AbstractThis study aimed to consider the street structure and the relationship between such structure and people’s route choices as well as people’s street structure concerns or preferences based on a common mapping tool. We used the map recording method to conduct a route choice experiment, captured the street structure–strolling route relationship, and performed a route choice experiment in actual space, preliminarily testing the results of the previous experiment. We also conducted interviews to determine how participant perceptions of street space on the map differed from perceptions of actual street space and how paths were chosen. The results show that the participants’ route choice tendencies change with street structure characteristics. Despite a rich built environment in real space, the participants give considerable attention to street structure. Based on the route choice strategies, the participants could recognize the streets on blank maps as real streets and imagine themselves walking on those streets. However, the map recording method has limitations, e.g., the influence of pedestrians’ physical strength on their route choices is not considered.IntroductionPeople are paying increasing attention to reducing environmental pollution (Blanco et al. 2009), maintaining their health (Sallis et al. 2004; Hoehner et al. 2005), and increasing social communication (Leyden 2003; Wood et al. 2010); therefore, there is increasing interest in walking behaviors that can actively address these issues. In recent years, research has been conducted on walking behavior (Sugiyama et al. 2012), and the research on route choice has produced an interdisciplinary field involving urban planning (Gim and Ko 2017), transportation (Marchand 1974), geography (Borgers and Timmermans 1986), and other disciplines. In the future, it will be important to construct urban streets that are favored by a large number of pedestrians; therefore, it is necessary to understand the characteristics and influential factors of pedestrians’ route choices.Street space is composed of the built environment (buildings, facilities, and plants) and street structure (street shape and type of connection), both of which will affect route choice behaviors. In this study, the street structure is defined as the connection between streets (topological attributes of streets) and the length, width, and shape of streets (morphological attributes of streets). Between these two components, the street structure plays an important role in determining the spatial structure of cities (Nishio et al. 2001; Sharifi 2019), which is difficult to modify after the completion of construction; otherwise, the surrounding buildings and facilities must be demolished, also causing damage to the existing landscape. Therefore, the influence of street structure should be considered in the study of route choice (Nishio et al. 2000). Therefore, the form of street structure should be carefully considered in the early stage of planning and design stages in addition to the needs and preferences of pedestrians including efficiency and walking pleasure.BackgroundMemory-based travel diaries (sent through the mail) are a traditional method of obtaining route choice data. However, the accuracy of the data is problematic because the method requires people to recall recently visited places and record their route, which is strongly influenced by an individual’s memory and problem-understanding ability (Prelipcean et al. 2018). Another method is to obtain data while moving along streets, mainly by (a) pedestrians moving and collecting data directly in the street space or (b) using auxiliary tools to collect the data.Method (a) is commonly used to analyze the influence of the built environment on pedestrian behavior by considering real space as the object. Although this method can obtain data that are closest to real situations, there are many factors in real space, such as street networks (Kang 2017), land use (Cervero and Kockelman 1997), shops (Borgers and Timmermans 2014), and traffic volume (Werberich et al. 2015). Moreover, it is difficult to obtain enough samples to accurately explain the effect of street structure (Shatu and Yigitcanlar 2018).To address this problem, researchers have proposed methods for collecting data using auxiliary tools such as simulated space or maps [Method (b)]. Computer-generated simulation space not only reproduces the characteristics of real space but also freely changes the spatial characteristics according to the research purpose (Togawa et al. 2006). However, simulated space has high requirements for equipment performance and technology; therefore, its universality is low. In particular, actions other than walking (such as operating a joystick or keyboard) that are similar to playing a game are often performed in a simulated space (Sueshige and Morozumi 2005; Togawa et al. 2006). The results are also influenced by participants’ familiarity with the equipment and their learning ability.In addition, the method of recording paths on paper maps, namely, map recording method, has also been used. Although this method does not allow participants to experience real streets directly, it can ensure a sufficient number and diversity of samples and can simplify the relationship between the street structure and the route choices. Osaragi and Tanaka (2017) asked participants to imagine walking along a street on a map and record their route and attempted to build a walking motion model with the walking distance and street width as the independent variables, which provided inspiration for this study. As a common tool, maps are widely used in daily life and have become a powerful information source for people to obtain knowledge of street space (Wakabayashi 2008). It has proven useful to consider the street structure and route choice behavior from the perspective of maps. Based on this background, to capture the relationship between route choices and street structure, to avoid the interference of other built environments in real space, and to remove the technical limitations of simulated spaces, we focus on maps that are familiar to people. However, although there is much less information on maps than in real space, it is still possible to infer a location based on the information shown on the map, and whether people know a location will have an impact on the route choice (Sueshige and Morozumi 2007; Nakada and Dohi 1982). For example, people who know a location are more likely to choose the facilities they typically use or to follow the routes to which they are accustomed, and it is, therefore, difficult to extract the influence of street structure.To solve this problem, our study used maps with simpler information than ordinary maps that include only the street structure (blank maps). We collected data using the map recording method and analyzed the influence of the street structure on route choices. The purposes of this study are: (1) to understand the objective characteristics of the street structure from the perspective of the whole region based on a map, which is a common tool; and (2) to aim to reduce the impact of other environmental factors to understand the relationship between the street structure and the people’s choice of route as well as their street structure concerns or preferences.The article consists of the following sections: In the “Case Study and Methodology” section, after summarizing the selection methods of the experimental areas, we introduce two route choice experiments and the indicators used in the analysis. In the “Characteristics of Object Spaces” section, we calculate and describe the characteristics of the street structure in each experimental area. In the “Results and Discussion” section, we compare the characteristics and strategies of route choice between blank maps and real space to verify the purpose of the study. In the “Conclusions” section, we summarize the results and problems.Case Study and MethodologySummary of Experiment on Blank MapsSelection Method of the Blank MapsBefore selecting the experimental areas, we carried out similarity evaluation experiments of the street structure. The experiments were carried out from October 10 to 16, 2015. We hoped that the participants could understand the street structure on the map, understand our experimental intention, and use their professional knowledge to help us classify the street network. Therefore, we solicited participants from people with architectural or related professional backgrounds, and we considered the following issues: If we randomly looked for participants from the street, it would be difficult to ensure that they had relevant professional backgrounds. If we looked for construction-related practitioners, it would be difficult to ensure that they would have enough time and patience to complete the map classification work. Therefore, choosing our undergraduate students or graduate students would be a more reliable method (we chose seniors and graduate students from Hiroshima University to carry out experiments in their spare time). After 4–6 years of professional study, we thought that they would have enough professional knowledge to complete the map classification work. Our experiment was a voluntary registration, and there were 15 participants (van Oel and van den Berkhof 2013) who signed up for the experiment, including 4 female participants. Considering that this experiment did not involve a comparative analysis of gender differences, we did not mandate the gender of the applicants when we recruited participants. Of course, we need to perform more detailed research in the future, including a comparison of gender differences. At that time, we will require the gender ratio of the participants to be close to the standard of 1:1.The content of the experiment is as follows. It is important to ensure the diversity of study areas (Koohsari et al. 2013). To collect various street structures, we choose experimental areas from 350 urban blocks introduced in the TV program Walking on the Streets of the World (NHK 2005-2015). First, we sent the maps of these 350 urban blocks to the participants and asked them to group the maps based on similar street structures. The scale of each map was 1/10,000, and each map was printed on B7 paper so that we could lay out all the maps in a classroom for the participants to examine.Then, according to the participants’ grouping results, for example, Participant 1 judged that Maps A, D, and F are a group, Maps B and E are a group, and Map C is a separate group, as shown in Fig. 1(a), we gave 1 point between Maps A, D, and F, 1 point between Maps B and E, and 1 point to map C itself. Participant 2 judged that Maps A, B, and C are a group, Maps D and E are a group, and Map F is a separate group, as shown in Fig. 1(b); we gave 1 point between Maps A, B, and C, 1 point between Maps D and E, and 1 point to Map F itself. We could obtain the score of Fig. 1(c) by adding the scores of the two groups. Based on this method, we added up the scores from the 15 participants, and the final score was used as the row–column distance, which was input into IBM SPSS statistics 20 software for cluster analysis (the longest distance method), and we divided the 350 maps into several groups.To ensure the diversity of street structures and to reduce the burden on the participants in the route choice experiment, we divided the maps into four groups. The results of the characteristics of each group are presented in Table 1. We choose a map from each group that best represents the characteristics of the group. There are some maps with seas, rivers, railways, and other elements that easily attract people’s attention; therefore, we needed to remove these maps. The city name and the cluster of each map are presented in Fig. 2, which shows a smaller scale version of these maps (Table 1).Table 1. Characteristics of each clusterTable 1. Characteristics of each clusterClusterNumber of maps included in each clusterNumber of streetsShape of the streetsShape of the street networkDirection of the streetsDistribution of the streetsNumber of Cul-de-sacs144ManyLinearLike a gridSame directionEvenly distributedFew262MediumLinearLike a skewed gridDifferent directionsEvenly distributedFew383ManyCurvyLabyrinthineDifferent directionsEvenly distributedFew4161MediumCurvyIrregularDifferent directionsEvenly distributedMediumTable 2. Date and weather conditions of the real space experimentTable 2. Date and weather conditions of the real space experimentDateDayTime rangeWeatherAir temperature (°C)October 21Sunday12:40–13:15Sunny10–21October 24Wednesday12:25–14:05Sunny9–22October 25Thursday12:40–13:50Sunny10–21October 26Friday12:30–14:00Cloudy12–18October 29Monday12:30–14:10Cloudy10–18October 30Tuesday12:40–14:05Cloudy6–17October 31Wednesday11:55–13:00Cloudy5–14The range of each map is set to 1.0 km2, and the result is obtained by calculating the average area of the 80 samples suitable for walking and posted on tourism websites. This area is more suitable for walking; too large of an area will reduce people’s willingness to walk. The scale of each map is 1/6,000, and the maps were printed on A4 paper. Of course, the four maps are insufficient to cover all street structure types. In this experiment, we first confirm the similarities and differences in people’s route choice and strategy under these different street structures. In the subsequent experiment, we add regional samples to enrich the research results.Route Choice Experiment on Blank MapsWe used the four blank maps selected in the “Selection Method of the Blank Maps” section for the route choice experiments. The experimental dates were August 27, 29, and 31; September 10, 12, and 19; and October 3–6, 2018. To minimize the influence of the participants’ experience, occupation, or other attributes, 24 participants were selected among students at Hiroshima University, including 19 males and 5 females aged approximately 20 years, who differed from the participants involved in the similarity evaluation experiment on the street structure.Walking is divided into transportation walking and recreational walking (Hasegawa et al. 2003), and the characteristics and influential factors of the route choices are different (Czogalla 2012; Cao et al. 2006). In the case of transportation walking, people will prioritize how to reach the destination; therefore, they are likely to choose the shortest route regardless of the type of street structure (Shatu et al. 2019). However, recreational walking involves paying more attention to pleasure and comfort, which is easily influenced by the street characteristics (Lee and Moudon 2006). Recreational walking can reflect the relationship between walking and the street from a different perspective. Therefore, we use recreation walking (i.e., strolling in this study) as the research object and aim to identify a clear relationship between the street structure and the route choice.In addition to the route choice behavior itself, the participants’ thoughts and focuses (strategies) should also be considered to be a part of their behavior. Interviews or questionnaires are common methods to obtain people’s strategies and environmental evaluations. In particular, if participants are asked questions immediately after walking, problems that are easy to ignore under normal circumstances can be identified (Kelly et al. 2011). Therefore, we asked the participants to identify their strategies as soon as they had chosen their route so that we could more easily determine the decisive factors of their route choice.The content of the experiment is as follows: (1) The participants were asked to imagine themselves strolling in the areas shown on each map and record their walking route on the maps, as shown in the example in Fig. 3 (5 min). (2) Then, the participants were asked to mark the places that interested them and that they wanted to see (places of interest) (1 min). (3) Immediately at the end of the route choice, the participants were asked to write down their route choice strategies. They were required to identify their overall strategies and those for each map separately (no time limit).During the experiment, the participants were told that the route choice behavior was for strolling without a specific purpose. To give them a general sense of distance, we sent them a map of Hiroshima University (including buildings) as a reference. When setting the starting point and ending point, we consider that if the starting point and ending point are set in different places, the strolling will become an exploration action toward the ending point; therefore, we set them as the same place. As shown in Fig. 3, there are six starting/ending points at the edge of each map, and four participants were randomly assigned to each place. In addition, the experiment was repeated three times to reduce the influence of experimental errors. The participants had different starting/ending positions in each experiment.Summary of Experiment in Real SpaceMethod for Selecting the Real Space Experimental AreaNext, we introduce the method for selecting the real space experimental area. The actual space does not come from the four previously noted cities. Instead, other urban blocks were chosen to avoid the interference of route choice habits and memory in the previous blank map experiment. In addition, as mentioned previously, there are many influencing factors in real space. For example, Guo and Loo (2013) analyzed the strong influence of traffic and open space on route choice for general urban streets. Some studies have analyzed route choice behavior for commercial streets and concluded that the shop type is also an influencing factor (Hahm et al. 2017). However, our study focused on the street structure and sought to eliminate the influence of other built environments as much as possible. Therefore, we chose a residential area with more uniform architectural forms and fewer factors [Fig. 4(a)]. To provide some comparability between the two experimental areas, for the chosen street structure, we sought to cover the main features of the four blank maps mentioned previously, for example, large blocks mixed with small blocks, wide boulevards mixed with small roads, and linear streets mixed with curved streets. In addition, if a real space area is too large, it will burden pedestrians. Therefore, we set the area at 0.54 km2 [Fig. 4(b), showing a smaller scale version of the map].Route Choice Experiment in Real SpaceWe used a real space for the route choice experiment, see Table 2 for experiment date, time period, weather, and other information. The participants in this experiment were the same as those from the blank map experiments.The content of the experiment is as follows: (1) The participants were asked to stroll in the experiment area. (2) The participants were asked to record places that were of interest to them. To identify the points that attracted the participants, they were asked to describe the specific content of their places of interest. (3) Immediately at the end of the chosen route, the participants were asked to write down their route choice strategies.The strolling time was limited to 0.5–2 h. Considering that walking is affected by physical strength, the participants were allowed to rest, and the rest time was limited to approximately 5 min; the participants were required to record their rest times and walking time. If they ended the walk due to fatigue, they were also required to record the time. The participants were not allowed to view the map on their mobile phones; therefore, to give them the scope of the experiment area, they were provided with a map of the experimental area printed on A4 paper [Fig. 3]. To ensure that they could identify their position at any time and not get lost, the participants were required to record their route on the map, as shown in the example in Fig. 5. At the same time, to confirm the correctness of the route record, the participants were required to carry a GPS device. In addition, if the participants were to start walking simultaneously, they might talk to each other, which could affect the route choice results (Suzuki and Okazaki 2001); therefore, each participant was asked to start walking 5 min after the previous participant had started walking. We set each starting/ending point to the same place. As shown in Fig. 5, there were six starting/ending points at the edge of the experiment area, and four participants were randomly arranged at each place. In addition, because it is time-consuming to travel to the subject area and the experiment itself is also time-consuming, increasing the number of experiments means that the participants were required to expend a lot of energy, which would reduce their motivation and fail to ensure that they would finish the experiment carefully. Therefore, this experiment was carried out once.Analysis IndexWe refer to the previous research results to select the following indicators, to analyze the characteristics of the experiment areas, and the results of the route choice experiments.Analysis Index of Street CharacteristicsTo identify the characteristics of the street structure, seven indicators were used. Three of the indicators (the total number of street units, the total length of the street units, and the grid axiality) represent the characteristics of whole streets. The other four indicators (length, width, Int.V (integration value), and curvature) represent the characteristics of the street units. In addition, a street unit is defined as a street from the center of one intersection to the center of the next intersection (Fig. 6).Grid axiality is used to express the deformation degree of the street grid when compared with the orthogonal grid. The result is a value between 0 and 1, in which values closer to 1 indicate a strong approximation to a perfect orthogonal grid, and values closer to 0 indicate a greater degree of axial deformation (Atsuyuki 2004).Int.V is an important indicator space syntax theory proposed by Hillier (Hillier and Hanson 1984) and represents the street structure. When its value is greater than 1, the space is integrated, it is easy to access from other spaces, and is close to the center in the region. When its value is less than 1, the space is in a deeper position, separated from other spaces and difficult to access.The curvature represents the degree of curvature of the street unit. The calculation method is based on the concept of the axial line (AL), which represents the longest line of sight and the straight-line moving distance in space syntax theory. As shown in Fig. 7, a large value of the curvature of a street means that the street is curved (Hu et al. 2018).Analysis Index of the Route ChoiceTo analyze the results of the participants’ route choice, seven indicators were used. The three indicators number of selected street units, length of the route, and tortuous extent of the route represent the characteristics of the participants’ overall route. The four indicators length, width, Int.V, and curvature represent the characteristics of the street units chosen by the participants. These seven indicators and the seven indicators mentioned previously have a one-to-one correspondence.Among them, the tortuous extent of the route is used to indicate the degree of change in the walking direction. The calculation method is based on the concept of AL in space syntax theory. When the walking route overlaps with the AL of the street, the number of ALs contained in the route is calculated. As shown in Fig. 8, a large value for the tortuous extent of the route means that people’s walking direction has changed substantially (Hu et al. 2018).Characteristics of Object SpacesStreet StructureThe calculation results of the characteristics of the street structure are presented in Table 3. Map 1 has the highest values for the total number of street units, the total length of the street units, and the grid axiality. The Int.V is also high, but the value of the curvature is low. It can be seen that there are many streets in this area, and many of them are straight; these streets constitute a grid-type street network. Map 2 has a high value of the width and a large standard deviation; we can also see from Fig. 3 that there are some wide and conspicuous streets in this area. Map 3 has the highest values for the length and curvature, and its streets are long and winding. Map 4 has the lowest value and standard deviation for its width and is made up of small streets.Table 3. Street characteristics of the object placesTable 3. Street characteristics of the object placesObject placesTotal number of street unitsTotal length of the street units (m)Grid axialityAverageStandard deviationAverageStandard deviationAverageStandard deviationAverageStandard deviationMap 1 Hiroshima40527,720.980.2968.4540.2312.4113.583.821.031.020.14Map 2 Fukuoka21916,507.920.2375.3846.0113.229.623.310.721.210.42Map 3 Lugang22919,348.730.1384.4962.978.132.642.850.821.280.53Map 4 Barcelona34524,991.370.1272.4446.025.351.133.090.861.190.46Real space30317,754.330.1358.2140.486.673.471.360.281.200.46Compared with the blank maps, the values of the total number of street units, the total length of the street units, and the grid axiality of the real space have no special values. It is a common irregular grid street plan. The values of length and Int.V are low, which indicates that the streets are short and have poor connectivity.Built EnvironmentNext, we outline the built environment of the real space. Photos of the street scenery around each starting/ending point are shown in Fig. 9. There are no landmarks on either side of the street and few shops, and most of the buildings are unique single-family housing in Japan. During the experiment, there were almost no vehicles or pedestrians along the routes. The landscapes of each street were similar, and the environmental characteristics were simple and uniform.Results and DiscussionComparison of the Route Choice Results between the Blank Maps and Real SpaceResults of the Overall Route ChoiceFirst, we analyze the characteristics of the overall route chosen by the participants. The results of the overall routes (the average number of selected street units, the average length of the route, and the average tortuous extent of the route) for each map were compared with the overall characteristics of the streets (the total number of street units, the total length of the street units, and the grid axiality). The results of the blank maps are given in Fig. 10.The average number of selected street units showed a linear tendency with a large total number of street units in the area and a large number of selected street units along the route. From the average length of the route, the participants imagined a distance of approximately 3,700–4,800 m, showing the following tendency: The longer the total length of street units, the longer the route. The average tortuous extent of the route showed a negative linear tendency.The results for the real space data are given in Fig. 11. The average number of selected street units and the average tortuous extent of the route showed a linear tendency similar to the results for the blank maps. However, as a result of the average length of the route, R2 decreases. The total length of the street units of the real space is approximately 17,000 m, which is close to Maps 2 and 3, but the average length of the route is only approximately 3,100 m.We speculate that this is because walking will be affected by physical strength. We collated the participants’ records of their rest times and walking time during walking (Table 4). One participant rested 0–3 times, 6 of the 24 participants had a rest along the way, and 5 of them ended their walk due to fatigue. It can be seen that physical strength has an impact on the route choice, especially on the length of the route.Table 4. Walking time and rest conditions (real space)Table 4. Walking time and rest conditions (real space)Number of people who took restNumber of people who ended their walk because of fatigueNumber of rest periods (1 person/times)Walking time (min)Total65——Proportion0.2500.208——Minimum——032Maximum——380Average——0.41749.542Results of the Places of InterestNext, the results of the participants’ places of interest are given in Fig. 12. The areas surrounded by black lines are places of interest; dense lines indicate that the participants’ choices are concentrated.From the results of the blank maps, wide roads and large blocks in Maps 1, 2, and 3 are marked frequently. In Map 4, the special shape of the upper-left corner and the circular areas are marked frequently. In other words, wide streets, large blocks, and irregularly shaped blocks easily attract the attention of the participants.However, in real space, compared with the blocks mostly composed of unit streets, the places of interest are small dots, and there are fewer streets along the place of interest. To explain this result, we collated the participants’ responses regarding their places of interest (Table 5).Table 5. Content of the places of interest (real space)Table 5. Content of the places of interest (real space)CategoryItemAnswer examplesFrequencyStreet structureStreet shapeA block of a different shape from its surroundings, street with special shape5BuildingsParkPark18BuildingHouse, post office12SchoolPrimary school, kindergarten7ShopFashionable shop2Street environment, sceneryStreet sceneryBeautiful street scenery11PlantTrees, flowers9Pedestrian special roadPedestrian special road8RampRamp1Street objectsObjectInteresting statues, object7Vending machineFashionable vending machine2There were some responses about blocks or street shapes but more about other built environment features such as buildings or plants. These environmental factors may be more interesting to participants in real space. Among them, parks are the most frequently identified (18 times). Parks or public spaces are very attractive factors, which is the same conclusion as that of the previous research (Sugiyama et al. 2010).Frequency of the Selected Street UnitsNext, we analyze the characteristics of the street units selected by the participants. Fig. 13 shows the number of selected street units. The selection frequencies of the streets are indicated by the thickness of the line; the wider the line, the more times the street is selected. The maximum frequency for a street is 72 times in the blank map experiments (24 participants × 3 experiments) and 24 times in the real space experiment (24 participants × 1 experiment).Based on the preceding results, we sought to identify the indicators of street units that have a greater influence on the selection frequency. Therefore, we adopt the selection frequencies of the street units as the dependent variable, and the independent variables are the length, width, Int.V, and curvature of the selected street units and the number of times a street unit is marked as a place of interest. We use the data for the blank maps to perform multiple linear regression (Table 6). As mentioned previously, because few streets are marked as places of interest in the real space, the places of interest index are removed, and the four indexes of the street unit are used as the independent variables; the data of the real space are used for the multiple linear regression (Table 6).Table 6. Influence of street unit indicators on the selection frequencyTable 6. Influence of street unit indicators on the selection frequencyIndicatorMap 1 HiroshimaMap 2 FukuokaMap 3 LugangMap 4 BarcelonaReal space Beta−0.011−0.053−0.106*−0.100−0.147* VIF1.5561.3211.7191.2382.013 Beta0.648**0.775**0.665**0.331**0.381** VIF1.3331.4081.6601.1371.546 Beta0.297**0.163**0.275**0.200**0.407** VIF1.1001.5541.9831.2751.489 Beta−0.0210.121**0.142**0.152**0.190** VIF1.2061.1501.6941.3601.951 Beta0.210**0.140**0.0420.364**— VIF1.1961.0651.0621.196—Number of samples400214224331264Multiple correlation coefficient0.8190.8940.8290.5610.652For the blank maps, in general, the fitting degree of the model is good; especially the value of the multiple correlation coefficient, which shows that Maps 1, 2, and 3 have higher values (above 0.800). We speculate that the street structure features of the three maps are more striking, and the differences between the very wide streets and the surrounding areas can attract the attention of the participants, thereby affecting their choice frequency. From the standard partial expression coefficient, width has a great influence of Maps 1, 2, and 3 (reaching a high value above 0.600), which proves that wide streets are easily selected in street structures without special shapes. Compared with other maps, Map 4 has a lower influence of width (only 0.331) but has a higher index value for places of interest (0.364, higher than other maps). According to Figs. 12 and 13, the frequently selected streets and frequently marked areas are concentrated in the same places on Map 4, and it can be confirmed that if there are streets with special shapes, these places easily attract people’s interest and will be chosen; however, these places seem to have limited influence compared with the wide streets.For real space, although one indicator is reduced, the value of the multiple correlation coefficient is also relatively high, indicating the influence of the street structure in real space. From the standard partial expression coefficient, all the indicators have explanatory power, especially the higher values of the width (0.381) and Int.V (0.407), confirming that they have a great influence on the selection frequency. The results have the same tendency as those of previous studies that broad (Guo 2009) and highly integrated streets (Koohsari et al. 2016; Baran et al. 2008) are more likely to trigger visiting. In addition, although we exclude the index of places of interest from the independent variables, we can see from the answer results in Table 5 that the blocks or street shapes that are different from the surrounding streets can arouse the participants’ interest. That is, in real space, if people can obtain information about the street patterns (through maps or mobile phone navigation), these special shapes may affect the selection frequency.Comparison of the Route Choice Strategies of Participants for Blank Maps and Real SpaceBy comparing the results of the two experiments, we analyze the objective characteristics of the street structure, confirming that there are some similar tendencies in the route choice on blank maps and in real space, which help in obtaining a general understanding of the relationship between the street structure and the participants’ choice of route.However, what are people’s subjective concerns over, or preferences for, the street structure? How do they think about their route choices? In particular, the blank maps have only street structure information, and it is necessary to confirm whether the participants recognize the streets on the blank maps as a figure and choose only conspicuous shapes or recognize them as streets and imagine themselves walking there. Therefore, we analyzed the participants’ route choice strategies.Strategies Regarding Blank MapsTable 7 shows some examples of strategies regarding the blank maps. There are strategies for the street structure (width and shape); walking experiences or whole-route planning (fatigue, the pleasure of walking, and change in location). Therefore, it can be concluded that the participants recognize streets on the blank maps as real streets and plan their strolling route and that the results of the blank map experiments are effective. In addition, it can be seen that the participants use different strategies based on the characteristics of the street structure.Table 7. Example of participants’ route choice strategiesTable 7. Example of participants’ route choice strategiesCategoryExamplesAll mapsConsiderations about the street structure:
•I chose main roads or roads that are easily accessible.•I am often attracted to roads with unique shapes (e.g., geometric shape).•I prefer to walk in narrow alleys.Strolling experience, overall route planning:
•I choose to walk for no more than 30 min because I would become tired if I took a long walk.•Always going straight makes me feel bored. Walking on winding roads is more interesting.•My strategy is to choose diverse streets or streets with different views along the way.Map 1 Hiroshima•I am interested in streets with high densities.•I chose the wide roads in the middle of the map.•Most areas are rectangular, in which I am less interested. Instead, I prefer walking in areas that are not as rectangular.Map 2 Fukuoka•The S-shaped road caught my attention.•I guess those curvilinear roads on the map probably follow the original landscape, and the scenery there must be beautiful, so I planned my route there.•My walk was all over the map, and I planned the route to learn about the street conditions in the whole area.Map 3 Lugang•The roads are narrow and monotonous, so I have no feeling about that.•I am not interested because there are only narrow roads.•The roads are monotonous without twists and turns.Map 4 Barcelona•It appears that the circular road would be incredible, and I want to see it.•I chose the circular road and the two parallel roads in the center of the map.•I think the two roads in the middle of the map are the center of the block, so I chose them.Next, to understand the participants’ route choice strategies in detail, we extracted nouns and adjectives (items) related to strategies from the participants’ responses and calculated the frequency of each item. The results are given in Fig. 14.From the results regarding the characteristics of the whole route, 1. Near was answered 5 times, and 2. Medium distance was answered 1 time. It is speculated that the participants will pay attention to walking distance. 3. Rough was answered 4 times, 4. Many places was answered 22 times, and 6. Never passed was answered 10 times. It is speculated that the participants will pay attention to the places they had already visited and try not to go there again, hoping to take a walk in various other streets. In addition, 5. Direction of the route and 8. Position were also influencing factors when the participants choose their route. The street unit characteristics were the most frequently mentioned, accounting for approximately 50% of the total number of answers. Among them, see 9. Main road and 10. Path; the two answers about street width were answered 34 times and 23 times, respectively, which indicated that the participants found it easy to pay attention to the width of the street when planning the route on the map. 12. Complex was answered 8 times; however, 13. Simple was answered only 1 time. We guess that complex areas were more interesting for the participants. 14. Curved was answered 10 times, and 15. Straight was answered 6 times. It seems that curved roads also easily attract people’s attention. Based on the results of the characteristics, there were 24 answers accounting for approximately 21% of the total, especially 16. Shape. We speculated that the participants focused on the specially shaped blocks, which they planned to visit.Then, we sorted out the participants’ responses to each map, extracted the nouns and adjectives (items) from each map, and calculated the frequency of each item. The results are given in Fig. 15.Compared with the overall strategies for all maps (Fig. 14), those for each map are less (Fig. 15); however, it can be seen that the participants’ strategies change on different maps. In Maps 1 and 2, 9. Main road and 16. Shape were answered more frequently. In Map 4, 16. Shape was answered 25 times, which is a high value because there are special geometric blocks such as circles on the map. In Map 3, the number of answers to each item is less, because this map lacks very wide streets or specially shaped areas compared with the other maps; therefore, the numbers of answers are scattered.Strategies Regarding Real SpaceThrough the analysis in the previous sections, we found that the participants chose wide streets, streets with high connectivity, and irregularly shaped streets in both experiments. However, there are rich built environments in real space. Did the participants actually make their choices based on the street structure or were their choices based on other features of the built environment? To clarify this question, we analyzed route choice strategies in real space in detail. We extracted the nouns and adjectives (items) related to strategies from the participants’ responses and calculated the frequency of each item. The results are given in Fig. 16.Of all the items, the characteristics of the street unit have the largest number of answers, accounting for approximately 50% of the total. It can be confirmed that even in real space, the characteristics of the street structure are also important attributes to which people pay attention when choosing their walking route. For the answers regarding the characteristics of the street area, the participants chose streets with special shapes, which was consistent with the results from the blank maps. That is, if people can obtain information on a street structure in real space (by maps or by navigation through mobile phones), places with special shapes may affect people’s route choices.However, we extracted some strategies that the blank maps could not reflect and that had nothing to do with the street structure, such as buildings and facilities, accounting for approximately 23% of the answers. This result shows that a good street view and special buildings will attract pedestrians’ attention. In addition, there are some elements in the answers that are not characteristics of the street itself but things that occasionally appear there such as the light and shadow of the sun and cars. These elements, which change over time, are also a factor that influences people’s route choice.It can be concluded that the participants in this study were able to identify street structures using a mapping tool and had their own preferences for street shapes (Table 7). In the case of blank maps, not only the width and connectivity of a street unit but also blocks composed of complex streets will affect the participants’ choices, which can be identified from the calculation results of multiple linear regression (Table 6). Width of a street unit, Int.V of a street unit, and A street unit is marked as a place of interest had significant influence. In the case of real space, despite a rich built environment, the characteristics of the street structure still have an impact on the route choice (Fig. 16). This result shows that the street structure is also an important attribute in actual space and warrants attention when considering the psychology and behavior of pedestrians.ConclusionsThe goal of this study is, after identifying the street structure based on a map, to understand the relationship between the street structure and the route choice as well as people’s street structure concerns and preferences and to conduct a brief preliminary verification in actual space. This study differs from previous studies in that (1) to clearly capture the influence of the street structure and avoid the interference of other built environments, a different method (map recording method using blank maps) was used, and the results were further compared with the results in real space. (2) We chose strolling as our research object. (3) We considered the route choice strategy as a component of walking behavior and conducted a detailed analysis. The main conclusions are as follows.Through the blank maps experiment, it is confirmed that the route choice tendency will change with the characteristics of the street structure. If an area has irregularly shaped streets, these streets are more likely to be selected (e.g., Map 4), and if there are no irregularly shaped streets in this area, the influence of the width or the connectivity of the streets is greater than the other factors (e.g., Maps 1, 2, and 3).Based on the route choice strategies on the blank maps, the participants could imagine walking on the street and could plan their route smoothly. In addition, they could have their own concerns and preferences for the shape and structure of streets; thus, it is worth trying to identify the street structure through maps.Regarding the route choice strategies in real space, although the responses to the built environment can be extracted (22.7%), responses about the street structure, such as the street unit and area characteristics, account for approximately 60% of the total, confirming that although there are plenty of built environment features in real space, the characteristics of the street structure are important attributes and can affect the participants’ choice when strolling.There are some errors in the results of the blank maps experiment and the real space experiment. In real space, the places of interest are small dots rather than large areas or streets. We speculate that because the built environment in real space will arouse the interest of the participants, and the blank map cannot show the built environment, the participants will be able to choose places of interest only according to the block and street shape. In addition, the route is shorter in real space because walking is also greatly affected by physical strength, which limits the map recording method.However, the results of the blank maps and real space also show some similar tendencies: The selected unit streets are concentrated in wide streets with high Int.V, and the participants are easily attracted by streets with special shapes.We hope that the results of this research will provide a new perspective on constructing urban streets that are favored by more pedestrians and help obtain a large amount of data and user opinions in preliminary investigations or designs. Improving curves and widening roads are common ways to improve street structure; from the perspective of convenience, a regular grid street structure has more advantages. However, we should consider not only the efficiency of traffic but also the design of urban streets from the perspective of pedestrians’ comfort and pleasure. For example, wide streets tend to gather walkers; widening certain major streets can not only increase walking comfort but also improve convenience and efficiency. However, if all streets are boulevards, the efficiency of pedestrian flow will be low; therefore, branches can be established that will strengthen the connection with other streets to guide pedestrians and optimize the efficiency of each street. The research results show that curved streets are more likely to attract people’s interest; therefore, long and straight streets can be avoided without affecting traffic efficiency. Today, mobile terminals such as mobile phones are very popular, and it is increasingly easy for people to view maps and obtain street structure information. Therefore, if we add irregularly shaped blocks in an area and include some important functional facilities in these places, it will attract people’s attention and bring vitality to the surrounding areas.This study has some limitations as follows: (1) In the similarity evaluation experiments of the street structure, it is necessary to increase the number of participants and increase participants with different attributes. The participants in this study are all college students. If the participants had been of different ages, educational levels, or occupations, we might have obtained different results. (2) In addition to the aforementioned limitations, there are other limitations in the blank map experiment. For example, the blank map could not show the effect of differences in street height, the flow of people, traffic flow, and weather, and the participants’ cognitive distance may have been biased when reading the map. (3) In the route choice experiment in real space, we took traditional residential areas in Japan as the object and did not discuss whether other types of streets would affect the participants’ judgment, nor did we consider the effects of day and night. We need to set more reliable experimental conditions to optimize the experiment. In addition, among the indicators of space syntax, only the most basic Int.V index was used in this study, and as space syntax theory develops, more indicators describing pedestrian routes should be further discussed in future studies.Data Availability StatementSome or all data, models, or code that support the findings of this study are available from the corresponding author upon reasonable request.References Atsuyuki, O. 2004. Islamic area studies with geographical information systems. London: Routledge. Baran, P. K., D. A. Rodríguez, and A. J. Khattak. 2008. “Space syntax and walking in a new urbanist and suburban neighbourhoods.” J. Urban Des. 13 (1): 5–28. https://doi.org/10.1080/13574800701803498. Blanco, H., M. Alberti, A. Forsyth, K. J. Krizek, D. A. Rodríguez, E. Talen, and C. Ellis. 2009. “Hot, congested, crowded and diverse: Emerging research agendas in planning.” Prog. Plann. 71 (4): 153–205. https://doi.org/10.1016/j.progress.2009.03.001. Cao, X. Y., S. L. Handy, and P. L. Mokhtarian. 2006. “The influences of the built environment and residential self-selection on pedestrian behavior: Evidence from Austin, TX.” Transportation 33 (1): 1–20. https://doi.org/10.1007/s11116-005-7027-2. Czogalla, O. 2012. “A model for pedestrian routing in walkable networks.” In Proc., 13th IFAC Symp. on Control in Transportation Systems. Laxenburg, Austria: The International Federation of Automatic Control. https://doi.org/10.3182/20120912-3-BG-2031.00062. Guo, Z. 2009. “Does the pedestrian environment affect the utility of walking? A case of path choice in downtown Boston.” Transp. Res. Part D Transp. Environ. 14: 343–352. https://doi.org/10.1016/j.trd.2009.03.007. Hahm, Y., H. Yoon, D. Jung, and H. Kwon. 2017. “Do built environments affect pedestrians’ choices of walking routes in retail districts? A study with GPS experiments in Hongdae retail district in Seoul, South Korea.” Habitat Int. 70: 50–60. https://doi.org/10.1016/j.habitatint.2017.10.002. Hasegawa, M., A. Kudo, S. Mori, and T. Oku. 2003. “Characteristics of walking in everyday life in urban space—Characteristics of action on difference of content of task.” J. City Plann. Inst. Jpn. 38 (3): 427–432. Hillier, B., and J. Hanson. 1984. The social logic of space. Cambridge, UK: Cambridge University Press. Hoehner, C. M., L. K. Brennan Ramirez, M. B. Elliott, S. L. Handy, and R. C. Brownson. 2005. “Perceived and objective environmental measures and physical activity among urban adults.” Am. J. Preventative Med. 28 (2): 105–116. https://doi.org/10.1016/j.amepre.2004.10.023. Hu, Y., D. Nishina, T. Tanaka, and W. Yamasaki. 2018. “A study on the relationship between strolling course selection and personal feature by a method of drawing on a map.” J. Environ. Eng. 83 (750): 547–656. Kelly, C. E., M. R. Tight, F. C. Hodgson, and M. W. Page. 2011. “A comparison of three methods for assessing the walkability of the pedestrian environment.” J. Transp. Geogr. 19: 1500–1508. https://doi.org/10.1016/j.jtrangeo.2010.08.001. Koohsari, M. J., J. A. Karakiewicz, and A. T. Kaczynski. 2013. “Public open space and walking: The role of proximity, perceptual qualities of the surrounding built environment, and street configuration.” Environ. Behav. 45 (6): 706–736. https://doi.org/10.1177/0013916512440876. Koohsari, M. J., T. Sugiyama, S. Mavoa, K. Villanueva, H. Badland, B. Giles-Corti, and N. Owen. 2016. “Street network measures and adults’ walking for transport: Application of space syntax.” Health Place 38: 89–95. https://doi.org/10.1016/j.healthplace.2015.12.009. Leyden, K. M. 2003. “Social capital and the built environment: The importance of walkable neighborhoods.” Am. J. Public Health 93 (9): 1546–1551. https://doi.org/10.2105/AJPH.93.9.1546. Marchand, B. 1974. “Pedestrian traffic planning and the perception of the urban environment: A French example.” Environ. Plann. A Econ. Space 6: 491–507. https://doi.org/10.1068/a060491. Nishio, K., H. Zaino, N. Matubara, and Y. Kurazumi. 2000. “A study on continuous recognition of street space from the view point of cognitive map.” J. Archit. Plann. Environ. Eng. 529: 217–223. https://doi.org/10.3130/aija.65.217_2. Nishio, K., H. Zaino, N. Matubara, and Y. Kurazumi. 2001. “Individual difference based on the ability to understand the street patterns—A study on the continuous recognition of street space from the view point of individual difference Part 1.” J. Archit. Plann. Environ. Eng. 540: 205–212. https://doi.org/10.3130/aija.66.205_1. Osaragi, T., and A. Tanaka. 2017. “Analysis of factors affecting route choice behavior and modeling of pedestrians’ movement.” J. Archit. Plann. 82 (734): 895–903. https://doi.org/10.3130/aija.82.895. Prelipcean, A. C., Y. O. Susilo, and G. Gidófalvi. 2018. “Collecting travel diaries: Current state of the art, best practices, and future research directions.” Transp. Res. Procedia 32: 155–166. https://doi.org/10.1016/j.trpro.2018.10.029. Sallis, J., F. Frank, L. D, B. Saelens, E. Kraft, and M. K. 2004. “Active transportation and physical activity: Opportunities for collaboration on transportation and public health research.” Transp. Res. Part A Policy Pract. 38: 249–268. https://doi.org/10.1016/j.tra.2003.11.003. Shatu, F., and T. Yigitcanlar. 2018. “Development and validity of a virtual street walkability audit tool for pedestrian route choice analysis—SWATCH.” J. Transp. Geogr. 70: 148–160. https://doi.org/10.1016/j.jtrangeo.2018.06.004. Shatu, F., T. Yigitcanlar, and J. Bunker. 2019. “Shortest path distance vs. least directional change: Empirical testing of space syntax and geographic theories concerning pedestrian route choice behaviour.” J. Transp. Geogr. 74: 37–52. https://doi.org/10.1016/j.jtrangeo.2018.11.005. Sueshige, Y., and M. Morozumi. 2005. “Assessment of the linked QTVR simulator for observing strolling activities of citizens—On the relationship of strolling activities and visual information given by the environment in downtown Kumamoto.” Plann. Environ. Eng. 597: 119–125. Sueshige, Y., and M. Morozumi. 2007. “The visual information which encourage or restrain citizens’ strolling activities in urban space: On the relationship of strolling activities and visual information given by the environment in downtown Kumamoto Part II.” J. Archit. Plann. 72: 191–197. https://doi.org/10.3130/aija.72.191_1. Sugiyama, T., J. Francis, N. J. Middleton, N. Owen, and B. Giles-Corti. 2010. “Associations between recreational walking and attractiveness, size, and proximity of neighborhood open spaces.” Am. J. Public Health 100 (9): 1752–1757. https://doi.org/10.2105/AJPH.2009.182006. Sugiyama, T., M. Neuhaus, R. Cole, B. Giles-Corti, and N. Owen. 2012. “Destination and route attributes associated with adults’ walking: A review.” Med. Sci. Sports Exercise 44: 1275–1286. https://doi.org/10.1249/MSS.0b013e318247d286. Togawa, S., Y. Kuboyama, and H. Endo. 2006. “Study of pedestrian wayfinding behavior characteristics in streets surrounding rivers, using virtual reality space.” Plann. Env. Eng. 601: 231–237. Wakabayashi, Y. 2008. “The role of maps in the cognition of geographic space.” Cognit. Stud. 15 (1): 38–50. Werberich, B. R., C. O. Pretto, and H. B. B. Cybis. 2015. “Calibration of a pedestrian route choice model with a basis in friction forces.” Transp. Res. Rec. 2519 (1): 137–145. https://doi.org/10.3141/2519-15.