Shared space: Motorists ’ perspective

A


Introduction
Shared spaces are intended to constitute an attractive urban space for all road users.There is no uniform definition of shared space, but it usually refers to a surface that is shared by unprotected and protected road users (Johansson & Rosén, 2011;Reid et al., 2009;Zalewski & Kempa, 2019).In Sweden, the shared spaces do not have to be, but are often, regulated as pedestrian areas, meaning that all traffic must behave on the pedestrians' terms.Vehicles, including bicycles, may, for example, not be driven or ridden faster than walking speed, and drivers must give way to pedestrians.According to Johansson and Rosén (2011), traffic safety is ensured by inducing a degree of uncertainty, which in turn is thought to prompt lower speeds and greater caution.This uncertainty is created by blending protected and unprotected road users by, for example, removing traffic markings and curbs.Most shared spaces in Sweden are, however, found in municipal locations that lack specific national design regulations, other than some basic criteria, and municipalities often depart from the original idea of avoiding the traditional division between protected and unprotected road users.This may be achieved by replacing markings and curbs with alternative designs such as paving or lighting installations, concrete traffic barriers, or flowerpots.The question is how these local solutions affect road users' experiences, as well as the traffic safety thought to be attained by blending protected and unprotected road users.
The experience of shared spaces has previously been studied to some extent (Cowan et al., 2018;Earl et al., 2018;Earl et al., 2019;Kaparias et al., 2012;Wallén Warner et al., 2022).To the authors knowledge, these studies have all focused on the pedestrian perspective and most often on specific groups such as people with cognitive disabilities or of different ages.It is therefore of interest to enhance our knowledge by focusing on motorists' perspective in a location already studied from a pedestrian perspective.In 2022, Wallén Warner et al. examined the effect of large flowerpots, used to recreate the traditional division between protected and unprotected road users, on pedestrians' experience of, and behaviour within, a shared space (e.g.Fisherman's Square) in the small town of Västervik in southeast Sweden.Q-methodology (Stephenson, 1953;Watts & Stenner, 2012) was used to examine pedestrians' subjective experience of the square without and with the large flowerpots deployed, while eye-tracking methodology was used to examine the large flowerpots' effect on pedestrians' visual scanning behaviour.The results showed that even with large flowerpots deployed, all age groups (young, middle-aged, and elderly) still experienced a degree of uncertainty as they thought that motorists did not always follow the traffic rules that applied to them.At the same time, the deployment of large flowerpots did not change the young or the middle-aged pedestrians' visual scanning behaviour.The elderly pedestrians did, however, change their visual scanning behaviour.Without large flowerpots deployed, the elderly pedestrians glanced more often at traffic-relevant objects than non-trafficrelevant objects.With large flowerpots, they glanced equally often at traffic-relevant and non-traffic-relevant objects.These results are somewhat concerning as they suggest that the deployment of large flowerpots may increase the risk of elderly pedestrians' involvement in a traffic incident or accidents with motor vehicles.
To enhance our knowledge of the motorists' perspective a follow-up study was conducted to further examine the effect of replacing markings and curbs with alternative designs, such as large flowerpots.In this follow-up study, Q-methodology was used to examine motorists' subjective experiences i.e. their views on a shared space and on their interaction with other road users there.This method was combined with video analysis to examine motorists' speed and placement of their vehicles within a shared space.

Aim
This follow-up study aimed to examine the effects of external factors, such as alternative design and pedestrian density, on motorists' subjective experiences, and the speed and placement of their vehicles within a shared space.

General method: The shared space
The study was conducted in the spring of 2022 in the small town of Västervik in southeast Sweden.In the central part of town, there is a small square called Fisherman's Square (see Fig. 1).The square is regulated as a pedestrian area, meaning that all traffic must behave on the pedestrians' terms.Västervik Municipality has, for many years, deployed large flowerpots on the square to create a degree of division between protected and unprotected road users.These flowerpots are made of terracotta, are about 100 cm high and 60-70 cm in diameter, and have flowers planted in them.To enable the current study to take place, the Västervik Municipality moved the large flowerpots on the square to accommodate our research schedule.Another shared space, the junction between the roads Drottning Kristinas väg and Lindstedtsvägen in the Swedish capital of Stockholm, was also examined within the same project.The results of that examination are not presented in the current paper.

Assessment 1: Motorists' subjective experience
Q-methodology (Stephenson, 1953) was used to examine motorists' subjective experience of the shared space with two alternative designs: without and with large flowerpots deployed.

Participants
By noting the number plates of vehicles crossing the square, it was possible to get the owners' contact details from the Swedish Transport Agency and send recruitment letters by mail.These letters were supplemented with posters on the square, flyers put under the windscreen wipers of parked vehicles in close vicinity of the square, advertising in social media, as well as personal contact with motorists crossing the square, people on site, and people employed by the insurance company Länsförsäkringar in Västervik that lent us the space where we conducted the study.Regardless of the communication channel used, interested persons were invited to contact our recruiter by telephone or e-mail, or alternatively to leave their contact details on a web page created for the purpose.
The recruitment was conducted without and with the large flowerpots deployed on the square.Before anyone was asked to participate in the study, we ensured that they had crossed the square during the right study conditions (without or with the large flowerpots deployed).
During the sorting of cards (see 3.1.3.Administering the Q-sort), the participants were offered coffee and snacks.After completion in the study, they were offered a choice between one cinema ticket or five lottery tickets, as compensation for their cooperation.
A total of 32 motorists aged 19-79 years participated in the study.Half of them participated without the large flowerpots deployed on the square, while the other half participated with large flowerpots deployed.Table 1 shows the number of participants, as well as their gender, age, and number of years they had held a driver's license.All participants regularly crossed the square in their vehicles, while most of them also crossed it by foot on a regular basis.Less than half of the participants regularly crossed the square by bicycle.

Material
The material comprised a pack with 44 cards, and a scale ranging from − 6 (strongly disagree), through 0 (neutral), to +6 (strongly agree).Each card contained a statement (see Table 2) thought to reflect the motorists' experience of the square, or of traffic situations in general.Each statement was combined with a representative image (see Fig. 2).The cards were based on material described by Earl et al. (2018), andWallén Warner et al. (2022) but the statements were rephrased to suit the motorists' perspective.Before the data collection started, a small pre-study was conducted with two of our colleagues.This was done to test the statements but also to train the colleagues in the Q-methodology before the data collection started.

Administering the Q-sort
The data collection without large flowerpots on the square was conducted between 4 and 8 April 2022, while the data collection with large flowerpots was conducted a few weeks later, between 25 and 29 April 2022.Although the material used was adjusted, the actual procedure for administering the Q-sorts was identical to that described in the previous study with pedestrians (Wallén Warner et al., 2022).
The participants first completed a consent form, as well as a few background questions about gender, age, and number of years they had held a driver's license.They were also asked how often they crossed the square by car, on foot, and by bicycle.The participants were then presented with the 44 cards and instructed to read all the statements and sort them into three provisional ranking piles: first, all statements that they definitely agreed with; second, all statements that they definitely disagreed with; and third, all statements that they felt indifferent or unsure about.When the participants had completed making the three piles, the experiment leader placed the first pile above the scale towards the participants' right-hand side, and hence towards the positive end of the distribution.The second pile was placed above the scale towards the participants' left-hand side, and hence towards the negative end of the distribution.Finally, the third pile was placed above the scale directly in front of the participant, and hence in line with the neutral area of the distribution.
The participants were then asked to spread all cards from the first pile (definitely agree) in front of them so that they could see all of them at the same time.They were then shown how to allocate each card starting with the statements they agreed with the most (to the far left) followed by the two statements they agreed with slightly less.They were also informed that the order within the columns was entirely irrelevant.When they had finished allocating the cards in the first pile, they were asked to repeat the procedure for the second

Table 3
The largest factor arrays (±4 to ±6) for the two viewpoints in the dataset for participants who sorted the cards without large flowerpots deployed on the square.
All 44 statements are presented in Table 2.
pile (definitely disagree), and finally the third pile (indifferent or unsure) to complete the Q-sort.The shape of the completed Q-sort (Fig. 2) should resemble a normal distribution turned upside down, with a continuum ranging from -6 (strongly disagree), through 0 (neutral), to + 6 (strongly agree).Like the statements, the shape of the Q-sorts, as well as the scale, was based on Earl et al. (2018).
The study ended with a post-interview that included questions about why the participants had chosen the particular statements they had as ± 6, whether they had not understood any statements, and whether they thought that any other statements should have been included.The interview also included questions on participants' general thoughts about the square.The results of these postinterviews improved and facilitated the interpretation process and confirmed the clarity and relevance of the statements.The specific results are, however, not presented in the current paper.

Analyses
The PQMethod 2.35 statistical program (https://schmolck.org/qmethod/pqmanual.htm)was used to analyse the data.The Q-sorts were divided into two datasets: one for the participants who sorted the cards without large flowerpots on the square and one for the participants who sorted the cards with large flowerpots on the square.
In Q factor analyses, as opposed to R factor analysis, the participants are seen as the variables.The Q-methodology therefore positively embraces studies with few participants, i.e., fewer than the number of statements (Watts & Stenner, 2012), which in this study was 44.However, it is important to remember that the aim of the Q-methodology is not to generalize to a wider population but to establish the existence of particular viewpoints.
In the dataset for the participants who sorted the cards without large flowerpots on the square, 16 Q-sorts were inter-correlated, and seven factors were extracted using Brown-centroid extraction.Two factors with eigenvalues of > 1 were then extracted and rotated (using varimax rotation); these explained 58 % of the study variance.Twelve (75 %) of the 16 Q-sorts loaded significantly on one of these two factors.The remaining four Q-sorts were excluded from further analysis as they did not load significantly on either factor.
In the dataset for the participants who sorted the cards with large flowerpots on the square, 16 Q-sorts were inter-correlated, and seven factors were extracted using Brown-centroid extraction.Two factors with eigenvalues of > 1 were then extracted and rotated (using varimax rotation); these explained 56 % of the study variance.Eleven (69 %) of the 16 Q-sorts loaded significantly on one of these two factors.The remaining five Q-sorts were excluded from further analysis as they did not load significantly on either factor.
As the same 44 statements were used, factor loadings of ± 0.39 were significant at the p = 0.01 level for both datasets.Q-sorts that load significantly on the same factor are those that share a similar sorting pattern.The factor exemplars were then merged, and a factor array was calculated for each factor using weighted averaging.Being a merged average, the factor arrays look like a single complete Qsort.
In 3.2 Results, the factors, here called viewpoints, are presented together with demographic information for those participants whose Q-sorts loaded significantly on each factor, along with the rankings of relevant items.For example, (18: -6) indicates that item 18 (Cyclists follow the traffic rules on the square) is ranked in the -6 (strongly disagree) position in the factor array Q-sort of the factor.Finally, consensus statements with no significant difference in scores across the factors are presented.See Watts and Stenner (2012) for a more detailed description of the Q-methodology for extracting, rotating, and interpreting factors.

Without large flowerpots deployed
Two viewpoints explained 58 % of the variation in Q-sorts made by the participants who sorted the cards without large flowerpots deployed on the square.

Viewpoint 1 without large flowerpots.
Viewpoint 1 without large flowerpots: I feel secure on the square and know how to act explained 41 % of the variation in Q-sorts and was defined by 9 (56 %) of the 16 participants.Two of the participants were women and seven were men.They were aged 22-79 years, with a mean age of 50 years.
The participants sharing this viewpoint felt confident making eye contact with people (see Table 3, Statement 28: Factor array + 4), and knew where (24: +4) and when (20: +5) to cross the square.They also knew when they should stop to let pedestrians (4: +6) and cyclists (5: +5) cross the square.They felt secure when they crossed the square themselves (29: +4), did not perceive the square as dangerous (26: − 6), and did not stay away from it (32: − 5).Furthermore, they did not agree that pedestrians (1: − 4) and cyclists (2: − 4) always stopped and let them cross the square, or that other drivers always followed the right-hand rule and let them cross (3: − 4).Finally, they did not think that the large flowerpots were in their way when driving (8: − 5).

Viewpoint 2 without large flowerpots.
Viewpoint 2 without large flowerpots: I prefer it when drivers and pedestrians are separated from each other explained 17 % of the variation in Q-sorts and was defined by 3 (19 %) of the 16 participants.Two of the participants were women and one was a man.They were aged 26-66 years, with a mean age of 48 years.
The participants sharing this viewpoint knew how to act towards pedestrians (see Table 3, Statement 36: Factor array + 6; 37: +4) and towards other drivers (40: +5; 41: +4) at pedestrian or road crossings both with and without traffic lights.They also knew when they should stop to let pedestrians cross the square (4: +4), but still thought that drivers and pedestrians should be separated from each other (12: +5).They did not like it when there are lots of road users around (30: − 5) but they did not feel that other drivers were driving too close to them on the square (35: − 4).They did not agree that pedestrians made eye contact with them on the square (21: − 4) nor that they always stopped and let them cross the square (1: − 5).Furthermore, they did not think that cyclists followed the traffic rules on the square (18: − 6).Finally, they did not think that the large flowerpots were in their way when driving (8: − 4).

Consensus statements.
There were five consensus statements with no significant difference in scores between the two viewpoints.Participants sharing both viewpoints thought it was easy for themselves (11: +3) and for other drivers (16: +2) to see drivers approaching on the square.However, they did not think that other drivers followed the traffic rules on the square (19: − 1) and they did not know when a cyclist was going to stop to let them cross the square (7: − 2).Finally, they did not think that large flowerpots made it easier or harder for them to know where to drive (27: 0).

With large flowerpots deployed
Two viewpoints explained 56 % of the variation in Q-sorts made by the participants who sorted the cards with large flowerpots deployed on the square.
3.2.2.1.Viewpoint 1 with large flowerpots.Viewpoint 1 with large flowerpots: I know how to act on the square and I don't think the large flowerpots are in my way when I drive explained 33 % of the variation in Q-sorts and was defined by 7 (44 %) of the 16 participants.Three of the participants were women and four were men.They were aged 24-75 years, with a mean age of 50 years.
The participants sharing this viewpoint knew when (Table 4, Statement 20: Factor array + 4) and where (24: +6) to cross the square, and perceived that it was easy to see other drivers approaching on the square (11: +4).They felt sure about how to act when crossing the square (31: +4), knew which traffic rules applied on the square (44: +5), and did not agree that the traffic rules were hard to understand (43: − 5).They knew when they should stop to let cyclists cross the square (5: +5) but did not agree that cyclists followed the traffic rules on the square (18: − 4) nor that they always stopped to let them cross the square (2: − 4).They did not perceive the square as dangerous (26: − 4) and did not stay away from it (32: − 6).Finally, they did not think that the large flowerpots were in their way when driving (8: − 5).

Viewpoint 2 with large flowerpots. Viewpoint 2 with large flowerpots: I know how to act at pedestrian, bicycle, and road crossings, but
it is more difficult on the square explained 23 % of the variation in Q-sorts and was defined by 4 (25 %) of the 16 participants.All four participants were women.They were aged 22-65 years, with a mean age of 43 years.
The participants sharing this viewpoint felt confident making eye contact with people (Table 4, Statement 28: Factor array + 4) and knew how to act towards pedestrians (36: +6; 37: +5) and other drivers (40: +5; 41: +4) at pedestrian or road crossings with and without traffic lights.They also knew how to act towards cyclists at bicycle crossings with traffic lights (38: +4).However, they did not like it when there were lots of road users around (30: − 6), nor did they agree that it was easy to see pedestrians approaching on the square (9: − 4).They thought that other drivers followed the traffic rules (19: − 5), including the right-hand rule (3: − 4), on the square.Finally, they did not agree that more zones regulated as pedestrian areas would make it easier for them (42: − 4), but they did not avoid the square (32: − 5).

Consensus statements.
There were two consensus statements with no significant difference in scores between the two viewpoints.Participants sharing both viewpoints did not think that other drivers were driving too close to them on the square (35: − 2).Furthermore, they did not know when a cyclist was going to stop to let them cross the square (7: − 3).

Assessment 2: Motorists' speed and placement of vehicles on the square
Video recordings were used to examine motorists' speed and placement of their vehicles, without and with large flowerpots deployed on the square.Even small differences in vehicle speed can have a significant effect on the relative fatality risk in the case of an accident (Kröyer et al., 2014, Elvik et al., 2019).Furthermore, Karndacharuk et al. (2014) have shown that vehicle speed changes with the number of interactions.They conclude that more pedestrian-vehicle interactions in shared zones result in lower speeds.Vehicle speed was therefore measured at three different levels of pedestrian density (low, moderate, and high) based on a pedestrian count made from the video.

Video recordings
Video footage was recorded on weekdays between 7 am and 7 pm for a period of 7 days without large flowerpots (28 March-4 April 2022) and for 10 days with large flowerpots (19-29 April 2022).Miovision Scout cameras were employed for this study, with a resolution of 720×480 pixels.

Analyses
The video analysis consisted of two steps.In the first step, a pedestrian count was manually taken during 10 minutes of each hour for 1 day of the video recordings, to estimate the number of pedestrians at the locations throughout a weekday.Each hour was then categorized based on quartiles, with all hours in Q1 considered having low pedestrian density, with Q2 and Q3 having moderate

Table 4
The largest factor arrays (±4 to ± 6) for the two viewpoints in the dataset for participants who sorted the cards with large flowerpots deployed on the square.
All 44 statements are presented in Table 2. pedestrian density, and Q4 having high pedestrian density.Table 5 shows which hours belong to which category, along with their estimated pedestrian counts.
In the second step of the analysis, 25 vehicles were randomly selected in each category (low, moderate, high) for each design (without/with large flowerpots), and trajectories were manually created using T-Analyst software (Johnsson et al., 2018).This software allows the user to first make a video calibration capable of translating pixel coordinates to coordinates in meters, and to then manually create trajectories tracking road users in the video recording.The maximum speed of each vehicle was taken as the highest measured speed throughout the path.The placement of each vehicle on the square was estimated as the individual difference in the mean path taken by all vehicle trajectories without and with large flowerpots deployed on the square (with regard to the direction of travel).Using this definition, a higher mean difference indicates that a vehicle is further away (more spread out) compared to the mean path.
To analyse motorists' speed and placement of their vehicles when crossing the square multivariate general linear model analyses of variance (ANOVAs) were computed.For motorists' speed a two (without/with large flowerpots) by three (low/moderate/high pedestrian density) univariate factorial, between-participant ANOVA was computed.To analyse the motorists' placement of their vehicles on the square a two (without/with large flowerpots) by three (low/moderate/high pedestrian density) by two (west/east direction) multivariate factorial, between-participant ANOVA was computed.

Motorists' speed
Table 6 shows the mean maximum speed of the vehicles (and standard deviation) depending on design (without/with large flowerpots) and on pedestrian density (low/moderate/high pedestrian density).Note that the total number of observed vehicles was 150 (25 per category).The ANOVA showed a significant main effect of design (without/with large flowerpots; F (1, 144) = 10.34;p < 0.01) whereby the vehicles' mean maximum speeds were significantly lower with (M = 17.2 km/h), compared to without (M = 18.8 km/h) large flowerpots deployed.There was also a significant main effect of pedestrian density (low/moderate/high pedestrian density; F (2, 144) = 4.64; p < 0.01).Pairwise comparisons revealed that vehicles' mean maximum speed was significantly lower with high (M = 17.2 km/h) compared to low (M = 19.0km/h) pedestrian density (p < 0.01), while there was no significant difference between vehicles' maximum speeds with moderate compared to low pedestrian density, or with moderate compared to high pedestrian density.The interaction effect was not statistically significant.
The data represent the mean maximum speed of the vehicles in km/h (and standard deviation).

Motorists' placement of their vehicles on the square
Fig. 3 shows the motorists' trajectory through the video recorded area.The thick lines show the mean path of the vehicles, with each individual track going west shown as thin green lines and each individual track going east shown as thin blue lines.
The ANOVA showed a significant main effect of design (without/with large flowerpots; F (1, 145) = 6.61; p < 0.05) and direction (west [green]/east [blue] direction; F (1, 145) = 26.97;p < 0.001) but not of pedestrian density (low/moderate/high pedestrian density).Table 7 shows the mean difference to the vehicles' mean path (and standard deviation) depending on design (without/with large flowerpots) and on direction (west [green]/east [blue] direction).Note that the total number of observed vehicles was 150 (see Table 7 for the number of observations per category).Pairwise comparisons revealed that the average difference to the vehicles' mean path was significantly shorter with (M = 0.7 m), compared to without large flowerpots deployed (M = 0.9 m).Furthermore, the mean difference to the vehicles' mean path was significantly shorter travelling west (green; M = 0.4 m) compared to east (blue; M = 1.0 m).The 2-way and 3-way interaction effects were not statistically significant.

General discussion
This follow-up study aimed to examine the effects of external factors, such as alternative design and pedestrian density, on motorists' subjective experiences, and the speed and placement of their vehicles within a shared space.
A basic idea underlying shared spaces is to enhance traffic safety by inducing a degree of uncertainty via the removal of traditional divisions between protected and unprotected road users (Johansson & Rosén, 2011).Most shared spaces in Sweden are found in municipal locations that lack specific national design regulations other than some basic criteria.In Västervik, as in many other municipalities, markings and curbs have been replaced by large flowerpots, thus departing from the original idea of avoiding the traditional division between protected and unprotected road users.The Q-sorts made by the participants who completed the study without or with the large flowerpots deployed on the square resulted in two viewpoints per group.Furthermore, for both designs, one of the viewpoints showed that the motorists experienced a degree of uncertainty, and preferred the traditional division between protected and unprotected road users (Viewpoint 2 without large flowerpots: I prefer it when drivers and pedestrians are separated from each other; and Viewpoint 2 with large flowerpots: I know how to act at pedestrian, bicycle, and road crossings, but it is more difficult on the square).The motorists' subjective experiences thus reflect the uncertainty previously found among pedestrians on the same square (Wallén Warner et al., 2022).
The previous study (Wallén Warner et al., 2022) also showed that elderly pedestrians glanced more often at traffic-relevant objects than non-traffic-relevant objects without large flowerpots deployed on the square.With large flowerpots, they glanced equally often at traffic-relevant and non-traffic-relevant objects.This causes some concern, as it suggests that elderly pedestrians focused less on traffic when the traditional division between protected and unprotected road users was recreated using large flowerpots.The increased risk associated with the lack of focus might however be counteracted by the reduction in vehicles' mean maximum speed and the decreased distribution of vehicles over the square found in the current study when the large flowerpots were deployed on the square.
Another factor shown to affect the mean maximum speed of the vehicles was the pedestrian density.Regardless of design (without or with large flowerpots deployed on the square), the vehicles' mean maximum speed was lower with high pedestrian density than with low pedestrian density.It should, however, be noted that the total number of pedestrians was fairly low even in the high density condition.In locations with a substantially larger number of pedestrians the results might be different.The current results showing a lower mean maximum speed with high pedestrian density does, however, support Jacobsen's (2003) theory of 'safety in numbers'.This theory implies that the likelihood for a collision between motorists and pedestrians decreases when the number of pedestrians increases.In addition, the likelihood of being seriously injured, in the event of an accident, also decreases as speed decreases.In order to achieve high pedestrian density, it is however crucial that pedestrians feel safe and secure.As mentioned above, deploying large flowerpots seemed to decrease the uncertainty felt by the pedestrians (especially the elderly) compared to no large flowerpots on the square (Wallén Warner et al., 2022).The data represent the mean difference to the mean path of the vehicles in meters (and standard deviation).The number of observed vehicles is presented for each category.
Taken together, the deployment of large flowerpots reduced the vehicles' mean maximum speed and decreased the distribution of vehicles over the square, while high pedestrian density only reduced the vehicles' mean maximum speed.This makes some intuitive sense since the large flowerpots always constituted additional obstacles making the motorists drive a bit more carefully, both regarding their speed and placement of their vehicles.Using the same logic, increased pedestrian density are expected to reduce motorists' speed but only alter their path over the square in the infrequent case of close interactions with pedestrians.
As with all studies, the current design has its limitations.The participants were divided into two groups, whereby the cards with statements were sorted either without or with the large flowerpots deployed on the square.All participants were, however, local residents of Västervik, and as the large flowerpots are deployed on the square for most of the year, they all had more experience of the square with than without the large flowerpots deployed.This limitation could possibly be overcome by recruiting participants without previous experience of the location studied (e.g., people from other areas) but it would still be hard to control for the participants previous experience of other shard spaces or similar design solutions.Furthermore, even though the Q-methodology positively embraces studies with few participants, i.e., fewer than the number of statements (Watts & Stenner, 2012), the viewpoints found cannot necessarily be generalized to a wider population.When appropriate, the Q-methodology would therefore benefit from being combined with more quantitative approaches such as questionnaires distributed to a larger sample.Finally, the number of vehicles included in the video analysis was also limited, and the pedestrian density was estimated for different time periods without taking the actual number of pedestrians in close proximity to each individual vehicle into account.Increasing the number of vehicles studied as well as using a more nuanced definition of pedestrian density could perhaps reveal more insight into how motorists interact with pedestrians and how these interactions are affected by the deployment of large flowerpots.

Conclusions
Shared spaces, such as the square in Västervik, are intended to constitute an attractive urban space for all road users.Traffic safety is enhanced by blending protected and unprotected road users, to create a degree of uncertainty, which is thought to prompt lower speeds and greater caution.However, it is important that all road users feel safe and secure on the shared space.The results from the current and previous (Wallén Warner et al., 2022) studies show that motorists and pedestrians prefer the traditional division between protected and unprotected road users, which in this case was recreated by large flowerpots deployed on the square.In addition, the results showed that the deployment of large flowerpots, as well as high pedestrian density, decreased the vehicles' mean maximum speed.The deployment of large flowerpots also decreased the distribution of vehicles over the square.These results indicate that the recreation of some type of division between protected and unprotected road usersfor example by deploying large flowerpotsis preferred.Further studies are needed to ensure that this solution works for all groups of road users, including cyclists and e-scooter riders not included in the current nor in the previous studies.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Fisherman's Square, in the small town of Västervik in southeast Sweden.The large flowerpots deployed provide a degree of division between protected and unprotected road users.
Fig. 2. Example of a completed Q-sort.

Fig. 3 .
Fig. 3.The motorists' trajectory through the video recorded area, with (left) and without (right) large flowerpots deployed on the square.

Table 1
Number of participants, gender, age and number of years driver's license held.

Table 5
The pedestrian count estimate for each hour and their respective category.

Table 6
Mean maximum vehicle speed, according to design and pedestrian density.

Table 7
Average difference to the vehicles' mean path, according to design and direction of travel.