Evaluation of the perceived social impacts of the Formula E Grand Prix of Santiago de Chile

The aim of this study is to analyse residents' perception of the social impact of the Formula E Grand Prix of Santiago de Chile. A sample of 414 residents was collected through a questionnaire made up of 46 items regarding possible positive and negative impacts. Descriptive analysis, confirmatory factorial analysis and cluster analysis were performed. The results showed two groups with different perceptions of this sporting event: realistic (n=152) and favourable (n=162). Sociodemographic variables referring to education level, civil status, income level, location of the family residence and political orientation contribute to significantly differentiating the clusters. The variables related to interest in the event, support for the celebration, attendance, participation and frequency of contact also contributed to significantly differentiating the groups. Citizens’ high degree of support for holding the sporting event could be identified by a positive social representation of the event.


Introduction
Sporting events are analysed from the perspective of different fields: economic, sociological or environmental. Many cities often use the rhetoric associated with "sustainable development" or "sustainable regeneration" to justify the hosting of sporting events (Smith, 2009). Within sporting events, the analysis of the impact on society and the quality of life of residents has gained special relevance in recent years, as the lack of support for the holding of sporting events can generate devastating effects on the host community, increasing social and political tensions (Gursoy, Yolal, Ribeiro and Panosso, 2017).
The social impact of sporting events has been analysed mainly through the study of residents' perception of various variables of interest (impacts on diverse areas, support for the celebration, identification with the event, linkage with the event, etc.). Papers that explore the perceptions of subgroups of host residents can facilitate a better understanding of the social impact of these events by identifying potential winners and losers (Ma, Ma, Wu and Rotherham, 2013). These kinds of works are few (Ma et al., 2013), and many insist on the need to make more contributions along this line to identify the factors that may define these groups (Chen and Tian, 2015). In this regard, several authors have pointed to the importance of assessing the reactions of subgroups of hosts at sporting events as a way of better understanding stakeholders (Chen and Tian, 2015;Fredline and Faulkner, 2000;Ma and Rotherham, 2015;Zhou and Ap, 2009).
The Formula E championship is made up of various events held on urban circuits in which electric single-seat race cars take part. This championship was born in the 2014-2015 season with eleven races, increasing the number of races to 13 in the 2018-2019 season. Because this is a sporting event with a little history, the contributions to the analysis of the impact on the host localities are limited.
Therefore, this paper analyses the social impact of the Formula E Grand Prix held in Santiago de Chile in 2018 by identifying groups of residents with different perceptions of the impacts associated with it. In addition, the variables that define these groups are identified to be able to determine the characteristics of the residents.

Social impact of motor sport events
The social impact of sporting events can be defined in terms of changes in the quality of life of residents as a consequence of hosting a sporting event (Parra, Calabuig, Núñez and Crespo, 2017). Studies analysing the social impact of sporting events tend to consult the perception of stakeholders affected by the event, with residents being one of the groups on which research in this area has focused the most (Parra, Alonso-Dos-Santos and Duclos, 2018a).
Motor sport events with cars, with the exception of rallies, in recent years, have tended to be held on urban circuits built specifically for the event or adapted for it (Lefebvre and Roult, 2011).
For this reason, urban circuits tend to use public roads and resources such as parks, trails and public areas that are not usually used for this purpose, all connected as circuits specifically designed for the motor sport events (Mackellar, 2013).
These events often generate a series of positive and negative impacts that can affect the host community. In this sense, most of the studies that have analysed the social repercussion of their celebration in urban circuits agree in highlighting the noise problems generated by singleseaters, traffic and the agglomeration of people in the areas in which they are celebrating as egregious negative impacts (Añó et al., 2012;Cegielski and Mules, 2002;Cheng and Jarvis, 2010;Fredline et al., 2013). Other significant negative impacts are related to economic costs (Kim et al., 2015) or concerns about the maintenance of public facilities (Mackellar, 2013).
On the other hand, several works highlight the positive impacts related to the improvement of the city image and its promotion as a tourist destination (Añó et al., 2012;Cegielski and Mules, 2002;Cheng and Jarvis, 2010;Kim et al., 2015;Mao and Huang, 2016;Zhou, 2010) or the stimulation of the local economy (Fredline et al., 2013). Other psychosocial and intangible aspects are related to the community's increased pride in hosting the event (Fredline et al., 2013;Mao and Huang, 2016), entertainment opportunities (Fredline et al., 2013;Kim et al., 2015), or the opportunity to meet new people (Cheng and Jarvis, 2010).

Explanatory theories of the social impact of sporting events
There are several theories that attempt to explain changes in residents' perceptions of the impacts of sporting events. In this paper, we will use the theory of social representations (Moscovici, 1984) as a theoretical frame of reference to analyse residents' perceptions of Formula E. This theory suggests that residents' feelings towards tourism and sporting events may be affected by their direct experiences, social interactions, personal and social values, and other sources of information, such as the media (Fredline, 2005, Ritchie, Shipway andCleeve, 2009). Social representations are difficult to change because they provide a framework through which new information is interpreted (Fredline, 2005).
This theory can help explain how various groups of people understand and respond to social issues. It is particularly appropriate when the topic of study involves multiple social perspectives or accompanies conflicts associated with possible change, such as the hosting of a sporting event (Zhou and Ap, 2009). In the research on the social impact of sporting events, several works have used this theoretical framework of reference (Chen and Tian, 2015;Cheng and Jarvis, 2010;Fredline, 2005;Fredline et al., 2013;Fredline and Faulkner, 2000;Zhou, 2010;Zhou and Ap, 2009).

Participants
In this study, 414 people living in Chile were interviewed about their perception of the impacts generated by the Formula E race in Santiago de Chile. The event being studied is the Santiago ePrix, which was attended by 25,000 spectators at the inaugural race held on February 3, 2018 at the Parque Forestal urban circuit and was the fourth race of the 2017-2018 season. Table 1 shows the sociodemographic characteristics of the sample. The majority of those interviewed are residents of the metropolitan region of Santiago de Chile (55.1%), aged between 25 and 54 (81.6%), men (79.2%) and have university-level education (88.4%).

Instrument
The instrument is composed of 46 items adapted from previous studies about positive and negative impacts associated with the celebration of sport events (Añó et al., 2012;Fredline et al., 2013;González-García et al., 2016;Parra, Aguado and Núñez, 2015;Parra, Añó, et al., 2016). The items proposed were classified by categories of positive and negative impacts. To make this classification, the impact categories proposed by Preuss and Solberg (2006) and Fredline (2004) were taken into account. In addition, the different contributions made to the dimensions proposed by the case studies are referred to at the beginning of this section. Thus, the proposed items were clustered, on the one hand, into five initial dimensions related to positive impacts (33 items in total), which were denominated as follows: socio-economic impact (7 items), impact on urban development and infrastructure (6 items), political and administrative impact (5 items), psychosocial impact (3 items), sports impact (6 items) and sociocultural impact (6 items). On the other hand, the negative impacts (13 items in total) were grouped into three dimensions: socioeconomic impact (7 items), environmental impact (3 items) and sociocultural impact (3 items).
A five-point Likert scale was used, where 1 means totally disagree and 5 means totally agree. The following sociodemographic variables were included: age, gender, education level, occupation, family residence, income level, civil status and political orientation. Other variables of interest were also included in the questionnaire to define the characteristics of the resident groups: interest in the sport, attendance at the event, support for the celebration, involvement in tourism or events, public participation as volunteers or workers in sporting events and frequency of contact with the area in which the event is held.

Procedure
A convenience sampling method such as that used in other work in this area was used (e.g., García-Pascual, Parra and González-García, 2019; Gursoy and Kendall, 2006;Oshimi and Harada, 2018; Parra, Alonso-Dos-Santos and Duclos, 2018b, 2019). The questionnaires were distributed and collected after the event (between February and June 2018).

Statistical analyses
First, two confirmatory factorial analyses (CFA) for the scale of positive impacts (33 items) and negative impacts (13 items) were carried out through the program for structural equation models EQS 6.2.
Several recommended goodness-of-fit indexes were used to check the overall fit of the model through the CFA (Kline, 2005): the normalised chi-square (χ 2 /df), the root mean square error of approximation (RMSEA), the non-normalised fit index (NNFI), the comparative fit index (CFI) and the incremental fit index (IFI). The reliability of the scale was checked through Cronbach's alpha, composite reliability (CR) and average variance extracted (AVE). Convergent validity was also analysed through the significance of factor charges (p<.05) in their respective dimensions and the associated t-test values. To contrast discriminant validity, the method suggested by Fornell and Larcker (1981) was used, which consists of checking whether the square root of the AVE value of a given factor is greater than the correlation coefficients between the factor and any other factor of the proposed scale.
Second, an analysis of clusters was carried out using the statistical program SPSS version 24.0 with the items derived from the CFA. Two methods of estimation (hierarchical and nonhierarchical) of the cluster solution were combined to optimise the results. The hierarchical cluster analysis was performed using the Ward Method grouping process and as a measure of similarity, the Euclidean distance squared. Based on the groups proposed in the previous analysis, a non-hierarchical analysis was applied using the K-means method, using as initial centres the means of the variables obtained for each cluster solution of the hierarchical analysis. To define the characteristics of group profiles and to evaluate predictive validity, ANOVAs and chi-square tests were performed with sociodemographic variables and variables of interest that were not included in the initial analysis. The value of the contingency coefficient (C) was also used to check the intensity of the association or the size of the effect of the related variables.

Confirmatory factor analysis of scales
The CFA results on the scale of positive impacts showed an adequate overall fit after several re-specifications of the initial model in which six indicators were removed from the initial 33: [(S-

Cluster analysis of differences between population groups
Cluster analysis allowed the identification of two clusters that presented a suitable theoretical interpretation for the study (see table 2). The significance level for the ANOVA results was set at p<.000, indicating significant differences and supporting the statistical validity of the research.
The two identified clusters were named "favourable" (36.7%) and "realistic" (63.3%): 1     (1) (2) The results are based on bilateral tests with a level of significance .05. In the results table, it is shown for each significant pair the key of the group of residents with the lowest column proportion below the group of residents with the highest column proportion.

Discussion and conclusions
This study contributes to analysing residents' perceptions of sporting events based on segmentation into groups with different reactions to the impacts of these types of events. First, the validity and reliability of the scales of positive and negative impacts were assessed, identifying 6 factors and 3 factors, respectively.
In line with previous investigations Chen and Tian, 2015;Ma and Rotherham, 2015;Zhou and Ap, 2009), in this work, two groups were found consisting of residents with different profiles. In the two groups identified (realistic and favourable), they show high percentages of support for the celebration of Formula E (67.5% and 95.9%, respectively). This is similar to what was observed in studies on other motorsports events, such as the Formula 1 Grand Prix in Australia or the Gold Coast Indy Car Faulkner, 2000, 2002). Nevertheless, in other works on this type of event, the inverse trend has been observed, as in the case of the Formula 1 European Grand Prix (Añó et al., 2012;Calabuig et al., 2014).
The residents classified within the group of realists seem to show a higher level of appreciation of the real impact of the event, highlighting some positive socioeconomic, political, administrative and sociocultural impacts. Nevertheless, there are also negative socioeconomic costs as a consequence of hosting the event. Thus, this group shows a more circumspect social representation of the event and perhaps are less conditioned or influenced by direct experiences or personal factors. In other studies, this group has been identified as realists (Chen and Tian, 2015;Fredline and Faulkner, 2000;Parra and Duclos, 2013), Ambivalent (Fredline and Faulkner, 2002) or Unconcerned (Fredline et al., 2013), Indifferent  or Moderate .
On the other hand, the residents designated as favourable show an optimistic trend in all the impact areas analysed: socioeconomic impact, impact on infrastructure and urban development, political and administrative impact, psychosocial impact, sporting impact and sociocultural impact. On the other hand, they tended to minimise negative impacts at the socioeconomic, sociocultural and environmental levels. This group presents a positive social representation of the event that is characterised by its high degree of support for the celebration of the event and interest in the sport. This group of residents has also been identified in all of the studies that analyse the perception of the impacts of sporting events Chen and Tian, 2015;Fredline and Faulkner, 2000;Fredline and Faulkner, 2002;Fredline et al., 2013;Parra, Añó et al., 2016;Parra et al., 2014;Parra and Duclos, 2013;Zhou, 2010;Zhou and Ap, 2009). As Cheng and Jarvis (2010) point out, this type of event provides an opportunity for the enjoyment and entertainment of the members of this collective; the benefits and costs associated with the event are not perceived in the same way by other groups with lesser identification with the event.
Sociodemographic variables such as gender, age, level of education or occupation show little or no effect in determining the subgroup of each resident. This finding is consistent with previous studies, which found that gender, age, and level of education (Chen and Tian, 2015;Ma et al., 2013;Parra et al., 2014;Zhou, 2010;Zhou and Ap, 2009) do not significantly affect such conglomerates. Nevertheless, other variables such as the level of income create significant effects in the differentiation of the groups, coinciding with what has been observed in other studies (Ma and Rotherham, 2015;Ma et al., 2013). Additionally, the civil status and the origin of the family residence generated significant effects in the differentiation of the groups, with the majority of the group with a more positive attitude towards the event having their family residence in regions other than Santiago de Chile. Finally, the political orientation variable generated significant effects in the differentiation of the groups, which is in line with what was observed in other works .
Regarding the variables of interest related to the event, a significant effect was observed in the differentiation of the groups in those variables related to support, interest in sports, attendance, participation as a worker or volunteer in sports events and frequency of contact with the zone where the event is held. This result is in line with what was observed in other studies in which it was found that the variables of support, interest and attendance Fredline et al., 2013;Ma et al., 2013;Ma and Rotherham, 2015;Parra and Duclos, 2013;Zhou, 2010), participation as a worker or volunteer and frequency of contact with the zone (Fredline and Faulkner, 2002) had a significant effect on group differences. With respect to the relationship with tourism or the economic activity derived from the events, these variables did not show a significant effect in the differentiation of the clusters, coinciding with what was observed in other works on diverse sport events (Ma and Rotherham, 2015;Parra, Añó, et al., 2016). Nevertheless, in other studies, it has been observed that these variables have an effect on the distinction between groups (Fredline and Faulkner, 2002;Parra et al., 2014).

Practical implications
From the point of view of the analysis of the social impact of events, it is important to consider residents as a heterogeneous group since the perception of a sporting event can vary according to different factors and variables . In this sense, the identification of groups of residents with different perceptions allows the organisers and the administration to improve the understanding of the different segments of citizens with the aim of better managing the impacts and trying to satisfy the needs of these groups (Deery and Jago, 2010).
Accordingly, organisers and administrations must bear in mind the characteristics and behaviours that define these groups to maximise the benefits of hosting this type of event . This work shows that there is a high degree of citizens' support for the celebration of the event that could be identified by a positive social representation of the sporting events.
Nevertheless, it is necessary for the organisers to pay attention to some of the costs highlighted by residents, such as those related to the inconvenience caused by the disruption of daily life, the distortion of the normal functioning of the city or restrictions on access to public facilities and services in the area. According to Ma et al. (2013), event organisers and local authorities should consciously plan events related to improving the quality of life in the host areas if they want a positive long-term relationship.

Limitations and future research lines
To conclude, it should be noted that these results should not be generalised to all city residents, as a consequence of the limitations of using a convenience sampling frame. It is therefore necessary to interpret them with caution and to increase the representation of some demographic groups in future studies.
Since the perceptions of the population may change as a consequence of various social conditioning factors, it is worthwhile to carry out consultations over time to observe possible changes in the configuration of clusters. In the same way, it would be advisable to compare perceptions about the impacts of the event in different cities that host Formula E events.
In addition, future studies could change the means of addressing some variables in a dichotomous way or with alternative responses on a Likert scale to better understand the behaviour of the population.