Elecciones a gobernador en Tlaxcala 2021: análisis espacial del voto in Tlaxcala 2021: Spatial Analysis of the Vote

The victory in the governorship of the Juntos Haremos Historia coalition in Tlaxcala, headed by the Morena parties, Partido del Trabajo (PT), Verde Ecologista de México (PVEM), Nueva Alianza Tlaxcala (Panalt) and Partido Encuentro Social Tlaxcala (PEST), introduced again in the entity a political alternation, after the Partido Revolucionario Institucional (PRI) was the winner in the 2010 and 2016 elections. For this work, the spatiality of the vote in the gubernatorial elections is investigated, which presents spatial dependence, registered since the 1998 elections. Methodologically, an exploratory analysis of electoral data by municipality is carried out. The local context of the election is also briefly presented, given that Lorena Cuéllar Cisneros, the second woman to occupy the main figure at the local level, represents the last generation of the local political elite formed in the ranks of the PRI.


Introduction
This work undertakes an exploratory analysis of spatial data whose space variable transitions from descriptive to explanatory, and takes into consideration that, in the case of the gubernatorial elections in Tlaxcala, a spatial dependency has been detected since 1998.
The premise is that the phenomenon of voting, seen through electoral participation (PE), does not have a random distribution, that is, it is not the result of a random process. For the treatment of electoral data, the GeoDa software was used, disaggregated at the municipal level.
Although the exploration of socioeconomic variables, as well as the analysis of the rest of the positions, has been a pending issue in Tlaxcala for a couple of years, spatial studies with electoral data at the local level have begun to be registered in different academic forums, despite the challenges in the accessibility and validity of the information.
The works on the spatial analysis of the vote help to explain the electoral situation of each electoral process, the evolution of political trends, the characteristics of the local party system, as well as the georeferencing of electoral behavior in the region.
One of the challenges mentioned, for example, are the information gaps that arise in the compilation of electoral results. For this election, blank rows were found in some sections of 11 municipalities. The cases were: Calpulalpan (section 80, 91), Chiautempan (section 124, 126, 137, 139,157) Party (PEST). ) with Lorena Cuéllar Cisneros as her candidate. In total, six candidates and one candidate at the head of coalitions and political parties (competing for the first time) 1 La información del dato electoral a nivel sección solicitada al Instituto Tlaxcalteca de Elecciones tenía la leyenda "Las filas en color rojo corresponden a los paquetes electorales que por algún motivo no fueron recibidos en los Consejos Distritales". led the main application in the entity. Regarding the latter, 10 political parties had national registration and five had local registration.
In the first part of this work, a brief analysis of electoral geography and spatial analysis is presented. The main mark of this approach is the use of the concept of region and space; that is, descriptive variable (region as container) and explanatory variable (space that impacts). In the second part of the work, the findings of the exploratory analysis of the data of the 2021 governor election in Tlaxcala at the municipal level with the Moran index (1948) and the local association index (LISA) are presented. , based on the vote understood as PE and the partisan preference of the two coalitions (Partido Revolucionario Institucional [PRI] and Morena). The third section presents the electoral results and a brief account of the context of the contest. Finally, it concludes with some very general reflections, given that this investigation is still in progress.

Electoral geography and spatial analysis
Electoral geography studies the phenomenon of voting based on aggregate data and considering the region as the social container, that is, the place where voting takes place. Peschard (1995, cited in Gómez andValdés, 2000) defines it as follows: Cartographic method to describe the regional distribution of partisan forces, which makes it possible to identify the evolution of political trends both in the areas where a party is rooted and in the areas of change in the political orientation of the electoral (p. 19).
But it would be the contributions of Cox (1969Cox ( , 1987, from political geography, that would promote the local context, considered an important element in the interpretation of voter decisions (Cox, 1969, cited in Broner, 2009, as well as well as economic conditions and local political affairs (Cox, 1987, cited in Vilalta, 2008b. The author Broner (2009) comments that many authors have accepted that voters are influenced as much by their immediate social and geographical environment as by their individual situation (Flint, 1995), while others have remained skeptical (King, 1996). To the list of works on the subject, Hernández (2015) identifies several investigations on the relationship between elections and space: On this subject, we find a wide variety of works that explicitly seek to explore the hypothesis of the non-randomness of electoral results. (Ward et al., 1996;O´Loughlin, 2002O´Loughlin, , 2003Darmofal, 2006;Klos, 2008;Tam Cho & Nicley, 2008;Soares & Terron,2008;Chen & Rodden, 2009;Seabrook, 2009;Lefebvre & Robin, 2009;Cutts & Webber, 2010;Sue Wing & Walker, 2010;Rodden, 2010;Crespin, Darmofal & Eaves, 2011) (p. 186).
In our country, Vilalta (2004Vilalta ( , 2006Vilalta ( , 2007Vilalta ( , 2008b and Lizama (2012) would be the ones who incorporate the theoretical and methodological discussion of the so-called spatial analysis. Since then, its analysis and application in social phenomena has begun to be integrated more frequently (Garrocho, 2016).
Although Vilalta (2006) himself comments that the use of the techniques was due in part to the following: 1) The diffusion in the use of geographic information systems (GIS), 2) the advances of the government to facilitate the public availability of updated statistical information and time series, and 3) the popularization of surveys and maps in sociodemographic matters, economic and political (p. 2).
The methodology proposed from spatial econometrics is mainly due to the work of Anselin (1995), as well as the tools developed in the GeoDa Center for Geospatial Analysis and Computation, 2 which has contributed to the improvement and dissemination of algorithms for spatial analysis, such as the Moran index (1948) and LISA. Vilalta (2006) clarifies that in the spatial analysis the use of the concepts autocorrelation and spatial dependence: They mean the same thing, but the difference in the use of words is that the first term refers simultaneously to a phenomenon and a statistical technique, and the second to a theoretical explanation. Specifically, spatial dependence exists when "the value of the dependent variable in a spatial unit is partially a function of the value of the same variable in neighboring units" (Flint, Harrower and Edsall, 2000: 4). This occurs for a theoretically important reason that summarizes Tobler's (1970) first law of geography: "Everything is related to everything, but things that are closer are more related than things that are distant." (p. 91).
The description of the concept of autocorrelation by the authors Siabato and Guzmán (2019) helps to understand its usefulness in electoral analysis: The essence of autocorrelation is to analyze the variability of a phenomenon through geographic space to determine spatial patterns and describe its behavior, that is, it can be understood as the means to understand how the phenomenon is distributed in the analyzed space and to what degree. local elements can be affected by their neighbors (p. 2).
For Sonnleitner (2013), the reflection on the individual and collective link that we as citizens have at the time of exercising universal suffrage, as well as the methodological tools provided by the disciplines of geography and cartography for the disaggregation of electoral data in different scales, are some of the substantial attributes that these works present. As a paraphrase: we vote like our neighbors. Likewise, she recognizes that the spatial analysis with electoral data contributes to the characterization of the region and micro-region in terms of territorial dynamics, concentration, fragmentation and spatial dispersion, among other characteristics.
Of more recent dates are the works of the study of the electoral geography of Charles, Torres and Colima (2018), who investigate the relationship between ideology and space.

Regional analysis with electoral data
In recent years, efforts to highlight the region variable have been notorious in electoral studies. In the words of Vilalta (2004): "For Mexico it is evident that the local or regional context is decisive for the vote" (p. 86). At the regional level, for the case of Tlaxcala, the work of González (1994) was the first document to outline an electoral geography based on indicators such as the behavior of the variables of participation and electoral abstention for the entity as a whole, according to the federal, local and municipal electoral results between 1979 and 1992, as well as the degree of consistency of the party system in the entity and the correlation between the electoral weight of the parties and the municipal socioeconomic development.
However, since 1998, when the first political alternation took place in the gubernatorial elections, the study of regional electoral analysis has been examined considering the region as a descriptive variable and space as an explanatory variable. This has made it possible to georeference electoral preference following the strict sense used as a synonym for voting (Anduiza and Bosch, 2004, p. 28), as well as some electoral indicators.
For the case that concerns this work, an exploratory analysis of spatial data from the 2021 elections in Tlaxcala is presented with the aim of explaining the spatial effects through techniques that "allow describing spatial distributions, identifying atypical locations (spatial outliers) , discover spatial association schemes (spatial cluster) and suggest spatial regimes or other forms of spatial instability (Moreno and Vayá, 2000, p. 29). It should be noted that it was Vilalta (2008)  The spatial data exploration was carried out under the premise that the vote, understood through the EP and the political preference for a political party or coalition, does not have a random distribution:

•
Hn : There is a spatial behavior of the vote.
• Ha : There is independence between the vote and the space. (1948) is the statistic that measures the existence or absence of dependency or spatial autocorrelation of a variable, whose values range between 1 and -1 (positive and negative autocorrelation), respectively, while the value 0 indicates the existence of random spatial patterns of the variable. "That is, the value that a variable takes in a region is not only explained by internal conditioning factors but also by the value of that same variable in other neighboring regions" (Moreno and Vayá, 2000, p. 21) His formula is expressed as follows:

Now, the Moran index
Vol. 11, Núm. 22 Julio -Diciembre 2022 As: • N: number of geographic units on the map (municipalities) • Hn: I = 0 → Hn: Moran's global coefficient (I) is equal to zero, so there is no spatial autocorrelation, rather a random distribution of the voting phenomenon is observed.
• Ha: I ≠ 0 → Ha: Moran's global coefficient (I) is not equal to zero, so there is spatial autocorrelation of the voting phenomenon.
Next, the Moran index is presented according to the neighborhood criteria. Of these, the queen criterion was decided in this work (table 1). Table 2 shows the four quadrants of the Moran scatterplot. Table 3 Table 3 presents the results of the Moran index from 1998 to 2021 for both EP and political preferences. As can be seen, there is none equal to zero; that is, the existence of spatial autocorrelation of the vote with a negative trend is statistically verified and, given that the data does not reach 3 the expected significance (0.05), the phenomenon of the vote analyzed from spatial econometrics does not allow advancing to the second phase, that is, proposing a spatial econometric model.

Therefore, only the relationship between what happens at a point in space and
another place is demonstrated. A. Knowing that spatial dependence "occurs when the values observed in a region depend on the values of neighboring regions" (Agudelo, 2010, p. 15), and since it is negative here, it means that the presence of an attribute hinder their neighborhoods.
The circles in each quadrant (figure 1) represent the municipalities observed according to the neighborhood criterion used, which for this exercise was the queen Regarding the LISA, it identifies in a cartographic way the spatial local dependency from the Moran index. (Lizama, 2012, p.19). Its formula is expressed as follows: As: • n: Number of municipalities on the map.  For the first time and derived from the electoral reform approved in August 2020, the criteria of gender parity, integrity by two blocks according to voting percentages, parity candidacies vertically, horizontally and transversally, among others, were going to be taken into account. , as well as the new provisions to punish political violence for reasons of gender, although some of these criteria did not apply to new political parties or coalitions that had not competed in previous elections (Zempoalteca, September 9, 2020). However, the Supreme Court of Justice of the Nation (SCJN) annulled, towards the end of November, the modifications made by the local legislature due to errors in the procedure. It left without effect the gender parity guidelines, the term of separation from the position and the integration of an inclusive language; In addition, candidates for a popularly elected position were also not required to prove that they were free of a record for the crime of political violence against women (Cruz, November 30, 2020).
For this election, there were six candidates and one candidate for the governorship. Two of them, at the head of historical and ironic coalitions, marked and concentrated electoral preferences. Both coalitions competed with this formula for said position and in some local districts, but not for the positions in city councils and community presidencies.

Some final thoughts
The 2021 elections meant the victory of Morena and her coalition led by Lorena Cuéllar Cisneros, but they also implied a territorial reconfiguration of the main political forces in the entity that, together with the new political parties (associated with traditional and local political actors) , have blurred the party system at the local level and whose incursion into political competition further atomizes electoral preference.
The analysis of the territorial reconfiguration of the vote is limited, as shown in figure 8 and table 8, since the region is used as a container where the electoral processes take place.
However, when using the spatial analysis of the vote, regions are presented whose neighborhood criterion draws and identifies the negative autocorrelation trend of both the PE and the two coalitions, with minimal clusters that change in each election. So the voting behavior in this election is not randomly distributed in space. This trend is added to the characteristics of the vote in the state for the election of governor: volatile, personalized, competitive and spatially dependent. This is a non-traditional work on voting and a contribution to voting studies in the entity, since although electoral behavior as a social phenomenon occurs in space and time, its incorporation presents comprehensive results (methodologically) with the spatial analysis.
The results are potentially interesting to expand in complementary works, as a future line of research, by incorporating the rest of the positions to be elected, as well as the incorporation of socioeconomic variables and electoral indicators, among others.