The main goal of this study was to investigate the changes in travel behavior following the implementation of a trial project, involving the reallocation of street space to vulnerable road users on an arterial street in Trondheim, Norway. The principal data collection method used in this study was a web-based survey, soliciting responses from users of the street and residents living in the area of the infrastructural intervention. The results indicated an increase in the frequency of cycling and a route choice shift to the intervention section. However, limited relationships were established between the user characteristics and the change in their travel behavior.
4.1. Mode Choice
It was hypothesized that the reallocation of space to bicyclists would result in an increase in the number of cyclists along the trial project. In the evaluation report commissioned by the Norwegian Public Roads Administration, it was found that the number of cyclists had increased by between 95 and 120% [32
]. However, it should be noted that the pre-intervention counts were performed in June, during the university students’ summer vacation, which could have negatively impacted cycling volumes given that roughly 20% of Trondheim’s population consists of students.
The survey conducted by the authors was launched around a year after the changes took place, a time period that allowed people to adapt to the new infrastructure so that they could be asked about possible mode and route choice changes. Having this information allowed us to get more in-depth insights into the reasons for the increase in bicycle usage. However, it should be noted that long-term impacts can continue to change even after one year. In the British iConnect study, which evaluated the impacts of new active transport infrastructure, living in proximity of the intervention had been related to changes in activity levels only after the second year, but not at the one-year follow-up [35
]. Future research could consider reasons for changes in the long-term use patterns.
One of the possible explanations for the observed increase in bicycle volumes is that some of the users of the street changed their transport mode to bicycling. The respondents were asked for the mode they had used for their most usual trip on or in the vicinity of Innherredsveien. It was found that 106 respondents or 29% of the 369 who used a bicycle in the after period had been previously using a different mode. This corroborates the findings from Standen et al. [15
] from Sydney, Australia, where 40% of the cyclists riding on the path had been previously using a different transport mode.
Nearly half (43.4% (46)) of those respondents who switched to cycling had previously been using public transport, which is again similar to the results from Standen et al. [15
], where 59% of the intercepted cyclists reported that they had used public transport prior to the intervention. However positive the increase in cycling is, the shift from transit to bicycle use is not the primary intention of the intervention, as it does not contribute to limiting car use, a principal goal for cities in the Norwegian National Transport Plan [36
In addition to the public transport to bicycle modal shift, Section 3.2
reveals three other modal shifts comprising over 4% of the sample population: private cars to bicycle (4.5%), public transport to walking (4.3%), and walking to bicycle (4.2%). A clear indication of the intervention’s success in terms of cycling is that it has attracted substantial numbers of users from all other modes; however, the intervention can also be seen to improve walking conditions, since there are also substantial numbers of users switching from public transport to walking. This could be explained by the freeing up of space on the relatively narrow footpath on the north side of Innherredsveien after cyclists received their own bicycle path. Public transport is the mode that contributes most to the growth of both walking and cycling. Transport mode loyalty or mode retention rate is the percentage of a transport mode’s users who continue to use the same mode following the intervention. This is visualized in Figure 3
, in which cyclists are observed to have the highest retention rate before and after intervention (92%), followed by pedestrians (66%), car/motorcycle users (51%), and public transport users (44%).
An insight into these findings was provided by Börjesson and Eliasson [16
] in a study related to the cost-benefit analysis of cycling investments. The cross-elasticity between car and bicycle use in Sweden was estimated to be low in an indirect way. A stated preference experiment was conducted where the respondents had to choose between bicycle and their second-best mode. It was found that public transport was the alternative for 87% of the respondents, whilst car was the preferred second option for the remaining 13%. Börjesson and Eliasson [16
] concluded that amongst transport users who may shift their mode to cycling as a result of bicycle infrastructure improvements, only 10–15% of them would have previously used a car. However, it was noted that their conclusions are context specific, as Stockholm, where the study was performed, has a well-functioning public transport system.
Evidence of the low cross-elasticity between car and bicycle use was also provided by Song et al. [37
], who investigated the modal shift resulting from bicycle infrastructure implementation in three cities in the UK using a quasi-experiment panel study. They found that about 20% of the respondents had switched from driving to active transport modes, but also that a similar percentage had done the inverse shift.
Van Goeverden [38
] reviewed studies that had assessed bicycle infrastructural interventions in the Netherlands and Denmark. One of the aspects that was considered was mode choice, for which seven studies that had used travel behavior surveys were found and summarized. The results indicated that the shifts from driving to cycling had been minimal. Van Goeverden et al. [38
] noted that in some of the referenced studies from Denmark, the changes from public transport were significantly larger than the shift from car-use. These findings collaborate the results about mode shift from the current study.
It has also been of interest to find out whether the improved cycling conditions may have had an impact on users’ overall frequency of cycling. Only ten respondents or 1.4% of the sample reported to have decreased their cycling frequency, whilst 37.8% (272) reported an increase, 41% (295) had not changed frequency, and the remainder used a bicycle less than once per month in the after period and were therefore not asked this question. The large number of respondents that reported an increase in their overall cycling frequency (47% of those who had received the question) corroborates findings from a similar study in the USA. The US study evaluated eight newly implemented protected bicycle lanes, with nearly half (49%) of the respondents reporting an increase in personal cycling frequency along the protected bicycle lanes, whilst nearly a quarter (24%) reported an increase in their overall frequency of cycling as a result of the protected lanes [14
The survey also reveals that 10 respondents amongst those 272 who reported an increase in cycling frequency (or 1.4% of the total sample) had begun using a bicycle as a direct result of the trial project. This finding is significantly lower than the aforementioned protected bicycle lane study, in which the proportion of users who began cycling after the interventions ranged from 6 to 21% across the eight examined locations in the USA (10% on average) [14
]. Different contextual conditions for cycling in Norway and the USA could explain some of this variation, as cycling rates in Norway are generally higher (8% versus 1% of all trips nationally) and the provision of bicycle infrastructure is more commonplace [25
The fact that the model using the demographic variables given in Table 2
could not predict the increase in the frequency of cycling of the respondents can be explained by the lack of a significant difference between the sample’s demographic subsets compared to the population as a whole. As mentioned in Section 3.2
, the only significant difference found was for the variable “age” (p = 0.037). This suggests that young people (less than 34 years of age) exhibit a significantly higher propensity to increase their frequency of cycling in response to the street redesign than the older respondents.
The high reported increase in cycling and low variation in change of cycling frequency across the sample’s different demographic groups can be indicative of a trial project that appeals equally to all users. In other words, women and men exhibit approximately equal propensity to increase their frequency of cycling after an implementation of this kind of infrastructure. This result differs from what was reported by Standen et al. [15
] for Sydney, Australia, where commuters who had changed mode to cycling in response to the opening of a separated bicycle path were more likely to be female. Similar conclusions were made in a study that used survey data from the evaluation of protected bicycle lanes in five cities in the USA [40
]. Dill et al. [40
] concluded that female cyclists have a greater propensity to increase their overall cycling frequency because of protected bicycle infrastructure. The discrepancy between the findings of the current study and those from the USA and Australia may again be explained by contextual differences between Norway and those countries. Utility cycling is more male-dominated in the USA (75% male) and Australia (79%), whilst in Scandinavian countries, it is nearly equal, with males making 56% of cycling journeys in Norway [25
4.2. Route Choice
A possible reason for the observed increase in bicycle volumes on Innherredsveien is that existing cyclists decided to change their route so that they could use the trial section. To test this hypothesis, survey respondents were asked to draw the route they had used for their most usual trip on or near Innherredsveien both before and after the street intervention. The length of the trial project section of Innherredsveien that was utilized by the participants before and after the changes were implemented was one means to approximately quantify the route choice changes. Participants who increased their utilization of the street post-intervention can be thought of as being attracted to the intervention. This was expected of cyclists who previously rode on alternative streets or without bicycle infrastructure in Innherredsveien and had now received a separated bicycle path. The opposite applies for those who decreased their utilization of the intervention (which could be expected for car users who now are forbidden from driving the full length of the street).
The significant increase in mean distance that bicyclists were riding on the trial section (379 m; 95% CI, 242 to 517 m) is indicative of an attraction to the Complete Streets initiative amongst cyclists. This supports findings from existing literature, in which between 24 to 48% of bicyclists have been found to change their routes to use the newly implemented bicycle infrastructure [14
The change in bicycle route choice is illustrated in Figure 4
, in which it is apparent that the intervention street Innherredsveien greatly increases in popularity amongst cyclists at the expense of neighbouring parallel streets. The parallel streets are mostly part of the existing bicycle network shown in Figure 1
and witness a decrease in usage (depicted in red). This change in route choice preference, also called route substitution, is a major contributing factor to the increase in volumes in the trial project. Figure 4
illustrates all bicycle journeys from the two time periods made by the panel respondents. Therefore, those cyclists who used a different mode earlier have only one route depicted in the figure. Thus, the overall increase in cycling is also illustrated to some degree in Figure 4
through the larger number of bicycle trips made post-implementation, which is represented by generally thicker lines in green than red (although this is not the primary intention). Given the clear impact the intervention has had on the route choice of existing bicycle users, it is recommended that intervention section traffic counts alone are not used to assess the impacts on bicycle mode choice (as was done in NPRA’s evaluation report). Should a traffic counting approach be used, parallel streets should always be considered at a minimum to gauge the extent to which route substitution is present [20
]. It is therefore recommended that traffic counts on an intervention street should not be used to assess the impacts on mode choice, given route choice is clearly shown to be affected as well.
The provision of high quality infrastructure on major arterials to the city center is important as it offers cyclists a more direct and faster route, while at the same time improving their perceived feeling of safety. Interventions like this project that deliver improved connectivity are likely to be of more value to users than projects which only improve isolated road segments. By offering a holistic bicycle network with minimal discontinuities, users with a lower tolerance for traffic interaction are empowered to switch to cycling.
Regarding the route choice of motorized vehicles, the evaluation report concludes that 16% of the former motorized traffic (9500 vehicles/day in total) has shifted from Innherredsveien to the Strindheim tunnel, which is a desired consequence of the project [32
]. This substitution of route assists in the minimization of congestion, something that is especially important for buses, for which the street is a key corridor. The tunnel was built in 2014 specifically to relieve the intervention street of traffic issues, and the change could therefore have been made upon the tunnel’s opening. The availability of bypass alternatives is an important prerequisite for the implementation of road restrictions in central urban areas, such as the Complete Streets solution investigated in this study. However, the report also notes that 500 vehicles/day or 5% of the former traffic volume (9500 vehicles/day) have transferred to streets other than the bypass tunnel, potentially due to having a destination not well connected by the tunnel following the through-traffic closure. In San Francisco, the replacement of two out of four traffic lanes with marked bicycle lanes was found to reduce car volumes by 10%, with the difference redistributed to parallel arterials [8
]. Sallaberry did not, however, investigate whether a mode shift had occurred [8
Despite the total decrease in motorized volume, it has been found that 800 vehicles/day did not conform to the prohibition [32
], something which could be regulated by the authorities in order to reduce traffic volumes to the extent intended.
Considering that participants could only draw one route and one mode for each time period, the mapped data in Figure 4
only reflects the primary mode and route for transport through the neighborhood. Shorter or less frequent journeys made by participants are thus not well represented by the collected data, and the low numbers of routes drawn with sufficient quality for mapping (from 211 of 719 participants) in both time periods limited the analysis scope.
4.3. Policy Implications
The policy implications of the study are related to the outcomes of planning process issues, the detected mode shift changes, and the public approval of the project.
Concerns have been expressed by various stakeholders during the political planning process. For instance, the public transportation authority was concerned that the reduction in the number of traffic lanes in Innherredsveien would result in a reduced capacity and therefore unnecessary delays for public transport. There have also been disagreements between city and county politicians concerning the role of the street. Therefore, after over four years of debate, the project was implemented temporarily so that it could be reversed in case it did not deliver the expectations.
The evaluation report revealed that the concerns were not realized as the average travel time for buses had decreased and the motorized traffic on the intervention street had reduced [32
]. Motorized traffic was to a large extent shifted to the bypass tunnel as originally intended. At the same time, the trial project provided a considerable improvement in the conditions for vulnerable road users and a resultant significant increase in their numbers.
The survey sample’s four largest modal shifts all involved a change in mode to cycling or walking. This mode shift, however, was not ideal from a sustainability perspective as the main contributor to the increase in the bicycle share was shifts from public transport rather than private car users. This suggests that initiatives such as the one presented in this study are not sufficient alone to achieve the sustainability goal of reduced car use. Such a shift away from cars is desirable, however, from an urban planning perspective due to improved public health outcomes, and reduced congestion and local air pollution. At the global scale, a modal shift away from cars contributes to decarbonizing the transportation sector, which is critical given the ever-mounting pressure to act on climate change. Other policy tools are likely needed to achieve a modal shift away from cars, such as the removal of parking places, higher parking fees, road pricing, higher fuel taxes, etc. The challenge remains, however, to achieve a combination of policy measures that are politically acceptable.
As the initial intention of the trial Complete Streets project was to create improved transport conditions and a more attractive street environment for the residents and other users of the street, it was expected that the general public would be generally satisfied with the changes that have taken place. The public response towards similar redistributions of space from motorized vehicles to bicyclists has been found to be generally positive in the US [8
]. The high approval rate of the project amongst survey participants (87%) and positive contribution towards sustainability goals can have policy implications, such as permanently redesigning the street in favor of vulnerable road users or encouraging other cities to consider similar initiatives.
The temporality of trial projects can be advantageous when political disagreement stops the initiative from being implemented, and has in this case demonstrated for stakeholders that the project did not result in unacceptable disadvantages for any transport modes. This does, however, require the trial project to be sufficiently well planned to ensure that the temporality itself does not adversely affect the traffic outcome and thus poorly represent a future permanent solution. It is, however, an additional cost in terms of time and materials to rebuild a road twice, even if the temporary solution has less impact than the permanent solution (presuming it is built).
It was assumed that the increase in the utilized length of the trial project section of Innherredsveien has been due to cyclists changing their route or mode because of the improved conditions of the street. However, there may have been other reasons for some of the respondents’ decision to change their route, such as a change of the most common trip origin or destination.
Another aspect of the current study that can have impacts on the results is that only the most common trips on Innherredsveien were analyzed, while the rest of their journeys were not studied. Future intervention investigations of route and mode choice can look into all the trips that users make in connection with changes in transport infrastructure. Heinen et al. [44
] investigated the modal shifts resulting from new bicycle infrastructure in Cambridge (UK) using a four-year quasi-experimental cohort study and found that partial modal shifts were more common than full modal shifts. It is reasonable to assume that people use different modes for their different trips and hence studying only the most usual one does not give complete information about the changes in users’ behavior. It should be noted that this study is based on the post-implementation evaluation of a single (and temporarily deployed) Complete Streets project, with its associated contextual factors, and is therefore only indicative of the impacts of this type of trial project. Further evaluations of trial projects are recommended so that a more complete picture of the effects of such projects can be acquired.
It should also be noted that cyclists were overrepresented in the sample, with 65% (468) of the respondents reporting that they use a bicycle at least once a week during the warmer half of the year. Having in mind that the cyclists are one of the major beneficiaries of the street reconfiguration, self-selection bias may have influenced the results of the study. The overrepresentation of cyclists was in part due to the nature of the recruitment process for the survey. Besides the delivery of flyers to households in the area around implementation and the other alternatives mentioned in Section 2.2
, the recruitment also included advertisement via a social media interest group for cycling in Trondheim and the manual distribution of flyers to bicyclists using the cycling path.
Respondents were asked to recall details from their travel behavior from a year earlier, before the implementation, which is associated with a poorer accuracy compared to journeys conducted more recently [29
]. Had it been known some time in advance of the intervention commencement, a before-and-after cohort or cross-sectional design could have been applied to achieve more accurate responses.