key: cord-0331969-3kg5egv4
authors: Pearson, L. K.; Dipnall, J.; Gabbe, B. J.; Braaf, S.; White, S.; Backhouse, M.; Beck, B.
title: The potential for bike riding across entire cities: quantifying spatial variation in interest in bike riding
date: 2021-03-17
journal: nan
DOI: 10.1101/2021.03.14.21253340
sha: d9a8dd5de955d42eebc125d757b55a62e4f85f0d
doc_id: 331969
cord_uid: 3kg5egv4

Background: Riding a bike is beneficial for health, the environment and for reducing traffic congestion. Despite this, bike riding participation in the state of Victoria, Australia, is low. To inform planning and practice, there is a need to understand the proportion of the population ("the near-market") that are interested in riding a bike, and how this varies across regions. The Geller typology classifies individuals into one of four groups, based on their confidence to ride a bike in various infrastructure types, and frequency of bike riding. The typology has been used at a city, state and country-wide scale, however not at a smaller spatial scale. We aimed to characterise and quantify the distribution of the Geller typology within Local Government Areas (LGAs) in the state of Victoria, Australia. Methods: An online survey was conducted in 37 LGAs in Victoria, including all LGAs in Greater Melbourne, and a selection of six key regional centres. Participants were recruited from an opt-in online research company panel with the objective of recruiting a representative sample of adults across each LGA. The Geller typology classified individuals as either: "Strong and Fearless", "Enthused and Confident", "Interested but Concerned", or "No Way No How". "Interested but Concerned" participants are those that would ride a bike if protected infrastructure were provided. Results: The survey was completed by 3999 individuals. Most participants owned a bike (58%), however only 20% rode at least once per week. The distribution of the Geller groups was: "Strong and Fearless" (3%), "Enthused and Confident" (3%), "Interested but Concerned" (78%), and "No Way No How" (16%). While variation in the distributions of the Geller groups was observed between LGAs within Greater Melbourne., the "Interested but Concerned" group, reflecting people who are comfortable riding only in protected lanes or off-road paths, was high across all LGAs and all demographic sub-groups. Even though the frequency of riding a bike was lower in women, interest in riding a bike was high and comparable to men. Participants who resided in the outer urban fringe regions of Greater Melbourne had high interest, but low participation in bike riding. Conclusions: While there was variation in interest in bike riding across an entire metropolitan region and across population groups, interest was high across all areas and demographics. Our results show the potential for substantial increases in cycling participation, but only when high-quality cycling infrastructure is provided. Further research is required to understand the policy and practice barriers to equitable provision of protected infrastructure.

least once per week. The distribution of the Geller groups was: 'Strong and Fearless' (3%), 'Enthused and Confident' (3%),

'Interested but Concerned' (78%), and 'No Way No How' (16%). While variation in the distributions of the Geller groups was 24 observed between LGAs within Greater Melbourne., the 'Interested but Concerned' group, reflecting people who are 25 comfortable riding only in protected lanes or off-road paths, was high across all LGAs and all demographic sub-groups. Even 26 though the frequency of riding a bike was lower in women, interest in riding a bike was high and comparable to men. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

Increasing participation in bike riding is well established as being beneficial for public health, the 36 environment and traffic congestion (1-5). The majority of populated areas of Australia exhibit ideal 37 conditions for riding a bicycle, including relatively flat topography and in most areas, a mild climate. Despite 38 this, participation in bike riding remains low in Australia compared to other international settings (6). To 39 increase bike riding participation and to ensure that planning and practice efforts are targeted to the whole 40 community, there is a need to identify specific sub-groups of the population that can be classified as the 41 bike riding 'near-market'; individuals who would like to start riding a bike. Bike rider typologies are 42 commonly used to achieve this, by segmenting populations into distinct groups with shared characteristics. 43

The most commonly used bicyclist typology is the 'Four Types of Cyclist' typology, first introduced in 2006 44 by Roger Geller (7) and later refined by Dill & McNeil (8) . This classifies people into one of four groups 45 based on their confidence to ride a bike in various infrastructure types, their interest in riding a bike and if 46 they had ridden a bike in the past month. One of the four categories are people who are 'Interested but 47

Concerned'. Interested but Concerned participants are those that would ride a bike if protected 48 infrastructure were provided. The typology has been used to quantify potentially latent groups of bicyclists 49 in cities, states, and countries. While the use of typologies to quantify the near-market of bicyclists across 50 large cities and regions has been helpful in informing bike riding strategies, prior studies have not explored 51 whether interest in bike riding varies within these geographies, such as across smaller spatial areas of cities 52 and states (8-11). The availability and quality of transport infrastructure often varies within cities 53 depending on zoning, distance from the city, topography, socioeconomic status and community 54 demographics (12-14). Identifying interest in bike riding at this small spatial scale is useful in understanding 55 geographical variation in bike riding potential and how protected infrastructure can be implemented to 56 maximise bike riding participation and address inequities in participation and infrastructure. To our 57 knowledge, previous literature has not quantified the Geller typology groups at a spatial scale smaller than 58 cities, and has not been applied in Australia. 59 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint 3 In this study, we aimed to quantify and identify the characteristics of the 'Four Types of Cyclist' in all Local 60 Government Areas (LGAs) within Greater Melbourne, and a selection of regional centres in the state of 61 Victoria, Australia. 62

We conducted a cross-sectional online survey in the state of Victoria, Australia. The objective was to recruit 65 a representative sample of adults (aged 18 years and older) across each local government area. Data 66 collection occurred over the period of August 12 th to September 10 th , 2020. Data collection occurred during 67 a period of restrictions due to the COVID-19 pandemic. Many Victorian workplaces were closed and 68 physical distancing was enforced (15), but there were no restrictions placed on leaving the home during 69 this period. 70

In line with values-based messaging for health promotion (16), we adopted the term "bike riding" rather 71 than "cycling", and their equivalents, in this study to ensure inclusivity and avoid association with 72 competitive cycling (17). 73

Victoria has a population of 6.7 million (18) of which 67% reside in the Greater Melbourne area (19). As of 75 2018, bike riding comprised 3% of all weekday trips in metropolitan Melbourne (20).

LGAs were chosen as 76 the geographical area for analysis. LGAs in Australia are defined subdivisions of states that are under the 77 jurisdiction of a particular Local Government (21). In Victoria, 85% of the road network is made up of local 78 roads, which are maintained by Local Governments (22). For this reason, Local Governments play an 79 essential role in the planning and maintenance of bicycling infrastructure. Data were collected in each LGA 80 within Greater Melbourne (n=31) and select regional LGAs (n=6) (Error! Reference source not found.). 81 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. LGAs were chosen specifically due to the high density of population (see Figure 1 , 85

Figure 2), and potential to improve bike riding participation rates and infrastructure. Similarly, data were 86 collected in six regional centres with higher populations, higher population density and relatively flat 87 topography. These areas also had the potential for increased participation in bike riding. Achieving a 88 representative sample in rural/regional LGAs other than those selected was not feasible due to small survey 89 panel sizes in these regions. 90

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint 6 comfortable riding a bike on any infrastructure, including on roads with no protected bicycling 105 infrastructure. Enthused and Confident participants are comfortable on non-residential streets with painted 106 on-road bike lanes (separating bikes from traffic without protection). Interested but Concerned participants 107 are comfortable riding a bike on protected bicycling infrastructure only, while No Way No How participants 108 are not comfortable riding a bike in any setting. 109

Adaptations were made to the Geller Typology for use in Australia. These included the conversion of miles 110 per hour to kilometres per hour, and change of the term "striped bike lane" to "painted bike lane". Each 111 question included a figure to allow the participant to better understand the scenario and context (see 112 Figure 3 ). Photos were not used as to avoid associations with particular areas, or other factors that could 113 influence bike riding comfort. The Geller typology was chosen for this study for its ability to inform policy 114 and practice, and to allow for comparisons to studies in other settings that have used the same questions. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint 

We recruited people 18 years of age or older who resided in one of the LGAs in Greater Melbourne or one 132 of the six regional LGAs. The aim was to achieve a representative sample size within each LGA, so that 133 typologies between LGAs could be compared. had partially responded to the survey or not entered the survey were sent a maximum of two reminders 143 over the data collection period to reduce non-response bias. In order to monitor the Victorian population 144 sample representativeness, the ORU were sent daily reports for the age and gender distributions for the 145 completed survey sample to date. Where particular groups were lower than the population, 'boosting' was 146 performed by sending further invitations and reminders those demographics not meeting the required 147 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint 8 proportions for sample representativeness. Age and gender representation at the LGA level was not 148 possible to monitor due to restrictions on information sharing restrictions between the ORU and Monash 149 University. Oversampling was performed for small LGA sub-populations to allow for more reliable estimates 150 and to avoid sampling bias due to under-coverage. When sample sizes were achieved, no further invitations 151 were sent to participants that resided in these areas. 152

Responses were checked for duplicates, of which none were found. Completion times were checked for 154 outliers to identify any responses completed in a time that was not feasible, and removed if so. Bureau of Statistics population density data (29), using the leaflet package in R (26, 30). 165

Descriptive statistics and frequency tables of weighted data were used to describe the bike riding 166 behaviours and frequency of each Geller Typology, and to make comparisons between age, gender and 167 region of residence. Chi-square tests were used to examine differences in the distribution of the Geller 168

Typology within age, income, and gender groups, and between Greater Melbourne and regional Victoria. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint 9 In total, 41,000 invitations and 20,000 reminders were sent by the ORU, with a response rate of 10%. Of the 173 4067 participants who started the survey, 3999 (98%) completed all sections and were included in the final 174 analysis. The age and gender of the weighted sample were generally consistent with the aggregated 175 population of the LGAs included in the study, with only the 65-74-year age group being 6.9% higher in the 176 sample than the target population (see Supplementary Materials C). The weighted sample had a higher 177 income than the target population, with 15% of participants earning over $104,000 per year compared to 178 7.2% of the target population. In the sample, 12% spoke a language other than English at home, compared 179 to 26% in Victoria (see Supplementary Materials C). 180

Over half of all participants owned a bike (57%), however only 20% rode their bike at least once per week. 181

Of participants that rode their bike at least once a year, most rode for recreation only (72%). 182

Over three quarters of the sample (78%) were classified as Interested but Concerned, 2.8% as Strong and 183

Fearless, 3.2% as Enthused and Confident and 16% as No Way No How (see Figure 4) . The Interested but 184

Concerned group comprised a higher proportion of women (48%) than the Strong and Fearless (28%) and 185

Enthused and Confident groups (25%) ( Table 1) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint 11 *For the counts of participants who identified as "non-binary" or "other", or reported "prefer not to say", robust analysis could not 193 be performed due to small sample sizes 194 **based on participants who had ridden a bike in the 12 months previous to COVID-19 restrictions 195

In Greater Melbourne, the Interested but Concerned group was consistently the largest group across all 197

LGAs, ranging from 66% to 88% ( Figure 5) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint Bike ownership was highest in the inner-West (67%), and lowest in inner-East areas, where in one LGA, only 205 40% of participants owned a bike. Participants who rode a bike at least once per week were most common 206 in the City of Melbourne (32%), and generally decreased with distance from the city (Figure 6 ). 207 208 

Proportions of Geller groups differed significantly between Greater Melbourne and regional Victorian LGAs 212 (x 2 = 19.4, p = <0.001). For both Greater Melbourne and regional Victorian LGAs, the Interested but 213

Concerned group made up the majority of the sample (79% and 71% respectively). A higher proportion of 214 participants were classified as No Way No How in regional Victoria (21%) than Greater Melbourne (15%), 215 however bike ownership was comparable between regional Victoria (60%) and Greater Melbourne (57%). 216

There was no association between how often participants rode a bike based on regionality (x 2 = 0.5, p = 3.5) 217

(Supplementary Material C). 218 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint

More than half of each sample were Interested but Concerned in all age groups (Figure 7) . There was an 220 association between age group and the Geller distribution (x 2 = 214.7, p = <0.001). The proportion of 221 participants classified as Interested but Concerned was lower in participants aged 55-74 years (72%) and 75 222 or more years (55%) compared to participants aged 18-34 years (84%) and 35 to 54 years (83%). The 

A higher proportion of men than women in the sample owned a bike (63% vs. 52%), rode a bike at least 229 once per week (28% vs. 12%) and rode a bike solely for transport (8.3% vs. 5.9%). The distribution of Geller 230 groups differed between women and men (x 2 =79.7, p=<0.001) (see Figure 8) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint

There is currently no data available on how interest in riding a bike differs between geographical regions, 251 and no data on the Geller typologies in Australia. This study quantified the proportion of each Geller 252 typology group across 37 LGAs in Victoria, Australia with complete data from 3999 individuals. Over half of 253 participants owned a bike, however only one in five rode a bike at least once per week. Most participants 254 were classified as Interested but Concerned, demonstrating a high latent demand for bike riding if 255 protected bicycling infrastructure were provided. Geller group distributions differed between LGAs within 256 Greater Melbourne, and between regional and metropolitan areas, supporting the need for analysis of 257

Geller groups at smaller spatial scales than cities or states. Previous published studies using the Geller Typology to quantify interest in bike riding have done so at a 271 multi-city, state and city level (8-11). To the authors' knowledge, our study is the first to quantify the Geller 272 typology proportions at a smaller spatial scale. We found that interest and confidence in riding a bike 273 ranged from 66% to 88% within the metropolitan area of Greater Melbourne. These differences may come 274 from differing available infrastructure, the demographics of an area, trip distance to central hubs and 275 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint physical setting (e.g. topography, primary land-use type) (35, 36). Further, it is important to identify these 276 participation, over two thirds of women in this study were interested in riding a bike. Importantly, many 296 existing bike riding environments are often designed for the needs and confidence levels of able-bodied 297 men (55-57). Infrastructure that enables women to ride a bike includes off-road paths, and bike-298 infrastructure that is physically separated from motor-vehicle traffic (55, 58). In future urban planning and 299 research, it is vital that a lack of participation is not seen as a lack of interest in bike riding. Understanding 300 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint the barriers to riding a bike for women, who are often under-represented in the bike riding community is 301 important to increasing participation. 302

In this study, a higher proportion of people in the lower income categories rode a bike for transport 304 purposes, and rode four or more days per week, compared to people in the higher income groups. Similar 305 findings have been shown both in Australia and internationally, where there was an association between 306 increased household income, and a decreased proportion of people riding a bike for transport (12, 59, 60). 307

People in lower income groups may ride a bike out of necessity for commuting, rather than as a 308 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint To our knowledge, this was the first study of Geller typologies in Australia, and the largest and most 331 population-representative for trip purpose and bike riding frequency. Further, this was the first application 332 of the Geller typology at a small spatial scale, demonstrating important differences between local 333 government areas. Despite this, some limitations were present. A recently published mixed methods study 334 raised concerns over the internal validity of the Geller typology, particularly that the category participants 335 were assigned to did not always represent their confidence in riding a bike (71). However, this typology was 336 used because of its ease of understanding with policy makers, and to enable comparisons to existing 337 literature and government bike riding strategies. Additionally, using an online research company may have 338 favoured participants with a higher income, with computer access, and whom are able to regularly check 339 their email, potentially introducing selection and response bias. To reduce response bias, invitations to 340 complete the survey were sent in batches and at different times during the week. This enabled people who 341 may have been occupied on certain days to receive a notification about the survey. As recommended by 342

Hay et al., convenience internet panels are useful when data can be weighted to compensate for coverage 343 errors (72). Some regional LGAs in this study had a low number of research company panel members. In 344 turn, only a small number of participants completed surveys and represented the population in those LGAs. 345

To reduce potential coverage errors, whole sample results were weighted by the population size of each 346

LGA in an effort to balance the effects of under and over sampling across LGAs. 347

To our knowledge, this study was the first application of the Geller typology groups at a small geographical 349 area level. We demonstrated that while there was variation in interest in bike riding across an entire 350 metropolitan region, interest was high across all LGAs, including outer urban areas that currently have 351 lower participation in bike riding. Similarly, despite participation in bike riding being lower in women, 352 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint interest remained high. It is vital that a lack of participation is not seen as a lack of interest in bike riding. 353

These findings are important in considering the role that bike riding can play in reducing health and 354 transport inequities and how protected bicycling infrastructure can be implemented to address these 355 inequities. Further research is required to identify what factors are preventing people in Victoria from 356 riding a bike and how these barriers vary across geographical areas. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint

Health benefits of cycling: a 367 systematic review

Health benefits of physical activity: the evidence

Benefits of shift from car to active transport. Transport policy

The 373 health and economic benefits of active transport policies in Barcelona

Evaluating active transport benefits and costs: Victoria Transport Policy Institute; 2015. 376 6. The Victorian Health Promotion Foundation. Victorian Cycling Strategy 2018-28

Four Types of Cyclists

Four types of cyclists? Examination of typology for better understanding of 380 bicycling behavior and potential

Influence of bike lane buffer types on perceived comfort and safety 382 of bicyclists and potential bicyclists

Area variation in 392 recreational cycling in Melbourne: a compositional or contextual effect?

Social (in) equity in access to cycling infrastructure

Health and Human Services, Victorian State Government. Victoria's restriction levels. 2020. 398 16. VicHealth. Values-based messaging for health promotion n.d

Perspectives and images of cycling as a barrier or facilitator of cycling. Transport 401 policy

Australian Bureau of Statistics. National, state and territory population 2020

Census QuickStats -Greater Melbourne

Victorian Inegrated Survey of Travel & Activity (VISTA)

RStudio Team. RStudio: Integrated Development for R

The emergence and characteristics of the Australian 473

Knight A. Cycling is the new golf. Sydney Morning Herald

Cycling provision separated from motor traffic: a 476 systematic review exploring whether stated preferences vary by gender and age. Transport reviews

Does the Built Environment Explain Gender Gap in Cycling? a Study of

Secondary Students in Toronto

I won't cycle on a route like this; I don't think I fully understood what isolation 481 meant": A critical evaluation of the safety principles in Cycling Level of Service (CLoS) tools from a gender 482 perspective

Promoting transportation cycling for women: the role of bicycle 484 infrastructure

Cycling for transport and recreation: associations with socio-486 economic position, environmental perceptions, and psychological disposition

Factors associated with proportions and miles of bicycling for 488 transportation and recreation in six small US cities

Equity and accessibility of cycling infrastructure: An analysis of 491

Riding tandem: Does cycling infrastructure investment 493 mirror gentrification and privilege in

Bicycle equity in Brazil: Access to safe cycling routes across neighborhoods in 496

The authors would like to acknowledge the contribution of the participants in this study. 359

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprintThe copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint 20 It is made available under a perpetuity.is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprintThe copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint It is made available under a perpetuity.is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprintThe copyright holder for this this version posted March 17, 2021. ; https://doi.org/10.1101/2021.03.14.21253340 doi: medRxiv preprint