performance expectancy, effort expectancy, social influence, and facilitating conditions.

In our study, the main indicators were derived from the UTAUT model. We identified appropriate transferable translations of the four determinants as defined within the UTAUT framework that can be adapted to the e-bike model. This represents our unique interpretation of applying the UTAUT model to the e-bike concept. Performance Expectancy is a critical factor in e-bike adoption in terms of trip frequency. It refers to what users expect from e-bikes when using the technology. To identify e-bike adoption while minimizing external influence, we remove the impact of new stations, analyzing data based on only those stations available throughout the entire time series of the data, e.g., the oldest stations in each city. At the same time, we removed "ramp-up" and "ramp-down" periods from the data, because the e-bike sharing system was shut down in winter months. Hence, only the data from April 1 to October 1 was kept each year, representing the peak usage season. Regarding COVID-19's impact, the whole time series (pre-and post-pandemic) was examined to see whether COVID-19 impacted the e-bike sharing concept. The demographic variables of age and gender are mainly related to how often bikes are used regularly. We henceforth assume that if the e-bikes met the users' expectations, they are more likely to be used repeatedly. Likewise, the number of trips over time was calculated for each city. Effort expectancy considers "user-friendliness" in terms of ease of use and effort required from the user. Our approach measures effort expectancy using the trip duration. For example, if the trips are too short, i.e., less than 5 minutes, we assume that people may be unlocking the bikes, finding their use difficult, and returning them to the same station. We note that if an e-bike user stops somewhere intentionally (e.g. running an errand) without docking the bike, the decreased average speed will complicate this effort expectancy analysis. Thus, we use the logged speed data for the bikes to identify intentional stops using a simple threshold technique (e.g., if the bike's is stationary for > 10 minutes, we say the trip includes an "errand stop"). We find that roughly 2.17% of all trips have a stop like this, and we simply discard them for the effort expectancy analysis. Social Influence concentrates the societal expectations within a demographic area, grounded in its perception of the technology's importance, utility or value. To ascertain the social influence, we evaluate whether users engage with the e-bike sharing system consistently and regularly. Specifically, we examine the number of trips in combination with demographics of these users and how it changed over time. Our approach for measuring the social influence is rooted in the social atmosphere of each city based on the population's demographics. Facilitating Conditions includes several factors that may influence the use of e-bike-sharing systems. We measure it as the number of trips against promoting factors in the specific area. One such contributing factor is the availability of charging stations, with more stations leading to increased usage. Another factor is the availability of protected bike paths, where we would expect e.g., a positive correlation between the number of miles of bike paths in a town and the usage of e-bikes. We define the normalized facilitating factors of a town as (miles of bike path × number of stations)/population.

To measure the acceptance and use of e-bike sharing within different social contexts, we use multiple subjective variables to identify their impact on e-bike use and acceptance, including median income, education (bachelor's degree or higher), median age, gender and voluntariness of use. We use U.S. Census data [1] to obtain the median income, education level, and age in each case study location. Finally, the voluntariness of use is an important factor in technology acceptance. When individuals feel they have a choice in whether or not to use a new technology, they may be more likely to adopt it. Based on the available data, including preferred e-bike routes, detailed bike maps, conclusions were drawn regarding the key factors, such as infrastructure.

For analysis, we use data from the e-bikes and their docking stations, along with publicly accessible census data. The e-bike data set provides information about each trip taken from 2018-2022, including the trip duration, starting and ending stations, and (anonymous) unique ID of the user. Each station in the network is located in one of the 5 cities under study, so Census data is integrated into our analysis by matching trips with the city in which they take place. At the beginning of 2019, the frequency of repeat users increased slowly. Year over year, there is a trend that the number of trips by repeat users increases up to 250 trips, while the bottom 50 percentile of users tends to decrease with a peak at 140 trips. Likewise, the middle users decrease with a number of 100 trips at peak. In 2020, the effects of the COVID-19 pandemic had a large impact on e-bike usage due to stay-at-home orders, as seen in the figure. E-bike usage restarted slowly in 2020 beginning in July, but did not reach the level of previous years. In 2021, the e-bike sharing system sees a resurgence, coinciding with the relaxation of COVID-19 restrictions. The trend shows that trips started by repeat users are rising sharply overall, with utilization peaking in September 2021. This trend is consistent in 2022, showing that repeated "heavy" users have the most trips, while new users or less frequent users have fewer trips.

To analyze the effort expectancy of the e-bikes, we analyzed the trip duration (i.e., the time elapsed from bike unlocking to redocking). Trip data spans July 2018 through October 2022. Using trip duration as a measure, Fig. 5 and Fig. 4 jointly illustrate the ease of e-bike usage, attributing the effort the users had to exert for their locomotion. Of particular interest here are short-term uses, especially those that take place for less than 5 minutes -for instance, the data set includes over 1000 trips that ended in less than 1 minute, returning the bike to the same station. In the early years of the bike share program, short trips (lasting between 1 and 5 minutes) typically returned to the same station. In later years, some of these short trips, despite brief duration, were between different stations. This trend may indicate an increase in the density of bike docking stations over time and an improvement in user experience, resulting in more efficient (faster) trips. Results for Social Influence: Regarding the relation of the median income with the number of trips, a rough trend can be recognized in the data that the higher the median household income, the higher the normalized number of trips. For example, in Holyoke and Springfield, both the median income and the number of trips are lower. Amherst is a clear outlier with respect to this factor, especially in comparison to Easthampton, which has almost the same median income -the number of normalized trips in Amherst is significantly higher than Easthampton. A strong trend can be equally seen in the correlation between the normalized number of trips and the education of the population. With increasing education (in terms of a bachelor's degree or higher), the amount of trips increases. Finally, plotting the normalized number of trips against the median age in a city, we find that the lower the median age, the higher the number of trips in general. An exception is Northampton, where the number of trips is high despite a median age of 40. Results for Facilitating Conditions: A notable outlier in terms of facilitating conditions is Northampton, driven by both a significant number of stations and strong bike path infrastructure. Interestingly, although Northampton has the strongest facilitating conditions by a significant margin, the normalized number of trips are higher in Amherst, which is a distant second in terms of facilitating conditions. On the other extreme of the graph, the other towns all have fairly low facilitating conditions. Notably, Holyoke is an exception. Despite very low facilitating conditions (explained mostly by a lack of bike path infrastructure), adoption as measured by the number of trips is comparatively high.

The introduction of e-bikes into the energy and mobility system of the U.S., particularly in Massachusetts, represents a whole system change. The findings show that the tradtionally car-centric U.S. has started adopting e-bikes as an innovative technology and alternative mode of transportation. A rising number of users over time indicates that the technology is experiencing promising growth in adoption, despite tempering effects due to the COVID-19 pandemic. Performance Expectancy: Results indicate that performance expectancy is particularly met by regular users, whose numbers have increased over time. The technology is especially beneficial to the top 10 percent of users, which accounts for the overall increase in regular use. During recovery from the COVID-19 pandemic (i.e., 2021 and 2022) the overall number of trips rose significantly, with a peak in September. However, comparing just 2021 and 2022 to each other, they show similar trend without further increases, suggesting a saturation among highly regular users. The behavior of non-frequent users implies that the technology is likely to be tried out, but not necessarily used again. This could implicate that for most users, there is not an overall acceptance of the system e.g., for routine usage. However, it can also be interpreted that the technology is useful to the smaller proportion of users that are using it regularly. Nevertheless, there is also a slight overall increase in irregular users with only a few trips. The percentile cutoffs in Fig. 3 confirm these observations of the overall increase in usage. The momentary downward trend in 2019 can be explained by an influx of new users joining, offsetting the stable returning users. Effort Expectancy: The results for effort expectancy show that irrespective of the trip duration, a significant number of trips started at one station and ended at another station. Since the stations are not close to each other, this indicates that users used the bike for a concrete purpose, such as commuting or traveling. In addition, Fig. 4 shows that the characteristics of the e-bike trip duration changed over the four years. While in 2018 short trips started and ended mostly at the same station, in 2022 short trips mainly ended at another station. Similarly, the data showed that trips from one station to another are not necessarily only for leisure -e-bike users increasingly use them for useful errands. Across all years, the most common trip duration is between 5-20 minutes, and this is also the category of trip that sees the greatest growth over 2018-2022. Social Influence: From the demographics, we could interpret that cities with higher educational attainment place more emphasis on environmental protection, the use of green technologies, and physical activity. Therefore, it could be assumed that there is a higher awareness of the e-bike system in these cities and that they are more likely to be championed by their governments. Northampton and Amherst are two examples that illustrate this particularly well. Facilitating Conditions: Facilitating factors such as the presence of bike paths and docking stations had a great influence on usage and adoption in the different cities. Most notable, cities with many bike paths had a higher adoption of the e-bike system within their town. For instance, university students showed high usage, likely helped by the bike paths on campus. We note that these facilitating factors are correlated with some other factors such as education and median income. This suggests that the infrastructure is generally very important and a valuable facilitating factor for the technology. Summary: While e-bike sharing has the potential to offer many benefits, there are several challenges to be addressed. These include increasing complexity of governance, effectiveness of current strategies, and issues related to connectivity and accessibility.

The success of e-bike sharing as an emerging mobility concept is contingent upon the technology meeting users' expectations, being equally accessible to all social groups, being suppored by its social surroundings, and having conducive facilitation conditions for its usage. The paper showed an analysis of the usage and acceptance of e-bike sharing as a new mobility concept in different demographic areas in Massachusetts, USA. For this purpose, the paper used the UTAUT model as an analytical framework to measure the performance expectancy, effort expectancy, social influence, and the facilitating conditions of e-bike sharing. Discussing various implications of e-bike sharing for the socio-technical transition of the energy system, our study reveals that a mobility concept can only be successful if the use of an e-bike can easily facilitate useful tasks (e.g., shopping) or be easily combined with other means of transport and therefore interact with existing systems. To conclude, e-bike sharing has the potential to become an essential pillar in a growing urban ecosystem of sharing, an important mobility concept, and a component of the energy transition.