Crowdsourcing the influence of physical features on the likely use of public open spaces

Miliasetal. Computational Urban Science (2024) 4:15 https://doi.org/10.1007/s43762-024-00126-0 ORIGINAL PAPER Open Access © The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Computational Urban Science Crowdsourcing theinfluence ofphysical features onthelikely use ofpublic open spaces Vasileios Milias1* , Roos Teeuwen1, Alessandro Bozzon1 and Achilleas Psyllidis1 Abstract The configuration of public open spaces plays a crucial role in shaping how different people use them. Nevertheless, our understanding of how the physical features of public open spaces influence the activities conducted within them, and the extent to which this impact differs across various individuals and population groups, is currently limited. In this study, we explore how the physical characteristics of public open spaces influence the likelihood of use among individuals, spanning different age and gender groups. By employing crowdsourcing, street-level imagery, statistical comparisons, and reflexive thematic analysis we uncover significant variations in the suitability of public open spaces for distinct activities, such as socializing or exercising. Greenspaces emerge as the preferred choice for almost all activities, whereas streets are consistently rated as the least suitable. Additionally, we identified various characteristics that influence the activities people are likely to engage in. These include the size of the space, the presence of seating, natural elements such as vegetation or water bodies, and the proximity to transport infrastructure. Surprisingly, we do not observe statistically significant differences in preferences among most age and gender groups. Overall, our study underscores the need for providing a diverse range of public open spaces tailored to accommodate different individuals, population groups, and activities. Keywords Public open space, Activities, Crowdsourcing, Street-level imagery, Demographics 1 Introduction Public open spaces, such as public squares, greenspaces, and streets, constitute a fundamental part of every city where different people can perform a wide range of activities. In this work, “public open space” refers to spaces open to the general public, located outdoors, and typically owned by governmental authorities or organizations. Various factors could impact how likely individuals are to engage in specific activities in public open spaces, with physical features playing a significant role (Jacobs, 1961; Ewing et al., 2006; Mehta, 2014). Physical characteristics include the size of the space, the presence of seating, the abundance of amenities, and the presence of natural elements such as trees, grass, or water bodies. Such characteristics influence how frequently public open spaces are used and shape their capacity to serve as hubs for social interaction and for engaging in different activities (Mehta, 2009; Whyte, 2012; Neutens etal., 2013; Putnam et al., 2000; Montgomery, 1998; Cohen etal., 2010; Lin etal., 2014; Wolch etal., 2014; Whyte etal., 1980). For instance, narrower streets with various amenities, wider sidewalks, and the availability of street furniture along sidewalks have been identified to encourage social interactions (Mehta, 2009). Similarly, the size of greenspaces has been found to influence the range of activities conducted within them. Larger greenspaces are often deemed more suitable for physical activities, while smaller ones are considered preferable for socializing and relaxation (Lee etal., 2015; Peschardt etal., 2012). Additionally, the preferences of different population groups may lead to varied choices in using public open *Correspondence: Vasileios Milias [email protected] 1 Delft University of Technology, Landbergstraat 15, Delft 2628CE, The Netherlands Page 2 of 17 Miliasetal. Computational Urban Science (2024) 4:15 spaces for specific activities (Sundevall & Jansson, 2020; Wen etal., 2018; Skelton, 2004). For example, research reveals that older individuals and women may perceive fewer spaces as suitable for exercise compared to men and younger adults (Foster & Giles-Corti, 2008). Seniors and women have also reported feeling vulnerable and having greater security concerns when walking (Basu etal., 2022). The importance of creating spaces that cater to the diverse preferences of various population groups is widely recognized (Mehta, 2009; Whyte, 2012; Neutens etal., 2013; Putnam etal., 2000; Montgomery, 1998), as emphasized by the United Nations’ Sustainable Development Target 11.7 (UN General Assembly, 2015), which underscores the necessity for inclusive spaces designed to accommodate everyone, with specific attention to the needs of women, children, and older individuals. To comprehend how the physical attributes of public open spaces influence their use, researchers have employed various methods such as participant observation (Sundevall & Jansson, 2020; Whyte, 2012; Mehta, 2009; Uslu etal., 2010), (online) questionnaires (Phillips etal., 2021; Jankowski etal., 2016;Talen etal., 2023), or crowdsourcing approaches (Salesses etal., 2013; Traunmueller etal., 2015). Often, these studies are focused on specific activities like exercising or walking (Koohsari et al., 2013; Lu, 2019), particular demographic groups such as the elderly (Levy-Storms etal., 2018) or children (Talen & Coffindaffer, 1999), or on specific qualities of a space like perceived safety or attractiveness for performing activities (Traunmueller et al., 2015). Nonetheless, our understanding of how the physical characteristics of public open spaces affect the suitability of various types of spaces for different activities, and the extent to which this suitability varies among different population groups, remains limited. This work employs a crowdsourcing approach to explore how physical characteristics of public open spaces influence the likelihood of their utilization among individuals of various ages and genders. We recruit 409 participants from 21 European countries in a thorough examination of people’s propensity to use public open spaces. To ensure a broad range of physical characteristics, we select various public open spaces such as public squares, open marketplaces, greenspaces, pocket parks, play spaces, and streets, sourced from three European cities: Rotterdam, Barcelona, and Gothenburg. We formulate three hypotheses and subject them to statistical testing: (H1) The likely use of public open spaces varies significantly by place type (e.g., parks, squares, streets); (H2) The likely use of public open spaces varies significantly across age groups; and (H3) The likely use of public open spaces varies significantly across gender groups. Subsequently, guided by the outcomes of the statistical analyses, we pinpoint cases where noteworthy differences were observed. For these instances, we employ reflexive thematic analysis to qualitatively assess the characteristics of public open spaces mentioned by participants as reasons for the identified disparities. In our approach, the collected ratings serve as a proxy for the probable use of space, indicating the likelihood that individuals would engage in activities there. By focusing on the likelihood of space usage rather than the observed behavior (i.e., real activities undertaken by people), we can account for factors unrelated to the physical characteristics of spaces that might discourage individuals from engaging in activities, such as not residing in close proximity (Neutens etal., 2013). Additionally, in contrast to conventional data collection approaches, such as participant observations or interviews, crowdsourcing serves as a time and resource-efficient method, allowing for the comprehensive study of various public open spaces and the recruitment of a diverse sample of participants in terms of age and gender. The remainder of this article is organized as follows. First, we explain our approach to capturing the likely use of public open spaces. Then, we describe the statistical tests conducted to scrutinize our hypotheses and how we qualitatively analyzed the collected data to identify the physical characteristics that influence the likely use of public open spaces. Next, we detail the data sources used in our empirical analysis and provide information about the participants of our study. Finally, we report the results of our study and discuss the empirical findings, implications, and limitations of our approach, as well as future lines of research. 2 Method Our methodology consists of four main steps: (1) sampling a variety of public open spaces in three European cities, (2) crowdsourcing the types of likely use, (3) testing three hypotheses using statistical methods, and (4) qualitatively exploring the physical characteristics of public open spaces that affect their use. 2.1 Selecting public open spaces We select public open spaces in three European cities: Barcelona (Spain), Rotterdam (Netherlands), and Gothenburg (Sweden). These cities represent urban environments in the Southern, Western, and Northern European regions. Consequently, our selection includes a wide range of spaces, enriching our study with a diverse set of spaces’ physical characteristics and types. The public open spaces included in this study were collected from OpenStreetMap (OSM) using the Overpass API and the OSMnx Python library (Boeing, 2017). OSM represents the physical features of the environment Page 3 of 17 Miliasetal. Computational Urban Science (2024) 4:15 using tags. To ensure a broad range of public open spaces we selected a variety of tags from OSM. We identified OSM tags related to public open spaces which reflect three main types of spaces: (1) vegetated spaces such as parks or forests, (2) play spaces dedicated to children’s activities such as playgrounds, and (3) other public open spaces such as squares and marketplaces. Since the tags related to the vegetated spaces encompass very different sizes of spaces, in accordance with recommendations by the World Health Organization Regional Office for Europe (2017), as well as the European Common Indicator for greenspace accessibility (Ambiente Italia, 2003), we divided vegetated spaces into two types: greenspaces that are larger than 0.5 hectares, such as parks, forests, and nature reserves; and pocket parks that reflect the vegetated spaces that are up to 0.5 hectares. Additionally, given that streets reflect the largest portion of public open space in every city, we also collected streets that are accessible to pedestrians. After defining these OSM tags, we collected all the spaces that fell under at least one of these five tags for all three case-study cities. Then, we investigated the spaces’ representation in Google Street-View. Using Google’s Street View Static API, we identified the spaces for which there is a street-level image within no more than 15meters, a reliable distance to observe and interpret an event according to Amiri and Crain (2019). Ultimately, we randomly sampled 420 public open spaces (140 per city), balanced in terms of OSM type. We then manually examined their street-level images and excluded spaces with images of poor quality, shot during nighttime, or not accurately representing the public open space because other urban objects obstructed the view (e.g., hedges or fences), and replaced these with other randomly sampled locations of the same type until all street-level images passed the test. In case a place is located in direct vicinity to multiple types (e.g., a public square located within a park), we assign it to both types. Table1 summarizes the types of public open spaces included in this study along with the number of collected spaces per type and city, and the tags used to collect them from OSM. 2.2 Capturing thelikely use ofpublic open spaces throughcrowdsourcing To collect information about how people of different ages and genders are likely to use public open spaces, we follow a crowdsourcing approach and use street-level images. Street-level imagery allows us to visually present the physical characteristics of spaces that potentially influence their use, such as seating, amenities, and trees. Regarding the likely use of public open spaces, we focus on the following five types of activities: socializing, relaxing, exercising, commuting, and children-related Table 1 Types of public open spaces, OSM tags used, number of collected spaces per type and city (Rotterdam (RTM), Barcelona (BAR), Gothenburg (GOT)), and examples of the activities examined Public Open Spaces Category Examples OSM Tags RTM BAR GOT TOTAL Greenspaces Parks, nature reserves forests (larger than 0.5 hectares) leisure:park | nature reserve, landuse:meadow | grass | village_green | forest, natural:wood | scrub | heath | grassland | fell | shrubbery 37 42 35 114 Pocket parks Parks, nature reserves (up to 0.5 hectares) Same tags as for greenspaces 31 30 30 91 Play spaces Playgrounds leisure:playground | schoolyard (access! = private) 28 29 27 84 Public squares & marketplaces Open public squares, open marketplaces place:square, amenity:marketplace, leisure:common 30 44 28 102 Streets Streets accessible to pedestrians network type = walk (OSMnx) 37 38 31 107 Activities Type Examples Socializing Picnics, meeting friends or others Relaxing Reading, or simply doing nothing Exercising Doing physical activities such as sports, walking/biking for fun Commuting Walk or bike to destinations Children-related Activities for children such as playing outdoors Page 4 of 17 Miliasetal. Computational Urban Science (2024) 4:15 activities. The selection of the activities considered in this work is aligned with and supported by Kruize etal. (2020) and Van den Berg etal. (2016), and aims to encompass a variety of activities that different individuals perform in public open spaces. In our experiments, we present to participants five different spaces and ask them to indicate to what degree, and why, they consider them suitable for any of the aforementioned activities. Our crowdsourcing campaign is implemented using the cloud-based research platform Qualtrics. The crowdsourcing task consists of four steps and requires 15 to 20minutes to be completed. First, we inform the Fig. 1 Main crowdsourcing task: participants first explore a public open space, represented as a 360◦ panoramic image, and then provide answers regarding the activities they would perform in that space Page 5 of 17 Miliasetal. Computational Urban Science (2024) 4:15 participant about the task and ask for their consent to participate. Second, we ask the participant their age and self-reported gender. The third step consists of the main crowdsourcing task, as illustrated in Fig.1. In this task, we initially show the participant a public open space represented as a 360◦ panoramic image and give them some time to pan around the image. Afterwards, drawing from the work of Kruize etal. (2020) and Van den Berg etal. (2016), we ask them to rate, on a 5-point Likert scale, if they find this place suitable for social, physical, relaxation, commuting, or children’s activities, and explain in their own words what characteristics influenced their ratings. We repeat this question for five different public open spaces. In the fourth step, we ask participants how important it is for them to have a space near their home where they can carry out these activities. We recruited participants using the Prolific platform. In total, we recruited 420 participants, evenly distributed across gender groups, as well as age groups categorized by decades (i.e., 18–30, 30–40, 40–50, 50–60, 60 +). We selected participants based on the following criteria. First, we only allowed participation through a laptop or desktop. Second, to ensure a similar level of familiarity with the shown spaces we only recruited participants residing in Europe. In addition, we ensured that people have not visited the shown spaces in real life, through a question we included in our crowdsourcing task. Third, we selected participants only if they had a high approval rate, based on the previous tasks they had contributed to. Fourth, we only selected participants who are proficient in English. All participants were above 18years old, provided informed consent to participate, and were compensated according to the minimum wage in [country hidden for anonymity due to the blind review process]. To ensure adequate quality of collected responses, we only kept responses from participants who passed a reCAPTCHA bot test, answered correctly to a simple attention check, and clicked at least four times inside the 360◦ images, as a proxy for panning around in the panoramic image. For each iteration, we ensured that the questions were displayed in a randomized order, and that the locations were selected at random from all public open spaces within one of the case-study cities. 2.3 Hypotheses testing After the crowdsourcing task, we have a set of spaces accompanied by the participants’ ratings reflecting how suitable these spaces are for social, relaxation, physical, commuting, or children’s activities (Fig.1). We use this information to examine how the characteristics of public open spaces influence how likely people are to perform activities there and the degree to which this varies among different age and gender groups. In particular, we formulate three hypotheses: • (H1) The likely use of public open spaces varies significantly by place type • (H2) The likely use of public open spaces varies significantly across age groups • (H3) The likely use of public open spaces varies significantly across gender groups With the first hypothesis, we explore the variation in the use of public open spaces based on their types, to examine queries such as: Does the likelihood of using a greenspace, public square, or pocket park for activities like socializing or exercise remain consistent, or do certain types of public open spaces naturally encourage specific activities more than others? With the second hypothesis, we examine whether there are variations in the likely use of public open spaces among different age groups, for instance: To what extent do younger adults consider the same spaces suitable for socializing or relaxing as older adults. With the third hypothesis, we investigate whether the likely use of public open spaces varies among different genders. We test these hypotheses using the ratings provided by the participants (i.e., 5-point Likert scale ratings). For H1, we aggregate all ordinal ratings of each space using the median rating per activity. Therefore, each space is assigned five median ratings, one per activity-type. To test if the different types of spaces received statistically different ratings we use the Kruskal–Wallis test. The Kruskal–Wallis test is suitable for our analysis as it is a non-parametric statistical test used to detect variations among three or more independently sampled groups based on a single non-normally distributed variable. This test is often employed for ordinal data (Kruskal & Wallis, 1952; McKight & Najab, 2010). We perform five Kruskal– Wallis tests and examine if the different types of spaces received significantly different ratings for each activity. To address the issue of multiple comparisons and reduce the likelihood of Type I errors, we applied the Bonferroni correction (Armstrong, 2014), setting our significance threshold to p-value = 0.05/5. It’s important to note that while the use of Bonferroni correction may elevate the risk of Type II errors, the decision to employ it reflects our emphasis on mitigating Type I errors, thus striving to minimize false discoveries. For H2, participants’ ratings were initially segmented based on age groups and then aggregated per space, again using the median value. To test H2 we compared the spaces’ ratings per activity between each pair of ages using the Mann–Whitney U test. The Mann–Whitney U test, is a non-parametric statistical method employed to Page 6 of 17 Miliasetal. Computational Urban Science (2024) 4:15 detect variations between two groups on a single ordinal variable (Mann & Whitney, 1947; Wilcoxon, 1992). For instance, we tested if the ratings of spaces for social activities differed significantly between the age groups [18–29] and [30–39], and conducted nine additional tests to compare all age groups pairwise. This process was then repeated for other activity types. We opted for pairwise tests using the Mann–Whitney U test instead of simultaneously testing all age groups (e.g., using the Kruskal–Wallis test), due to the insufficient number of spaces for which we obtained ratings from all age groups, limiting statistical comparisons. Thus, for each activity, we performed ten pairwise comparisons (five age groups compared with each other), resulting in fifty tests across all five activities. To address the issue of multiple comparisons, we adjusted our significance threshold using the Bonferroni correction, setting the p-value to 0.05/50. For H3, participants’ ratings were first divided based on gender groups and then aggregated per space, employing the median value. Similar to H1, to test H3 we compared the ratings of spaces per activity between genders using the Mann–Whitney U test. As our participant pool included two gender groups and five activities, we conducted one test per activity to assess differences between the two groups, totaling 5 tests. Once again, to address multiple comparisons, we employed the Bonferroni correction, setting the p-value to 0.05/5. For H2 and H3, participants’ ratings were first segmented based on age or gender groups and then aggregated per space, using the median value. To test H2 and H3, we compared the ratings of space per activity between every pair of age or gender groups using the Mann–Whitney U test. For instance, we assessed whether the ratings of spaces for social activities differed statistically between the age groups [18–29] and [30–39], and conducted nine additional tests to compare all age groups with each other. This process was then repeated for other activity types. We opted for pairwise tests using Mann–Whitney U test instead of simultaneously testing all age groups, because the number of spaces for which we received ratings from all age groups was insufficient to facilitate statistical comparisons. 2.4 Thematic analysis To provide further insight into our quantitative results, we qualitatively explore the reasons individual participants provide to explain their ratings. Specifically, we focus on the cases for which we find significant differences in the likely use of public open spaces, whether those pertain to the types of space (H1) or to the demographic groups (H2, H3). We employ reflexive thematic analysis (Clarke & Braun, 2013), using iterative inductive coding followed by identifying common themes, and document our analysis in Atlas TI. 3 Results 3.1 Descriptive statistics onparticipants andspaces Among all recruited participants, 409 participants met our quality standards as described in 2.2. These 409 participants completed our task in March—May 2023 and reside in 21 different European countries. The duration for participants to complete their task varied between 15 and 20minutes. Participants were evenly distributed across age groups categorized by decades with 80–82 participants per group (i.e., 18–30, 30–40, 40–50, 50–60, and 60 +). Similarly, they were evenly spread among genders, including only female and male groups since most participants self-identified as such and we did not have enough data to perform statistical analyses on the other gender groups. Following the exclusion of data due to technical issues (such as delayed panorama loading) or in cases where participants showed no interaction with the panorama (i.e., no clicks to pan or zoom), 413 places were included in this study, representing 102 public squares and marketplaces, 107 streets, 114 greenspaces, 91 pocket parks, and 84 play spaces (Table1). Each participant provided input for five different public open spaces, resulting in 9700 ratings and 6388 short explanations of these ratings. All participants stated that they found the task clear, 93% stated they turned the 360 images to look around as requested, and 95% answered that they were not familiar with the spaces they were asked to rate, as planned. For most participants (> 70%), having space for nearly all activities is considered important or very important. Exceptions are observed for social and children-related activities. Having access to space for social activities is deemed (very) important by 60.8% of respondents, while 10.5% consider it not important (at all). Moreover, having space for children-related activities is viewed as (very) important by 75.2% of participants with children, with 10.3% rating them as not important at all. Among respondents without children, such spaces are considered as (very) important by only 21.1%, while 56.5% regard them as not important at all. Overall, participants would often/always use a relatively large proportion of public open spaces for commuting activities (41.4% of spaces), a smaller proportion for physical and children-related activities (≈ 20%), regardless of whether considering participants with or without children, and an even smaller for social and relaxation activities (≈ 12%). Page 7 of 17 Miliasetal. Computational Urban Science (2024) 4:15 3.2 Hypotheses testing In this section, we present the results of our three hypotheses. In case the hypotheses are accepted, we perform exploratory analyses to gain deeper insights into the identified statistical differences. 3.2.1 H1: The likely use ofpublic open spaces varies significantly byplace type Overview With this hypothesis, we examine if the activities people would perform in a given public open space vary per type of space (e.g., public square, greenspace, street). To test our hypothesis, we perform five KruskalWallis tests, one for each activity-type (i.e., social, relaxation, physical, commuting, children-related). For each activity, we test if participants’ ratings are significantly different for different types of public open spaces. Result H1 was accepted for social (p = 3.32 × 10−11), relaxation( p = 5.04 × 10−11), physical (p = 2.71 × 10−10), and children-related activities (p = 1.26 × 10−26). However, it was rejected for commuting activities (p = 5.90 × 10−1). In other words, the degree to which people are likely to use a space for social, relaxation, physical, or childrenrelated activities varies for different types of public open spaces. For commuting activities, we did not find evidence for such variation. Exploratory analysis Based on our results, we further explore (1) which types of spaces would participants more often use for social, relaxation, physical, or children-related activities (Fig.2) and (2) what variety of activities people are likely to perform in the same space (Fig.3). Greenspaces, as illustrated in Fig.2, emerge as highly favored locations for most activities. An exception to this is found for the children-related activities for which, unsurprisingly, play spaces are preferred the most. In particular, participants’ ratings suggest that they would often use green spaces for commuting, sometimes for socializing, relaxing, and exercising, and rarely/never for children-related activities. Notably, pocket parks are less favored than greenspaces for all types of activities. Participants would sometimes use pocket parks for commuting and exercising and rarely for socializing, relaxing, and children-related activities. Concerning public squares and marketplaces, we note that participants gave them relatively low ratings for most activities. In particular, participants would never or rarely use these spaces for relaxation or childrenrelated activities, and would rarely or sometimes use them for social and physical activities. Public squares and marketplaces are predominantly preferred for commuting. We observe similar results for streets. Notably, streets received the lowest ratings among all types of spaces and for all activities except for commuting, for which participants would sometimes use them. Finally, regarding play spaces, apart from being often considered to be used for children-related activities, they were also considered sometimes suitable for socializing, exercising, and commuting. The least appropriate activity to perform in a play space, as indicated by the participants, is relaxing. Furthermore, to identify the different activities participants would perform in the same public open space, we measure the correlations among the spaces’ ratings for each pair of activities using the Spearman’s rank correlation coefficient. The corresponding correlation matrix is presented in Fig.3. Overall, only positive correlations were found. A strong and statistically significant correlation was found between social and relaxation activities (ρ = 0.76, p < 0.05), indicating that spaces that are considered suitable for socializing are also deemed suitable for relaxing and vice versa. Children-related activities were also found to have a moderate and significant correlation with both relaxation (ρ = 0.67, p < 0.05) and social activities (ρ = 0.57, p < 0.05). Moreover, physical activities exhibit a significant weak or moderate correlation with all other activities. Notably, commuting is the sole activity type lacking a significant correlation with all the other activities, excluding exercising. 3.2.2 The likely use ofpublic open spaces varies significantly acrossage groups Overview With this hypothesis, we examine if the activities participants are likely to perform in public open spaces vary per age group. To test our hypothesis, we perform Mann–Whitney U tests: one for each activity and each pair of age groups. As explained in Section.2.2, the age groups included in our tests are [18–29], [30–39], [40–49], [50–59], [60 +]. Result H2 was accepted for physical activities between the age groups [18–29]–[60 +] and for commuting activities between the age groups [30–39]–[60 +]. However, it was rejected for all other activities and age groups. That is, several age groups were found to differ in how likely they would use public open spaces for physical activities and commuting activities, but we do not have sufficient evidence to conclude that the degree to which different age groups would use a space to perform social, relaxation, or children-related activities differs significantly. Page 8 of 17 Miliasetal. Computational Urban Science (2024) 4:15 Fig. 2 Ratings of the likely use of public open spaces per activity Page 9 of 17 Miliasetal. Computational Urban Science (2024) 4:15 3.2.3 H3: The likely use ofpublic open spaces varies significantly acrossgenders Overview With this hypothesis, we examine if the activities participants intend to perform in public open spaces differ between gender groups. Out of all participants, 49% self-identified as males, 49% as females, and 2% as non-binary, third gender, or prefer to self-describe or not to say. To statistically test our hypothesis we limit to males and females since we do not have sufficient data to draw statistical conclusions for the other groups. In particular, we performed four Mann–Whitney U tests, one for each type of activity, between the ratings we received from male and female participants. Result H3 was rejected for all types of activities. That is, we do not have sufficient evidence to conclude that the likely use of public open spaces between males and females differs significantly. 3.3 Thematic analysis The aim of the thematic analysis is to explore what characteristics people consider promoting or obstructing Fig. 3 Correlation of activity-based ratings Table 2 Most prevalent characteristics of public open spaces that positively or negatively influenced participants’ ratings Positive Negative Social amenities, sitting spaces, calm, open, nice for picnic, vibrant, nature, nice meeting place, view lack of amenities, lack of sitting spaces, unattractive, lack of space, road/traffic/cars, lack of nature, industrial, noisy, busy, dedicated to certain age groups, not calm Relaxation nature, calm, quiet, sitting spaces, nice for reading, sun/shade, open, view, little traffic/cars, attractive, private noisy, busy, lack of sitting spaces, high traffic/cars, no nature, industrial, unattractive, no calm, lack of space, unsafe Physical space, nature, dedicated to specific sports (e.g., walking, biking, pools), safe, open, calm, little traffic, attractive, quiet, view little space, busy, unsafe, unattractive, high traffic/cars, industrial, little nature, narrow, pollution, bad view Children play equipment, safe (supervised/unsupervised), amenities, nature, space, open, quiet, little traffic, sitting spaces unsafe, high traffic/cars, nothing for children to do, nearby roads, little space, lack of amenities, industrial, blue spaces (considered unsafe), isolated, little nature Page 16 of 17 Miliasetal. Computational Urban Science (2024) 4:15 gender groups, regarding the activities they would engage in across a variety of public open spaces, and the characteristics that matter to them. Our findings reveal significant differences regarding the suitability of different types of public open spaces for different activities. Greenspaces emerged as the most favored for nearly all activities, while pocket parks were less preferred, and streets were considered the least suitable for engaging in various activities. Additionally, our findings suggest that the suitability of a space for one activity implies its suitability for other activities. Contrary to expectations, most instances did not reveal significant differences among different age and gender groups in their preferences for engaging in specific activities in public open spaces. When variations were observed, they stemmed either from distinct preferences for specific characteristics of public open spaces among various groups or from differing perceptions of shared, recognized characteristics. Our findings underscore the importance of ensuring diversity in public open spaces to accommodate the preferences of different individuals, activities, and population groups. Acknowledgements We thank all participants for contributing to our study. Authors’ contributions Vasileios Milias: Conceptualization, methodology, data curation, formal analysis, software, validation, visualization, writing – original draft, writing – review and editing. Roos Teeuwen: Conceptualization, methodology, data curation, writing – original draft. Alessandro Bozzon: Funding acquisition, methodology, supervision. Achilleas Psyllidis: Conceptualization, methodology, funding acquisition, resources, supervision, writing – original draft, writing – review and editing. Funding This research has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 874724. Availability of data and material The data collected from the participants and the code used for the analysis are publicly available on https:// zenodo. org/ recor ds/ 10974 191. The street-level imagery obtained by Google Street-View cannot be publicly shared due to the company’s data-sharing regulations. Declarations Competing interests The authors declare no potential conflicts of interest. Received: 11 March 2024 Revised: 13 May 2024 Accepted: 31 May 2024 References Ambiente, Italia. (2003). European common indicators: Towards a local sustainability profile. Ambiente Italia Research Institute: Report. Amiri, S., & Crain, D. R. (2019). Quantifying jacobs’ notion of ‘eyes upon the street’ in 3-dimensions. Journal of Urban Design, 25(4), 467–485. Armstrong, R. A. (2014). When to use the b onferroni correction. Ophthalmic and Physiological Optics, 34(5), 502–508. Barnett, D. W., Barnett, A., Nathan, A., Van Cauwenberg, J., Cerin, E., & on Environment, C., and working group, P. A. C.-O. A. (2017). Built environmental correlates of older adults’ total physical activity and walking: A systematic review and meta-analysis. International Journal of Behavioral Nutrition and Physical Activity, 14, 1–24. Basu, N., Oviedo-Trespalacios, O., King, M., Kamruzzaman, M., & Haque, M. M. (2022). The influence of the built environment on pedestrians’ perceptions of attractiveness, safety and security. Transportation Research Part f: Traffic Psychology and Behaviour, 87, 203–218. Bertolini, L. (2020). From “streets for traffic” to “streets for people”: Can street experiments transform urban mobility? Transport Reviews, 40(6), 734–753. Boeing, G. (2017). Osmnx: New methods for acquiring, constructing, analyzing, and visualizing complex street networks. Computers, Environment and Urban Systems, 65, 126–139. Bubalo, M., van Zanten, B. T., & Verburg, P. H. (2019). Crowdsourcing geo-information on landscape perceptions and preferences: A review. Landscape and Urban Planning, 184, 101–111. Clarke, V., & Braun, V. (2013). Successful qualitative research: A practical guide for beginners. Sage publications. Cohen, D. A., Marsh, T., Williamson, S., Derose, K. P., Martinez, H., Setodji, C., & McKenzie, T. L. (2010). Parks and physical activity: Why are some parks used more than others? Preventive Medicine, 50, S9–S12. Coley, R. L., Sullivan, W. C., & Kuo, F. E. (1997). Where does community grow? the social context created by nature in urban public housing. Environment and Behavior, 29(4), 468–494. Ewing, R., Handy, S., Brownson, R. C., Clemente, O., & Winston, E. (2006). Identifying and measuring urban design qualities related to walkability. Journal of Physical Activity and Health, 3(s1), S223–S240. Foster, S., & Giles-Corti, B. (2008). The built environment, neighborhood crime and constrained physical activity: An exploration of inconsistent findings. Preventive Medicine, 47(3), 241–251. Goličnik, B., & Thompson, C. W. (2010). Emerging relationships between design and use of urban park spaces. Landscape and Urban Planning, 94(1), 38–53. Jacobs, J. (1961). The Uses of Neighborhood Parks, chapter 5, pages 116–145. Modern Library, New York. Jankowski, P., Czepkiewicz, M., M-lodkowski, M., & Zwolin´ski, Z. (2016). Geoquestionnaire: A method and tool for public preference elicitation in land use planning. Transactions in GIS, 20(6), 903–924. Koohsari, M. J., Karakiewicz, J. A., & Kaczynski, A. T. (2013). Public open space and walking: The role of proximity, perceptual qualities of the surrounding built environment, and street configuration. Environment and Behavior, 45(6), 706–736. Kruize, H., van Kamp, I., van den Berg, M., van Kempen, E., Wendel-Vos, W., & Ruijsbroek, A., Swart, W., Maas, J., Gidlow, C., Smith, G., Ellis, N., Hurst, G., Masterson, D., Triguero-Mas, M., Cirach, M., Grazuleviciene, R., Van den Hazel, P., and Nieuwenhuijsen, M. (2020). Exploring mechanisms underlying the relationship between the natural outdoor environment and health and well-being–results from the phenotype project. Environment International, 134, 105173. Kruskal, W. H., & Wallis, W. A. (1952). Use of ranks in one-criterion variance analysis. Journal of the American statistical Association, 47(260), 583–621. Lee, A. C. K., Jordan, H. C., & Horsley, J. (2015). Value of urban green spaces in promoting healthy living and wellbeing: prospects for planning. Risk management and healthcare policy, 8, 131–137. Levy-Storms, L., Chen, L., & Loukaitou-Sideris, A. (2018). Older adults’ needs and preferences for open space and physical activity in and near parks: A systematic review. Journal of Aging and Physical Activity, 26(4), 682–696. Lin, B. B., Fuller, R. A., Bush, R., Gaston, K. J., & Shanahan, D. F. (2014). Opportunity or orientation? who uses urban parks and why. PLoS ONE, 9(1), e87422. Linday, N. (1978). It all comes down to a comfortable place to sit and watch. Landscape Architecture, 68(6), 492–497. Lu, Y. (2019). Using google street view to investigate the association between street greenery and physical activity. Landscape and Urban Planning, 191, 103435. Page 17 of 17 Miliasetal. Computational Urban Science (2024) 4:15 Mann, H. B. & Whitney, D. R. (1947). On a test of whether one of two random variables is stochastically larger than the other. The annals of mathematical statistics, pages 50–60. McKight, P. E. and Najab, J. (2010). Kruskal-wallis test. The corsini encyclopedia of psychology, pages 1–1. Mehta, V. (2009). Look closely and you will see, listen carefully and you will hear: Urban design and social interaction on streets. Journal of Urban Design, 14(1), 29–64. Mehta, V. (2014). Evaluating public space. Journal of Urban Design, 19(1), 53–88. Milias, V., Sharifi Noorian, S., Bozzon, A., and Psyllidis, A. (2023). Is it safe to be attractive? disentangling the influence of streetscape features on the perceived safety and attractiveness of city streets. AGILE: GIScience Series, 4:8. Montgomery, J. (1998). Making a city: Urbanity, vitality and urban design. Journal of Urban Design, 3(1), 93–116. Navarrete-Hernandez, P., Vetro, A., & Concha, P. (2021). Building safer public spaces: Exploring gender difference in the perception of safety in public space through urban design interventions. Landscape and Urban Planning, 214, 104180. Neutens, T., Farber, S., Delafontaine, M., & Boussauw, K. (2013). Spatial variation in the potential for social interaction: A case study in flanders (belgium). Computers, Environment and Urban Systems, 41, 318–331. Peschardt, K. K., Schipperijn, J., & Stigsdotter, U. K. (2012). Use of small public urban green spaces (spugs). Urban Forestry & Urban Greening, 11(3), 235–244. Phillips, A., Khan, A. Z., & Canters, F. (2021). Use-related and socio-demographic variations in urban green space preferences. Sustainability, 13(6), 3461. Putnam, R. D., et al. (2000). Bowling alone: The collapse and revival of American community. Simon and schuster. Rees-Punia, E., Hathaway, E. D., & Gay, J. L. (2018). Crime, perceived safety, and physical activity: A meta-analysis. Preventive Medicine, 111, 307–313. Salesses, P., Schechtner, K., & Hidalgo, C. A. (2013). The collaborative image of the city: Mapping the inequality of urban perception. PLoS ONE, 8(7), e68400. Skelton, T. (2004). ‘nothing to do, nowhere to go?’: teenage girls and ‘public’space in the rhondda valleys, south wales. In Children’s geographies, pages 69–85. Routledge. Sullivan, W. C., Kuo, F. E., & Depooter, S. F. (2004). The fruit of urban nature: Vital neighborhood spaces. Environment and Behavior, 36(5), 678–700. Sundevall, E. P., & Jansson, M. (2020). Inclusive parks across ages: Multifunction and urban open space management for children, adolescents, and the elderly. International Journal of Environmental Research and Public Health, 17(24), 9357. Talen, E., Choe, K. W., Akcelik, G. N., Berman, M. G., & Meidenbauer, K. L. (2023). Street design preference: an on-line survey. Journal of Urban Design, 28(1), 1–24. Talen, E., & Coffindaffer, M. (1999). The utopianism of children: An empirical study of children’s neighborhood design preferences. Journal of Planning Education and Research, 18(4), 321–331. Thompson, C. W. (2002). Urban open space in the 21st century. Landscape and Urban Planning, 60(2), 59–72. Traunmueller, M., Marshall, P., and Capra, L. (2015). Crowdsourcing safety perceptions of people: Opportunities and limitations. In Social Informatics: 7th International Conference, SocInfo 2015, Beijing, China, December 9–12, 2015, Proceedings 7, pages 120–135. Springer. UN General Assembly (2015). Transforming our world: the 2030 agenda for sustainable development. Resolution adopted by the General Assembly on 21 October 2015, A/RES/70/1. Uslu, A., et al. (2010). Social interaction in urban transformation areas and the characteristics of urban outdoor spaces: A case study from turkey. African Journal of Agricultural Research, 5(20), 2801–2810. Van den Berg, M., Van Poppel, M., Van Kamp, I., Andrusaityte, S., Balseviciene, B., Cirach, M., Danileviciute, A., Ellis, N., Hurst, G., Masterson, D., Graham, S., TrigueroMas, M., Uzdanaviciute, I., de Wit, P., van Mechelen, W., Gidlow, C., Grazuleviciene, R., Nieuwenhuijsen, M., Kruize, H., & Maas, J. (2016). Visiting green space is associated with mental health and vitality: A crosssectional study in four European cities. Health & Place, 38, 8–15. Wen, C., Albert, C., & Von Haaren, C. (2018). The elderly in green spaces: Exploring requirements and preferences concerning nature-based recreation. Sustainable Cities and Society, 38, 582–593. Whyte, W. H., et al. (1980). The social life of small urban spaces. DC: Conservation Foundation Washington. Whyte, W. H. (2012). City: Rediscovering the center. University of Pennsylvania Press. Wilcoxon, F. (1992). Individual comparisons by ranking methods. In Breakthroughs in statistics: Methodology and distribution, pages 196–202. Springer. Wolch, J. R., Byrne, J., & Newell, J. P. (2014). Urban green space, public health, and environmental justice: The challenge of making cities ‘just green enough.’ Landscape and Urban Planning, 125, 234–244. World Health Organization Regional Office for Europe (2017). Urban green spaces: a brief for action. Report. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.