Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 Contents lists available at ScienceDirect Transportation Research Interdisciplinary Perspectives journal homepage: www.sciencedirect.com/journal/transportation- research-interdisciplinary-perspectives Built environment bikeability as a predictor of cycling frequency: Lessons from Barcelona Oriol Codina a, Monika Maciejewska b, Jordi Nadal a, Oriol Marquet a, c, * a Geography Department, Autonomous University of Barcelona – Edifici B, Campus de Bellaterra, Cerdanyola del Vallès, 08193 Barcelona, Spain b Cosmopolis Centre for Urban Research, Vrije Universiteit Brussels, Brussels, Belgium c Institute of Environmental Science and Technology (ICTA), Autonomous University of Barcelona, Barcelona, Spain A R T I C L E I N F O A B S T R A C T Keywords: Background: Many cities are putting cycling at the centre of their sustainable transportation policies after the Bikeability COVID pandemic. Cycling is seen as a desirable mode of transport in dense and compact areas and needs to be Built environment promoted accordingly. However, to date, only a handful of different bikeability indexes exist attempting to map Modal choice biking conditions and the built environment’s potential to promote biking as a modal choice on a city scale. Cycling frequency Methods: In this article, we use objective GIS data to map bikeability potential in the city of Barcelona. To do so we extracted the main bikeability components from an adhoc cycling survey and then create an index using ten spatial indicators. This bikeability index is mapped at a 100 × 100 m scale in the city of Barcelona. We then use actual travel behavior data extracted from a local representative travel survey to test the reliability of the index in predicting daily bike use. Results: Results confirm the validity of the bikeability index as a predictor of the frequency of cycling. People living in areas with higher levels of built environment features associated with bikeability such as dedicated infrastructure, low accident rates and small slopes are more likely to use the bike more often. Conclusions: Results validate our approach providing new methods to be used in further biking studies and a useful tool for policy and decision making. The use of our new bikeaiblity index is especially indicated for highly- dense, compact, Mediterranean-style cities. 1. Introduction that encourage traveling by bike. Building city-level bikeability indexes can help support decision-makers, monitor progress over time and In recent years, there has been a shift in the mobility patterns of facilitate communication with the general public (Kamel et al., 2020). many cities around the world, as cycling is gaining popularity as an This makes them a useful tool in the process of deciding interventions to everyday mode of transport (Pucher and Buehler, 2017). Among the improve cycling conditions as well as being a great way of displaying the many factors partially explaining the preference for using a bike, envi­ distribution of areas in need of improvements to policymakers and the ronmental awareness, functional factors such as convenience or time general population. management, the need to exercise, or affordability are frequently highly Since bikeability is a complex phenomenon, bikeability indexes can cited (Bhandal and Noonan, 2022; Charreire et al., 2021). This surge in encounter issues when trying to incorporate the necessary number of the popularity of bikes has brought about a growing interest in the sci­ built environment attributes. Previous literature has identified a sig­ entific community to create methods to assess the bike-friendliness of a nificant number of built environment and contextual attributes that can particular built environment (Krenn et al., 2015a; Wysling and Purves, be linked to cycling behaviour and they include bikeway density, 2022). Among these methods, bikeability indexes are particularly popu­ bikeway width, bikeway exclusiveness, slopes or nearby green and blue lar, as they score the appropriateness of the built environment to support areas, among others (Krenn et al., 2015b; Lin and Wei, 2018; Naess, bicycling, considering both the factors that influence choosing the bi­ 2012; Nielsen and Skov-Petersen, 2018; Pucher and Buehler, 2008; cycle as the main mode of transport as well as the environmental factors Winters et al., 2013). Additionally, biking has been found to be a means * Corresponding author at: Institute of Environmental Science and Technology (ICTA), Autonomous University of Barcelona – Edifici ICTA-ICP, Carrer de les Columnes s/n, 08193 Cerdanyola del Vallès, Barcelona, Spain. E-mail address: oriol.marquet@uab.cat (O. Marquet). https://doi.org/10.1016/j.trip.2022.100725 Received 2 September 2022; Received in revised form 7 November 2022; Accepted 11 November 2022 Available online 25 November 2022 2590-1982/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 of transport that is susceptible to dynamic environmental factors, such et al., 2016). The availability of cycling infrastructure has been as weather conditions or time of day (Hyland et al., 2018; Parkin et al., repeatedly found to be a key element in making an area more bikeable 2008; Wang et al., 2018; Winters et al., 2007). This makes measuring (Dill, 2009; Ma and Dill, 2017; Muhs and Clifton, 2016; Winters et al., bikeability extremely challenging and geographically specific, as 2010a; Winters and Teschke, 2010). Pucher et al. (2010) found that extrapolating the methods used in one city to another may not match cyclists prefer having dedicated bike lanes to riding in mixed traffic, to reality, and factors affect each place differently. which McNeil (2011) added that is especially true if they are segregated In this article we aim to assess the spatial distribution of bikeability from other road users. Traffic calming measures around the trip origin, conditions in Barcelona by creating a bikeability index adapted to the such as road markings or signage, are also associated with higher rates of urban morphology of Mediterranean and south European cities. Our cycling (Pucher and Buehler, 2008; Winters et al., 2010a). However, a index is built based on two data sources: (1) previous literature findings single bike lane will not increase bikeability in an area per se, as it must on the main determinants of biking, and (2) self-reported answers to an be part of a wider, more complete and more varied network of bicycling adhoc local survey on biking preferences and infrastructure needs. The infrastructure (Dill, 2009; Muhs and Clifton, 2016; Pucher et al., 2010). validity of the index is later tested by modelling it as a predictor of bike In this regard, Cervero et al. (2009) and Winters et al. (2010b) found use, employing the geolocated answers to an official travel survey. By that a high intersection density was associated with a greater likelihood modelling whether bikeability levels around the habitual residence are of cycling in their studies of cycling behaviour in Bogotá and Vancouver. associated with biking habits, we also provide the growing research in This effect is especially relevant in grid-pattern street networks (Dill, the field with a framework that makes it easy for policy makers and 2009; Marshall and Garrick, 2010). The association between cycling and experts to replicate other bikeability indexes in other local geographical intersection density, however, is still debated as Van Dyck et al. (2013) contexts. To do so, section 2 of this article reviews the literature to date, and Mitra and Nash (2019) found this relationship to be negligible section 3 shows the main variables and methods to create our bikeability across different cities in North America and Europe. Muhs and Clifton index, section 4 presents the resultant mapping of the city and tests its (2016) for their part, suggested that the reason behind that may be that power to predict likelihood of cycling, and sections 5 and 6 are left for having too many intersections also means more stops, effectively slow­ the discussion and conclusion of the research, respectively. ing travel speeds. The role of parking spaces has also been considered in studies of 2. Literature review bikeability (Lin and Wei, 2018; Van Dyck et al., 2013). To date, how­ ever, it is not clear whether an increase in parking facilities can Over the last few years, biking has seen an upsurge in cities around encourage more bicycling, or if it is the lack of proper parking spaces the world as a healthy and environmentally-friendly alternative mode of what effectively dissuades from bike use (Pucher et al., 2010). In their transportation (Kellstedt et al., 2021; Krenn et al., 2015b; Pucher et al., study in Beijing, Shu et al. (2019) found that cycling was incentivized 2010). Advances in technology in the past two decades have brought when the building entrance or exit was reasonably close to a bicycle about innovative methodologies in many fields to assess communities parking rack. and their support for bicycle (Kellstedt et al., 2021), a phenomenon Existing literature has also paid attention to the fact that cycling known as “bikeability”. However, to this date, there is still no consensus infrastructure does not only provide comfort to bike users but also over a universal definition of this term (Kellstedt et al., 2021; Muhs and safety, as they allow bike users not to have to share space with the Clifton, 2016). On the one hand, bikeability is defined as the “assess­ automobile (Muhs and Clifton, 2016). While some studies have found ment of an entire bikeway network for perceived comfort and conve­ that the relationship between perception of safety and cycling is unclear nience and access to important destinations” (Lowry et al., 2012), and it (Parkin et al., 2008; Van Dyck et al., 2013), others have found that is mainly employed as a tool for comparison between cities. On the perception of danger can hinder bicycle commuting significantly (Handy other, bikeability can also be defined as the “extent to which the actual and Xing, 2011) or at least influence route choice (Winters et al., and perceived environment is conducive and safe for bicycling” (Kell­ 2010b). These findings suggest that better perceptions of safety may stedt et al., 2021), a definition that focuses more on the suitability that a foster increased biking. At the same time, having more bikes in the particular built environment offers for cycling. streets has been found to collectively increase security levels, improving The literature surrounding bikeability is still rather recent, and only the perception of safety and lowering the actual rate of bike accidents a handful of different bikeability indexes still exist. These have been put (Pucher et al., 2010). This trend, commonly labelled as “safety in to use in cities such as Vancouver (Kamel et al., 2020; Winters et al., numbers” (Elvik and Bjørnskau, 2017) is explained by the fact that 2013), Graz (Krenn et al., 2015b), Taipei (Lin and Wei, 2018) and increased number of bikes on the streets contribute to making them Barranquilla (Arellana et al., 2020), among some others to assess the more visible to other modes of transport, an the same time that other spatial distribution of bikeability on a city level. The limited number of road users learn how to safely interact with bikes. Perceptions of safety bikeability index limits its reproducibility, as indexes respond to the is a key component of bike modal choice and because of that a number of morphological features of the urban areas where they have been tested/ authors support safety to be included in bikeability indexes (Kamel et al., created (Winters et al., 2016). Given the sensitivity of biking behavior to 2020). local conditions and local biking culture however, a larger variation of Atmospheric wheather and climate are factors also commonly indexes is clearly needed better representing cycling determinants in included in bikeability indexes, with rainfall and lower temperatures other urban areas with different built environment characteristics. being associated with less willingness to cycle (Parkin et al., 2008; In general, bikeability has been found to align with walkability, but Winters et al., 2007). In addition, population density and land mix-use that is not always the case, as cycling has specific requirements on the have been found to have a positive impact on cycling activity (Dill, personal level – equipment, experience – and on the infrastructure level 2009; Dill and Voros, 2007; Handy and Xing, 2011; Winters et al., – parking places, access to bike lanes –. In the case of biking, infra­ 2010a), as they generate shorter trip distances, which are more readily structure usually plays a more important role than for walking (Kellstedt covered by bicycle (Pucher et al., 2010). et al., 2021). Additionally, previous studies have found that biking de­ However, conditions of the area – summarized in bikeability scores – terminants can also be motive-dependent and thus vary depending on cannot explain bike usage by themselves, as socio-demographic factors, the trip purpose (Beenackers et al., 2012). attitudes and social environments have been found to determine the Previous research has identified several additional factors that in­ actual amount of bike use as well. In that context, women have been fluence the decision to bike. Hilliness is considered a pivotal element in found to be less prone to cycling than men (Handy and Xing, 2011; Ma biking environments as hilly areas effectively dissuade from using active and Dill, 2017; Miralles-Guasch et al., 2022; Mitra and Nash, 2019; transport in general (Ma and Dill, 2017; Parkin et al., 2008; Winters Parkin et al., 2008; Van Dyck et al., 2013; Winters et al., 2007). Older 2 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 people are also more reluctant to cycling than their younger counter­ generate shorter trip distances, which are more readily covered by bi­ parts (Ma and Dill, 2017; Pucher and Buehler, 2008; Winters et al., cycle (Pucher et al., 2010). Furthermore, the city has a Mediterranean 2007). Regarding social status, some studies consider that low-income climate, which is warm and dry most days, conditions that favour people have a more negative perception of bikeability and lower cycling (Parkin et al., 2008; Winters et al., 2007). Additionally, the grid cycling rates than wealthier individuals (Ma and Dill, 2017; Parkin et al., pattern, which is exemplified in Barcelona with the Eixample district, 2008), while others have found that cycling rates are negatively asso­ provides consistently high levels of connectivity (Dill, 2009) (Fig. 1). ciated with income (Van Dyck et al., 2013; Winters et al., 2007). Simi­ larly, it is also important to take into account the attitudes of commuters 3.2. Variables and data sources themselves, particularly their comfort level with bicycling and how much they enjoy biking as an activity in itself (Handy and Xing, 2011). In order to select the main variables that play a role in cycling in Additionally, we must bear in mind that trip distance is a fundamental Barcelona from a user perspective and later assess their relative consideration in mode choice (Winters et al., 2010a), especially importance in the overall bikeability of the city, we used responses to a commuting cyclists, who are more sensitive to distance and less sensitive local bike-user survey)Enquesta Baròmetre de la Bicicleta) (Generalitat to most other variables compared to other cyclists (Broach et al., 2012), de Catalunya, 2019). The survey is carried out annually since 2006 at which indicates that not all variables in a bikeability index have the the request of the Generalitat de Catalunya, and comprises a sample of same importance for every individual. 800 people, interviewed via telephone, as well as an oversample of 200 cyclists, surveyed online. Within the survey, bike users (n = 290) were 3. Methods asked to state the main difficulties and perceived barriers when using a bicycle as a daily mode of transport (Generalitat de Catalunya, 2019). 3.1. Study area: Barcelona We used the answers to the survey to identify the basic variables that needed to be included in the index (Table 1). We then crossed the ob­ Barcelona lies on the western Mediterranean coast in north-eastern tained variables with the most commonly utilized variables in previous Spain. It gathers a population of around 1.6 million inhabitants in an bikeability indexes available in the literature. Both the answers to the area of circa 100 km2, making it one of the densest cities in Europe, with survey and the findings from the literature review were used to assign an intense land-use mix in relatively short distances (Delclòs-Alió and weights to each variable within the final bikeability index. That method Miralles-Guasch, 2018; Marquet and Miralles-Guasch, 2015). These are allowed us to attain to the perceived needs of local bike users while also key elements in making a city attractive for biking (Dill, 2009; Dill and incorporating previous findings from the most common international Voros, 2007; Handy and Xing, 2011; Winters et al., 2010a), as they bikeability indexes. Fig. 1. Study area map: Barcelona. 3 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 Table 1 bikes with motor vehicles. It is computed as the ratio of bike collisions Results of the answer to the question “In a city, which is the main difficulty when per bicycle trips in any given cell. Using the gauging data from different moving around cycling?” in the Enquesta Baròmetre de la Bicicleta (2019), vari­ stations, we estimated the gauging for all the city using the interpolation able and associated weight assigned. tool in ArcMap. The value of each cell is the standardized z value of the Original variable % of Variable Weight coefficient of the total bike collisions in a 5-year period (as reported by respondents (abbreviation) Guàrdia Urbana – Barcelona’s police department) occurred within a 150 Interaction with motor transit 22,1% Traffic (Ti) 40 % m around the centroid of each cell divided by the estimated gauge of Insufficient adequate spaces 15,9% Infrastructure (Ii) 15 % bicycle traffic in the same area. This buffer size is commonly used in the Connectivity (Ci) 15 % literature when assessing biking collisions (Urban Systems, 2015). As Theft risk 15,6% fewer accidents mean better bikeability, the resulting z values have been – – Orography conditions 11,2% Topography (Si) 20 % Lack of habit 9,4% – turned into negative values. Atmospheric conditions 8,7% – The quality of the Infrastructure (Ii) has been estimated by calcu­ Insufficient parking facilities 6,6% Parking spaces (Pi) 10 % lating the distance from the centroid of each cell to the nearest biking Lack of accessibility in public 4,2% – infrastructure using the Near tool in ArcMap. Distance to segregated transit Others 0% – cycle paths was assigned three times more weight than distance to Not answered 0,4% – painted cycle lanes (also known as “sharrows”), as suggested by the literature (Walk Score, 2021). Additionally, all distances further than Source: Enquesta baròmetre de la bicicleta 2019 (Generalitat de Catalunya, 2019). 400 m have been re-coded to 400.1 m to avoid a cell close to one type of infrastructure but far from another getting a misleading score. As in this case lower values represent better bikeability levels, z values have been Variables identified as a significant barrier to bike use such as theft turned into negative values. risk, atmospheric conditions, lack of habit and difficulties regarding Connectivity (Ci) is referred as the density of intersections. We have their carriage on public transit could not be included due to lack of counted the number of cycling infrastructure intersections inside a 400 objective and geolocated associated data. The variable “insufficient m buffer from the centroid of each cell following the method by Winters adequate spaces” was divided into two equally-weighted variables, as et al. (2013). The weight has also been modified depending on the na­ some studies consider availability of cyclable infrastructure and its ture of the infrastructure (Urban Systems, 2015), with intersections of connectivity to have a different impact on bikeability (Lin and Wei, segregated cycle paths being assigned the largest weight, followed by 2018; Winters et al., 2013). intersections between segregated cycle paths and sharrows and finally Table 2 summarizes the included final bikeability index and their intersection between sharrows being assigned the smallest weight. Re­ correspondent indicators. All the data has been collected from the Bar­ sults have been transformed into z values. celona Open Data website (Ajuntament de Barcelona, 2021d), except for The estimate count of Parking spaces (Pi) within each cell has been the topography variable, which has been calculated using the digital calculated by counting the number of bike racks and Bicing stations elevation model available from the Institut Cartogràfic i Geològic de (Barcelona’s public bicycle-sharing system) within a 240 m buffer from Catalunya (Institut Cartogràfic i Geològic de Catalunya, 2021). All the centroid of each cell. These 240 m are estimated to be the maximum variables were calculated using a grid map with cells (i) of 100 m × 100 distance a person would walk to access a bike (Shu et al., 2019). Results m that covers all the municipality of Barcelona. are expressed also as z values. The Traffic (Ti) indicator aims to represent the level of interaction of Topography (Si) has been estimated based on a digital elevation model. Using the Slope tool in ArcGis, we have created a slope map of Barcelona and, later, with the Zonal Statistics tool, we have measured Table 2 the mean percent rise in a 120 m-radius buffer from each cell centroid, a Diagram of the variables and indicators used. distance used by the Manual for the Design of Cyclepaths in Catalonia Variable Definition (measured in a zone to be assessed) (Medina and Hernández, 2008) and also roughly the size of a block in Indicator Eixample district. A rise higher than 25 % has been recoded as being Traffic (Ti) 25.01 % in order to limit the range of subsequent z values, which have Collisions involving bicycles1 Number of total bike collisions in a 5-year-period been later turned into negative. within a 150 m radius Cyclist volume 2 Number of daily bicycles within a 150 m radius Thus, to calculate the Bikeability value for each cell we have used the Infrastructure (Ii) following formula: Nearest cycle path3 Proximity to cycleway with exclusiveness (in m) Nearest cyclable lane3 Proximity to cycleway without exclusiveness (in Bi = (0.4 × Ti ) + (0.15 × Ii ) + (0.15 × Ci ) + (0.1 × Pi ) + (0.2 × Si ) m) Connectivity (Ci) The resulting bikeability index ranges from 0 to 1, 1 expressing the Intersections of cycle paths3 Number of intersections between cycle paths most bikeable areas and 0 the least bikeable ones. within a 400 m radius Intersections of cyclable lanes3 Number of intersections between cyclable lanes 3.3. Validating the index within a 400 m radius Intersections of cyclable paths Number of intersections between cycle paths and and cyclable lanes3 cyclable lanes within a 400 m radius To validate the index, we used mobility data from the EMEF18 and Parking spaces (Pi) EMEF19 travel surveys. These are annual CATI (Computer Assisted Distance to Bicing stations4 Proximity to bicycle-sharing system stations (in Telephone Interviews) surveys carried out by the Autoritat del Transport m) 5 Metropolità and comprise a sample of 10,100 individuals. These surveys Distance to bike racks Proximity to open-air public bike racks (in m) Topography (Si) have been repeatedly used not only by transportation management in Percent rise6 Mean percent rise within a 120m radius the Barcelona area but also by academic literature (Cubells et al., 2020; 1 Maciejewska et al., 2019; Marquet et al., 2017; Marquet and Miralles- (Ajuntament de Barcelona, 2019b). 2 (Ajuntament de Barcelona, 2019a). Guasch, 2018). Their main objective is to assess the habits and atti­ 3 (Ajuntament de Barcelona, 2021a, 2021b, 2021e). tudes of the population in the Barcelona metropolitan region regarding 4 (Ajuntament de Barcelona, 2021c). their mobility and commuting patterns during working days. We 5 (Ajuntament de Barcelona, 2015). assigned bikeability values to EMEF18 and EMEF19 participants based 6 (Institut Cartogràfic i Geològic de Catalunya, 2021). on the average bikeability values found in a 500 m buffer around their 4 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 home address. We then tested the association between bikeability levels problematic. and the odds of choosing bike as a modal choice. First, we used an As we can see, hilly areas – Collserola, Montjuïc and Tres Turons – ANOVA to test the associations between living in a high-bikeable envi­ are some of the least suited for cycling. These areas perform well in one ronment and cycling as the main mode of transportation. Then we run a of the most important components of bikeability (Traffic), but the lack of binary logistic regression model to test the association between bike­ flat areas and cycling infrastructure is detrimental to their overall score. ability and choosing a bike as a modal choice after adjusting for key On the opposite side, flatter, more coastal areas, as well as neighbour­ socioeconomic variables such as gender, age, education level and job hoods with very small blocks and abundance of pacified streets, such as status. Ciutat Vella, Gràcia or the north-western part of Sants-Montjuïc, concentrate high bikeability scores. The map also highlights the exis­ 4. Results tence of some problematic intersections. The intersections along the Diagonal Avenue, especially those in the centre of Eixample are critical, The process of creating and testing the bikeability index is structured and the northern districts of Nou Barris, Sant Andreu and Sant Martí also in two parts. First, we have mapped the distribution of the index values have intersections that present more danger to cyclists than other parts in the city of Barcelona, as well as the distribution of the five main of the city. environmental features that compose the index based on the 100 × 100 While the composite map is a great way of summarizing the infor­ m grid. On a second stage, we have tested the validity of our index mation, it does not include the nature of the diversity in bikeability. In comparing it with the results obtained from the local official travel other words, it does not differentiate between the many problems that survey (EMEF2019). can affect the bikeability of an area. On the other hand, component maps, shown in Fig. 3, offer a much better insight into the potential and challenges that each area may face when evaluating its suitability for 4.1. Bikeability index cycling. In the spatial distribution of the variable “Traffic” we can see that When we assess the distribution of the bikeability index in Barcelona overall, the city appears to be relatively safe for cyclists. The only areas (Fig. 2) we can see that most built areas are bikeable, although a where the interaction between cyclists and motorists are problematic are considerable number of spaces present some challenges. Areas in green in the districts of Eixample and Sant Martí, which are also the districts are the ones that score well in the index and therefore have better with wider roads and more motorized traffic. This is especially true for conditions for cycling, while areas in yellow to red need improvement. the more central red area, in the vicinity of Diagonal Avenue and Gran The map shows significant disparities in areas in the city of Barcelona Via. Meridiana Avenue, one of the main traffic arteries of the city, also and makes it easy to identify those places where cycling can be Fig. 2. Bikeability index for the city of Barcelona. 5 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 Fig. 3. Component maps of the bikeability index. scores as a risky area, concentrating a higher frequency of accidents. it is understandable that the demand for cycling facilities there is lower. However, Diagonal Mar, a recently-developed residential area located in Nevertheless, the variable “Connectivity” nuances the real utility of the the north-eastern part of the Sant Martí and away from the more con­ cycling infrastructure. Only a few areas in Eixample, Sants and Ciutat gested streets, has a worrying number of collisions that calls for further Vella stand out for having well-connected cycling infrastructure, while intervention. the rest of the city still has room for improvement. Collisions, however, cannot solely be attributed to a lack of infra­ As for “Parking Spaces”, there is a lack of appropriate and dedicated structure. As we can see on the “Infrastructure” map, most of the city is parking spaces in vast areas of the city, as only Eixample and Ciutat Vella well provided of cycling infrastructure, except for the mountainous area appear as areas where leaving the bike on the street is relatively easy. of Collserola and the industrial area of Zona Franca, in the southern part Even though some areas such as the hills of Collserola and Tres Turons as of Sants-Montjuïc, as well as a smaller area near the northernmost area well as the industrial part of Zona Franca (southern Sants-Montjuïc) do in Sant Martí. Except for the latter, these areas are scarcely populated, so not require large parking facilities because they are not residential areas, 6 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 districts such as Nou Barris are heavily populated and yet there is a Table 4 significant lack of bike racks or bike sharing systems. Binary logistic regression model assessing odds of being a frequent bike user per Finally, the variable “Topography” confirms one of the main culprits main socio-demographic and bikeability variables. of the lower bikeability score in some areas in places such as Montjuïc, Variables Exp(B) Sig. 95 % CI Collserola or Tres Turons. The hilliness of the terrain makes it difficult Lower Upper for cyclists to climb. Nevertheless, we must bear in mind that Barcelona Socio-demographic variables has an increasingly steepness from the sea to the mountain, which Women1 0.390 0.000 0.327 0.465 overall makes cycling a challenge for those living in the upper Ages 30–642 0.769 0.031 0.605 0.976 neighbourhoods. Ages ≥ 652 0.345 0.000 0.206 0.580 Secondary education3 1.439 0.016 1.071 1.932 Higher education3 2.082 0.000 1.569 2.762 4.2. Testing the predictive capacity of the index Unemployed4 0.939 0.688 0.689 1.278 Retired4 0.363 0.000 0.230 0.573 To test the predictive capacity of our bikeability index, we use an Homemaker4 0.518 0.136 0.218 1.231 official travel survey that asked about the frequency of cycling in the Student4 0.808 0.301 0.539 1.210 Bikeability variables Barcelona area (n = 3620). Using the bikeability values within a 500 m Bikeability index 1.830 0.001 1.300 2.576 network buffer from the geolocated address of each survey respondent we can observe a high association between bikeability values and fre­ Constant 0.367 0.000 quency of cycling. Table 3 displays the average bikeability values 1 Reference category = Men. around the residence of those who report biking always, almost always, 2 Reference category = Ages 16–29. often, sometimes, or never/hardly ever. The higher the bikeability 3 Reference category = No studies / primary education. index, the more frequent bike use is – a significant association that is 4 Reference category = Occupied. confirmed by the ANOVA test further validating the predictive capacity of our bikeability index. This means that those living in more bikeable Our results show that there is an evident gender gap in cycling in environments are generally more prone to cycling more often. There­ Barcelona, as men are more willing to cycle for transport than women. fore, it confirms that a more bicycle-friendly environment helps increase Our model estimates that women in Barcelona have 0.39 the chance of the likelihood of cycling as a means of transportation. cycling with respect to men. Young people (aged 16–29) are more likely Table 3 also includes the results for each of the individual indicators to cycle than their older counterparts. This happens with adults aged that compose the bikeability index. People who cycle more often do not 30–64 (OR = 0.769; p = 0.031), but more significantly with elderly necessarily seem to live in places where traffic interaction is safer. The people (OR = 0.345; p = 0.000). same can be said about the distance to cycling infrastructure. Never­ Education level was used as a proxy for income and the results show theless, those who cycle often or almost always do tend to live in areas that despite cycling is a very affordable means of transport, it is more that have well-connected cycle infrastructure and that have more prevalent among more affluent people. In Barcelona, people with higher accessible parking facilities. This underlines the importance of a well- education use the bike in a significant higher rate than people with connected network to increase cycling, as well as having convenient elementary or no education (OR = 2.082; p = 0.000). Even people with places to park the bike, be it shared or private. When it comes to secondary education also cycle significantly more (OR = 1.439; p = topography, it also appears that those living in flatter areas cycle more 0.016). Therefore, besides the gender and age gap, there is also a class often. These results complement the aggregated bikeability index and and/or educational gap. The more educated a person is, the more likely show that bikeability does not come down to a single element of the built they are to bike. environment, but it is rather a sum of its parts. Lastly, when using job status as a sociodemographic variable, the Additionally, we have tested our index through a binary logistic results are not significant, except for retired people. However, we cannot regression model that adjusts for basic sociodemographic variables. In infer that their lower likelihood for cycling (OR = 0.363; p = 0.000) is Table 4 we can see that a positive association between bikeability and due to their condition as retired people. It is more likely that this is due the odds of being a frequent bike user exists (OR = 1.830; p = 0.001). to their age, as we have previously seen. The fact that most of the included sociodemographic variables are also When including all the components of walkability along with statistically significant, suggests that they also have an influence on adjusting for basic sociodemographic variables (Table 5) results do not cycling rates. Table 3 Bivariate associatios between bikeability components and frequency of bike use. Frequency of bike Bikeability index Traffic Infrastructure Connectivity Parking spaces Topography usea Mean (SD) p- Mean (SD) p- Mean (SD) p- Mean (SD) p- Mean (SD) p- Mean (SD) p- valueb valueb valueb valueb valueb valueb 0.000 0.000 0.009 0.000 0.000 0.000 Never / hardly ever 0.170 − 0.495 0.767 0.736 0.637 0.395 (0.24) (0.66) (0.36) (0.85) (0.92) (0.51) Sometimes 0.191 − 0.569 0.807 0.837 0.747 0.489 (0.25) (0.66) (0.30) (0.88) (0.93) (0.44) Often 0.233 − 0.586 0.783 1.001 1.004 0.496 (0.28) (0.65) (0.36) (0.97) (1.05) (0.49) Almost always 0.227 − 0.763 0.864 1.168 1.141 0.564 (0.27) (0.75) (0.20) (0.90) (0.92) (0.43) Always 0.246 − 0.563 0.749 0.963 1.064 0.540 (0.33) (0.67) (0.39) (1.08) (1.12) (0.47) Total 0.179 − 0.519 0.776 0.781 0.694 0.419 (0.18) (0.66) (0.35) (0.87) (0.94) (0.50) a Self-reported frequency of bike use: “Evaluate your use of cycling”. b One-way ANOVA test. 7 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 Table 5 et al., 2015b; Lin and Wei, 2018; Winters et al., 2013) had already Binary logistic regression model 2 assessing odds of being a frequent bike user shown that “bikeability” is a multidimensional concept that encom­ per main socio-demographic and bikeability variables. passes the sum of multiple factors. Our results confirm this as shown by Variables Exp(B) Sig. 95 % CI the fact that the predictor capacity of the aggregate index was found Lower Upper significant while each of the individual components of the index did not. However, we have observed that some variables are indeed individually Socio-demographic variables related with higher rates of cycling. For instance, people with nearby Women1 0.386 0.000 0.323 0.460 Ages 30–642 0.754 0.022 0.593 0.960 parking spaces have a higher probability of cycling, a phenomenon that Ages ≥ 652 0.331 0.000 0.197 0.557 had previously been hinted in Barcelona (Cole-Hunter et al., 2015). Secondary education3 1.411 0.023 1.048 1.898 Nonetheless, it is topography the variable that consistently influences Tertiary education3 2.016 0.000 1.517 2.679 bikeability on its own in all our models. Our finding coincides with a Unemployed4 0.925 0.624 0.678 1.262 Retired4 0.363 0.000 0.229 0.573 wide array of previous research (Arellana et al., 2020; Cole-Hunter et al., Homemaker4 0.530 0.152 0.223 1.264 2015; Kamel et al., 2020; Krenn et al., 2015b; Lin and Wei, 2018; Student4 0.768 0.203 0.511 1.153 Winters et al., 2013), indicating the need to overcome the difficulties Bikeability variables emerging from hilly terrain. Our finding is important as it demonstrates Traffic 1.007 0.933 0.851 1.193 that objectively measured orography is having a negative effect over Infrastructure 0.833 0.298 0.591 1.175 Connectivity 1.086 0.420 0.889 1.326 willingness to cycle. As e-bikes become more popular (Galatoulas et al., Parking spaces 1.180 0.052 0.998 1.394 2020; Roig-Costa et al., 2021) it is essential to continue monitoring the Topography 1.344 0.025 1.039 1.740 influence of this variable to bikeability. Hilly areas in Collserola, Tres Turons and Montjuïc are deemed particularly inappropriate for cycling. Constant 0.361 0.000 However, because these areas are relatively unpopulated areas and 1 Reference category = Men. away from the main centres of economic activity, their impact on the 2 Reference category = Ages 16–29. city’s overall daily mobility is rather limited. 3 Reference category = No studies / primary education. Additionally, we have observed that presence of infrastructure and 4 Reference category = Occupied. connectivity do not go together. No matter how good and safe cycle paths are, a poorly connected network eventually makes cycling less vary greatly compared to those with the aggregated index. Gender, age, convenient, as it may force cyclists to take detours and leave designated and education remain significant predictors of cycling levels, while job bike paths, increasing the risk of accidents (Olmos et al., 2020). Bike­ status does not. Most of the individual components of bikeability are not ability indexes should thus consider connectivity indicators as basic core significantly associated with the odds of being a frequent bike user. components, something that was already pointed out by Cervero et al. Topography is the only exception (OR = 1.344; p = 0.025), in a way that (2009) and Winters et al. (2013). for every increase in 1 in our topography index (calculated in z values), In Barcelona, the main deterrent factor is interaction with motorized there is a 1.344 increase in the likelihood of being a frequent cyclist. traffic, and it is especially critical in the most congested streets, as well as the area around the eastern part of Sant Martí. This supports the need 5. Discussion and conclusion for providing infrastructure that not only enable cycling but also en­ courages it by making it safe and having cyclists perceive it as such The present analysis assesses cycling conditions in the city of Bar­ (Handy and Xing, 2011; Parkin et al., 2007). Previous research such as celona by creating and validating a new bikeability index. To do so we Kamel et al. (2020) had already established that safety from motorized have used objective spatial indicators that stem both from the literature traffic was an important attribute to incorporate to the index, and our and a local survey on self-reported relevant factors for cycling in the city. results fully agree with that assessment. The resulting index and distribution map has been later tested by using When it comes to sociodemographic groups we have also corrobo­ an official geolocated travel survey and has been validated as a predictor rated some points made in previous research. The gender gap in cycling, of bike use. The proposed index is thus both a valuable tool to represent where women cycle less than men, has been observed in a myriad of the hotspots and problems of the capacity of Barcelona’s built envi­ studies (Handy and Xing, 2011; Ma and Dill, 2017; Mitra and Nash, ronment to sustain cycling, as well as to reliably predict the frequency of 2019; Parkin et al., 2008; Van Dyck et al., 2013; Winters et al., 2007), cycling. This index can therefore also be exported to other cities, espe­ and stands in stark contrast with the generally greater representation of cially with similar urban morphology features, such as those in the women in active transport (Maciejewska and Miralles-Guasch, 2020). Mediterranean region. The cause of this inequality in cycling rates between genders has been Previous attempts to create bikeability indexes have tried to assign widely discussed, but it can be attributed to insufficient access to dedi­ weights to spatial indicators based on theoretical criteria (Krenn et al., cated bike infrastructure and a high presence of high-speed traffic (Mitra 2015b; Winters et al., 2013). The present analysis goes beyond that and and Nash, 2019; Van Dyck et al., 2013). In a similar vein, young people similarly to Arellana et al. (2020), it wheights the variables by are more likely to cycle than their older counterparts (Ma and Dill, 2017; measuring their relative importance according to subjective data (the Pucher and Buehler, 2008; Winters et al., 2007), a phenomenon that we factors that are reported by local cyclists to be detrimental to cycling). have confirmed to also be the case in Barcelona. This may be attributed Bikeability indexes can often end up including too many variables to the better physical condition of young people, but it can also respond leading to difficulties in interpretation or replicability for the general to a generational gap, where younger citizens prefer alternatives to public, such as in the case of Lin and Wei (2018). Given the results of our motorized transport for their daily mobility due to different factors, study, we would argue that using some kind of validation criteria based ranging from environmental concerns to a postponement of certain life on either self-reported barriers to cycling or on actual bike use would be milestones (Wang et al., 2018). On the other hand, education level was more beneficial than trying to incorporate too many indicators. Bike­ used as a proxy for income and the results show a paradox: despite the ability indexes should be informative enough to pinpoint the main fact that cycling is a very affordable means of transport, it is more problems of cycling in a city as well as simple enough for them to be prevalent among more affluent people, as Ma and Dill (2017) and Parkin easily replicated elsewhere. Additionally, simplicity can also tolerate the et al. (2008) already observed. inclusion of more place-specific variables better as each city can adapt The present study is the first to map bikeability and its distribution in the index to suit its own circumstances. Barcelona. In this regard, other studies have worked on the city’s Previous research (Arellana et al., 2020; Kamel et al., 2020; Krenn walkability (Marquet and Miralles-Guasch, 2015) and vitality levels 8 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 (Delclòs-Alió and Miralles-Guasch, 2018). The spatial correlation be­ support was provided by Government of Catalonia. tween walkability and bikeability levels appears to be weak which is probably due to the incomplete network of dedicated bike lanes and Data availability uneven distribution of bike infrastructure. Bikeability seems to spatially correlate with urban vitality. When assessing bikeability levels, we must Data will be made available on request. bear in mind that conditions might change throughout the day or the week. As Broach et al. observed (2012), traffic congestion can have Acknowledgements enormous variations throughout the day, thus influencing the attrac­ tiveness of biking in certain areas at different times. Similarly, cycling Not applicable. has both utilitarian and recreational uses, and the consideration of what makes an environment “bikeable” may be different in both cases (Bee­ nackers et al., 2012; Cervero et al., 2009; Kellstedt et al., 2021; Ma and Ethics Approval and Consent to Participate Dill, 2017). At the local level, the financial and political dimensions are crucial to It is not applicable, because this manuscript does not report on or increase its bikeability capacity (Alm and Koglin, 2022). However, city involve the use of any animal or human data or tissue. planners should be cautious when increasing levels of bikeability if they desire to increase cycling rates, as the attitudes, liking and comfort level Consent for Publication with bicycling of the population are also important factors (Handy and Xing, 2011). This means that it is very difficult to change the willingness It is not applicable, because this manuscript does not contain any to cycle of a population just with infrastructure, there is a need to also individual person’s data in any form. change the attitudes and the mobility culture (Haustein et al., 2019). Therefore, the role of other forms of public policy is crucial in this sense, and a more comprehensive approach is highly recommended. Pro- Availability of Data and Material bicycling initiatives, supportive land use planning and restrictions on car use are some of the complimentary policies that can help increase the The datasets used and/or analysed during the current study are number of daily cyclists and leave behind the marginalisation that available from the corresponding author on reasonable request. bicycling once suffered in urban transport planning in comparison to other forms of transportation (Koglin, 2013; Koglin and Rye, 2014; References Nielsen et al., 2013; Pucher et al., 2010; Pucher and Buehler, 2008). Alm, J., Koglin, T., 2022. (In)capacity to implement measures for increased cycling? Experiences and perspectives from cycling planners in Sweden. J. Urban Mobility 2, 6. Limitations 100029. https://doi.org/10.1016/j.urbmob.2022.100029. Arellana, J., Saltarín, M., Larrañaga, A.M., González, V.I., Henao, C.A., 2020. Developing an urban bikeability index for different types of cyclists as a tool to prioritise bicycle Our study is not without limitations. Our index only comprises the infrastructure investments. Transp. Res. A Policy Pract. 139, 310–334. https://doi. municipality of Barcelona, but the urban continuum of the city extends org/10.1016/j.tra.2020.07.010. beyond its borders. When testing our index with the EMEF survey, we Beenackers, M.A., Foster, S., Kamphuis, C.B.M., Titze, S., Divitini, M., Knuiman, M., Van Lenthe, F.J., Giles-Corti, B., 2012. Taking up cycling after residential relocation: have considered the coordinates of the place of residence to see if cycling Built environment factors. Am. J. Prev. Med. 42 (6), 610–615. https://doi.org/ habits were greater in addresses with higher bikeability scores. It has 10.1016/j.amepre.2012.02.021. often been discussed that eagerness to cycle is also dependent on the Bhandal, J., & Noonan, R. J. (2022). Motivations, perceptions and experiences of cycling for transport: A photovoice study. Journal of Transport and Health, 25(July 2021), levels of bikeability of the origin, route and destination (Winters et al., 101341. 10.1016/j.jth.2022.101341. 2010a). Broach, J., Dill, J., Gliebe, J., 2012. Where do cyclists ride? A route choice model developed with revealed preference GPS data. Transp. Res. A Policy Pract. 46 (10), 1730–1740. https://doi.org/10.1016/j.tra.2012.07.005. Funding Institut Cartogràfic i Geològic de Catalunya. (2021). Model d’elevacions del terreny de 5x5 m. Institut Cartogràfic i Geològic de Catalunya. https://www.icgc.cat/ca/Desc This project has received funding from the Generalitat de Catalunya arregues/Elevacions/Model-d-elevacions-del-terreny-de-5x5-m. (2020 PANDE 00023) and Spain’s Agencia Estatal de Investigación, Cervero, R., Sarmiento, O.L., Jacoby, E., Gomez, L.F., Neiman, A., 2009. Influences of built environments on walking and cycling: Lessons from Bogotá. Int. J. Sustain. Prueba de Concepto projects (MOVACTES PDC2021-120820-I00). Oriol Transp. 3 (4), 203–226. https://doi.org/10.1080/15568310802178314. Marquet is funded by a Ramón y Cajal fellowship (RYC-2020-029441-I) Charreire, H., Roda, C., Feuillet, T., Piombini, A., Bardos, H., Rutter, H., Compernolle, S., awarded by the Spanish Ministry of Science and Innovation. Mackenbach, J.D., Lakerveld, J., Oppert, J.M., 2021. Walking, cycling, and public transport for commuting and non-commuting travels across 5 European urban regions: Modal choice correlates and motivations. J. Transp. Geogr. 96 (August), CRediT authorship contribution statement 103196 https://doi.org/10.1016/j.jtrangeo.2021.103196. Cole-Hunter, T., Donaire-Gonzalez, D., Curto, A., Ambros, A., Valentin, A., Garcia- Aymerich, J., Martínez, D., Braun, L.M., Mendez, M., Jerrett, M., Rodriguez, D., de Oriol Codina: Conceptualization, Methodology, Formal analysis, Nazelle, A., Nieuwenhuijsen, M., 2015. Objective correlates and determinants of Writing – original draft, Writing – review & editing. Monika Macie­ bicycle commuting propensity in an urban environment. Transp. Res. Part D: Transp. jewska: Conceptualization, Data curation, Writing – review & editing, Environ. 40 https://doi.org/10.1016/j.trd.2015.07.004. Cubells, J., Marquet, O., Miralles-Guasch, C., 2020. Gender and Age Differences in Funding acquisition. Jordi Nadal: Conceptualization, Writing – review Metropolitan Car Use. Recent Gender Gap Trends in Private Transport. Sustainability & editing, Supervision. Oriol Marquet: Conceptualization, Formal 12 (18), 7286. analysis, Resources, Data curation, Writing – original draft, Writing – Ajuntament de Barcelona. (2019b). Vehículos implicados en accidentes gestionados por la Guardia Urbana en la ciudad de Barcelona—Conjuntos de datos. Open Data Barcelona. review & editing, Supervision, Project administration, Funding https://opendata-ajuntament.barcelona.cat/data/es/dataset/accidents-vehicle acquisition. s-gu-bcn. Ajuntament de Barcelona. (2021a). Carriles Bici de la ciudad de Barcelona—Conjuntos de datos. Open Data Barcelona. https://opendata-ajuntament.barcelona.cat/data/es/da Declaration of Competing Interest taset/carril-bici. Ajuntament de Barcelona. (2021b). Carriles Bici en construcción en la ciudad de The authors declare the following financial interests/personal re­ Barcelona—Conjuntos de datos. Open Data Barcelona. https://opendata-ajuntament. lationships which may be considered as potential competing interests: barcelona.cat/data/es/dataset/carrils-bici-en-construccio. Ajuntament de Barcelona. (2021e). Vías Ciclables en la ciudad de Barcelona—Conjuntos de Oriol Marquet reports financial support was provided by Spain Ministry datos. Open Data Barcelona. https://opendata-ajuntament.barcelona.cat/data of Science and Innovation. Carme Miralles-Guasch reports financial /es/dataset/vies-ciclables. 9 O. Codina et al. Transportation Research Interdisciplinary Perspectives 16 (2022) 100725 Ajuntament de Barcelona. (2015). Puntos de anclaje de bicicletas de la ciudad de Medina, L., Hernández, S., 2008. Manual for the Design of Cyclepaths in Catalonia, first Barcelona—Conjuntos de datos. Open Data Barcelona. https://opendata-ajuntament. ed. Generalitat de Catalunya. barcelona.cat/data/es/dataset/punts-ancoratge-bicicletes. Miralles-Guasch, C., Roig-Costa, O., Marquet, O., 2022. Patinetes eléctricos y bicicletas Ajuntament de Barcelona. (2019a). Detall dels aforaments de mobilitat de la ciutat de compartidas. Nuevos transportes urbanos, nuevos usuarios. El caso de Barcelona. Barcelona—Conjunts de dades. Open Data Barcelona. https://opendata-ajuntament.ba Papeles de Economía Española 171, 148–158. rcelona.cat/data/ca/dataset/aforaments-detall. Mitra, R., Nash, S., 2019. Can the built environment explain gender gap in cycling? An Ajuntament de Barcelona. (2021c). Información de las estaciones del nuevo Bicing de la exploration of university students’ travel behavior in Toronto, Canada. International ciudad de Barcelona—Conjuntos de datos. Open Data Barcelona. https://opendata Journal of Sustainable Transportation 13 (2), 138–147. https://doi.org/10.1080/ -ajuntament.barcelona.cat/data/es/dataset/informacio-estacions-bicing. 15568318.2018.1449919. Ajuntament de Barcelona. (2021d). Open Data BCN. https://opendata-ajuntament.barce Muhs, C.D., Clifton, K.J., 2016. Do characteristics of walkable environments support lona.cat/ca. bicycling? Toward a definition of bicycle-supported development. J. Transp. Land Generalitat de Catalunya. (2019). Enquesta baròmetre de la bicicleta 2019. Use 9 (2), 147–188. https://doi.org/10.5198/jtlu.2015.727. Delclòs-Alió, X., & Miralles-Guasch, C. (2018). Looking at Barcelona through Jane Naess, P., 2012. Urban form and travel behavior: Experience from a Nordic context. Jacobs’s eyes: Mapping the basic conditions for urban vitality in a Mediterranean J. Transp. Land Use 5 (2), Article 2. https://doi.org/10.5198/jtlu.v5i2.314. conurbation. Land Use Policy, 75(November 2017), 505–517. 10.1016/j. Nielsen, T.A.S., Skov-Petersen, H., 2018. Bikeability – Urban structures supporting landusepol.2018.04.026. cycling. Effects of local, urban and regional scale urban form factors on cycling from Dill, J., 2009. Bicycling for Transportation and Health: The Role of Infrastructure. home and workplace locations in Denmark. J. Transp. Geogr. 69, 36–44. https://doi. J. Public Health Policy 30, 95–110. https://doi.org/10.1057/jphp.2008.56. org/10.1016/j.jtrangeo.2018.04.015. Dill, J., Voros, K., 2007. Factors affecting bicycling demand: Initial survey findings from Nielsen, T.A.S., Skov-Petersen, H., Agervig Carstensen, T., 2013. Urban planning the Portland, Oregon, region. Transp. Res. Rec. 2031 (2031), 9–17. https://doi.org/ practices for bikeable cities – the case of Copenhagen. Urban Res. Pract. 6 (1), 10.3141/2031-02. 110–115. https://doi.org/10.1080/17535069.2013.765108. Elvik, R., Bjørnskau, T., 2017. Safety-in-numbers: A systematic review and meta-analysis Olmos, L.E., Tadeo, M.S., Vlachogiannis, D., Alhasoun, F., Espinet Alegre, X., Ochoa, C., of evidence. Saf. Sci. 92 (0349), 274–282. Targa, F., González, M.C., 2020. A data science framework for planning the growth Galatoulas, N.F., Genikomsakis, K.N., Ioakimidis, C.S., 2020. Spatio-temporal trends of e- of bicycle infrastructures. Transportation Research Part C: Emerging Technologies bike sharing system deployment: A review in Europe, North America and Asia. 115 (March), 102640. https://doi.org/10.1016/j.trc.2020.102640. Sustainability (Switzerland) 12 (11), 4611. https://doi.org/10.3390/su12114611. Parkin, J., Wardman, M., Page, M., 2007. Models of perceived cycling risk and route Handy, S.L., Xing, Y., 2011. Factors correlated with bicycle commuting: A study in six acceptability. Accid. Anal. Prev. 39 (2) https://doi.org/10.1016/j.aap.2006.08.007. small U.S. cities. Int. J. Sustain. Transp. 5 (2), 91–110. https://doi.org/10.1080/ Parkin, J., Wardman, M., Page, M., 2008. Estimation of the determinants of bicycle mode 15568310903514789. share for the journey to work using census data. Transportation 35 (1), 93–109. Haustein, S., Koglin, T., Nielsen, T.A.S., Svensson, Å., 2019. A comparison of cycling https://doi.org/10.1007/s11116-007-9137-5. cultures in Stockholm and Copenhagen. Int. J. Sustain. Transp. 14 (4), 280. Pucher, J., Buehler, R., 2008. Making cycling irresistible: Lessons from the Netherlands, Hyland, M., Frei, C., Frei, A., Mahmassani, H.S., 2018. Riders on the storm: Exploring Denmark and Germany. Transport Reviews 28 (4), 495–528. https://doi.org/ weather and seasonality effects on commute mode choice in Chicago. Travel Behav. 10.1080/01441640701806612. Soc. 13, 44–60. https://doi.org/10.1016/j.tbs.2018.05.001. Pucher, J., Buehler, R., 2017. Cycling towards a more sustainable transport future. Kamel, M.B., Sayed, T., Bigazzi, A., 2020. A composite zonal index for biking Transp. Rev. 37 (6), 689–694. https://doi.org/10.1080/01441647.2017.1340234. attractiveness and safety. Accid. Anal. Prev. 137 https://doi.org/10.1016/j. Pucher, J., Dill, J., & Handy, S. (2010). Infrastructure, programs, and policies to increase aap.2020.105439. bicycling: An international review. Preventive Medicine, 50(SUPPL.), S106–S125. Kellstedt, D.K., Spengler, J.O., Foster, M., Lee, C., Maddock, J.E., 2021. A Scoping 10.1016/j.ypmed.2009.07.028. Review of Bikeability Assessment Methods. J. Community Health 46 (1), 211–224. Roig-Costa, O., Gómez-Varo, I., Cubells, J., Marquet, O., 2021. La movilidad post- https://doi.org/10.1007/s10900-020-00846-4. pandemia: Perfiles y usos de la micromovilidad en Barcelona. Revista Transporte y Koglin, T., 2013. Vélomobility—A critical analysis of planning and space. Lund Univ., Territorio 25, 72–96. Faculty of Engineering, Department of Technology and Society. Walk Score. (2021). Bike Score Methodology. Walk Score. https://www.walkscore.com Koglin, T., Rye, T., 2014. The marginalisation of bicycling in Modernist urban transport /bike-score-methodology.shtml. planning. J. Transp. Health 214–222. https://doi.org/10.1016/j.jth.2014.09.006. Shu, S., Bian, Y., Rong, J., Xu, D., 2019. Determining the exact location of a public Krenn, P.J., Oja, P., Titze, S., 2015a. Development of a Bikeability Index to Assess the bicycle station—The optimal distance between the building entrance/exit and the Bicycle-Friendliness of Urban Environments. Open Journal of Civil Engineering 05 station. PLoS One 14 (2), 1–18. https://doi.org/10.1371/journal.pone.0212478. (04), 451–459. https://doi.org/10.4236/ojce.2015.54045. Urban Systems. (2015). Cycling safety study. Final Report. In City of Vancouver. Krenn, P.J., Oja, P., Titze, S., 2015b. Development of a Bikeability Index to Assess the 1166.0024.01. Bicycle-Friendliness of Urban Environments. Open. J. Civ. Eng. 05 (04) https://doi. Van Dyck, D., Cerin, E., Conway, T.L., De Bourdeaudhuij, I., Owen, N., Kerr, J., org/10.4236/ojce.2015.54045. Cardon, G., Frank, L.D., Saelens, B.E., Sallis, J.F., 2013. Perceived neighborhood Lin, J.-J., Wei, Y.-H., 2018. Assessing area-wide bikeability: A grey analytic network environmental attributes associated with adults’ leisure-time physical activity: process. Transp. Res. A 113, 381. Findings from Belgium, Australia and the USA. Health Place 19 (1), 59–68. https:// Lowry, M., Callister, D., Gresham, M., Moore, B., 2012. Assessment of communitywide doi.org/10.1016/j.healthplace.2012.09.017. bikeability with bicycle level of service. Transp. Res. Rec. 2314 https://doi.org/ Wang, K., Akar, G., Chen, Y.J., 2018. Bike sharing differences among Millennials, Gen 10.3141/2314-06. Xers, and Baby Boomers: Lessons learnt from New York City’s bike share. Transp. Ma, L., Dill, J., 2017. Do people’s perceptions of neighborhood bikeability match Res. A Policy Pract. 116, 1–14. https://doi.org/10.1016/j.tra.2018.06.001. “reality”? J. Transp. Land Use 10 (1), 291–308. https://doi.org/10.5198/ Winters, M., Friesen, M.C., Koehoorn, M., Teschke, K., 2007. Utilitarian Bicycling. A jtlu.2016.796. Multilevel Analysis of Climate and Personal Influences. Am. J. Prev. Med. 32 (1), Maciejewska, M., Marquet, O., Miralles-Guasch, C., 2019. Changes in gendered mobility 52–58. https://doi.org/10.1016/j.amepre.2006.08.027. patterns in the context of the Great Recession. J. Transp. Geogr. 79. Winters, M., Brauer, M., Setton, E.M., Teschke, K., 2010a. Built environment influences Maciejewska, M., Miralles-Guasch, C., 2020. Evidence of gendered modal split from on healthy transportation choices: Bicycling versus driving. J. Urban Health 87 (6), Warsaw, Poland. Gend. Place Cult. 27 (6), 809–830. https://doi.org/10.1080/ 969–993. https://doi.org/10.1007/s11524-010-9509-6. 0966369X.2019.1639631. Winters, M., Teschke, K., Grant, M., Setton, E.M., Brauer, M., 2010b. How far out of the Marquet, O., & Miralles-Guasch, C. (2018). Resilient territories and mobility adaptation way will we travel? Built environment influences on route selection for bicycle and strategies in times of economic recession. Evidence from the Metropolitan Region of car travel. Transp. Res. Rec. 2190, 1–10. https://doi.org/10.3141/2190-01. Barcelona, Spain 2004-2012. European Urban and Regional Studies, 25(4), 345–359. Winters, M., Brauer, M., Setton, E.M., Teschke, K., 2013. Mapping bikeability: A spatial Marquet, O., Hipp, J.A., Miralles-Guasch, C., 2017. Neighborhood walkability and active tool to support sustainable travel. Environ. Plann. B. Plann. Des. 40 (5) https://doi. ageing: A difference in differences assessment of active transportation over ten years. org/10.1068/b38185. J. Transp. Health 7, 190–201. Winters, M., Teschke, K., 2010. Route preferences among adults in the near market for Marquet, O., Miralles-Guasch, C., 2015. The Walkable city and the importance of the bicycling: Findings of the cycling in cities study. Am. J. Health Promot. 25 (1), proximity environments for Barcelona’s everyday mobility. Cities 42 (PB), 258–266. 40–47. https://doi.org/10.4278/ajhp.081006-QUAN-236. https://doi.org/10.1016/j.cities.2014.10.012. Winters, M., Teschke, K., Brauer, M., Fuller, D., 2016. Bike Score®: Associations between Marshall, W.E., Garrick, N.W., 2010. Effect of street network design on walking and urban bikeability and cycling behavior in 24 cities. Int. J. Behav. Nutr. Phys. Act. 13 biking. Transp. Res. Rec. 2198, 103–115. https://doi.org/10.3141/2198-12. (1), 1. https://doi.org/10.1186/s12966-016-0339-0. McNeil, N., 2011. Bikeability and the 20-min neighborhood: How infrastructure and Wysling, L., Purves, R.S., 2022. Where to improve cycling infrastructure? Assessing destinations influence bicycle accessibility. Transp. Res. Rec. 2247 (2247), 53–63. bicycle suitability and bikeability with open data in the city of Paris. Transportation https://doi.org/10.3141/2247-07. Research Interdisciplinary Perspectives 15, 100648. https://doi.org/10.1016/j. trip.2022.100648. 10