 547 in the before condition, decreased from 11 to six children. This 45 percent decrease may reflect exogenous effects, most obviously the termination of school patrol protection, but possibly enrollment varia- tions as well. Subtracting out adults walking schoolchildren, the adjusted 3-hour adult pedestrian count approaching the intersection along MD 547 increased from four to 16 persons. Thus the MD 547 adult pedestrian count increased between 320 percent (raw count) and 300 percent (adjusted to remove adults accompanying schoolchildren). Total pedestrians approaching via the cross-street, with no sidewalks but low vehicular volumes, increased from three to five, up 67 percent. The 3-hour two-way vehicular count on the side street went from 49 to 34 vehicles on the north leg and from 84 to 111 (including schoolchild drop-offs but adjusted for double-counting) on the south leg, a 9 per- cent increase over all—relatively insignificant compared to the adult pedestrian count increases. General observation suggests that sidewalk usage is substantially higher in warmer weather, including the block with no prior sidewalk and virtually no pedestrians in the before situation because of patently unsafe walking conditions. It is not known how much, if any, of the new pedes- trian traffic is diverted from parallel low volume residential streets mostly without sidewalks. In any case, this instance of a quadrupling in the adult and adolescent pedestrian count cannot be readily explained by other than the sidewalk addition and improvement. More—Sidewalk Indirectness. A townhouse condominium close by to the Grosvenor-Strathmore Station of Metrorail was built with sidewalks immediately adjacent to the curbs. The curbs and sidewalks jog inward at several locations to accommodate indented perpendicular parking, intro- ducing four right-angle turns. A 36-foot total of deviation from a straight line is produced at the particular location observed and counted. It imposes 27 percent extra in walking distance to get around the parking, or 7 percent extra distance as measured along the entire 500-foot condo- minium access roadway up to the first townhouses. Figure 16-10 illustrates the general layout and the primary results of the 5:00 to 6:30 PM weekday count. 16-411

The 90-minute count total of 46 pedestrians is inclusive of the peak evening outflow from the Metrorail station into the townhouse development. Most were walking into the development, but irrespective of direction, 80 percent chose to walk in the street behind perpendicularly parked cars rather than use the sidewalk. A total of five cars came into or left the 14 parking spaces during the count. Even with the relatively low vehicular volumes in the street, this is a design situation that should have been avoided. It is a clear demonstration of the oft-articulated observation that pedes- trians want to walk directly toward where they are going and will not stick to sidewalks not designed toward that end. In other Montgomery County examples, paths chosen by pedestrians to avoid indirectness can be observed in the form of existing or former dirt paths. Here are four such cases: • Just beyond the situation described above, the access roadway turns right, and the sidewalk goes around two sides of a single-bay parking lot. Pedestrians walk diagonally across the park- ing aisle, tracing a path through bordering grass, now afforded flagstones. The pedestrian- selected route is 200 feet while the sidewalk route is 236 feet (with two stair steps), such that the avoided sidewalk route suffers from 15 percent indirectness. • At the Grosvenor-Strathmore Metro Station itself, the original design provided a sidewalk on the most direct possible path north to a cross street and the nearest main intersection. However, a walk- way continuing north on the other side of the cross-street was introduced with a 101-foot offset along the cross-street. Metro riders forged a dirt path directly to a point opposite the walkway north. A short diagonal sidewalk reducing the 101-foot offset by 36 feet was to no avail. Ultimately a 260-foot sidewalk was built along the dirt path, resolving a 39 percent indirectness in the original route as measured from the point of path divergence, or 15 percent as measured from the station entrance. The new walk also avoids an unnecessary climb entailing roughly a 5-foot extra gain in elevation. • When Elm Street Park in Bethesda was refurbished, a 400-foot walkway was paved diagonally across it, more or less along the shortcut pedestrians headed for the downtown district had long been following. The new walkway resolved a 12 percent indirectness in the shortest pre- viously available paved pedestrian route. • An artistically designed zigzag walkway along a “paper street” right-of-way in Potomac requires 490 feet to cover a 420-foot distance. Dirt path traces give evidence that neighborhood 16-412 Figure 16-10 Example of pedestrian route choice in response to sidewalk indirectness. Note: PM peak 90-minute count along Cloister Drive entrance- way into Stoneybrook community, Monday, October 22, 2007, 5:00 to 6:30 PM, in and out traffic combined. 9 Pedestrians via Sidewalk 37 Pedestrians in Street 80 Vehicles and 2 Bicycles 137’ 18’ 4’ Sidewalk

walkers and cyclists reduce their distance to 446 feet at worst, a 6 percent indirectness as com- pared to the 17 percent indirectness of the paved walkway. Results—Mid-Block Crossing. The Maryland State Highway Administration (SHA) has of late been installing mid-block crossings at selected state highway locations. Sites picked are where jay- walking was rampant and the cause seemed to be that the “safe” route was circuitous enough to encourage violations. Some of these crossings are not truly mid-block but are at “T” intersections. Nevertheless, they do represent a departure from prior SHA practice of seeking to always locate pedestrian crossings at prominent traffic intersections. An example is the 2004 installation of a crosswalk and signal in Silver Spring on Georgia Avenue, a major arterial running nominally north-south, where Ellsworth Drive has a “T” intersection. The area, an older suburban downtown, was under redevelopment at the time. Adjacent Georgia Avenue intersections are Colesville Road, a major arterial approximately 400 feet to the north, and Wayne Avenue, a minor arterial 300 feet to the south. Ellsworth Drive splits what would otherwise be a 700 foot superblock on the east between Colesville and Wayne. It is a narrow street fronted by restaurants and other retail, and as part of the redevelopment, was made pedestrian-only during retail hours. The office superblock across Georgia Avenue on the west is triangular, coming to a point at the Wayne Avenue intersection. Office development was sited to provide a pedestrian cut- through across the southern point of the block, aligned with Ellsworth. Prior to crosswalk and sig- nal installation, Ellsworth Drive was treated essentially as an alley. Crossing Georgia Avenue at Ellsworth was considered jaywalking, but anecdotally, “A lot of people were crossing here any- way.” The Georgia Avenue median was and remains functionally continuous at Ellsworth Drive. The available counts contribute to understanding the pedestrian response to the new crossing, but because of seasonal differences, lack of completeness, and ongoing area redevelopment, offer only a partial view of crosswalk impact. A count of jaywalking in the before condition was not made. The earlier (2001/2002) of two all-day winter counts at Colesville Road (see “More—Volume Variability” below) was selected as the preferred before count, it having been made as part of a detailed pedestrian analysis. The other count (2003) seemed strangely low. For the Wayne Avenue intersection, a single summer count (2003) was available. The south Georgia Avenue crosswalk at Colesville Road handled 2,143 pedestrians in 13 hours in December 2001, including 213 in the noon hour. The north crosswalk at Wayne Avenue had 218 pedestrians in 13 hours in July 2003, includ- ing 40 in the noon hour. In the after condition, the one full-scale count was taken at the Ellsworth Drive crossing on August 3, 2004, about one month after opening ceremonies. The total 13-hour volume crossing Georgia Avenue at the new signalized crossing was 1,357 pedestrians including 243 in the noon hour. The new crossing saved most users almost 250 feet of walking indirectness out of 600 feet, a 40 percent savings in a block. Short-duration counts made later that August showed the Georgia Avenue south crosswalk to be serving 530 noon hour pedestrians (relative to 213 before, in the winter), with 42 noon hour pedestrians in the Wayne Avenue north crosswalk (relative to 40 before).92 Limitations in the data, and lack of any interviews, preclude any firm estimate of pedestrians diverted from other crosswalks, pedestrians not making the walking trip in the before condition, or pedestrians previously jaywalking. Nevertheless, the use of the new crosswalk together with 16-413 92 As a matter of record, it should be noted that the time of the August 2004 noon hour counts, a new multistory building on the east side of Georgia Avenue between Ellsworth and Wayne was not yet occupied, and the north side Wayne Avenue sidewalk east of Georgia was closed partway down the block because of construction.

the increase in both of the noon-hour after counts on the pre-existing proximate crosswalks, slight in one case but much greater in the other, is certainly suggestive of new walk trip crossings and improved mobility. It is obvious that the new signalized crossing meets a real need, significantly enhances downtown pedestrian circulation, and is supportive of the downtown revitalization. More—Volume Variability. One of the downtown Silver Spring intersection counts utilized in the mid-block crossing analysis above, the December/January 2001/2002 Colesville Road and Georgia Avenue count, illustrates how strongly pedestrian volume characteristics reflect the events of the day and/or the nature of nearby land development, and how major the differences can be among nearby counts, even within the same basic area type. In comparison with a count taken 1 year later, it also illustrates how much variability there can be between two counts at the same location, for whatever reason. The intersection in question is located at the core of the Silver Spring business district, but slightly off-center to the north. Immediately east of Georgia Avenue, the combined office and residen- tial density is greater to the north of Colesville Road, while the restaurant and other retail density is greater to the south. A Washington Metrorail station lies two blocks to the west with exits on both sides of Colesville Road, and sidewalk provisions are comparable all around. Either the land use differences or the fact that the 2001 count was taken just 5 days before Christmas was enough to make the lunchtime-crowd and afternoon-shopping pedestrian flow dominant in the south crosswalk, parallel to Colesville Road. On the other hand, AM and PM peak commuter traffic pedestrian flows parallel to Colesville Road were easily discernible on the north crosswalk, in January 2002. As in many such instances, it is not certain to what degree these results are related to the land use differences as compared to the seasonal difference (pre/post Christmas). Table 16-128 highlights the different time-of-day pedestrian flow patterns, and the even greater differences between the 2001/2002 counts and the 2003 counts, taken with different objectives and using different protocols. The earlier counts focused exclusively on pedestrians, in an examination of crossing behavior vis-à-vis pedestrian signal indications, while the drastically lower pre-Christmas 2003 counts were taken as an adjunct to vehicular traffic counts intended for capacity analysis purposes. 16-414

Results—Path Connection to Transit. Construction of a pedestrian/bicycle path and stream cross- ing connecting Montgomery County’s Randolph Hills neighborhood to the north with the Garrett Park MARC commuter rail station to the south provides a documented example of enhancing non- motorized transportation (NMT) access to public transit by means of trail or walkway construc- tion. Even with the small numbers involved, and without formal before counts, the Randolph Hills share of Garrett Park station MARC ridership illustrates the importance such connections can have. The 800-foot path was constructed in the mid-1980s across what had earlier been overgrown pri- vate property adjoining an abandoned coal/oil yard. A major function of the path is to provide neighborhood interconnection and to link Garrett Park and other neighborhoods south of Randolph Hills to trail and recreational facilities to the north, in Rock Creek Park. After the path proved attractive to rail commuters, it was illuminated for use after dark. Prior to path develop- ment, the Garrett Park Station was for all practical purposes inaccessible for residents of Randolph Hills and other northerly neighborhoods. Auto access involves a roundabout 9,500-foot (1.8-miles) drive just to go 800 feet (0.15 miles). There was no known MARC ridership from Randolph Hills except for an occasional intrepid soul braving the kudzu and stream crossing in dry weather. On May 22, 2008, the 33 passengers alighting in Garrett Park on the six outbound trains serving the station were counted according to their egress mode and direction of travel.93 Of the 33 passengers, 16-415 Hourly Counts 13-hour Temporal Distributions Hour Starting North 1/9/02 South 12/20/01 North 12/2/03 South 12/2/03 North 1/9/02 South 12/20/01 North 12/2/03 South 12/2/03 6:00 AM n/a n/a 1 3 n/a n/a 0.2% 0.7% 7:00 60 51 9 25 4.8% 2.4% 2.0% 6.2% 8:00 89 72 22 45 7.0% 3.4% 5.0% 11.2% 9:00 86 73 30 8 6.8% 3.4% 6.8% 2.0% 10:00 50 81 4 9 4.0% 3.8% 0.9% 2.2% 11:00 99 127 40 3 7.9% 5.9% 9.0% 0.7% 12:00 PM 143 213 26 17 11.3% 9.9% 5.9% 4.2% 1:00 112 166 39 53 8.9% 7.8% 8.8% 13.2% 2:00 96 210 68 135 7.6% 9.8% 15.4% 33.5% 3:00 108 203 21 34 8.6% 9.5% 4.7% 8.4% 4:00 102 271 51 16 8.1% 12.6% 11.5% 4.0% 5:00 PM 141 277 81 34 11.2% 12.9% 18.3% 8.4% 6:00 112 243 51 21 8.9% 11.3% 11.5% 5.2% 7:00 62 156 n/a n/a 4.9% 7.3% n/a n/a 8:00 42 a 119 a n/a n/a — a — a n/a n/a 9:00 26 a 128 a n/a n/a — a — a n/a n/a Total 1,260 a 2,143 a 443 403 100% 100% 100% 100% Note: a Pedestrian volumes from 8:00 to 10:00 PM are not included in the totals or the percentage calculations in order to maintain 13-hour count comparability. Table 16-128 Comparison of North-side and South-side Crosswalk Pedestrian Volumes by Hour, Intersection of Georgia Avenue and Colesville Road, Silver Spring 93 Train P879 (5:58 PM arrival) used the inbound track on 5/22/08, blocking the observer’s view of passengers walking into neighborhoods to the south and west. Train P879 alightings were recounted in similarly clear, warm weather on 5/29/08. After determining that other count components were identical, the 5/29/08 count of Train P879 passengers walking south and west was substituted.

14 (42 percent) walked into Garrett Park and adjoining neighborhoods to the south and west. Another 11 (33 percent) drove away in cars parked at the station. The other eight passengers (24 per- cent) walked north on the path to Randolph Hills. They constituted 36 percent of the 22 walk-egress MARC passengers. It may be inferred that the Garrett Park MARC station ridership is over 30 per- cent more than it would be without the connection. The path also regularly serves Garrett Park bus patrons accessing the nearest Randolph Hills stop. More—Off-Street Versus On-Street NMT User Mix. There appears to be a tendency, affecting many bicycling travel demand and safety research and advocacy papers, to overlook the different user mixes attracted by different types of facilities and the implications thereof. Even where user mix differences are discussed, little quantification has been encountered. The ideal might be to quantitatively compare path usage with parallel bike lane usage. That opportunity did not pres- ent itself in the Montgomery County special mini-studies. However, it did prove possible to inves- tigate the user mix of the Rock Creek hiker-biker trail versus the on-road NMT user mix of parallel Beach Drive, using short-duration concurrent classification counts. Beach Drive is a low-speed, curving, two-lane, urban scenic highway with infrequent intersections, flanked by parkland. It has no shoulders or bicycle lanes, but it is signed “Bike Route” and “Share the Road” (on the same posts), and the 25 mile-per-hour speed limit is periodically enforced. Both the parallel trail and the on-road bike route lie entirely within riparian parkland. At the time of the 2005 user-mix investigation, Beach Drive pavement quality was fair-to-good but not excellent, while trail pavement quality was fair. Beach Drive is level. The hiker-biker trail is a low-speed, 8-foot-wide paved facility through partially forested rolling terrain, with frequent, often tight, hor- izontal and vertical curves. Two locations where both facilities could be clearly seen were observed and counted. Three 1-hour counts were taken, on a Friday, Saturday, and Sunday in warm early November weather with fall foliage still in its prime. The counts were done in early afternoon, and thus did not pick up NMT com- muter traffic, which is thought to be limited given facility orientation, design, and peak traffic condi- tions. Walkers and cyclists were identified as to sex (except for younger than teenage children), accompaniment by children in conveyances (regular and jogging strollers, carriages, bicycle seats, trailers, and adult/child tandems), accompaniment by dogs, and dress. A binary classification system was used for dress, namely, persons wearing special cycling outfits (“cycling gear”) and persons wear- ing ordinary pants, jeans, shorts, etc. (“street clothes”). Groups of walkers/cyclists were noted. Trail pavement conditions were not supportive of in-line skating. Results are summarized in Table 16-129 and the two paragraphs which follow. 16-416

NMT traffic totals for the Rock Creek off-road hiker-biker trail compared to the parallel Beach Drive on-road bike route over the 3 hours/days were very similar, with roughly 10 off-road trail users counted for every nine on-road cyclists. The distributions of user types, however, varied con- siderably. On the off-road trail there were three adult walkers or joggers for every two cyclists, while on the on-road route there was only one person on foot, a female jogger. About 5 percent of trail traffic was users walking dogs. Among bicyclists, only 38 percent of adults choosing the trail wore cycling gear, suggestive of moderate levels of involvement and a broad range of skill levels. In contrast, 89 percent choosing the on-road route wore special cycling outfits, suggesting avid involvement and concomitant skill. On the trail, 37 percent of adult and teenage cyclists were female, while 50 percent of adult and teenage trail traffic overall (walkers and joggers included) was female. Younger children constituted 15 percent of trail traffic, with a percentage breakdown of 34 percent on their own bicycles, about 21 percent in bicycle seats and trailers (including one on an adult/child tandem bike), 7 percent walking, and 38 percent in strollers and the like. On the road, just 20 percent of cyclists and of NMT users overall were female, and there were no children other than one male teenager. Of trail traffic, 55 percent appeared to be in groups, with almost 1/2 of group members appearing to be composed of family groups inclusive of children (26 percent of all trail traffic). Only 18 per- cent of on-road traffic seemed to be in groups, about 2/3 of which (12 percent of all NMT road users) were male-female pairings. It may reasonably be concluded, at least in the context of the 16-417 User Category (Male/Female shown in data columns) Fri. 11/4/05 2:10-3:10 PM Sat. 11/5/05 1:10-2:10 PM Sun. 11/6/05 2:00-3:00 PM Three Count Total Trail Road Trail Road Trail Road Trail Road Walker, adult, no child/dog 2M 3F 0 10M 18F 0 4M 5F 0 16M 26F 0 Jogger, adult (1 with stroller) 4M 1F 0 7M 4F 1 F 10M 13F 0 21M 18F 1F Walker with dog 1M 1F 0 2F 0 1M 5F 0 2M 8F 0 Walker w/child in stroller, etc. 2F 0 2M 4F 0 0 0 2M 6F 0 Children in strollers, etc. 4C 0 7C 0 0 0 11C 0 Walker, child 0 0 1C 0 1C 0 2C 0 Cyclist, adult, in cycling gear 0 13 M 3F 8M 4F 59 M 10F 4M 9F 48M 17F 12M 13F 120M 30F Cyclist, adult, in street clothes 4M 1F 3M 8M 7F 3M 1F 14M 1F 9M 3F 26M 9F 15M 4F Cyclist w/child in seat/trailer 0 0 3M 1F 0 1M 1F 0 4M 2F 0 Children in cycle seats/trailers 0 0 4C 0 2C 0 6C 0 Cyclist, child (Teens included as M or F) 0 0 1M 1F 7C 0 3C 1M 1M 1F 10C 1M Total, Male, Female, Child (Teens included as M or F) 11M 8F 4C 16M 3F 39M 41F 19C 62M 12F 34M 34F 6C 58M 20F 84M 83F 29C 136M 35F Grand total 23 19 99 74 74 78 196 171 Notes: M = Male, F = Female, C = Child (among non-adults only teenagers are identified by gender). Bicycle trailers were each assumed to contain one child. The 11/5/05 count was made just south of the Puller Drive connecter trail. The other two counts were made east of Franklin Street. Table 16-129 NMT User Mix of Rock Creek Hiker-Biker Trail Compared to Parallel Beach Drive Bike Route, Thee One-Hour Counts

count timing and weekend dominance of count totals, that the off-road trail alternative was the one serving a broad range of NMT modes and user types. The overall male-female balance was essentially equal. Family groupings inclusive of children were common, and all bicycle training for children occurred on the trail. In contrast, the Beach Drive on-road option attracted users— mostly male—focused on bicycling for sport and sustained exercise. The societal functions of the two facility types, although both facilitated exercise, were basically quite different. Neither one alone met needs fulfilled by the other. Sources. Short-duration counts (AM-, PM-, or midday-only), field observations, distance measure- ments, and all conclusions, by the Handbook authors. • Ujifusa, A., “Officials want amenities added to Elm Street Park.” The Gazette (March 18, 2009). • Harvey, P., “Pedestrian education continues in Silver Spring.” Gazette Regional News (July 14, 2004). • Watkins, C. K., State Highway Administration, Maryland Department of Transportation, letter to the Handbook authors Re. “US 29 (Georgia Avenue) at Ellsworth Drive Mid-block Pedestrian Crossing Studies” with attached 2001–2004 traffic counts at the Georgia Avenue intersections with Colesville Road, Ellsworth Drive, and Wayne Avenue (August 17, 2004). Pedestrian Activity Effects of Neighborhood Site Design—Seattle Situation. A series of research projects have been undertaken on the relationship between site design and pedestrian travel in mixed-use medium density neighborhoods of the greater Seattle area. Twelve neighborhoods, each containing a small to medium size neighborhood commercial center, were the focus of the analysis reviewed here. The neighborhoods varied in site design char- acteristics. These were described in terms of block size and length and completeness of the side- walk systems. Sites characterized as “urban” by the research had small blocks the equivalent of 300 by 400 in dimension, a complete and continuous public sidewalk system on both sides of all streets, averaging 38 miles total in length per site, and on-street parking together with off-street parking in small lots. Sites characterized as “suburban” had large blocks the equivalent of 1,000 by 1,300 feet in size, an incomplete and discontinuous public sidewalk system lining less than one- half the streets, averaging 8 miles total in length per site, and only off-street commercial parking, in large lots. Retail stores in the urban sites tended to face directly onto one main street, while in the suburban sites retail was located in large blocks of private land containing broad areas of sur- face parking. Actions/Analysis. A quasi-experimental methodology focusing on a single point in time was used to study pedestrian volumes into each neighborhood commercial center. The 12 research sites were selected to have substantially different pedestrian environments, one-half with extensive pedes- trian facilities and one-half with quite limited facilities. This difference in neighborhood site design constituted the study’s independent variable, while the dependent variable was the volumes of pedestrians crossing from residential areas into the central commercial area across a survey cor- don line. Each survey was manned during representative hours of the day totaling some 16 hours. Sites were matched for population density, land use mix and—to the extent possible—income, in order to minimize the effect of these factors. Auto ownership per person, although not per dwelling unit, was found to be similar for all sites, at 0.6 to 0.8 automobiles per person. All sites were described by a 1/2-mile pedestrian travel catchment area around their neighborhood commercial center, with a gross population density of about 10 people to the acre, creating an average population of 6,000 people for each site. Dwelling types within this average density ranged from apartments and condo- miniums to single family houses. Statistics for each site were normalized to remove the effect of unde- veloped sectors resulting from topological features such as bodies of water. 16-418

Results. The researchers concluded that the measures traditionally employed to predict pedestrian volumes—population density, income, land use distribution and intensity—are insufficient to explain the variation in pedestrian volumes. Pedestrian volumes were found to be also related to neighborhood site design and pedestrian facilities design, as reflected in block size and extent of pedestrian facilities provided. Absolute size of the neighborhood commercial development was shown not to explain the observed pedestrian volumes. The urban sites exhibited, on average, three times the pedestrian volumes of suburban sites. The urban site volumes averaged 38 pedes- trians per hour per 1,000 neighborhood residents walking between residences and commercial cen- ters, while the corresponding suburban volumes averaged between 12 and 13 pedestrians per hour per 1,000. The one site not in Seattle proper among those accorded an “urban” classification pro- duced 24 pedestrians per hour per 1,000 residents, still twice the suburban neighborhood average. The suburban sites were clearly not without pedestrians. The suburban volume range was between eight and 16 pedestrians per hour per 1,000 neighborhood residents moving between the subur- ban residential areas and the commercial centers. The majority of these suburban pedestrians were found to use streets with sidewalks where available. Suburban pedestrians are more likely to jay- walk than the urban pedestrians (32 versus 20 percent) and more likely to use marked crosswalks (60 percent versus 14 percent). These seemingly contradictory results reflect the lesser availability of legal walking options at suburban sites on the one hand, and the apparently high perceived risk, on the other hand, of crossing wide suburban streets without a marked crosswalk. The mean distance between points where pedestrians can enter the commercial area was found to be twice as long for suburban sites as for urban sites. Suburban pedestrian route options were found to be constrained not only by the large blocks, but also by apartment and school campus fences/gates. The length of suburban-site actual walking routes was found to be 66 percent longer than airline distance, as compared to 27 percent longer for urban sites. This represents a difference averaging 600 feet, enough, in the opinion of the researchers, to suppress pedestrian activity. More . . . Young people and non-whites, together with persons judged to be of Hispanic origin, were over-represented proportionally among the suburban pedestrians when compared to the local area population. This was taken to indicate that the suburban pedestrians may represent in substantial measure people who do not have the option of driving. At the suburban sites, an average of 41 per- cent of pedestrians were under 18 years of age, 180 percent higher than in the neighborhood popula- tion. In urban sites the proportion of young pedestrians, 16 percent, was similar to the percentage of youth in the neighborhood population. Non-whites and Hispanics were over-represented among pedestrians at both suburban and urban sites. The suburban site over-representation was by 240 per- cent, relative to the tributary population, compared to 200 percent for urban sites. The researchers note that the high proportion of young pedestrians combined with lack of appropriate pedestrian facilities in the suburban sites raises troubling safety issues, as does the observed presence of pedestrians with impairments at three of the suburban sites. Sources. Hess, P. M., Moudon, A. V., Snyder, M. C., and Stanilov, K., “Site Design and Pedestrian Travel.” Transportation Research Record 1636 (1998). • Moudon, A. V., Hess, P. M., Snyder, M. C. and Stanilov, K., “Effects of Site Design on Pedestrian Travel in Mixed-Use, Medium-Density Environments.” Transportation Research Record 1578 (1997). 50 Years of Downtown NMT Facility Provisions—Minneapolis Situation. Minneapolis is in the upper ranks of cities for active-transportation tripmaking. In 2007 it was 10th out of the 50 largest cities in walking to work (6.4 percent mode share), 2nd in bicycling 16-419

to work, (3.8 percent), and 10th in taking transit to work (13.4 percent). The city has, in recent years, been progressing forward pursuant to goal-driven pedestrian and bicycle master plans. Even before NMT goal-setting was the norm, however, downtown Minneapolis took ground-breaking actions supportive of pedestrian activity and connectivity. The first pedestrian bridge of the ulti- mate 8-mile downtown Skyway system was opened in 1962, and the Nicollet transit mall was brought on line in late 1967. Actions. The first two downtown Skyway links were opened in 1962 and 1963, connecting financial district buildings to the city’s first mixed-use building, the Northstar Center. From this start, the Minneapolis Skyway system has grown to 82 bridges in 2004. The cumulative number of links open each year is given in the right-most column of Table 16-130 under “Results.” The present-day system connects the office and retail core with various mixed-use buildings, hotels, apartment complexes, close-in and fringe-area parking ramps (i.e., garages), and a west-side I-294 bus transit terminal. All entrances to the Skyway system are through buildings. There are no direct sidewalk connections. The Nicollet Mall was constructed as an eight-block transit mall in 1966–67. The roadway was nar- rowed to 24-feet and restricted to buses, emergency vehicles, and taxis. For most of the distance the roadway follows a serpentine path, producing a sidewalk 20 to 36 feet in width, as compared to the original 15-foot width. Bus shelters and street furniture were concentrated in the bulb-outs, leaving a clear path for pedestrians of 15 feet. Original amenities included sidewalk heating for snow removal, fully equipped bus waiting shelters each block on both sides, and bike racks. Bicycles have been allowed on the mall from the beginning. The transit mall was extended by four blocks in 1982, and refurbished in 1991. Attention to bicycle provisions mostly came later. Among off-road shared use trail projects between 1990 and 2000, two—both 8 miles long—provided important through connections from the southwest and southeast, with the latter connecting into extensive trail systems serving both the Minneapolis and St. Paul sides of the Mississippi River. During the same 1990–2000 decade a 4-mile pair of bike lanes tying the south side of the city to downtown, plus an approximately 1-mile pair to the east, were implemented. Crossing the Mississippi, two pedestrian/bicycle bridges were opened and bike lanes were added to two road bridges, raising from two to six the number of bridges with dedicated bicycle facilities serving downtown. There has also been a major program of adding bicycle lanes to downtown streets. The entire city has supportive NMT infrastructure, with sidewalks on both sides of 80 percent of all streets, sidewalk provisions deemed appropriate on another 12 percent, 18 miles of off-road bicycle/pedestrian trails with connections to suburban facilities, and 35 major pedestrian/bicycle bridges over highways, streets, railroads, and the Mississippi River. Analysis. A 12-hour manual cordon count around the Minneapolis CBD has been conducted peri- odically. There were 18 cordon counts between 1958 and 2003. The cordon circumscribes an area some nine by 12 blocks in extent, bounded by 1st Street N./S. (Mississippi River), 5th Avenue S., 12th Street S., and 2nd/4th Avenues N. (I-394 and railroad). Each count tallies both vehicles and people entering and leaving the CBD at each count station, by mode, by 15-minute intervals, from 6:30 AM to 6:30 PM. The counts have been uniformly taken on the 2nd Wednesday of September. This case study compares mode share and volume trends over time with development of the Skyway, transit mall, and bicycle lane and path infrastructure. It is important to note that a cordon count necessarily intercepts through traffic; indeed, most through trips are counted twice relative to each CBD-generated trip. This circumstance tends to inflate the auto mode share, and introduces a probable discontinuity in the 1970s as freeway links were completed bypassing downtown, diverting some of the through vehicular traffic. Other known exogenous influences of importance 16-420

are commented on in the “Results” and “More . . .” presentations. Individual-facility count data available at more than one point in time are introduced under “More . . .”. Additional analysis of 1968 through 1974 Skyway count data, and also 2002 Skyway counts, is found in the “Response by Type of NMT Strategy” section under “Pedestrian Zones, Malls, and Skywalks”—“Pedestrian Skywalks”—“Skywalk Impacts on Walking.” Results. Table 16-130 presents 1958–2003 findings of the Minneapolis CBD Cordon Count, focus- ing on the transportation of people. Both the cordon counts and other time-series data indicate that economic conditions and development trends have strongly influenced the ups and downs of downtown Minneapolis pedestrian and other person-flows. The first 7 years shown are annual counts, before the Nicollet Mall and before any more than two Skyway bridges. They serve to illus- trate the variabilities affecting such counts, with the walk and bike share totals oscillating between roughly 6 percent and roughly 7 percent. Separate walk and bike shares are not available for years prior to 1974, but it is thought that bicycle shares were very low, as suggested by the 0.2 percent 1974 share. It will be noted that the pedestrian (and bicycle) count and mode share reached a low point in the 1961 cordon count, dipping again in the period concurrent with I-35W freeway com- pletion and economic downturn in the 1970s. 16-421 Year Total Persons NMT Persons Percent Mode Shares at Cordon Employ- ment Sky- waysWalk Bike Bus Auto Other 1958 564,992 41,511 7.3% — 21.7 % 61.9 % 9.1% n/a 0 1959 555,569 36,863 6.6 — 22.5 61.6 9.3 n/a 0 1960 558,194 39,320 7.1 — 20.1 64.0 8.8 n/a 0 1961 522,021 30,824 5.9 — 19.8 65.5 8.8 n/a 0 1962 526,228 37,041 7.0 — 20.4 63.7 8.9 n/a 1 1963 532,543 33,639 6.3 — 19.6 65.1 8.8 n/a 2 1964 514,425 34,289 6.7 — 20.0 64.2 9.1 See 2 1970 548,307 38,123 6.9 — 16.5 66.3 10.3 Notes 5 1972 516,059 33,402 6.4 — 18.8 64.3 10.3 n/a 5 1974 475,278 27,026 5.4 0.2% 22.8 61.0 10.5 n/a 10 1975 489,765 30,698 5.8 0.5 23.9 61.1 8.7 n/a 10 1977 492,173 29,475 5.5 0.4 26.9 57.2 10.0 n/a 13 1981 460,822 33,862 6.6 0.8 25.7 58.5 8.5 141,304 16 1984 494,540 35,532 6.7 0.4 23.0 62.6 7.2 143,562 30 1987 515,543 44,133 7.9 0.8 22.0 62.3 7.1 151,780 38 1990 494,188 42,228 7.7 0.9 23.0 60.0 8.5 155,932 45 1998 541,195 44,941 7.4 0.9 19.8 64.2 7.6 164,463 68 2003 522,815 39,578 6.7 0.8 20.8 64.2 7.4 153,732 82 Notes: Downtown retail sales declined sharply from 1957 to 1963. Circa 1967 CBD employment was about 95,100 (area definition unknown). The 1990 and 2000 employment totals in the area roughly encompassed by the freeway loop and the Mississippi River were 132,617 and 146,474, respectively. The 1981-2003 employment data within the table are for a larger “downtown area” and are presented simply to show trends, including the post-9/11 (2001) downturn. The “Skyways” column tallies Skyway bridge crossings of streets. All data in Table 16-130 are from before opening of light rail transit (LRT) in 2004. Table 16-130 Total Persons Entering and Leaving the Minneapolis CBD, with Percentages by Mode, 6:30 AM–6:30 PM

Although the 1987 through 2003 cordon counts indicate no growth in bicycle shares, 1990 and 2000 Census mode share tabulations suggest otherwise, at least for journey-to-work trips headed for the downtown area. The downtown Minneapolis destination bicycle commute share grew from 2.27 percent in 1990 to 2.58 percent in 2000, a 0.31 positive percentage point shift (and a 14 percent increase) concurrent with the extensive bicycle facility provisions enumerated above under “Actions.” In St. Paul, where there were few if any downtown area cycling facility improvements, the corresponding downtown bicycle commute shares declined from 0.64 percent in 1990 to 0.59 percent in 2000, despite overall bicycle commuting increases in both cities. More . . . Asked why they were on the Nicollet Mall in a 1977 survey, with multiple answers allowed, 57 percent reported shopping, 42 percent were walking for pleasure, 24 percent were there because of their work, 16 percent were there because it was their bus stop location, another 16 percent were headed to some place off the mall, and 5 percent had other reasons. Asked to select among specified alternative locations the place they were most likely to walk or browse, 85 per- cent chose the Nicollet Mall (of course they were on the Mall), 10 percent selected the Skyway sys- tem, 2 percent chose parallel Hennepin Avenue, and 1 percent picked parallel Marquette Avenue. Pedestrian and other counts are available for Nicollet Avenue at various points in time. In September of 1958, well prior to 1966–67 transit mall construction, the 12-hour pedestrian volume average between 4th and 10th Streets was 12,800 per side, per block. Individual block-face pedestrian volumes ranged from 23,600 to 4,700. Shortly after initial opening of the mall the average had increased to 13,600, up 6 percent despite the sharp 1957–1963 retail decline noted with reference to Table 16-130. By 1976, how- ever, the 4th to 10th Street 7:00 AM to 6:00 PM average, per side, per block had declined to 7,400 pedestrians. (The comparable 1958, 11-hour average was 12,400.) The 1972–1981 period had exhib- ited an overall decline in downtown NMT activity, as seen in the Table 16-130 tabulation, but of greater importance to Nicollet Mall walking was the burgeoning development of the Skyway sys- tem. By 1976, all major department stores on the transit mall were connected by Skyways. With 1/3 to over 2/3 of pedestrians shifting to available Skyways, depending on weather, the combined 1976 pedestrian flow in the Nicollet corridor may well not have reflected a decline at all. Nicollet Mall counts, supplemented by estimates, are also available for 2002. The highest 11-hour side- walk volume on a single block face was 14,550. For computation of averages, data availability requires shifting the coverage by one block, to between 5th and 11th Streets—a shift that is consistent with the southward movement of major retail establishments. Full 11-hour counts cover 3 blocks, and extrap- olations from 3-hour midday 2002 or 2000 counts cover the other 3 blocks. On this basis, the 5th to 11th Street 7:00 AM to 6:00 PM average, per side, per block was 7,200 pedestrians, virtually the same as 1976 despite additional Skyways. The most closely parallel Skyways crossing 6th through 10th Streets, five Skyways in total, averaged 11,500 pedestrians. If one adds the Skyway and block face averages, an approach that presumes Skyway volumes to each side of the Nicollet Mall to be roughly equal, the average 2002 pedestrian volume per side of the Nicollet corridor is found to be 18,700 pedestrians.94 This is on the order of 50 percent more than the 1958 pre-mall, pre-Skyway, 11-hour pedestrian flow. Nicollet corridor counts over time are summarized in Table 16-131. 16-422 94 In developing the average parallel Skyway volume, only the one nearest Skyway per cross-street was included, as counts are not available for some second-nearest parallel Skyways. Counts are primarily 11-hour, September counts, with some 12-hour counts at retail centers. The assumption that Skyway volumes to each side of the Nicollet Mall are roughly equal to the nearest-Skyway 11,500-pedestrian average is rather crude. Three of 10 block-pair combinations bordering the Nicollet Mall between 5th and 11th Streets have two north- south Skyway connections instead of one, while one block-pair has none. The two second-nearest parallel Skyways for which counts are available average only 5,400 pedestrians. If one utilizes this figure as the “other side of the mall” Skyway average, then the corridor pedestrian count increase since 1958 is on the order of 25 percent. The lower figure is the basis for the low-end-of-the-range entry for 2002 in Table 16-131.

Vehicle volumes displaced from Nicollet Avenue averaged 6,800 per direction per hour. There was essentially no congestion accompanying this shift. Peak hour bus volumes increased from 20 to 60 per hour in each direction. Two-way, 12-hour Nicollet Avenue bus volumes at the cordon line increased by 422, from 188 in 1964 to 623 after transit mall construction and subsequent bus rerout- ings. This was only partially counterbalanced by a reduction of 138 buses on Hennepin Avenue to the northwest and 33 buses on Marquette and Second Avenues to the southeast. The Nicollet Mall bus count at 12th Street in 2003 was 580 buses of all types. As measured at the cordon line north of 12th Street, the total 12-hour person-flow on Nicollet Avenue (walking, biking, and riding in buses, cars, and other vehicles) increased from 17,246 per- sons in 1964 to 23,708 persons in 1970 after opening of the mall, increasing further to 25,184 in 1975. Person-volumes on nearby parallel streets declined. With extension of the transit mall in 1982, almost all persons recorded at the Nicollet count station are walking, bicycling, or riding on buses, a travel mode almost always involving NMT access at least one end of the trip if not both. Cordon counts in 1998 and 2003 show the Nicollet Avenue 12-hour person volumes at 12th Street under these more recent conditions to have been 23,223 and 24,140, respectively. Six published financial/office district 2002–2007 Skyway count comparisons show 5-year increases ranging from 12 to 39 percent and averaging 24 percent. City of Minneapolis Nicollet Mall pedestrian data for 2007 suggest typical variation but no growth in at-grade sidewalk counts. Between 6th and 7th Streets 13,415 pedestrians were counted during 12 hours in 2007 versus 13,000 for 11 hours in 2002, and between 11th and 12th Streets 7,228 pedestrians were counted during 11-3/4 hours in 2007 versus 8,686 for 12 hours in 2003. Bicycle count increases from 2003 to 2007 at nine locations in downtown ranged 16-423 Year Count Parameters/Events Avg. Nicollet Sidewalk Count (per Side) Avg. Parallel Skyway Count (to One Side) Avg. Nicollet Corridor Count (per Side) 1958 Nicollet Ave., 4th to 10th Sts., 12 hrs. 12,800 0 12,800 1958 Same, 11-hour count, 7 AM - 6 PM 12,400 0 12,400 1962-69 First 5 Skyways opened 1966-67 Nicollet Mall opened for 8 blocks 1973 Nicollet Mall, 4th to 10th Sts., 12 hrs. 13,600 n/a n/a 1973-75 5 additional Skyways opened 1976 Nicollet Mall, 4th to 10th Sts., 11 hrs. 7,400 1,800 - 7,400+ a 9,200 - 14,800+ 1976- 2002 81 additional Skyways opened for a Minneapolis system total of 82 crossings 1982, ’91 Nicollet Mall extended, refurbished 2002 Nicollet Mall, 5th to 11th Sts., 11 hrs. 7,200 8,400 - 11,500 15,600 - 18,700 Note: See preceding text, including Footnote 94, for data limitations, assumptions, and discussion. a Calculated at 1/3 - 2/3+ of corridor total per side, but then discounted by 50 percent for two to three cross-street Skyway coverage (parallel and adjacent to Nicollet) out of five cross- streets total. Table 16-131 Nicollet Corridor Pedestrian Flows—1958–2002—Six-Block, Per-Side Averages

from 13 to 96 percent, with a weighted average of 51 percent.95 These indicators in combination suggest a return, following recovery from the post-9/11 downturn, to the irregular but overall gradual upward trend in Minneapolis downtown NMT volumes (roughly 1/2 of 1 percent per year) since the mid-1960s. Economic downturns and upturns exhibit the strongest influence, but Minneapolis has clearly suc- ceeded in stabilizing and enhancing its downtown area and its NMT attractiveness. Circumstantial evi- dence supports a likely correlation with development of the Skyway system, and more recently with implementation of the various downtown-focused bicycle facilities, while the Nicollet Mall has pre- sumably played a supporting role. Sources. Alliance for Biking & Walking, “Bicycling and Walking in the United States: 2010 Benchmarking Report.” Washington, DC. http://peoplepoweredmovement.org/site/index. php/ site/memberservices/alliance_2010_benchmarking_report_information_findings (2010) • City of Minneapolis Public Works Department, “Minneapolis Pedestrian Master Plan.” Draft for Public Review. Minneapolis, MN (June 8, 2009). • City of Minneapolis, MN, “Report on Bicycle & Pedestrian Counts.” City of Minneapolis Department of Public Works (October 22, 2007). • Corbett, M. J., Xie, F., and Levinson, D., “Evaluation of the Second-Story City: The Minneapolis Skyway System.” TRB 87th Annual Meeting Compendium of Papers DVD. Washington, DC (January 13–17, 2008). • Barnes, G., Thompson, K., and Krizek, K., “A Longitudinal Analysis of the Effect of Bicycle Facilities on Commute Mode Share.” TRB 85th Annual Meeting Compendium of Papers CD-ROM. Washington, DC (January 22–26, 2006). • Koffman, D., and Edminster, R., Streets for Pedestrians and Transit: Examples of Transit Malls in the United States. Final Report—Phase I. Prepared for the Urban Mass Transportation Administration, U.S. Department of Transportation, by Crain and Associates, Menlo Park, CA (August, 1977). • Edminster, R., and Koffman, D., Streets for Pedestrians and Transit: An Evaluation of Three Transit Malls in the United States. Final Report—Phase II. Prepared for the Urban Mass Transportation Administration, U.S. Department of Transportation, by Crain and Associates, Menlo Park, CA (February, 1979). • Robertson, K. A., Pedestrian Malls and Skywalks—Traffic separation strate- gies in American downtowns. Avebury—Ashgate Publishing Limited, Brookfield, Vermont (1994). • SRF Consulting Group, Inc., “2003 City of Minneapolis Central Business District Cordon Count.” Prepared for the Public Works Department, Traffic & Parking Services Division, City of Minneapolis, MN (December, 2003). • Carlson, R., Metropolitan Council, email to the Handbook authors with attached table, “Minneapolis Downtown area covered employment” (August 24, 2004). • Bruce, P., Nicollet Mall Pedestrian Count—September 2002 Daily Volumes. Prepared by Community Enhancement/Pedestrian Studies/www.pedestrianstudies.com, Minneapolis, MN [2002c]. • Bruce, P., 2002 Minneapolis Downtown Pedestrian Count and Analysis. Prepared by Community Enhancement/Pedestrian Studies/ www.pedestrianstudies.com, Minneapolis, MN [2002a]. • SRF Consulting Group, Inc., “City of Minneapolis Central Business District 1998 Cordon Count.” Prepared for the Department of Public Works, Transportation Division, City of Minneapolis, MN (November, 1998). • Bruce, P., Community Enhancement and Pedestrian Studies, email to the Handbook authors with attached map, “2007 Downtown Minneapolis Count Project—Skyway Level Daily Volumes—2002–2007 percent change” [redacted version] (July 27, 2009). • Summary computations for 2002, supplemental growth calcula- tions, and overall conclusions by the Handbook authors. 16-424 95 The collapse of the I-35W bridge across the Mississippi just in advance of the 2007 count program could conceiv- ably have had some effect on bicycle counts, even though the freeway bridge itself did not accommodate cyclists. The 2002–2007 bicycle traffic increase at the three counted Mississippi River crossings was 62 percent while the increase at the other six downtown sites was 44 percent.

Bicycle Lanes in the Downtown Area—Toronto, Canada Situation. Since 1993, the city of Toronto, Canada, has fine-tuned their established policy and process for implementing bicycle lanes along arterial streets mostly in the downtown area. The first bicycle lane in Toronto was installed in 1979 on Poplar Plains Road, a narrow residential street that had just been converted to one-way operation. Between 1990 and 1991, the city added approxi- mately 5 miles of bicycle lanes along two arterial streets and one residential street, respectively: Queens Quay (1990), Bloor Street Viaduct (1991), and Russell Hill Road (1991), the one-way cou- plet to Poplar Plains Road. By 1993, bicycle traffic entering and leaving the downtown area appeared to be growing and had become a noticeable presence at about 17,000 bicycles per week- day. Bicyclists were also being over-represented in vehicle crashes, with nearly 15 percent of all reported collisions causing injuries involving bicyclists, compared to bicycle volumes represent- ing 3 percent of wheeled traffic. At the same time, motor vehicle traffic levels appeared to be sta- tic. Given these factors and a strong, official plan in support of bicycling, the city of Toronto decided to embark on an expanded bicycle lane program. Actions. In total, about 25 miles of bicycle lanes were constructed on six downtown arterial streets between 1993 and 1998. The bicycle lanes were added to an already congested street network (most arterial streets were carrying about 15,000 to 20,000 vehicles per day) to improve the safety of bicy- clists and to encourage bicycling. Along many of these two-way, four-lane arterial streets, the motor vehicle lanes were reduced to one lane per direction, with left turn lane provisions at most signalized intersections. A bicycle lane and an all-day parking lane were introduced in each direc- tion of travel. Analysis. Before-and-after studies of bicycle and motor vehicle traffic volumes were conducted to gauge the potential changes after bicycle lanes were installed. The reported motor vehicle traffic volumes are expressed as annual average weekday traffic (AAWT). Bicycle volumes were season- ally adjusted to represent annual average weekday volumes. The “after” bicycle counts were typ- ically performed 2 years after opening of the bicycle lane. Results. Table 16-132 shows that, on average, bicycle volumes on the streets to which bicycle lanes were added increased by 23 percent while motor vehicle volumes remained static. The bicycle vol- ume increases along the six routes with bike lanes ranged from 4 to 42 percent. For motor vehicles, there was one route where volumes dropped by 6 percent and another where volumes increased by 7 percent. Motor vehicle traffic volumes on the other four routes were unchanged after instal- lation of bicycle lanes. Despite the cycling increases on the streets with bicycle lanes, bicycle traf- fic levels city-wide since 1994 appeared to have remained static or to have declined by as much as 4 percent in a year. This trend was attributed to either declining employment in the central area or an aging population less likely to bicycle. The authors note that the declines in bicycle traffic lev- els have been most noticeable on streets without bicycle lanes. More . . . By paying careful attention to design details, the vehicle lane reductions (from two lanes to one through lane in each direction) resulted in only minor reductions in vehicular traffic capac- ity. For example, parking is prohibited near key intersections, permitting the addition of left-turn lanes. As indicated above and as illustrated in Table 16-132, the lane reductions did not signifi- cantly impact motor vehicle volumes on most arterial streets. 16-425

Source. Macbeth, A. G., “Bicycle Lanes in Toronto.” ITE Journal, Institute of Transportation Engineers, Washington, DC (April, 1999). Anderson Road Bicycle Lanes—Davis, California Situation. The city of Davis, California, has a well-established network of bicycle facilities and is home to a campus of the University of California. In the initial design of the city’s bicycle lane sys- tem, a general travel grid for bicycles was laid over the existing street network. On-street bicycle lanes were placed on some, but not all, of the streets in the designated grid. Initial plans for Anderson Road included bicycle lanes but they were not immediately constructed. The addition of bicycle lanes in 1974 provided an opportunity to evaluate the impact on bicyclists’ route choice. This opportunity along Anderson Road was unique for several reasons. Davis already had a rela- tively high percentage of bicycle commuters as compared to typical small college towns. A mature bicycle lane system was already in place, thereby lessening any novelty effect the Anderson Road bicycle lanes might have on bicycle trip generation, mode choice, or lane usage. Lastly, knowledge of the bicycle lane implementation was available enough in advance to design a behavioral exper- iment on bicyclists’ route choices. Actions. In 1974, the city of Davis converted Anderson Road from its original configuration (four lanes plus parking on 64 feet of pavement) to two motor vehicle lanes, a center two-way left-turn lane, and two bicycle lanes, one on each side. The on-street parking remained in the new configuration. Analysis. A total of 254 bicyclists living within two blocks of Anderson Road were interviewed in their homes, before installation of the Anderson Road bicycle lanes, about their route choice selec- tion. Based on the results, a partially different set of 108 bicyclists were home-interviewed after 16-426 Facility Installation Date Motor Vehicle Traffic a Bicycle Traffic b Before After c % Change Before After c % Change Davenport Road (North of Dupont Street) May 1995 22,000 22,000 0% 600 850 42% Gerrard Street (West of Sherbourne St.) Aug. 1995 18,000 18,000 0% 800 900 13% Sherbourne Street (North of Gerrard Street) Sept. 1996 16,000 15,000 -6% 550 570 4% Harbord Street (West of Bathurst Street) Aug. 1997 15,000 16,000 7% 1,100 1,500 36% St. George Street (North of College Street) Aug. 1993 16,000 16,000 0% 1,500 1,650 10% College Street (West of St. George Street) Oct. 1993 20,000 20,000 0% 1,450 1,900 31% Average 17,800 17,800 0% 1,000 1,230 23% Notes: a Annual average weekday traffic volume. b Seasonally adjusted (year-round) average weekday traffic volumes. c Typically surveyed 2 years after installation. Table 16-132 Before and After Traffic Volumes for Selected Streets with Bicycle Lanes

installation. Bicyclists living further from Anderson road were added, as far out as the parallel roads with previously installed bicycle lanes. This provided survey coverage equivalent to roughly five normal city blocks on each side of Anderson Road. A group of bicyclists living very close to Anderson Road were omitted, as they had been shown to already exhibit approximately 90 per- cent use of Anderson Road in the before condition. The subjects were categorized by gender and age group to discern differences in route selection among these groups. Route choices were deter- mined both for the after condition and, retrospectively, for the before condition. Respondents were also asked to rank bicycling conditions before and after. Bicycle traffic volumes were also collected along Anderson Road as well as the two parallel alternate routes with existing bicycle lanes, Sycamore Lane and Oak Avenue. The bicyclist volumes were counted manually on all three streets for two consecutive days several weeks before and 1 week after installation of the bicycle lanes. The counts were taken from 7:30 to 8:30 AM and 3:30 to 5:30 PM. The traffic observers also categorized each bicyclist by gender and age group as best possible. Results. Interviews conducted with 108 bicyclists after installation of the bicycle lanes revealed that 44 percent (25 of 57) of the surveyed bicyclists previously using other streets in the Anderson Road area shifted to the Anderson Road bicycle lanes. No cyclist reported changing from Anderson Road to other streets. The route choice shift to Anderson Road was most pronounced in the 25 and older age category, which accounted for the majority of the surveyed bicyclists. Interview results are detailed in Table 16-133. The actual counts, on the other hand, showed increases in bicycle volumes on all major routes. These increases, seen in the count data presentation of Table 16-134, were ascribed to seasonal variations related to weather and school schedules. The counts indicated that total bicyclist volumes along Anderson Road did not increase proportionately more than on nearby alternate routes after addition of the Anderson Road bicycle lanes; indeed, they increased slightly less. The 25-and-older age category did, however, show a significantly greater increase in bicyclist volumes (87 percent as compared to 52 percent on Oak Avenue or 8 percent on Sycamore Lane). The majority of the bicyclists counted appeared to be in the 18 to 24 age category, but the changes in total bicyclist volumes after implemen- tation of the bicycle lanes show no clear pattern for Anderson Road or the two alternate routes. 16-427

16-428 Age and Sex Class 0 to 11 12 to 17 18 to 24 25 and up Total Route Selection M F M F M F M F M F All Residence East of Anderson Using other routes before 2 3 1 7 10 12 11 23 Using Anderson before 1 3 3 5 1 1 4 5 9 14 23 Using Anderson after 1 3 3 5 1 2 8 9 13 19 32 Change from other route to Anderson a +1 +4 +4 +4 +5 +9 of 23 (39%) Residence West of Anderson Using other routes before 2 6 3 1 3 1 7 11 15 19 34 Using Anderson before 1 2 3 2 3 12 7 18 12 30 Using Anderson after 1 4 6 3 4 16 12 26 20 46 Change from other route to Anderson a +2 b +3 b +1 +1 +4 +5 +8 +8 +16 of 34 (47%) Total (East and West of Anderson) Using other routes before 4 6 3 1 6 2 14 21 27 30 57 Using Anderson before 2 5 6 5 3 4 16 12 27 26 53 Using Anderson after 2 7 9 5 4 6 24 21 39 39 78 Change from other route to Anderson a - +2 b +3 b - +1 +2 +8 +9 +12 +13 +25 of 57 (44%) Notes: The interviewees lived between Anderson Road and the next-over parallel bike lanes. Bicyclists living very close to Anderson Road were omitted. For additional background see text under “Analysis.” a Percentages calculated relative to those using other routes before. No one reported changing from Anderson Road to another route. b These children changed from using the Anderson Road sidewalks to using the Anderson Road bicycle lanes. Source: Lott, Tardiff, and Lott (1979). Table 16-133 “After” Survey Interview Data on Selection of Anderson Road as a Travel Route Before and After Bicycle Lane Installation Age and Gender Class 0 to 11 12 to 17 18 to 24 25 and up Route Selection M F Total M F Total M F Total M F Total Grand Total Sycamore Lane Before 82 13 95 28 14 42 526 389 915 118 16 134 1,186 After 98 33 131 26 31 57 552 443 995 130 15 145 1,298 Change +20% +154% +38% -7% +121% +36% +5% +14% +9% +10% -6% +8% +9% Anderson Road Before 6 3 9 29 14 43 617 550 1,167 223 32 255 1,474 After 2 5 7 33 8 41 488 564 1,052 395 82 477 1,577 Change -60% +67% -22% +14% -43% -5% -21% +3% -10% +77% +156% +87% +7% Oak Avenue Before 2 1 3 27 18 45 277 139 416 206 34 240 704 After 5 1 6 24 16 40 232 157 389 284 80 364 789 Change +150% 0% +100% -11% -11% -11% -16% +13% -6% +38% +135% +52% +12% Notes: Bicyclist volumes are from 7:30 to 8:30 a.m. and 3:30 to 5:30 p.m. and are average values for two days. Age and gender class estimated by count crew. Source: Lott, Tardiff, and Lott (1979). Table 16-134 Bicyclist Volume Count Data on Three Parallel Routes Before and After Installation of Anderson Road Bicycle Lanes

More . . . The interviews showed no primary route shifts for bicyclists under age 18, but five of 23 children (22 percent) reporting use of the Anderson Road bicycle lanes had formerly used the sidewalk. All interviewees were asked to rate Anderson Road for bicycle use before and after bicy- cle lane implementation. The rating scale ran from 7 (very bad conditions) to 1 (very good condi- tions). The average ranking was lowered 3.7 points between the before and after conditions, more than three-fifths the span of the scale, indicating substantial perceived improvement. There was a rough correspondence between perception of improvement and the decision to change route, the other factor in play apparently being directness of travel route. Source. Lott, D. F., Tardiff, T. and Lott, D. Y., “Evaluation by Experienced Riders of a New Bicycle Lane in an Established Bikeway System.” Transportation Research Record 683 (1979). Six Urban, Suburban, and Semi-Rural Trails—Indiana Trails Study Situation. Indiana is among the states that have committed significant federal and state funds to shared use, off-road trail development in local communities. The Indiana Trails Study was under- taken to address decision-maker need for comprehensive information on trail use and attitudes of users and trail neighbors. Six urban, suburban, and rural trails were studied. Although facilities classified as “rural” are generally beyond the scope of this “Traveler Response to Transportation System Changes” Handbook, the trails deemed rural in the Indiana study are included not only for comparison purposes but also because part of their alignments lie within small cities. Actions. Table 16-135 lists the location, trail type, pavement type, width, and construction dates for each of the six trails. From the construction dates one may infer the time the trails were in ser- vice prior to the year 2000 counts and surveys. 16-429 Location Trail Name Type Pavement Length, Width Construction Dates Fort Wayne Rivergreenway Trail riverfront hard surface 15 mi., 8-12’ 1980’s Goshen Maple City Greenway mill-race a crushed stone 10 mi., 10’ ± 1996, 2000 Greenfield Pennsy Trail rail-trail asphalt 3 mi., 12’ 1998 Indianapolis Monon Trail rail-trail asphalt 7½ mi., 10-12’ 1995, 1997 Muncie Cardinal Greenway rail-trail asphalt 10 mi., 12’ 1998 Portage Prairie Duneland rail-trail asphalt 6 mi., 12’ 1996 Note: Trail characteristics and extent are as of the survey/study timeframe. The Rivergreenway Trail (Ft. Wayne) and Monon Trail (Indianapolis) are classified urban, the Maple City Greenway (Goshen) and Prairie Duneland trail (Portage) are classified suburban, and the Pennsy Trail (Greenfield) and Cardinal Greenway (Muncie) are classified rural. a The Maple City Greenway is a trail network that includes rail-trails, trails built in parkland and utility easements, a trail alongside an 1860’s mill race (man-made open water conduit), and city street segments. The Goshen study concentrated on Mill Race Trail use. Table 16-135 Locations, Characteristics, and Construction Dates for Studied Indiana Trails

Analysis. Study methods applied on each of the six trails included user counts at selected trail seg- ments made with infrared trail counters, entering/exiting user interviews with follow-up mail- back questionnaires, a mail survey of “trail neighbors” (adjacent property owners), and telephone interviews with local realtors. A counter was positioned on each trail: at one location on the 3-mile Pennsy Trail, two locations on the 6-mile Prairie Duneland Trail, and three locations on the other, longer trails. They were in place during August and September, 2000, and on each trail were rotated among locations every 10 days. Infrared counter results were adjusted on the basis of “hand” counting done for validation purposes. Trail user surveys were conducted at four locations on each trail for 1 week each in July and August. One location at a time was surveyed, with 4-hour rotation so that 7:00 AM to 7:00 PM cov- erage was obtained. Every nth adult was asked to participate in a 3-minute interview with a mail- back questionnaire covering additional detail. Mail-back survey response rates were high, ranging from 70 to 97 percent for all trails except in Portage (50 percent) and Fort Wayne (39 percent). The sample of trail users intercepted ranged from 108 in Muncie to 585 in Fort Wayne. Totals of usable survey returns ranged from 72 in Greenfield to 200 in Fort Wayne. Even with the excellent survey return rate, the response for minorities may have been below aver- age, most notably among urban blacks and populations for whom English would likely be a sec- ond language (see comparative data provided in the bottom rows of Table 16-137). An important detail to note is that the trail user surveys were conducted at trail access points, and the inter- views/surveys were of trail users entering or leaving the trail.96 Results. Table 16-136 presents the volume data obtained, with population and trail distance listed first as a point of reference. All count-based information in the table is shown as ranges where the first number is based on the September 2000 count and the second number is based on the October count. The Indianapolis Monon Trail October count information involved extrapolation to cover about 2 weeks when actual counts were not successfully obtained. There was a shift forward of peak hours and a general reduction in total trail traffic as the days shortened in October relative to September. For example, the monthly total Fort Wayne Rivergreenway trail count dropped 10 per- cent from 26,914 to 24,231 while the weekday peak hour moved forward 1 hour on weekdays and 2 hours on weekends. However, the peaks sharpened as the daylight hours for trail activity com- pressed. It can be seen that in some cities the count for the highest single hour of the month actu- ally increased in October. As discussed elsewhere, new facilities such as those covered in the Indiana Trails Study present an analytical problem in that there is no “before” data with which to compare. Impacts must be either determined from screenline data (not done here and rarely obtained) or from retrospective ques- tions asked in surveys (see below). 16-430 96 This process of interviewing and handing-out of surveys to persons beginning and ending trail use rather than persons intercepted on the main trail itself has important implications. It means that, as a survey of users, it obtained results not biased by trip length differentials. If an on-trail intercept had been used, given that average bicycle trip distances are longer than walk trip distances, a typical bicycle trip would have been more likely to have been picked up than a typical walk trip. The generally available classification counts taken on trails reflect user mix at points along the trail, useful for operational analyses, whereas interception of persons starting or ending trail use is best for analysis of the mix of users taking advantage of a trail. In transportation planning terms, the method used in these Indiana surveys provides actual trip-based user data, as would a trip attraction survey.

The demographic information obtained for users of the Indiana trails is contained within Table 16-137. The distribution of males versus females on the trails ranges all the way from 68 percent male and 32 percent female on the rural Cardinal Greenway Trail radiating out from Muncie to 46 percent male and 54 percent female on the urban Monon Trail within Indianapolis. The Muncie trail is very bicyclist oriented, while the Monon Trail has many more walkers and runners than cyclists in terms of individ- ual trail users. Other trail user demographics also vary among locations, with a definite slant toward the most educated and highest income users in Indianapolis. Table 16-138 gives trail use characteristics for the Indiana trails. As indicated in the “Analysis” sec- tion above and Footnote 96, these data are derived from actual trip-based user attraction surveys taken of persons entering or leaving the trails, rather than from classification count observations or mainline-trail-based survey results. Distributions of uses normally involving less or more mileage, such as walk trips versus bicycle trips, are best represented by trip-based data such as presented here. It is also important to recall that the information covers all days of the week, including weekdays and weekend days. 16-431 Fort Wayne Goshen Greenfield Indianapolis Muncie Portage Population 205,727 29,383 14,600 1-½ million [MSA] 67,430 33,496 Trail Length 15 miles 10 miles 3 miles 7-½ miles 10 miles 6 miles Sept. – Oct. Count 26,914 – 24,231 10,530 – 9,107 5,218 – 6,108 55,148 – 45,606 9,275 – 9,063 12,766 – 8,430 Average Weekday 835 – 684 310 – 251 166 – 175 1,618 – 1,133 270 – 252 376 – 243 Peak 1 Hour Starts: 6 PM – 5 PM 5 PM – 5 PM 6 PM – 5 PM 6 PM – 5 PM 5 PM – 4 PM 6 PM – 5 PM Percentage of Day 13.2% – 14.2% 11.6% – 14.9% 15.0% – 12.5% 17.9% – 19.4% 10.7% – 11.0% 12.5% – 14.0% Avg. Weekend Day 1,025 – 1,017 447 – 430 192 – 252 2,352 – 2,181 408 – 372 541 – 398 Peak 1 Hour Starts: 4 PM – 2 PM 4 PM – 2 PM 6 PM – 4 PM 4 PM – 4 PM 3 PM – 3 PM 5 PM – 11 AM Percentage of Day 9.9% – 11.5% 11.2% – 14.2% 11.5% – 13.9% 10.0% – 12.6% 12.0% – 15.3% 9.4% – 11.3% Highest Single Hour 377 – 247 162 – 148 74 – 108 554 – 635 114 – 192 109 – 94 Note: Peak 1 hour start times are the average per city/trail for all weekday or weekend days surveyed, as appropriate. Table 16-136 Indiana Trails NMT Traffic Count Information (September–October 2000) Fort Wayne Goshen Greenfield Indianapolis Muncie Portage Male / Female Distribution 57% / 43% 57% / 43% 50% / 50% 46% / 54% 68% / 32% 51% / 49% Age Dist.: 25/26-45/46-65/ 66 11/49/32/8 19/34/37/10 16/39/36/9 12/50/32/6 18/36/35/11 18/36/36/10 Income Dist.: <$40K/$40-80K/>80K a 35/48/17 39/45/16 33/46/21 22/45/33 33/51/16 33/48/18 Percent College Graduates a 60 57 33 79 52 32 Race Dist.: White/Black/Hispanic 86/10/4 92/1/7 98/1/1 92/6/2 95/5/0 92/3/5 Pct. White from Mail-back Survey a 94% 98% 100% 97% 96% 96% Notes: a From mail-back survey responses. Both intercept observations and survey determinations of white race are listed to illustrate possible bias despite high mail-back rates. Table 16-137 Indiana Trails Demographic Percentage Distributions from Intercept Interviews and Mail-back Survey

Users on foot dominate especially on the urban trails, with walkers and runners totaling 64 percent in both Fort Wayne and Indianapolis. Wheeled users (bicyclists and skaters) constitute the 36 per- cent remainder. The percentages shift somewhat for the trails classified as suburban, those in Goshen and Portage, with a simple average of 54 percent users on foot. The dearth of skaters in Goshen—the “1%” of Table 16-138 is identified as “other”—may simply be reflective of trail surface conditions (crushed limestone). The trend continues on the mostly rural trail in Muncie, with only 16 percent users on foot. The Greenfield rural trail is a statistical outlier, with 68 percent users on foot, perhaps because it is only 3 miles long and fails to reach wide-open country in that distance. All six trails have uniformly high use for fitness and recreational activities. Use for utilitarian trans- portation purposes may be somewhat higher than meets the eye, however, a possibility explored below under “More . . .” Responses identifying whether the user’s trail entry and exit points are identical give an alternative indicator of utilitarian use. The 2 to 19 percent of trail respondents for whom entry and exit points were not the same are likely, although not certain, to have been using the trail at least in part for utilitarian travel purposes. It is notable that the majority of access is via auto except in Goshen, even though 50 percent or more of trail users were found to live within 2 miles of their trail. Behavioral impacts of all-new facilities such as these shared-use trails typically must be deter- mined from retrospective queries, or “what if?” questions, asked in interviews and surveys. While there are reliability issues with such survey approaches, they at least offer some insight. The first part of Table 16-139 contains response data suggesting that 14 to 19 percent of walkers, runners, cyclists, and skaters on the trails are engaging in their chosen activity only by virtue of the pres- ence of the trail. From 70 to 87 percent believe they are walking, running, cycling, or skating more 16-432 Fort Wayne Goshen Greenfield Indianapolis Muncie Portage Trail activities Walk 49% 39% 54% 51% 11% 39% Run 15% 20% 14% 13% 5% 11% Bicycle 30% 40% 25% 23% 77% 40% Skate/other 6% 1% 7% 13% 7% 10% Mean distance on trail 6 miles 3 miles 4 miles 8 miles 15 miles 7 miles Median time 35 min. 35 min. 40 min. 60 min. 90 min. 60 min. Purpose of trail use Health/exercise 66% 64% 79% 71% 56% 74% Recreation 32% 32% 19% 23% 39% 26% Commute 2% 4% 1% 5% 3% – Other – – 1% 1% 1% – Entry/exit points same 88% 81% 89% 91% 93% 98% Mode of access Walk 24% 27% 19% 29% 6% 9% Bicycle 17% 30% 19% 14% 27% 15% Auto 56% 40% 61% 52% 66% 71% Other 3% 3% 1% 5% 1% 5% Median access distance from home 2 miles 1 mile 1-½ miles 1 mile 2 miles 2 miles Note: Seeming discrepancies between distance and time on trail may result from one being reported as the mean and the other being reported as the median. Table 16-138 Indiana Trail-Use Information from Entry/Exit Intercept Interviews

because of the trail. (Less than definitive clarity is suggested by the fact that the two percentages for Muncie add to more than 100 percent.) In general, responses such as these suggest a substan- tive positive effect on incidence and frequency of physical activity. Alternative activity possibili- ties must be taken into account, however, as is done next in the “More . . .” subsection using findings from the second half of the table. 16-433 Fort Wayne Goshen Greenfield Indianapolis Muncie Portage First-time trail users a 9% 7% 11% 4% 9% 6% W/r/c/s now because of trail 19% 14% 14% 16% 19% 17% W/r/c/s more with trail available 79% 70% 74% 81% 87% 82% W/r/c/s time spent because of trail availability (median, weekly) 120 min. 100 min. 120 min. 180 min. 200 min. 180 min. Without the trail available would have participated in the same activity: b On streets or sidewalks 68% 68% 86% 59% 62% 59% In park or other outdoor place c 18% 15% 1% 16% 7% 12% In gym, mall, or other 2% 1% 0% 3% 2% 6% No, would have done something different 12% 6% 13% 2% 29% 22% Remainder d – 10% 0% 20% – 1% Note: W/r/c/s = walk, run, cycle, or skate. a Users who, on the survey day, had never been on the trail before. b From mail-back survey responses as tabulated in the individual trail reports (values not so indicated are from intercept interviews as tabulated in the Summary Report). c Including other trails or linear greenways. d “Stay Home” for Goshen and Greenfield, not explained for Indianapolis or Portage. Table 16-139 Impact of Indiana Trails on User Activity as Reported in Intercept Interviews and Mail-back Survey More . . . Some of the Indiana Trails user survey questions offer a look behind effects observed “on the surface.” The second half of Table 16-139 covers responses from probing what activities trail users would have engaged in had the trail not been there. Some 71 to 88 percent of survey respondents advised that they would have walked, run, cycled or skated somewhere else, mainly on streets or sidewalks. This indicates a high level of commitment to the chosen activity, although it is not to say respondents would have been active to the same extent. Another 2 to 29 percent would have done something different. It is not clear how much transportation or physical activity that would have involved. Finally, there were those who would have remained at home (10 percent in Goshen) or whose alternative activity was not accounted for in the survey response reporting (20 percent in Indianapolis). Overall, there is imprecise but strong indication that many trail users became more physically active as a result of trail development. In addition, the user survey asked about both “main” and “other” purposes of visiting the trail. The purpose information in Table 16-138 pertains to the main use. As an example, among users of the Monon Trail in Indianapolis, 5 percent reported “commute” as their main purpose. Of those who answered the other-purpose question, 12 percent reported “commute,” and another 3 percent reported various utilitarian secondary purposes ranging from dining to business. Although many users undoubtedly combine health/exercise with recreation, it nevertheless appears that combinations of

health/exercise or recreation with commuting and other utilitarian transportation must be taken into account when interpreting trail use. The proportion of trips on the Monon Trail that serve to accom- plish the work commute could, for example, theoretically lie somewhere between the raw figure of 5 percent and a maximum of 17 percent (5 percent primary-purpose plus 12 percent secondary-purpose) if assessed solely on the basis of “main” and “other” purposes. However, it would seem illogical for trips with a commute purpose to enter and exit at the same trail access point. Thus in the Monon trail example, with 91 percent of users reporting the same entry and exit points, no more than 9 percent can reasonably be true commute trips unless, perhaps, the entry/exit point question was misunderstood. The trail user surveys also included attitudinal questions. The median attitude toward the trail on all six facilities was one of being “very satisfied” (5 on a 6-point scale). The city was viewed more favorably by 76 to 100 percent of users as a result of the trail, with no report of viewing the city less favorably. Only a minority of users, however, found the trail to be a reason for choice of housing location. Surveyed persons living adjacent to the trail were more guarded in their responses, but were nevertheless satisfied (five cities, 5 on a 7-point scale, median ranking) or neutral (one city, 4 on a 7-point scale) with regard to having the trail as a neighbor. Among those having purchased their home after trail opening, the reaction was one of being “very supportive” (6 on a 7-point scale) or, in one city, “extremely supportive.” Estimated household trail use by trail neighbors was 117 to 139 days annually. Roughly 90 percent perceived that the trail had modestly added to or not affected their property value. Interviewed realtors were just slightly more circumspect, advising that they saw no major increases in property value or ease of making sales. Sources. Patten, R. S., Derry, A., Hiemstra, H., and Fowler, M., “ISTEA and Trails: Merging Transportation Needs and Recreation Values.” Published by Rails-to-Trails Conservancy and American Trails for the 12th National Trails Symposium, Anchorage, Alaska. http://ntl.bts. gov/DOCS/mtn.html (September, 1994). • Indiana University, “Indiana Trails Study—A Study of Trails in 6 Indiana Cities.” Summary Report and individual trail reports (Rivergreenway Trail—Ft. Wayne, IN; Maple City Greenway Trail—Goshen, IN; Pennsy Rail Trail—Greenfield, IN; Monon Trail—Indianapolis, IN; Cardinal Greenway Trail—Muncie, IN; Prairie Duneland Trail—Portage, IN). Prepared by Epply Institute for Parks & Public Lands, Indiana University, Bloomington, IN (November/December, 2001). • “Without trail” distributions in Table 16-139 and certain interpreta- tions in the activity and purpose discussions include elaborations by the Handbook authors. Variations on Individualized Marketing in the Northwest United States Situation. “Individualized marketing” seeks to modify travel choices by delivering tailored infor- mation on walking, bicycling, and public transit options for meeting daily travel needs. The “Individualized Transit Marketing in Europe” case study in Chapter 11, “Transit Information and Promotion,” describes procedures and outcomes for the original IndiMark™ individualized transit- marketing protocol developed by Socialdata GmbH. Chapter 16’s “Response by Type of NMT Strategy” section provides an update keyed to active transportation applications in the United States, the United Kingdom, and Australia (see “Individualized Marketing” within the “Walking/Bicycling Promotion and Information” subsection). Over one-half of these applications have followed the essential IndiMark protocol while the remainder are variants. Programs in Northern California and the Pacific Northwest illustrate newer developments that range from major departures from the original protocol to incremental but substantive enhance- ments to the IndiMark approach. In adjudged order of increasing interactivity of personalized intervention, the four programs covered below are the “In Motion” demonstrations in Seattle; the “Way to Go Sausalito” pilot program within Marin County, California; the post-2004 SmartTrips 16-434

campaigns in Portland, Oregon; and the Whatcom Smart Trips IndiMark application in Bellingham, Whatcom County, Washington. Actions. Seattle’s “In Motion” campaign was initiated in 2004 with demonstration programs situ- ated in neighborhoods with reasonable sidewalk availability, access to nearby services, and at least half-hourly bus frequencies all day. The demonstrations, in terms of a “passive,” “active,” and “inter- active” classification system, exhibited characteristics of a passive targeted marketing campaign. Individualized information materials and incentives were innovatively advertised and delivered, but the intended recipients had to proactively react to receive them. They had to react again, more or less on their own, to put alternative travel mode use into practice. The “In Motion” approach put sub- stantial emphasis on blanket neighborhood promotion, such as catchy telephone-pole posters and direct mail, to prompt inquiries about the additional information available. Those choosing to respond received materials typical of individualized marketing programs, via mail, and also were asked to take an alternative-mode-use pledge. A website, neighborhood displays, a neighborhood Transportation Action Team, and local events were part of the community-focused effort. The “Way to Go Sausalito” pilot program of 2008 was an element of “Walk Bike Marin,” the Marin County component of the multi-faceted national Nonmotorized Transportation Pilot Program. Sausalito is an historic, mostly upscale, waterfront city of some 7,000 residents in the San Francisco com- mutershed. The Sausalito approach was arguably intermediate between “passive” and “active” in char- acter. To obtain their “Go Kit” of customized information, residents had to respond to a mass-mailed newsletter, reminder postcard, other conventional community communication media, or the project website. “Go Kits” were delivered by bicycle in “Way to Go” tote bags. The Sausalito program is per- haps most notable for having a full menu of events, though as quantified below under the “More . . .” heading, attendance at many was quite small. Portland, following a demonstration in 2003 and their first full-scale sector program in 2004 (both IndiMark based), worked to handle subsequent programs with mostly in-house staff and volun- teers and to broaden enticements and events. (Table 16-143 under “More . . .” illustrates the array of materials and activities offered.) At the same time, conduct of any one-on-one alternative mode assistance sessions at the residence was apparently dropped. Portland’s post-2004, annual, sector- by-sector SmartTrips campaigns thus probably fall in the “active” category of individualized mar- keting, with contact calls and personalized delivery of materials, but with almost full reliance on the receiver for ultimate action. The 2008 “Neighborhood Smart Trips” component of Whatcom Smart Trips was an enhanced IndiMark application. It targeted the central area and coastal corridors of Bellingham between Bellingham Bay and the I-5 freeway. Extending from north to south city limits, some newer sub- urban-style development was included, but coverage was dominated by older areas with grid streets, sidewalks and/or light traffic, and substantial shared use trail and bike lane infrastruc- ture. Just over 10,000 households were included, representing about 1/3 of the city’s population. This IndiMark application was the first large-scale use of three different contact methods to address decline in listed land-line telephone numbers and ensure all possible target area house- holds were reached. Telephoning, after an introductory mailing, was the preferred initial contact. If telephone contact proved impossible and outreach via U.S. Mail produced no response, dwelling-to-dwelling door knocking was employed in selected walkable areas. In all areas, “dif- fusion” was also relied upon, counting on residents to see the involvement of neighbors and become interested themselves. Representative of the “interactive” approach, dialogue was estab- lished, requested materials were delivered, and follow-up interaction—both group and one-on- one—was encouraged. 16-435

The overall ongoing Whatcom Smart Trips program includes an interactive web-based Smart Trips Diary where adults living or working in Whatcom County (including British Columbia residents) can maintain a record of walking, cycling, transit, and ridesharing trips made. The diary automat- ically calculates statistics such as pollution prevented and is tied to a system of rewards (gift cer- tificates and the like) and recognition for reaching Smart Trips milestones. An emergency ride home program (a.k.a. guaranteed ride home or GRH) is provided for commute trips, and a Smart Trips Employer Partners program offers assistance to state-mandated and volunteer worksite trip reduction programs. Targeted outreach to seniors and women provides education in the use of the bus system and bicycles, respectively, while School Smart Trips offers middle school classroom activities. An EverybodyBIKE educational program provides cycling mentors, skill rodeos for chil- dren, and safety classes. Analysis. Table 16-140 lists the primary evaluation parameters for each of the four programs. Seattle’s “In Motion” demonstrations did not include surveys of target area residents overall. The only response data are for the 6 to 10 percent who actually became participants in the initial three demonstrations and thus are not included here for lack of compatibility with the survey informa- tion from other cities. Target-neighborhood bus boarding increases were compared with boarding statistics for a control neighborhood. 16-436 Program Before Sampl e After Sampl e Before/After Response Rate Control Group External Evidence Seattle “In Motion” None — a n/a Bus boardings Bus boardings “Way to Go Sausalito” 1,525 1,500 18%/11% b None reported None reported Portland (2005) E Hub 300 c 300 c n/a 1/2 of samples Walk, bike counts Portland (2006) NE Hub 600 c 600 c n/a 1/2 of samples None reported Portland (2007) SE 600 c 600 c n/a None reported Bike counts Portland (2008) SW 692 d 288 d n/a/64% d None reported Bike counts Portland (2009) N/NW No specifics reported on the before and after surveys None reported Whatcom Smart Trips 7,495 e 3,863 e 76%/78% Yes – included Anecdotal Note: a A self-assessment survey was conducted of “In Motion” participants only. b The self-administered-survey response rate drop-off in the randomly-selected Sausalito samples, to the notably low 11% “after” survey response rate, introduces above-average concerns of possible response bias (Handbook authors’ assessment). c Number of randomly selected telephone interviews completed. d Of 692 respondents to September 2007 and April 2008 surveys prior to the 2008 TravelSmart, 449 households agreed to participate in a September 2008 follow-up panel survey, and 288 (64%) were reached and did so. e Number of actual respondents, not the entire randomly selected target group sample. Table 16-140 Seattle “In Motion,” “Way to Go Sausalito,” Portland “SmartTrips,” and Whatcom “Smart Trips” Target Group Evaluation Parameters The Sausalito evaluation surveys covered in Table 16-140, including Note B, were conducted immediately preceding and after the individualized marketing and related activities. Therefore, effects identified may or may not have been short term only. There was no control group or collec- tion of external evidence.

The Portland Table 16-140 entries do not include the 2003 and 2004 IndiMark-based programs. The variety of post-2004 evaluation approaches and reporting in Portland may be judged from both Table 16-140 and the results descriptions to follow. It is rather obvious from the statistics in Table 16-140 that the most robust survey-based analysis potential is offered by the large Whatcom Smart Trips before and after surveys with their 76 and 78 percent response rates. The Whatcom Smart Trips application in Bellingham followed the basic IndiMark protocol includ- ing follow-up prompts to increase response rates for the relatively large surveys. These surveys were scheduled roughly 1 year before and 1 year after the 2008 individualized marketing. Both the before and after survey samples were random picks, as contrasted to the panel approach less often used. The high response rate, use of trip diaries, lack of reference to the Smart Trips endeavor, and response rate factoring supported the minimization of potential for bias in the findings. Factoring was separately done for the standard IndiMark “Interested,” “Regular [user],” and “Not Interested” individualized marketing groupings. The control sample component of the survey populations was used to identify trends, which were in turn employed to project mode shares for a hypothetical “without IndiMark” target population. These were then used as a base to compare “with IndiMark” outcomes against. For example, control-group public transportation bus riding increased 29 percent from 2007 to 2009 in presumed response to increased service and a Western Washington University bus pass. Thus the target group 2007 public bus mode share of 3 percent was adjusted to 4 percent before comparing 2009 “with IndiMark” out- comes against it. Results. Results documentation for these innovative Northwest U.S. and Northern California programs has mostly been too limited, and lacking in discernible outcome differentials relative to other individu- alized marketing projects, to allow firm conclusions as to effectiveness of the broadening of information and activity menus. Portland auto driver trip reductions have not varied from the norm established prior to the 2005 addition of tours and workshops, except in 2006 during marked increases in gasoline prices. The Whatcom Smart Trips undertaking comes with extensive survey data and very favorable compar- isons, but caution should be used in transferring findings to other areas, given Bellingham’s location in the midst of the environmentally conscious Pacific Northwest. In any case, broadened menus of support actions are unlikely to detract from the individualized marketing core approach—unless they replace proven protocols—and may support other community objectives. Overall target area mode shifts were not surveyed in Seattle, but the demonstration study-area bus-boarding increase of 11 percent for up to 9 months after program implementation—compared to 1 percent in a control area—was notable. The percentage of target area households actually par- ticipating in the initial “In Motion” demonstrations was, however, only 6 to 10 percent. The lesser emphasis on proactive contact and interactive, dialogue-based follow-up may have shifted out- comes toward the lesser response typically found with mass marketing approaches. The “Way to Go Sausalito” program results, where the proportion of households requesting “Go Kit” information packets was just over 15 percent, may similarly reflect an emphasis shift toward event- and website-based mass marketing techniques as compared to priority emphasis on highly proactive multi-pronged individualized dialogue/contact. Based on the before-and-after surveys addressed in Table 16-140, including Note B, the effect on Sausalito trip making was reported as a 9.5 percent rela- tive increase in resident share of trips via active transportation (12.8 percent for walking and bicy- cling and no change in transit riding) and a 4.7 percent resident decrease in auto share. The absolute mode shift equivalents were +3.3 percentage points for walk/bike, zero shift for transit use, and −3.0 percentage points for auto use. Given the survey timing, these were short-term effects, with longer-term effects unknown. 16-437

Portland’s IndiMark individualized marketing results for 2003 and 2004 were covered in the “Response by Type of NMT Strategy” section along with selected information for subsequent years (see “Walking/Bicycling Promotion and Information”—“Individualized Marketing”—“U.S. Home/Community-Based Program Mode Share Results.”) Following post-2004 program changes, full mode shift detail has been reported only for certain years, but auto driver trip reduction results are available for each annual program. The 2005 Eastside Hub “Options” project resulted in before-and-after survey-based relative mode share changes of +7 percent for walking, +41 percent for transit use, −8.6 percent for auto driving, no change in carpooling, and an insignificant shift for bicycling. Before-and-after two-hour AM, midday, and PM peak counts at selected Eastside Hub intersections showed a 7 percent walking increase in confirmation of the survey results. Corresponding 10-intersection bicycle counts indi- cated a 23 percent increase in average cycling volumes.97 The 2006 program covered the Northeast Hub area, introducing the city of Portland’s own “SmartTrips” branding. The 2006 results were explicitly adjusted for rather significant mode shifts, identified in control group survey findings, thought likely to be the result of gasoline price increases. The adjusted absolute gains for environmentally friendly modes were 5 percentage points walk mode share gain, 2 percentage points bike share gain, and 1 percentage point bus- and light- rail-transit-share gain. The drive-alone adjusted shift was an 8 percentage points absolute decline (with carpooling increasing 3 percentage points) or a 12.8 percent decline in relative terms. The 2007 SmartTrips Southeast project reported a 17.5 percent overall relative increase in use of environmentally friendly travel modes among southeast residents. Peak-hour 3-day AM, Noon, and PM bicycle counts at four key southeast intersections, obtained in September of 2006 and again in 2007, averaged a 26.5 percent increase. This was higher than the 18 percent citywide increase seen in the annual reporting of bicycle counts. The 2008 SmartTrips Southwest project found weekday walking shares to have increased 36 percent, from 8.3 percent before individualized marketing to 11.3 percent after. Drive alone trips decreased 9.0 percent in relative terms. Bicycle work-purpose trips increased 38 percent, from 3.1 to 4.3 percent, but across all trip purposes the bike mode share remained constant. However, a count program sim- ilar to the previous year, covering three southwest intersections, found a 42 percent average growth in bicycle volumes—one-and-one-half times the citywide annual increase of 28 percent. In Portland’s 2009 North/Northwest project, approximately 7,500 target area households (about 25 percent) ordered materials, participated in one or more of the program events, or stopped by tables manned at other neighborhood events. The percentage specifically ordering materials was 12 to 13 percent. A survey-based relative increase estimate of 10.5 percent in environmentally friendly mode usage was accompanied by a 9.3 percent decrease in drive-alone trips. As indicated in the “Response by Type of NMT Strategy” section, Portland appears to have achieved somewhat larger environmentally friendly mode shifts than observed in the 2003–2006 FTA IMDP National Demonstrations, which did not include Portland. Comparing 2005–2009 Portland results with the earlier 2003–2004 Portland outcomes, however, there is no sound basis for concluding that the protocol changes and innovations introduced in 2005 and enhanced in suc- 16-438 97 City-wide secular trends were not reported in connection with the pedestrian and bicycle count results for the 2005 Eastside Hub individualized marketing analysis.

ceeding years have either increased or decreased program effectiveness as measured by travel mode shift outcomes. The Whatcom Smart Trips 2008 Bellingham application of an enhanced IndiMark protocol has pro- duced among the largest, if not the largest, of shifts to environmentally sustainable travel modes of any major individualized marketing project to date. The results have included substantial shifts to walking and bicycling. A 15 percent reduction in auto vehicle miles of travel (VMT) has been computed. Of 10,037 households targeted in the individualized marketing outreach, 8,880 (88 percent) were successfully contacted. Among these, 356 (4 percent) were regular environmentally friendly mode users with no information needs, 847 (9 percent) were regular users who wanted more informa- tion, 3,963 (45 percent) were interested in exploring use of environmentally friendly modes, and 3,714 (42 percent) were not interested. With 90 percent of interested households following up with an individualized marketing materials request, combined with regular users who wanted more information, roughly 50 percent of households reached and 44 percent of target area population obtained offered materials. Table 16-141 presents the changes in mode shares measured in the control group. As noted earlier, the increase in public transit use from 2007 to 2009 is thought to reflect effects of bus service increases and Western Washington University bus pass availability and use. The control group changes observed were applied to the 2007 target group shares, as previously explained, to obtain adjusted “before” shares as shown in Table 16-142. The “after” shares in Table 16-142 are best com- pared with the adjusted “before” shares. That is how the relative and absolute mode shifts attrib- utable to the Smart Trips project (last two columns) are computed. 16-439 Travel Mode 2007 Mode Shares 2009 Mode Shares Relative Changes (Percent Up/Down) Absolute Changes (Percentage Points) Walk 7% 7% -1% 0 Bicycle 3% 3% +2% 0 Public Transit 2% 3% +29% +1% Auto Driver 63% 62% -1% -1% Auto Passenger 24% 23% -3% -1% Notes: Absolute changes calculated by the Handbook authors from before/after mode share percentages reported in integers. Motorcycle mode omitted (1 percent or less). School bus mode omitted (1 percent throughout). Table 16-141 Whatcom Smart Trips 2007 to 2009 Bellingham Control Group Mode Changes

Target group shifts to active transportation modes were fairly evenly distributed across all purposes of travel, including leisure, judging by auto trip reductions by trip purpose. There was perhaps a moderately elevated impact on education and “other” purpose trips. Quantification of physical activ- ity effects is provided in the “Response by Type of NMT Strategy” section (see “Walking/Bicycling Promotion and Information”—“Individualized Marketing”—“Home/Community-Based Program Effects on Physical Activity”). More . . . Portland’s annual SmartTrips materials distribution and event participation tallies offer an indication of participant interest levels in different forms of information and outreach. A core ele- ment of the distributions has been the “Ten Toe Express” walking campaign kit. A popular item in the kit is a discount coupon book for and supported by businesses within walking distance. Local businesses are said, anecdotally, to have gained new customers from this outreach. The city has also succeeded in attracting health maintenance organization support in the form of pedestrian/bicycle map, “Ten Toe Express” kit, and guided walk sponsorship. Table 16-143 lists items and activities made available to interested individuals in the 2009 campaign and gives the number of requests or participants for each. The numbers reflect more than one distribution protocol and some event participation overlap, thus the indication of interest levels is imprecise. Nevertheless, the requests/ distributions for walk/bike maps covering neighborhoods outside of the 2009 North/Northwest tar- get area clearly suggest that such maps are very much in demand. 16-440 Travel Mode 2007 Target Group Shares Adjusted “Before” IndiMark Shares 2009 Target Group “After” Shares Relative Mode Shifts (Percent Up/Down) Absolute Mode Shifts (Percentage Points) Walk 16% 16% 20% +22% +4% Bicycle 8% 8% 11% +35% +3% Public Transit 3% 4% 4% +11% <+1% Auto Driver 51% 50% 44% -13% -6% Auto Passenger 21% 20% 19% -3% -1% Notes: Absolute changes calculated by the Handbook authors from before/after mode share percentages reported in integers. Motorcycle mode omitted (1 percent or less). School bus mode omitted (1 percent throughout). Table 16-142 Whatcom Smart Trips 2007 to 2009 Bellingham Target Group Mode Shifts

Comparable information for the “Way to Go Sausalito” pilot program of 2008 reflects a much smaller-scale operation, but is of interest because all reported distributions were apparently upon participant request only. Respondent interest levels are thus more directly represented. Order forms for materials were mailed to 5,402 households and 844 orders were received. Materials and activities requested or participated in were (listed in decreasing order of interest with number of requests or participants in parenthesis): Way to Go (WTG) Sausalito map (701), WTG event calendar (675), WTG coupon book (673), Golden Gate Bus and Ferry Guide (498), pocket ferry brochure (474), 511 Getting There on Transit Guide (453), WTG Guide to Your Ride 16-441 Materials/Activities Number Materials/Activities Number Ten Toe Express walking kits 5,500 a TriMet transit info., maps, sched., etc. 5,940 d Portland by Cycle kits 4,700 a Transit TrackerTM bus stop IDs 1,005 NW Portland Walk/Bike Map 3,000 b CarpoolMatchNW.org materials 270 N Portland Walk/Bike Map 4,000 b Zipcar brochure 506 NE Portland Walk/Bike Map 1,000 Smart Driver brochure 652 SE Portland Walk/Bike Map 860 AAA Safe Driving for Seniors booklet 376 Outer SE Portland Walk/Bike Map 590 SmartTrips umbrellas (incentives) 1,483 SW Portland Walk/Bike Map 840 Bandana bicycle maps (incentives) 1,106 Citywide Bicycle Map 6,000 c Walk There! booklets (incentives) 872 Downtown Bike Map 1,300 Ten Toe Express walks (17 walks) ~200 Portland by Cycle flyer 8,000 c Senior Strolls (22 strolls) n/a e Portland by Cycle Guide 6,000 c Portland by Cycle rides (19 rides) 230 f Women on Bikes flyer 5,500 c Portland by Cycle workshops (12) ~100 Senior Stroll flyer 2,750 Women on Bikes clinics and rides 190 g Notes: Order forms were mailed to some 28,000 of the 29,500 Northwest and North neighborhood households. A total of 3,656 households ordered materials. This 12 to 13 percent rate was smaller than previous campaigns, thought to reflect bundling of order forms with junk mail at the numerous apartments and condominiums. To compensate, Ten Toe Express and Portland by Cycle kits were also distributed through libraries, schools, community events, and other venues. The numbers of kits and items contained in the kits reflect this augmentation. a Of Ten Toe Express walking kits, 3,900 were ordered or distributed at neighborhood events, 1,100 were made available to libraries, schools, and non-profit groups, and 500 were not accounted for in the documentation. The Portland by Cycle kits were distributed similarly. b Distribution included those inserted in Ten Toe Express and Portland by Cycle kits. c Distribution included those inserted in Portland by Cycle kits. d Number appears to be a count of items distributed. Some kind of transit information was ordered by 2,048 individual households. A popular item separately listed/counted is the personalized Transit TrackerTM card with the ID numbers of nearby bus stops, with which real-time bus arrival times can be obtained via telephone or a number of web-based options. e Attendance in 2008 (SmartTrips Southwest) for 22 Senior Strolls averaged 20 persons per stroll, with many repeat attendees from prior years and other neighborhoods. There were 50 first-time strollers overall. f Attendee total for 19 rides was 230, representing 136 different riders. g Signup total for rides and clinics was 190, representing 175 individuals. Table 16-143 Portland 2009 SmartTrips North/Northwest Campaign Transportation Materials, Incentives, and Activities and Acceptance/Attendance Totals