Introduction

Urban design focused on improving walkability has received attention as a method of increasing physical activity among the population1,2,3. People living in high-walkability neighborhoods have higher levels of physical activity and social capital, which benefit physical and mental health4,5,6. Moreover, highly walkable areas demonstrate increased neighborhood retail sales7, higher property values8,9,10, and greater urban sustainability11. Improving urban walkability involves enhancing land-use diversity, traffic safety, street design, population density, and green cover ratio12,13,14,15. However, only limited methods are available for improving walkability in the urban centers of highly motorized suburban cities16. In the urban central areas of suburban cities, increasing land-use diversity by opening a multifunctional facility is considered one of the most effective strategies for an architecture-scale intervention6,17. Previous cross-sectional studies have shown a significant relationship between land-use diversity and walking time18,19,20,21. Furthermore, the relationship was associated with differences in gender and age subgroup22. However, among the many land uses, the impact of opening a multifunctional facility on the residents is not clearly understood.

This means a research gap exists between the correlation and causality of opening a multifunctional facility and promoting actual walking time in specific subgroups. Natural experiments, which take advantage of changes in the built environment without researchers’ direct intervention, have been effective research designs for elucidating causality. Previous studies analyzed the natural experiments in the case of district-scale interventions, street-scale interventions, and park interventions at site-scale23. The district-scale interventions reported that metro development did not promote the walking revel of older adults in Hong Kong24, but opening new stations increased older adults’ average daily walking steps by approximately 400 steps/day in Tokyo25. The district-scale interventions suggest that some natural experiments focused on older adults, not on younger age subgroups. Moreover, as a street-scale intervention, installing new sidewalks increased the number of people who walk to more than 60 min/week26 and opening greenways significantly increased walking time27,28. Meanwhile, park improvements at the site scale did not promote walking29,30,31. Their findings indicate that people’s daily walking time was affected by street-scale interventions, not by site-scale interventions of park improvements. For the site-scale interventions, there has never been a natural experiment analyzing the impact of the opening on residents’ walking by opening a multifunctional facility on an architectural scale. This issue highlights the significance of natural experiments studying the effect of walking time on opening a multifunctional facility for built environment research.

This study aimed to clarify the effect of opening a multifunctional facility on the residents’ average daily walking time. In addition, the effects were analyzed by subgroups of gender (male, female) and age (minors, young adults, middle-aged adults, and older adults). The research design adopted the natural experiment using the case of the Ibaraki City Cultural and Childcare Complex “ONIKURU.” The natural experiment of this study would provide theoretical contributions to the causality of promoting walking by opening a public multifunctional facility at an architecture-scale intervention in specific subgroups. This study used Global Positioning System trajectory (GPS-trajectory) data using GPS tracking techniques, which is anonymized location data for smartphone users. Using GPS-trajectory data, this study conducted propensity score matching to adjust for confounders and analyzed the change in average daily walking time before and after opening a multifunctional facility for visitors and non-visitors. The difference-in-differences analysis elucidated the causal impact of opening a multifunctional facility on the residents’ average daily walking time.

Ibaraki City Cultural and Childcare Complex, a well-known multifunctional facility nicknamed “ONIKURU,” was selected for this study. The ONIKURU was located in Ibaraki City, which is a typical suburban city in the Osaka metropolitan area. The ONIKURU project was initiated in December 2015, when the Ibaraki City Civic Hall was closed due to building deterioration32. Then, the citizens conducted a series of workshops and social experiments, commonly known as the “IBA-lab” project. Toyo Ito & Associates and Takenaka Corporation designed the multifunctional facility based on these workshops and experiments. Toyo Ito is a famous Japanese architect and the winner of the Pritzker Prize33. ONIKURU opened on November 26, 2023, as a public multifunctional facility that integrates a hall, library, childcare support, planetarium, and civic activities. The public multifunctional facility can be used by all generations, including minors, young adults, middle-aged adults, and older adults. Figure 1 shows the satellite photographs of ONIKURU in Ibaraki Central Park before and after its opening. In addition, Fig. 2 shows the images of ONIKURU in Ibaraki Central Park. Subsequently, nearly 100,000 people visited ONIKURU within 15 days of its opening34, making it a widely recognized successful example of a multifunctional facility in a suburban city.

Fig. 1
figure 1

Satellite photos of the Ibaraki City Cultural and Childcare Complex “ONIKURU” in the Ibaraki Central Park before and after its opening. (a) A satellite photo captured in December 2014 before the Ibaraki City Civic Hall closed. (b) A satellite photo captured in May 2024 after ONIKURU opened. The satellite photos were obtained by Arc GIS PRO 3.0 with the copyright of ArcGIS of the World Imagery (https://services.arcgisonline.com/ArcGIS/rest/services/World_Imagery/MapServer).

Full size image
Fig. 2
figure 2

Images of the Ibaraki City cultural and childcare complex ONIKURU in the Ibaraki Central Park. (a) Perspective. (b) Café on the first floor. (c) Childcare support center on the second floor. (d) Multipurpose studio on the third floor. (e) Hall on the fourth floor. (f) Library on the fifth and sixth floors. (g) Citizen activity center on the seventh floor. (h) Planetarium on the seventh floor. Picture (a) shows a copyright-free image provided by the Ibaraki City Government. Pictures (b)–(h) were captured by the corresponding authors.

Full size image

Methods

This study analyzed the average daily walking times of visitors and non-visitors at ONIKURU using GPS-trajectory data. The study was approved by the Ethics Committee of the Graduate School of Life and Ecology, Osaka Metropolitan University (No. 24 − 13). Additionally, all methods were performed in accordance with the relevant guidelines and regulations. In detail, this study’s methods were carried out following the “Guidelines for the Use of Device Location Data,” a common regulation for GPS data analysis in Japan35. These guidelines prohibit the use of GPS-trajectory data for any purpose that involves identifying individual users to protect the privacy of their GPS location histories. All participants were informed about the type of data collected, the intended use, potential disclosure to third parties, and privacy policies. Informed consent was obtained from all participants. In addition, the participants could stop sending location history data at any time by adjusting the application settings on their mobile phones.

Research design

The research design of this study employed a natural experiment to investigate the effect of opening ONIKURU on the average daily walking time of Ibaraki citizens. In the natural experiment, this study analyzed the change in average daily walking time before and after opening a multifunctional facility for visitors and non-visitors. The natural experiment was also analyzed by gender and age subgroups. The gender subgroup includes males and females. In addition, the age subgroup includes minors (people aged 17 years old and under), young adults (people aged from 18 to 44 years old), middle-aged adults (people aged from 45 to 64 years old), and older adults (people aged 65 years old and older).

Traditionally, pedometers and questionnaires have been used to measure walking time. However, these survey methods are limited in terms of data accuracy and sample size. Therefore, this study used GPS-trajectory data to estimate “walking” as a mode of transportation as it captures location information and speed. The GPS-trajectory data provide a large and diverse sample, allowing for robust analyses. The GPS-trajectory data was collected from all users who logged in to Ibaraki City. In this study, the logs of people living in Ibaraki City were extracted as analysis samples based on their place of residence. In addition, this study excluded citizens with no attribute data and those who had not logged into Ibaraki City before or after the opening of ONIKURU. The analysis sample was categorized as the visitor group that logged into ONIKURU and the non-visitor group that did not log into ONIKURU. This study analyzed the changes in average daily walking time from baseline to the follow-up period for each group.

The analysis period spanned from November 11 to December 11, 2023, covering 15 days before and after the opening of ONIKURU. The baseline period was set from November 11 to November 25, 2023, and the follow-up period was set from November 26 to December 11, 2023.

Dataset

This study used GPS-trajectory data collected by the data provider company of Geo-Technologies. GPS-trajectory data consists of anonymized location big data for each smartphone user using GPS tracking techniques. Geo-Technologies obtained consent from smartphone users who subscribed to a particular mobile operator or installed a particular application36. The Geo-Technologies GPS-trajectory data are characterized by shorter logging intervals compared with the GPS-trajectory data of other companies. The logging interval for the GPS-trajectory data obtained by Japanese companies is usually approximately 15 min37,38,39,40. However, the median logging interval during walking obtained by Geo-Technologies was approximately 1 min36. This makes Geo-Technologies’ GPS-trajectory data particularly suitable for analyzing the detailed walking time of visitors to a specific urban facility.

The GPS-trajectory data were obtained in Japan from October 11 to November 11, 2023. The variables in the GPS-trajectory data included the user ID, year, month, day, hour, minutes, latitude and longitude, log-location city code, activity type, and accuracy. In addition, user IDs contain attribute data, including home city codes, age, gender, occupation, commuting transportation, marriage, housing, educational background, and personal annual income, reflecting demographic characteristics and personal preference. The accuracy variable indicated the horizontal accuracy of the smartphone’s location information (in meters). The accuracy variable tends to be higher at locations where the GPS satellite signals are difficult to reach or reflect, such as inside buildings. Therefore, this study extracted log data with an accuracy of less than 50 m below the neighborhood block scale and the length of the ONIKURU building. This extraction allowed us to analyze only highly accurate log data. Furthermore, this study identified Ibaraki citizens. The GPS-trajectory data used in this study included all logs from Japan. Therefore, this study used the attributes of the home city codes of users to extract the citizens of Ibaraki City. The activity types were categorized as on-foot, walking, running, on-bicycle, on-vehicle, still, or unknown. Among these seven types, walking activity is associated with “on foot” recorded in iOS and “walking” recorded in Android. This study calculated daily walking time as the sum of the time to the next log of “on foot” or “walking” activity type in each day. This analysis was possible owing to the GPS-trajectory data with short logging intervals.

In summary, this study analyzed effective analysis samples of people who live in Ibaraki City. Therefore, this study excluded respondents who did not live in Ibaraki City, data with unknown attributes, and no GPS-trajectory data before or after the opening. For effective analysis samples, this study extracted log data with an accuracy of less than 50 m and on-foot/walking activity types. Using the log data, each user’s daily walking time data is averaged for the period before and after opening ONIKURU. This study analyzed each user’s average daily walking time changes, divided into visitor and non-visitor groups. The visitor group consists of those who logged in to ONIKURU during the analysis period, and the non-visitor group consists of those who did not. Figure 3 shows the images of the analysis using GPS-trajectory data.

Fig. 3
figure 3

Images of analysis using GPS-trajectory data. (a) Map with visitor (blue) and non-visitor (red) logs, which are analyzed in this study as valid analysis samples. (b) Analysis of a user’s walking time on a specific day in the case of the visitor group. (c) Analysis of the user’s daily walking time of visitor and non-visitor groups over one month. The data shown here are not actual data but image data. The satellite photo of the map (a) was obtained by Arc GIS PRO 3.0 with the copyright of ArcGIS of the World Imagery (https://services.arcgisonline.com/ArcGIS/rest/services/World_Imagery/MapServer).

Full size image

Statistical analysis

A propensity score matching analysis was conducted to estimate the effect adjusted for confounding factors. Propensity scores were calculated using a logistic regression model. The logistic regression model set the dependent variable as the visiting group and the independent variable as the demographic characteristics. The result of the logistic regression indicates the propensity score. The demographic characteristics included gender, age, occupation, commuting transportation, marriage, housing, educational background, and personal income. The non-user group was matched to the user group in a 1:1 ratio using the propensity score. In the propensity score matching, the nearest neighbor method with a caliper of 0.05 was used.

The user and non-user groups were analyzed for changes in average daily walking time using difference-in-differences (DID) analysis. This method uses data observed at two-time points to estimate the difference between the “average change in the visitor group” and the “average change in the non-visitor group” as the exposure effect. Considering that daily walking time likely varies seasonally41,42, DID analysis was deemed appropriate for this study. Shapiro–Wilk test confirmed the normality of the average daily walking time data. Based on the results, the Wilcoxon rank-sum test was used to compare groups. The significance levels were set at 1% and 5%.

Study sample

Figure 4 shows the sampling flowchart of this study. During the analysis period, 17,284 users were recruited from Ibaraki City. Among these users, logs from 7,555 Ibaraki citizens were extracted based on their place of residence. Then, citizens with no attribute data (n = 1755) and those who were no GPS-trajectory data in Ibaraki City before or after the opening of ONIKURU in Ibaraki City (n = 983) were excluded. Finally, this study analyzed samples from 4,817 citizens.

For this study, the samples logged at ONIKURU were set as visitor samples (n = 883). The remaining samples were classified as non-visitor samples (n = 3932). However, the visitor and non-visitor samples showed different characteristics. Therefore, propensity score matching was performed using gender, age, occupation, commuting transportation, marital status, housing, educational background, and personal income as covariates. Consequently, among the 3932 non-visitor samples, 883 non-visitors were statistically selected through the propensity score matching technique to match the demographics of 883 visitor counterparts. This study analyzed the change in average daily walking time between visitor and non-visitor groups after propensity score matching.

Fig. 4
figure 4

Flowchart of the sample selection process.

Full size image

Results

Demographic characteristics

Supplementary File 1 shows the demographic characteristics of the visitors and non-visitors before and after the propensity score matching. Before matching, the visitors and non-visitors differed significantly in occupation, commuting transportation, marriage, and educational background. After matching, no significant differences were noted in these demographics, as shown in Table A. After matching, the average age of the visitor and non-visitor groups was approximately 40 years old, with a slightly higher proportion of women in both groups. This may be related to the presence of a childcare support center in ONIKURU.

Difference-in-differences of average daily walking time

Table 1; Fig. 5 show the average daily walking times of visitors and non-visitor groups before and after the opening of ONIKURU. Before the opening of ONIKURU, the visitor group walked on average 78.010 [74.615, 81.404] min/day, while the non-visitor group walked on average 85.478 [80.215, 90.741] min/day. In this manuscript, the numbers in parentheses indicate the upper and lower limits of the 95% confidence intervals. Meanwhile, after the opening of ONIKURU, the visitor group walked on average 82.910 [79.173, 86.647] min/day, while the non-visitor group walked on average 87.214 [81.803, 92.624] min/day. With regard to the changes that occurred before and after the opening of ONIKURU, the average daily walking time of the visitor group increased to 4.900 [1.463, 8.338] min/day. Meanwhile, the average daily walking time of the non-visitor group increased to 1.735 [− 1.702, 5.173] min/day. This result means that the visitor group increased their average daily walking time to 3.165 [− 1.697, 8.027] min/day compared to the non-visitor group. The Wilcoxon rank-sum test was used to compare the changes in average daily walking time between the visitor and non-visitor groups. A significant difference was observed in the changes in the average daily walking time between the visitor and non-visitor groups.

Fig. 5
figure 5

Difference-in-differences of average daily walking time. The red line indicates the changes in the average daily walking time of visitors. The blue line indicates the changes in the average daily walking time of non-visitors. Error bars indicate the 95% confidence intervals.

Full size image
Table 1 Change in average daily walking time before and after the opening of ONIKURU.
Full size table

Subgroup analysis

Table 2 indicates the average daily walking times of visitors and non-visitor groups in each subgroup before and after the opening of ONIKURU.

As a result, Table 2 shows that the average daily walking time was increased in the female older adults’ subgroup (19.807 [− 11.918, 51.532] min/day), the male minors’ subgroup (8.854 [− 13.737, 31.445] min/day), and the male older adults’ subgroup (8.295 [− 12.842, 29.432] min/day). In addition, the Wilcoxon rank-sum test was used to compare the changes in average daily walking time between the visitor and non-visitor groups. The result shows a significant difference in the female young adults’ subgroup (3.385 [− 4.906, 11.676] min/day). However, the average daily walking time did not change significantly in the other seven subgroups. The result suggests that opening ONIKURU significantly increased female young adults’ average daily walking time to 3.165 [− 1.697, 8.027] min/day.

Table 2 Change in average daily walking time of each subgroup.
Full size table

Discussion

This study examined the effects on the average daily walking time of the opening of ONIKURU (Ibaraki City Cultural and Childcare Complex). ONIKURU opened in the southern part of the Ibaraki Central Park as a multifunctional facility that integrates a hall, library, childcare support, planetarium, and civic activity center. This study’s research design was a natural experiment using GPS-trajectory data based on GPS tracking techniques. As a result, this study clarified that the opening of ONIKURU significantly increased visitors’ average daily walking time to 3.165 [− 1.697, 8.027] min/day compared with that of non-visitors. The novel findings of this study were discussed with those of some relevant natural experiments. The walking time effect of opening a multifunctional facility was slightly lower than that of greenway interventions (36.4 min/week, approximately 5.2 min/day)28. However, as a site-scale intervention, many previous studies reported that park improvement did not promote residents’ average daily walking29,30,31. Compared with those studies, our findings contributed novelty theoretical contributions to a health-promoting built environment significantly affecting walking at an architecture-scale intervention. In other words, this research has succeeded in calculating the effect of increasing one multifunctional facility among many land uses from the perspective of average daily walking time. This contribution provided significant insights into the causality between land use and physical activity, which we should prioritize in our research agenda43. The theoretical contributions implicated urban planners in guiding public multifunctional facilities in suburban centers following Japan’s “compact plus network” policy9,44,45.

Another important point is that visitors’ average walking time increased from 78.010 [74.615, 81.404] min/day to 82.910 [79.173, 86.640] min/day after the opening of ONIKURU. Increased physical activity through walking is important for healthy living as it extends life expectancy46,47,48. In contrast, non-visitors walked on average 85.478 [80.215, 90.741] min/day before the opening of ONIKURU and 87.214 [81.803, 92.624] min/day after its opening. Therefore, visitors to ONIKURU improved their walking to a level comparable with that of non-visitors after the opening of ONIKURU. Adults are generally recommended to walk more than approximately 80 min/day49,50. Based on the criteria, the average daily walking time of visitors was lower than the recommended value before its opening. Therefore, opening a multifunctional facility possibly changed visitors’ behaviors, contributing to health benefits.

In addition, this study also clarified that opening ONIKURU significantly increased female young adults’ average daily walking time to 3.165 [− 1.697, 8.027] min/day. The average daily walking time effect was not observed in subgroups of males, minors, middle-aged adults, and older adults. The surprising result was obtained because the public multifunctional facility “ONIKURU” includes a library, childcare support, and citizen activity center for all age groups. Some natural experiments analyzed the impacts of older adults’ walking24,25. However, the effects of other subgroups have not been studied well, regardless of their importance22. In particular, there was a lack of effective intervention methods to promote physical activity among female young adults51,52. Therefore, this study’s finding contributes to the health-promoting intervention for female young adults from the built environment research field.

This study has four limitations. First, the analysis period was relatively short (15 days before and after opening ONIKURU). Although the GPS-trajectory data enabled a high-accuracy analysis of user travel behavior, the large amount of data made it difficult to perform a long-term analysis. ONIKURU reached 1 million visitors on June 6, 2024, 194 days after its opening. A long-term analysis would reduce the risk of bias in the results. Second, the study considered a limited number of places visited. In urban design, it is important to consider not only the walking time, but also examine the impact of the surrounding area. The Ibaraki City Government has begun planning to improve the walkability of urban centers around ONIKURU. Future research should also examine the changes in the areas where visitors stay after the opening of ONIKURU. The third limitation is the limit of causality. This study conducted DID analysis using propensity score matching based on natural experiments. This is an excellent analytical method for quasi-experiments. However, we cannot rule out the possibility that factors other than the characteristics included in the propensity score matching, such as personal walking preferences, may have influenced the results. Future research should use experimental methods such as randomized controlled trials. The fourth limitation is the risk of sampling bias. The smartphone penetration rate in Japan is over 75%53. However, the GPS-trajectory data does not sample residents who do not have smartphones. Future research should analyze complex data that includes data other than GPS-trajectory data.

Conclusion

This study examined the effect of opening a multifunctional facility in an urban center on the walking time of citizens, focusing on ONIKURU in a suburban city. In conclusion, the opening of ONIKURU significantly increased visitors’ average daily walking time to 3.165 [− 1.697, 8.027] min/day compared with that of non-visitors. In detail, the walking time of visitors improved to a level comparable to that of non-visitors after the opening of ONIKURU. In addition, opening ONIKURU significantly increased female young adults’ average daily walking time to 3.385 [− 4.906, 11.676] min/day. Our findings contributed theoretical contributions to a health-promoting built environment significantly affecting walking at an architecture-scale intervention. The theoretical contributions highlight essential considerations for urban planners aiming to design health-promoting built environments in urban centers.