The theory of diffusion innovations [1] has been used to explain the adoption of ideas and technologies in different realms including among others, science, policies, marketing, and transport [2, 3]. This theory describes how in a given community during a specific period, members are likely to adopt a similar device (convergence), which responds to a specific unmet need. Convergence however may be slow but also may not occur at all. Indeed, social networks, whereby technologies or ideas are channelled, can impede spreading. This may be because the innovation is not properly communicated or understood, does not match a need, or means to adopt it are not yet available [4, 5]. Considerable part of diffusion of innovation research has focused on classifying what makes individuals to adopt before convergence could be observed. Broadly, when an innovation has finally spread, five overlapping groups of users could be identified: innovators, early adopters, early majority, late adopters and laggards or sceptics [6].
The functioning of transport systems can be positively or negatively altered depending on how rapidly transport providers and/or users accept new technological innovations [7]. This has been the case for instance of mobile ticketing service in public transport [8], models of bike sharing [9], alternative fuel vehicles [10], seat belts [11] or child restraints [12]. Ultimately, the introduction of these technologies can be associated with less polluted environments after alternative fuel vehicles cope the market [13], or considerable decreases in traffic fatalities after seat belts or child restrains are both massively and properly used [14–16]. Yet, the adoption and further use of transport-related technologies may not be equally accepted by transport providers or users [17]. Understanding these multifaceted technological processes is critical when studying multiple outcomes in the realm of transport systems.
Ridesharing, a service that connects drivers with potential passengers through a mobile application, is a type of technological innovation, which has shown a remarkable case of diffusion and convergence within transport systems. For instance, Uber, one of many ridesharing application providers, has since 2011 spread to at least 900 cities in 84 countries, and during 2018 it supplied over 14 million journeys per day globally [18]. In the Middle East, North Africa, and South Asia, Careem has spread to more than 50 cities [19]. In United States, Lyft operates in more than 300 cities, and in China, DiDi Chuxing is used by more than 450 million users, and spread to more than 400 cities [20]. Regarding the diffusion of ridesharing applications and its multiple impacts on transport systems, two elements should be highlighted. First, studies have suggested that early adopters of ridesharing are both highly educated and from high-income households [21–25]. Further, the initial use of this service was associated with ease of payment since trips could be charged to credit cards, which were integrated into cell phone applications [22]. To extend its use, many ridesharing applications have been complemented with in-cash payments [26, 27]. Second, another group of nascent studies have indicated that ridesharing services could be also associated with positive and negative transport outcomes such as congestion [28], changes in public transit use [29], air quality [30], and alcohol-related crashes [31–36]. More specifically, in terms of decreasing alcohol-related crashes, it has been argued that ridesharing provides a valuable alternative to drinking and driving when access to public transport is limited (i.e., low frequency of provision during late hours) and/or traditional taxis costs may be unknown (i.e., high fluctuation of prices for similar trips).
Despite the interest in and growth of ridesharing studies and alcohol-related crashes, there is a surprising gap on understanding how changes in the diffusion of ridesharing use could be associated with variation of these outcomes. This is important because part of the literature has reported mixed results. Greenwood and Wattal [32], Peck [33], Morrison et al. [34], and Martin-Buck [37] have indicated reductions in alcohol-related crashes after ridesharing increased over time. Conversely, Brazil and Kirk [31, 38], Dills and Mulholland [39], and Nazif-Munoz et al. [35] did not observe any change in these outcomes, even after exploring use increments. These null results possibly mask either short terms associations or changes in specific groups—such as innovators and/or early adopters. However, to better understand ridesharing application associations, it should be acknowledged that the use of this technology follows over time a gradient of socioeconomic status, whereby innovators and early adopters are more likely to belong to high-income groups, and late adopters and laggards may be linked with lower-income groups [21–25]. As such, when ridesharing applications begin emerging in a given urban community, transport-related changes would be expected in high-income groups more prominently [40], and then, depending on whether convergence of this application is reached, positive outcomes at the population level, regardless of the socioeconomic gradient, should be observed.
In this work, we put lens on understanding how the early adoption of one ridesharing model—Uber—in Chile’s capital city, Santiago, could be associated with alcohol-related crash variations. The case of Chile is interesting due to four reasons. First, early adopters and frequent users of ridesharing applications, particularly Uber, are from high-income groups [41]. Second, credit cards are not equally distributed across income groups in Chile [42, 43]. Indeed, individuals at the highest income decile, and relative to lowest, third- and sixth-income deciles have respectively 48.2, 30.0, and 12.9 more chances of having a credit card from a bank [44]. Further, during the first two years of implementation (2014 to 2016), Uber limited its use to credit card holders only [41]. This reinforces the notion that high-income individuals were more likely to be Uber users than individuals from lower income groups. Third, in terms of alcohol consumption while no differences across socioeconomic groups are observed regarding heavy drinking and heavy episodic drinking, individuals from higher socioeconomic status are more likely to drink higher volumes of pure alcohol weekly than any other socioeconomic group [45]. Lastly, Santiago is a highly socially segregated city, its east central sector, made up of seven municipalities, contains more than 80% of the main population of the richest quintile. Under these characteristics it could be assumed from an ecological perspective that ridesharing in its genesis could be associated with reductions in alcohol-related outcomes in municipalities where high-income individuals inhabit, and no associations would be necessarily observed in municipalities where individuals from lower incomes live.
Exploiting information from 34 municipalities of Santiago, we combined a two-fold statistical approach, spatial statistical analysis and random effects meta-analyses, to assess the association between alcohol-related crashes, and the absence and presence of Uber in Santiago, Chile. Focusing on the slow pace of Uber’s implementation in Chile during its first year, we took advantage of this highlight segregated city to explore differences across time and municipalities. We took this approach because we considered that the innovators and early adopters of prospective Uber passengers would be individuals from high-income groups. We hypothesized that, in municipalities where high-income individuals inhabit, the beginning of Uber operations relative to its absence, would be associated with lower risks of alcohol-involved crashes.