Social Identity and Voting Behavior in a Deeply Divided Society: The Case of Israel

Abstract

This paper examines what best explains political behavior in a deeply divided society. Despite the democratic nature of the Israeli political system, we suggest that Israel’s society is characterized by social affiliations mainly defined by social identity that include race, ethnicity, and religion and amount to tribalism. Based on the results of the 2020 election, we examined whether these social affiliations or other socioeconomic characteristics better explain voting behavior. First, we found a significant correlation between tribalism and voting behavior in Israel and socioeconomic characteristics and voting behavior in Israel. When comparing tribalism’s correlation versus other socioeconomic characteristics, we found that the correlation between tribalism and voting behavior in Israel trumps the correlation between socioeconomic characteristics and voting behavior in Israel, except for the Ashkenazi Jews. In other words, voting behavior is better explained by affiliation to the social group, that is, by kinship, race, ethnicity, and religion, than by other socioeconomic characteristics such as education level, economic status, or geographic location. This extends the idea of “in-group favoritism”.

1. Introduction

This study investigates the factors influencing voting behavior in deeply divided societies while focusing on social identity. Drawing from the social identity theory [1], our research explores the concept of “in-group favoritism”, which refers to the tendency to favor members of one’s in-group over out-group members [2]. This phenomenon can be observed across various social identities, such as work-related groups [3], immigration-related status [4], political parties [5], age [6], gender [7], nation [8], race [9], ethnicity [10], religion [11], culture [12], or geographic location [13]. Using the case of Israel, our study aims to understand how this “in-group favoritism” manifests in the context of voting behavior in a deeply divided society.

The connection between social affiliation and politics has already been mentioned concerning political coalitions [14] or political competition over resources and power [15]. Politics, in particular, elicits firm in-group commitments [16]. Political in-group favoritism can take the form of preferring to vote for an in-group candidate—hence, “identity voting” [17]. Such political in-group favoritism has been documented globally. Preferring to vote by race and gender affiliation was shown in the USA [18]. Voting by ethnicity was the primary determinant of the party vote in South Africa [19]. Religious belonging, behavior, and belief influenced voting behavior in Britain [20]. Preferring to vote by cultural similarities was documented in the parliamentary elections of the Basque Autonomous Community when language and Basque sentiment were found to have a significant correlation with voting [17]. The effect of geographical affiliation of the candidate on voting has been witnessed in Turkey, the USA, and Japan. In Turkey, favoring the local candidate was documented in legislative elections [21]. In the USA, an advantage has been shown for the “native son presidential candidate” [22]. In Japan, an advantage was suggested for constituency candidates [23].

Voting behavior in democracies has been under examination by scholars. A more complex situation exists in deeply divided societies, since they have long been seen as challenging for maintaining stability and democracy [24]. Low [25] provides examples from Africa and Asia to show that tribalism and ethnic division in newly independent states are impossible to maintain as a democratic system. Others bring examples from Eastern Europe and countries of the former Soviet Union [26]. However, some scholars have taken a different approach.

In any case, the problem exists. Divided societies are made of a multilayer of major groups that are different in race, ethnicity, culture, language, and religion, to name a few. Geographic areas of living, economic status, and institutions might also separate such groups. Agarin et al. [27] extend the problem in deeply divided societies to the “Others”. They argue that the systematic exclusion of others is a central democratic problem in a divided society. This further strengthens the importance of identifying all relevant social groups and studying their voting patterns.

A more extreme part of social formation in deeply divided societies is tribalism, which is defined as being organized in a tribe or tribes. The definitions of what constitutes a tribe have shifted over history. Definitions include an “autonomous community”, “separate political community”, and “distinguished by cultural-regional criteria” [28]. Tribalism is a form of organizing social groups based on tribes with various levels of kinship [29], such as within the race [30], ethnicity [31], religion [32], political coalitions [14], or even groups with shared property, worldview, and ideologies [32]. Tribalism includes pride in belonging and loyalty to the tribe [33]. It is often used to describe societies in African, Arab, and Muslim Middle Eastern countries [28,34,35]. Also, tribalism has been used to describe one of the stages of people’s development toward statehood [36]. It sometimes explains failure in advancing democratic institutions [37]. However, in this study, we offer an understanding of tribalism in a different context. We suggest that tribalism exists in Israel and extends to the Jewish majority, not solely being a phenomenon among Arab citizens; it is the opposite of the state-building process. Tribalism might have been a result of colonialism, but the Israeli case will show that it is possible for this phenomenon to have been copied into a society self-proclaimed as western.

We will build on the established scholarship on Israel as a deeply divided society and extend the model to claim that Israel’s society is organized as tribal, based on various levels of kinship such as race, ethnicity, religion, and political coalitions [14,29,30,31,32]. In addition, we will test how tribalism affects political behavior in Israel by dictating political in-group favoritism [17]. Such groups are organized as tribes based on kinship, race, ethnicity, religion, culture, or geographic location [17,18,19,20,21,22,23]. The underlying assumption of this attitude is that changes in democratic societies have increased the importance of sub-societies and subcultures [38]. Hence, the aim is to examine social affiliation and political in-group favoritism in the case study of Israeli society. We extend previous suggestions to examine political behavior in Israel through its cleavages [39]. Others have also researched the topic while focusing on the role of economy, education, ethnicity, religious affiliation, age, and gender [40,41].

This study seeks to explore what explains voting behavior in deeply divided societies. Hence, the research questions are the following: Do members of a kinship-based social group favor a fellow member over out-group members, specifically when voting for an in-group candidate in elections? Do socioeconomic characteristics affect voting behavior? In other words, this study explores whether voting behavior is better explained by social affiliation or other socioeconomic characteristics.

2. Background: Israel as a Deeply Divided Society

This section explains why Israel was chosen as the case of a deeply divided society. It is a suitable case to examine political behavior and, more specifically, the effect of social structure on in-group political favoritism. In its early years, Israel embraced the “melting pot” policy that was aimed at creating a national society through the ingathering of the exiles [42]. Current scholarship has debated the successes and failures of the “melting pot” policy in Israel [43]. Some bring evidence that the melting pot policy did not achieve its declared targets [44]. The term “tribe” has already been used in the Israeli Jewish context to describe social differences, race, ethnicity, and origin by geography [45,46,47]. However, it was coined by Israeli President Rivlin in his speech titled “The Four Tribes”, when he voiced his concerns about the future of Israeli identity [48]. President Rivlin suggested that Israel’s society is divided into four distinct “tribes”: secular, national religious, Haredi (Ultra-Orthodox), and Arab.

This suggestion is in line with some of the current scholarship defining Israel as a deeply divided society [49,50,51,52]. With some differences between scholars, it is suggested that five major divisions exist in Israel’s society: first, the national divide between Jews and Arabs; second, the religious divide between the secular and religious; third, the political divide over Zionism, the future of Israel, and the Jewish people; fourth, ethnic divide between Ashkenazi and Sephardic/Mizrahi Jews; and fifth, socioeconomic divide that overlaps with the previous three divides of the Israel society [53,54].

The first divide is national, which distinguishes Arabs as a minority from the Jewish majority by national affiliation, culture, and religion. The religious divide categorizes three social groups among Jewish Israelis: “Secular Jewish”, a “Jewish national-religious” group, and “Jewish Haredi”. They are distinct from the Arabs by culture, national affiliation, and religion, and different from one another by their levels of religiosity. The political divide over Zionism further distinguishes the Arabs from those three Jewish social groups. However, it also helps distinguish the “Jewish Haredi” from the other two Jewish social groups due to their different attitudes towards Zionism. Further divides exist and overlap with these four social groups—the gender divide, the political divide between the left and right, and the socioeconomic divides between the rich and poor and between the center and periphery. All these four social groups, Arabs, Secular Jewish, the Jewish national-religious, and Jewish Haredi, comprise the first level of social affiliation.

Within those social groups, further coherent cleavages produce the second level of social affiliation, further explaining Israel as a deeply divided society. Arabs constitute about 21% of Israel’s population—within Israel proper as of 2019 [55]. Most Arabs in Israel see their Arab and Palestinian identity as a defining factor distinguishing them from Jewish Israelis [56]. However, they are divided into three religious affiliations, with Muslims as the majority among them, with 82% out of the Arabs [57]. A distinct social group among Muslims in Israel are Bedouins [58]. Their social structure and traditions differ from other Muslims [59]. Most Bedouins reside in southern Israel, in the Negev, making them segregated from other Arabs or Muslims, hence being a distinct sub-group of Muslims in Israel. Christians comprise 9% of Arabs in Israel as of 2018 [60]. The fourth coherent cleavage among Arabs in Israel is the Druze, comprising 8% of Israeli Arabs [61]. Each group in these four cleavages among Arabs uses their group as a source of social self-identification [62]. Hence, we will label them as sub-social groups of Arabs in Israel and as the second level of the divide.

The majority of Israeli society is Jewish by their cultural, religious, and national affiliation, with about 79% of the population being Jewish and others [55]. Most of them rank their Israeli Jewish identifications first before any subethnic identification. However, they do not eliminate differences [63]. The first factor in helping distinguish between the different groups in Jewish Israeli society is the level of religiosity—secular, traditional, religious, or Haredi [64].

The Secular Jewish group, including the secular and part of the traditional group, comprises about 51% of Israel’s population [55,65]. It is ethnically divided into four coherent cleavages between the majority that formed the Jewish Yishuv before the state of Israel—Ashkenazi Jews—and, on the other hand, all the immigrants that were absorbed into the state of Israel after its establishment [66,67], including Sephardic Jews native to Palestine pre-1948, as well as new immigrants, and Mizrahi immigrants since the early years of the state, also including those from the massive Russian immigration in the 1990s and the smaller-in-scale Ethiopian immigration. Each group holds their ethnic background as part of their social self-identification, although at various levels [68,69,70].

Moving forward with religiosity, the religious camp is categorized into the secular and traditional. It includes those who identify as religious or Haredi. However, they do not make one coherent social group. The political divide between them creates two coherent groups. The first is the “National-Religious” group that adopts Zionist ideas [71]. The second is the “Haredi”, who do not adopt Zionist ideas [72]. The National-Religious group constitutes about 17% of Israel’s population [65]. It includes three main streams of thought represented by coherent groups: mainstream national-religious, modern-liberal, and National Haredi Jews [65]. The Haredi comprise about 11% of Israel’s population [73]. This group is divided into two coherent groups, mainly along ethnic lines, Ashkenazi and Sephardic/Mizrahi Jews [74]. Within the Ashkenazi Haredi, further division distinguishes those from Lithuanian, Hasidic, Edah HaChareidi, and other places of origin [75]. Among the Mizrahi, further division is seen mainly by their countries of origin, such as Iraqi Jews, Moroccan Jews, or Yemenite Jews [76].

Considering all these divisions in Israel’s society, we suggest that these form four social groups at the first level of division, which are then divided into 13 sub-divisions, as presented in Figure 1.

To help maintain this segregation, there are state policies and contributing societal practices [77]. Segregation in housing is a societal trend common in Israel. Most of the members of these social groups and subgroups prefer to live in localities or neighborhoods with other people that are close to their social group and subgroups as much as possible (for the Ashkenazi–Mizrahi divide, see Tzfadia [78]; for Russians and Ethiopian immigrants, see Offer [79] and Mesch and Mano [80]; for Arabs, see Falah [81]; for the religious–secular divide, see Flint et al. [82]). Additionally, endogamy marriages further help make this social affiliation in Israel’s society a reality (for the Ashkenazi–Mizrahi divide, see Okun [83] and Gshur and Okun [84]; for the Arab–Jewish divide, see Raz and Atar [85]; for the religious–secular divide, see Lehmann and Siebzehner [86]). Cultural preservation, such as language, food, and religious traditions, and segregation in synagogues by country of origin, such as Latvian or Moroccan, add additional components to help maintain divisions in Israeli society.

This suggested division of Israel’s society into social groups and subgroups is directly connected to and maintained by the separate education streams within Israel’s education system [87]. The Ministry of Education divides the education sectors into Hebrew education, Arabic education, public education, and private education [88]. It has four separate education streams: the Arabic public education stream, Hebrew public education stream, Hebrew public education religious stream, and Haredi stream [89]. These education streams are further divided and help maintain the above tribal and sub-tribal affiliations, such as the separate education stream for the Druze or the separate education system for Sephardic Haredi Jews, Ashkenazi Haredi Jews, and the National-Religious education steam.

3. Research Setting

3.1. Research Model and Hypothesis

This study focuses on voting for parties per locality, following studies such as Afriat and Dahan [90]. Afriat and Dahan [90] examined the impact of social cleavages on voting patterns, emphasizing the role of social identity in political preferences. Their model included testing social affiliation at one level and socioeconomic characteristics, such as education and salaries. Our model extends their work by incorporating additional socioeconomic factors, extending social affiliation to three levels, and evaluating their relative influence on voting behavior. Therefore, the baseline of our model is shown in Figure 2, and our three primary hypotheses posit the following:

H1: 

In deeply divided societies, the more the residents in a locality belong to a social group at the first level (tribe), the more the party that represents that social group (tribe) receives votes, regardless of socioeconomic characteristics.

H2: 

In deeply divided societies, the more the residents in a locality belong to a social group at the first (tribe) and the second level (subtribe), the more the party that represents that social group (subtribe) receives votes, regardless of socioeconomic characteristics.

H3: 

In deeply divided societies, the more the residents in a locality with a hometown candidate belong to a social group at the first (tribe) and the second level (subtribe), the more the party that represents that social group (subtribe) receives votes, regardless of socioeconomic characteristics.

These hypotheses encompass three levels of examination—H1 for social division at the first level with affiliation to a tribe, H2 for social division at the second level with affiliation to a sub-tribe, and H3 for a hometown candidate at the third level. Each set of hypotheses encompasses four types of socioeconomic correlations with political behavior: Geography, Population, Economy, and Education. We compare the correlation of social affiliation versus socioeconomic characteristics on voting behavior. By doing so, we claim that social cleavages (tribe, sub-tribe, and hometown candidate at the locality level) trump socioeconomic characteristics in correlation with voting behavior. The merit of these hypotheses is that they test all variables within one examination while controlling for the other independent variables. This extends the findings of scholars who have already found a significant correlation between voting patterns with demographic transition and economic levels in Israel [91].

Figure 2. The research model. (adaptation of Zeedan and Hogan [88] and Zeedan [92]).

Figure 2. The research model. (adaptation of Zeedan and Hogan [88] and Zeedan [92]).

3.2. The Dependent and Independent Variables

This study examined the correlation between twenty-two independent variables and eight dependent variables. The independent variables are divided into two major groups: social affiliation and socioeconomic characteristics. For social affiliation, we used three sets of variables. The first set represents the belonging to one of the four social affiliations at the first level: Arab (TR1), Secular Jewish (TR2), Jewish National-Religious (TR3), and Jewish Haredi (TR4). The second set represents the belonging to one of the subgroups—the second level of affiliation (subtribe)—Muslim (ST1), Christian (ST2), Druze (ST3), Bedouin (ST4), Russian Jews (ST5), Ethiopian Jews (ST6), Ashkenazi Jews (ST7), and Mizrahi Jews (ST8). The third set of independent variables from the social affiliation group represents the native hometown candidate—the number of candidates who live in the locality and are likely to be elected in the list (NC1).

The second group of independent variables is socioeconomic characteristics. We used four sets of socioeconomic characteristics: Geography, Population, Economy, and Education. For the first set, Geography, we used the following variables: District (GE1), Municipal status (GE2), and Settlement type in a regional council (GE3). For the second set, Population, we used the following variables: Density per square meter in 2018 (PO1) and Percentage of residents under the age of 18 in 2018 (PO2). For the third set, Economy, we used the following variables: Average monthly salary for employees (EC1) and Percentage of employees with wages below the average salary (EC2). For the fourth set, Education, we used the following variables: Percentage of degree holders among the population aged 33–55 (ED1) and Percentage of students in higher education out of all population (ED2).

The dependent variable is the percentage of the vote in the localities to each one of the eight parties that obtained seats in the Knesset: Labour-Gesher-Meretz (V1), United Torah Judaism (V2), Joint List (V3), Yamina (V4), Yisrael Beiteinu (V5), Likud (V6), Blue and White (V7), and Shas (V8).

3.3. The Data

For these variables, data were collected from two primary sources. For the variables of social affiliation and socioeconomic characteristics, the data were collected from the 2018 publication on the localities in Israel by the Central Bureau of Statistics (CBS). The data on the native hometown candidate (NC1) were collected from the personal information of all 120 elected members of the Knesset (MKs), as posted on the Knesset website. The percentage of the vote in the localities of each party was collected from the databases of the Israeli Election Central Committee (ECC). Supplemental data were collected from other official and NGO websites.

Table A1. Descriptive statistics of the research variables.

Variable Type		Variable	Min	Max	Mean	SD
Independent variables	Socioeconomic characteristics	GE Geography
		GE1: District	1	7	3	2
		GE2: Municipal status	1	4	2	1
		GE3: Settlement type in regional councils	1	5	1.76	0.74
		PO Population
		PO1: Density per square meter (in localities with 5000 residents or more in 2018)			68.3	26,368.4	3331.6	3349.6
		PO2: Percentage of residents under the age of 18 in 2018			20.9%	64.4%	34.8%	7.7%
		EC Economy
		EC1: Average monthly salary of employees (NIS in 2016)			4612.0	17,116.7	8535.0	2751.2
		EC2: Percentage of employees with wages below the average salary (2016)			24.6%	65.7%	41.8%	9.9%
		ED Education
		ED1: Percentage of degree holders among the population aged 35−55 in 2017			1.9%	69.1%	25.8%	17.0%
		ED2: Percentage of students in higher education out of the entire population in 2017			0.4%	11.1%	3.2%	1.3%
	Social affiliation	TR The social affiliation at the first level
		TR1: Arab (percentage of Arabs in the locality)			0%	100.0%	39.3%	47.9%
		TR2: Secular Jewish			0%	100.0%	48.7%	46.8%
		TR3: Jewish National-Religious			0%	100.0%	81.2%	34.7%
		TR4: Jewish Haredi (percentage of Haredi people in the locality)			2.0%	100.0%	29.6%	31.5%
		ST The sub-group: social affiliation at the second level
		ST1: Muslims (percentage of Muslims, including Bedouins, in the locality)			0.0%	100.0%	30.0%	44.2%
		ST2: Christians (percentage of Christians in the locality)			0.0%	99.8%	2.7%	11.7%
		ST3: Druze (percentage of Druze in the locality)			0.0%	100.0%	5.4%	21.4%
		ST4: Bedouin			0.0%	100.0%	15.6%	30.8%
		ST5: Russian Jews (percentage of Russians of the 1990s immigration in the locality)			0.0%	45.5%	7.5%	9.0%
		ST6: Ethiopian Jews (percentage of Ethiopian Jews in the locality)			0.0%	16.3%	0.4%	1.6%
		ST7: Ashkenazi Jews (percentage of Ashkenazi Jews in the locality)			0.0%	70.3%	29.9%	11.4%
		ST8: Mizrahi Jews (percentage of Mizrahi Jews in the locality)			0.0%	54.2%	21.3%	7.1%
		NC Hometown native candidate
		NC1: Number of candidates who live in this locality are likely to be elected in the list—all Israel
		Labour-Gesher-Meretz			0.00	3.00	0.04	0.24
		United Torah Judaism			0.00	5.00	0.03	0.33
		Joint List			0.00	2.00	0.06	0.25
		Yamina			0.00	1.00	0.03	0.18
		Yisrael Beiteinu			0.00	1.00	0.03	0.16
		Likud			0.00	7.00	0.12	0.54
		Blue and White			0.00	6.00	0.12	0.51
		Shas			0.00	4.00	0.03	0.27
Dependent variables	Percentage of vote for the party:	V1 Labour-Gesher-Meretz			0%	33%	6%	6%
		V2 United Torah Judaism			0%	80%	3%	10%
		V3 Joint List			0%	100%	31%	43%
		V4 Yamina			0%	70%	6%	11%
		V5 Yisrael Beiteinu			0%	29%	4%	5%
		V6 Likud			0%	63%	22%	19%
		V7 Blue and White			0%	74%	23%	21%
		V8 Shas			0%	51%	5%	8%

shows the descriptive statistics of the research variables.

3.4. Procedure

To validate the research model, we statistically examined the correlation between the independent and the dependent variables. Pearson correlation was used for those sets of independent/dependent variables, when both are numeric. When the correlation is significant and positive, it means that the independent variable is correlated with voting behavior, while the higher the value of the dependent variable, the higher the value of the variable of the voting behavior. Other sets of variables that we used included one dichotomic and one numeric variable. We examined these sets’ correlations with the dependent variables using the Biserial correlation co-efficient. If the Biserial correlation coefficient is positive and significant, the value (1) is more associated with voting behavior than the other value (0).

The first step was to examine the correlations between social affiliation and socioeconomic variables versus voting behavior. In the second step, we examined the hypotheses with Pearson correlations of all variables in the study, followed by a regression analysis. We used the log of our dependent variables (percentage of votes for the party) to examine the correlation of the various independent variables to the dependent variables. Standardized coefficients allowed us to bring all variables to a standard scale, making comparisons much easier. We included social affiliation and socioeconomic variables in the models so that the regression analysis could help us determine which could be used to better predict political behavior in the form of the chosen party. To test the hypotheses, we examined how the co-efficiencies with the vote for the party compare social affiliation (first level, second level, third level) versus Geography, Population, Economy, and Education. For the statistical procedures to analyze the data, we used the IBM SPSS 28.0 software package.

4. Findings

Table 1. Pearson correlations between social affiliation (first level, second level, and third level) versus socioeconomic characteristics, and voting behavior.

Dependent Variables/Independent Variables	N	V1 Labour-Gesher-Meretz	V2 United Torah Judaism	V3 Joint List	V4 Yamina	V5 Yisrael Beiteinu	V6 Likud	V7 Blue and White	V8 Shas
Socioeconomic characteristics (geography, population, economy, education) GE1: District (Biserial—the chosen district versus the others)	276		0.159 **		0.252 ***	−0.141 *		−0.121 *
GE2: Municipal status (Biserial—the chosen status versus the others)	276	0.183 **	−0.168 **	0.228 ***		−0.403 ***	−0.334 ***		−0.290 ***
PO1: Density per square meter (in localities with 5000 residents or more in 2018)	162	−0.133 *	0.183 **		−0.139 *		0.121 *		0.264 ***
PO2: Percentage of residents under the age of 18 in 2018	255		0.348 ***	−0.130 *	0.202 ***	−0.138 *			0.283 ***
EC1: Average monthly salary of employees (NIS in 2016)	255	0.491 ***		−0.655 ***	0.171 **	0.143 *	0.424 ***	0.730 ***
EC2: Percentage of employees with wages below the average salary (2016)	255		0.223 ***					−0.206 ***	0.212 ***
ED1: Percentage of degree holders among the population aged 35–55 in 2017	255	0.559 ***	−0.138 *	−0.608 ***	0.386 ***		0.296 ***	0.714 ***	−0.145 *
ED2: Percentage of students in higher education out of all the population in 2017	255	0.359 ***	−0.162 **	−0.491 ***	0.462 ***	0.153 **	0.325 ***	0.441 ***	−0.140 *
TR The social affiliation at the first level (H1) TR1: Arab (percentage of Arabs in the locality)	124	−0.298 ***	−0.250 ***	0.896 ***	−0.391 ***	−0.267 ***	−0.763 ***	−0.506 ***	−0.396 ***
TR2: Secular Jewish (Biserial versus other Jewish social groups)	1105	0.453 ***		−0.730 ***		0.344 ***	0.681 ***	0.666 ***
TR3: Jewish National-Religious (Biserial versus other Jewish social groups)	154	−0.154 *		−0.201 ***	0.845 ***	−0.129 *	0.164 **	−0.151 *
TR4: Jewish Haredi (percentage of Haredi people in the locality)	40	−0.210 ***	0.930 ***	−0.230 ***				−0.208 ***	0.786 ***
ST The sub-group: social affiliation at the second level ST1: Muslims (percentage of Muslims, including Bedouins, in the locality)	89	−0.382 ***	−0.211 ***	0.935 ***	−0.341 ***	−0.443 ***	−0.723 ***	−0.611 ***	−0.350 ***
ST2: Christians (percentage of Christians in the locality)	37			0.237 ***			−0.218 ***	−0.158 ***
ST3: Druze (percentage of Druze in the locality)	25	0.144 *				0.293 ***		0.228 ***
ST5: Russian Jews (percentage of Russians of the 1990s immigration in the locality)	192			−0.509 ***		0.710 ***	0.654 ***		0.226 ***
ST6: Ethiopian Jews (percentage of Ethiopian Jews in the locality)	20	−0.129 *		−0.265 ***		0.288 ***	0.458 ***		0.293 ***
ST7: Ashkenazi Jews (percentage of Ashkenazi Jews in the locality)	542			−0.540 ***		0.580 ***	0.596 ***	0.228 ***	0.233 ***
ST8: Mizrahi Jews (percentage of Mizrahi Jews in the locality)	542			−0.567 ***			0.743 ***	0.196 ***	0.423 ***
NC Hometown native candidate NC1 Number of candidates who live in this locality and are likely to be elected in the list—all Israel	276		0.256 ***	0.186 **	0.254 ***		0.131 *	0.206 ***	0.145 *

Notes. Significance: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001; Other variables were tested; however, they did not provide any additional information: GE—Israel proper versus the settlements in the West Bank and peripheral index; PO—population by thousands; EC—average monthly income of the self-employed, percentage of the self-employed with wages below the average, Gini index, and Socioeconomic index; Ed—percentage of twelfth graders starting college education within eight years. ST4: Bedouin (Biserial versus other Muslims) was insignificant with all variables; For the number of candidates who live in this locality are likely to be elected in this list, we used the expectation based on the last projected number of seats for the party and one extra, Labour-Gesher-Meretz: 9 (+1); United Torah Judaism: 7 (+1); Joint List; 15 (+1); Yamina: 8 (+1); Yisrael Beiteinu: 6 (+1); Likud: 33 (+1); Blue and White: 33 (+1); and Shas: 8 (+1).

shows the correlations of the research variables, including social affiliation, socioeconomic characteristics, and party voting. Correlations exist between socioeconomic variables and voting and between social affiliation and voting. Hence, this study is essential for comparing these correlations against each other.

4.1. Findings Related to H1: In Deeply Divided Societies, the More the Residents in a Locality Belong to a Social Group at the First Level (Tribe), the More the Party That Represents That Social Group (Tribe) Receives Votes, Regardless of Socioeconomic Characteristics

At the first level of social affiliation (the “tribe”), Arab-majority localities, thus belonging to the Arab social group (TR1), are positively correlated with voting for the “Joint List”. The higher the percentage of Arabs in a locality, the higher the vote for the Joint List is (r = 0.896, p < 0.001). The vote in these Arab-majority localities negatively correlates with the other parties. The higher the percentage of Arabs in a locality, the lower the vote for all other parties than the Joint List (r ranges from −0.267 to −0.762, p < 0.001).

Localities with a majority of the population being Secular Jews, thus belonging to the Secular Jewish group (TR2), are positively correlated with voting for five parties: Labour-Gesher-Meretz, Yisrael Beiteinu, the Likud, and Blue and White. Compared to the other two Jewish social groups, the National-Religious and the Haredi, the vote in the Secular Jewish localities is higher for the Labour-Gesher-Meretz, Yisrael Beiteinu, Likud, and Blue and White (0.453, 0.334, 0.681, and 0.666 respectively, p < 0.001). There is a negative correlation between the Secular Jewish group (TR2) and the vote for the Joint List (−0.730, p < 0.001). Localities with a majority of National-Religious Jews, thus belonging to the National-Religious social group (TR3), are positively correlated with voting for Yamina and the Likud. Compared to the Secular Jewish and the Haredi, the vote in the National-Religious localities is higher for Yamina and the Likud (0.845 and 0.164, p < 0.001). Compared to the Secular Jewish and the Haredi, the vote in the National-Religious localities is lower for the Labour-Gesher-Meretz, Yisrael Beiteinu, and Blue and White (−0.154, −0.129, −0.151, p < 0.01), as well as with the Joint List (−0.201, p < 0.001). Lastly, Haredi-majority localities, thus belonging to the Haredi group (TR4), are positively correlated with voting for United Torah Judaism and Shas. The higher the percentage of the Haredi in a locality, the higher the vote for United Torah Judaism and Shas (0.930 and 0.786, p < 0.001).

Not surprisingly, for social affiliation, we found that Labour-Gesher-Meretz voters are mainly Secular Jewish. The United Torah Judaism voters are Haredi Jewish. The Joint List voters are mainly Arabs. Yamina’s voters are from the National-Religious social group. Voters for Yisrael Beiteinu are from the Secular Jewish social group. The Likud voters are from the Secular Jewish and the National-Religious groups. Blue and White voters are from the Secular Jewish group. Shas voters are from the Haredi Jewish group.

Next, we will report the correlations between socioeconomic characteristics and voting. In addition to the findings in Table 1, in this section, we report additional findings that are listed in

Table A2. Additional data on the relationships between socioeconomic characteristics (geography) and voting behavior in Israel’s 2020 election.

Independent Variables: Socioeconomic Characteristics for H2		N	V1 Labour-Gesher-Meretz	V2 United Torah Judaism	V3 Joint List	V4 Yamina	V5 Yisrael Beiteinu	V6 Likud	V7 Blue and White	V8 Shas
H2a: GE Geography
GE1: District (Biserial—the chosen district versus the others)
	Northern	93	ns	−0.162 ***	0.230 ***	−0.212 ***	0.111 **	−0.192 ***	ns	−0.107 **
	Haifa	30	ns	ns	ns	−0.102 **	ns	ns	ns	ns
	Central	52	ns	ns	−0.198 ***	ns	ns	0.153 *	0.254 ***	ns
	Tel-Aviv	12	ns	ns	ns	ns	ns	ns	ns	ns
	Jerusalem	6	ns	ns	ns	ns	ns	ns	ns	ns
	Judea and Samaria	24	−0.172 ***	0.251 ***	−0.227 ***	0.635 ***	ns	0.187 ***	−0.130 *	ns
	Southern	59	ns	ns	0.176 ***	ns	ns	ns	−0.158 ***	ns
GE2: Municipal status (Biserial—the chosen status versus the others)
	City	77	−0.197 ***	0.190 ***	−0.213 ***	−0.106 **	0.376 ***	0.355 ***	ns	0.292 ***
	Local council	124	ns	ns	0.199 ***	ns	ns	−0.214 ***	ns	−0.104 **
	Regional council	54	0.467 ***	ns	−0.258 ***	0.234 ***	−0.190 ***	ns	0.310	ns
	Not recognized	21	−0.204 ***	ns	0.382 ***	ns	−0.204 ***	−0.287 ***	−0.267 ***	−0.167 ***
GE3: Settlement type in a regional council (Biserial—the chosen type versus the others)
	Moshav	416	−0.333 ***	ns	−0.193 ***	−0.166 ***	−0.099 ***	0.457 ***	ns	0.300 ***
	Moshav Shitufi	38	−0.069 *	−0.027 *	ns	0.145 ***	ns	ns	ns	−0.054 **
	Kibbutz	262	0.667 ***	−0.123 ***	−0.120 ***	−0.157 ***	0.061 **	−0.458 ***	0.389 ***	−0.296 ***
	Community settelment	192	−0.184 ***	ns	−0.096 ***	0.363 ***	0.106 ***	0.086 ***	−0.221 ***	ns
	Village	70	−0.175 ***	ns	0.760 ***	−0.073 ***	−0.089 ***	−0.190 ***	−0.260 ***	ns

Notes. Significance: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001; Other variables were tested; however, they did not provide any additional information: GE—Israel proper versus the settlements in the West Bank and peripheral index.

. When examining the differences between the seven districts (GE1), we found that voting behavior is the most differentiated from the other districts by locality in the Northern and Southern districts. In a locality situated in the Northern District, the vote is higher compared to the other districts for the Joint List and Yisrael Beiteinu (0.230, p < 0.001, and 0.111, p < 0.01) and lower for United Torah Judaism and Likud (−0.162, −0.192, p < 0.001). In a locality situated in the Central district, the vote is higher for the Likud and Blue and White (0.153, p < 0.05; 0.245, p < 0.001) and lower for the Joint List (−0.198, p < 0.001). In localities situated in the Judea and Samaria district, we found that the vote is higher, compared to the other districts, for Yamina (0.635, p < 0.001) and lower for other parties. In a locality situated in the Southern district, the vote is higher compared to the other districts for the Joint List (0.176, p < 0.001) and lower for Blue and White (−0.158, p < 0.001). For the following three districts, we found no significant correlations with political behavior that is different from the more peripheral ones: Haifa, Tel-Aviv, and Jerusalem.

Under Geographic characteristics, we examined the effect of the type of municipality on voting behavior. First, we examined the differences between localities designated as cities, local councils, regional councils, or not recognized by the Ministry of the Interior (GE2). In cities, the vote is higher compared to the other types for these parties (p < 0.001): United Torah Judaism (0.190), Yisrael Beiteinu (0.376), Likud (0.355), and Shas (0.292). In local councils, the vote for the Joint List was higher than for the other types of local governments (0.199, p < 0.001). In regional councils, the vote was higher (p < 0.001) for the Labour-Gesher-Meretz (0.467), Blue and White (0.310), and Yamina (0.234).

We also examined the different correlations of the types of settlements within regional councils (GE3) and voting. The most significant correlations are the following (p < 0.001): in a Moshav, a higher vote for the Likud (0.457); in a Moshav Shitufi and Community Settlement, a higher vote for the Yamina (0.145 and 0.363); in a Kibbutz, a higher vote for the Labour-Gesher-Meretz (0.667) and Blue and White (0.389); and in a Village that is part of a regional council, the vote is higher for the Joint List (0.760).

The findings point to a significant correlation between population size and voting behavior. Density per square meter (PO1) was positively correlated with a vote for United Torah Judaism, the Likud, and Shas (0.183, 0.203, and 0.264). The higher the density (PO1), the higher the vote for United Torah Judaism, the Likud, and Shas. Furthermore, the percentage of residents under 18 (PO2) was positively correlated with the vote for United Torah Judaism, Yamina, and Shas (0.348, 0.202, 0.283, p < 0.001).

These findings indicate a significant correlation between the economy and voting behavior, except for United Torah Judaism, the Joint List, and Shas. The higher the average monthly salary for employees in a locality (EC1), the higher the vote for Labour-Gesher-Meretz (0.491, p < 0.001), Yamina (0.171, p < 0.01), Yisrael Beiteinu (0.143, p < 0.05), Likud (0.424, p < 0.001), and Blue and White (0.730, p < 0.001). Also, the higher the percentage of employees with wages below the average salary in a locality (EC2), the higher the vote for United Torah Judaism and Shas (0.223 and 0.212, p < 0.001).

The findings point to a significant correlation between education and voting behavior. The higher the percentage of degree holders among the population aged 35–55 in a locality (ED1), the higher the vote for Labour-Gesher-Meretz, Yamina, Likud, and Blue and White (0.559, 0.386, 0.296, and 0.714, p < 0.001). On the other hand, the higher the percentage of degree holders among the population (ED1), the lower the vote for United Torah Judaism (−0.138, p < 0.01), the Joint List (−0.608, p < 0.001), and Shas (−0.145, p < 0.01). Similarly, we found such correlations with the percentage of students in higher education out of the entire population (ED2). When the percentage of students is higher, the vote for Labour-Gesher-Meretz, Yamina, Yisrael Beitein, Likud, and Blue and White is higher (0.359, 0.462, 0.153, 0.325, and 0.441, p < 0.001). A negative correlation was found between the percentage of students in higher education out of the entire population (ED2) and the vote for United Torah Judaism, the Joint List, and Shas.

In conclusion, we found two major groups based on socioeconomic characteristics. The first group is for voters of Labour-Gesher-Meretz, Yamina, Yisrael Beiteinu, Blue and White, and Likud. The Labour-Gesher-Meretz voters are mainly from Kibbutzim in regional councils, are from cities that have a low percentage of residents under 18, are highly educated, and are high-earning employees. Yamina’s voters are from the Judea and Samaria district or community settlements and Moshav Shitufi that are part of regional councils and that have a higher percentage of residents under 18, are highly educated, and are high-earning employees. Yisrael Beiteinu’s voters are from the Northern district or cities with a low percentage of residents under 18 and high-earning employees. Blue and White voters are from the Central district, or a Kibbutz in a regional council, with a low percentage of residents under 18, who are highly educated, and who are high-earning employees. The Likud voters are from the Central and the Judea and Samaria districts, or from cities or Moshav in a regional council, with a low percentage of residents under 18, who are high-earning employees and highly educated. Therefore, for this camp, voters are not from the peripheral districts but mainly from cities with a low percentage of residents under 18, who are highly educated, and who are high-earning employees.

The other group of voters based on socioeconomic characteristics are voters of United Torah Judaism, the Joint List, and Shas. The United Torah Judaism voters are from the Judea and Samaria district, or cities with high population density and higher percentages of residents under 18, who are less educated, and who are low-earning employees. The voters for the Joint List are mainly from the Northern District, local councils, or villages within regional councils with low population density, a higher percentage of residents under 18, who are less educated, and who are low-earning employees. Shas voters are from Moshav in a regional council or cities with high population density and a higher percentage of residents under 18, who are less educated, and who are low-earning employees. Therefore, this camp’s voters are mainly from cities with a higher percentage of residents under 18, who are less educated, and who are low-earning employees.

Following the findings above, which focused on the Pearson correlations of all variables in this study, the second step was to examine the hypotheses with a regression analysis. This will help determine which has a more significant correlation with voting behavior: is it social affiliation at the first level (tribe) or other socioeconomic characteristics? In H1, we expected a positive relationship between social cleavages in deeply divided societies like Israel and voting. The more the residents in a locality belong to a social group (tribe), the more the party representing that social cleavage (tribe) receives votes, regardless of socioeconomic characteristics.

We tested this hypothesis with four different sets of variables: Geography (location) affects voting behavior, with a positive correlation between the location of a locality and the vote for a particular party; Population (size) affects voting behavior, with a positive correlation between the size of the population in a locality and the vote for a particular party; Economy (wages) affects voting behavior, with a positive correlation between high wages in a locality and the vote for a particular party; and Education affects voting behavior, with a positive correlation between the level of education in a locality and the vote for a particular party. To test H1,

Table 2. Correlations with voting behavior: social affiliation at the first level (the tribe) versus socioeconomic characteristics (H1).

Dependent Variables	V1 Labour-Gesher-Meretz	V2 United Torah Judaism	V3 Joint List	V4 Yamina	V5 Yisrael Beiteinu	V6 Likud	V7 Blue and White	V8 Shas
Constant (B)	−0.813 *** (t = −4.203)	1.242 *** (12.376)	−0.1815 *** (−15.176)	2.652 *** (19.670)	0.564 * (2.519)	1.923 *** (15.665)	−0.554 *** (−3.600)	1.719 *** (11.444)
GE2: Municipal Status	0.459 *** (6.789)				−0.320 *** (−4.085)	−0.213 *** (−4.958)
PO1: Density per square meter in 2018						0.067 * (2.064)
PO2: Percentage of residents under the age of 18 in 2018					−0.330 ** (−2.648)
EC1: Average monthly salary of employees				0.385 *** (5.464)		0.426 *** (5.383)	0.963 *** (11.651)
EC2: Percentage of employees with wages below the average salary							0.467 *** (4.546)
ED1: Percentage of degree holders	0.792 *** (7.657)			0.179 * (2.487)		0.793 *** (12.084)	0.912 *** (11.087)	−0.433 *** (−5.387)
ED2: Percentage of students in higher education				0.169 * (3.023)	0.251 ** (2.698)	0.184 *** (3.610)	0.138 * (2.159)
TR1: Arab	−0.604 *** (−5.087)	−0.577 *** (−8.172)	0.923 *** (23.738)	−0.266 *** (−3.782)	−0.421 *** (−5.735)	−0.955 *** (−19.716)	−0.159 ** (−2.863)	−0.860 *** (−14.077)
TR2: Secular Jewish	0.742 *** (7.369)		−0.871 *** (−19.729)	0.251 *** (5.036)	0.528 *** (6.380)	1.187 *** (26.185)	0.227 *** (3.992)	0.502 *** (9.044)
TR3: Jewish National-Religious	−0.307 *** (−5.155)		−0.448 *** (−12.169)	0.855 *** (20.591)	−0.202 * (−2.923)	0.649 *** (17.149)	−0.212 *** (−4.472)
TR4: Jewish Haredi		0.932 *** (36.131)	−0.362 *** (−11.790)					0.727 *** (18.839)
Adjusted R2	0.528	0.873	0.820	0.771	0.367	0.810	0.701	0.715
F	31.778	190.142	125.956	93.443	16.950	118.162	65.495	70.124

Notes. Presenting standardized coefficients (Beta) for independent variables; t statistics are in parentheses; N = 276; significance: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001; GE1: The district was found to be insignificant in this model with all parties.

presents the standardized coefficients of the correlations on voting behavior: social affiliation at the first level (the tribe) versus socioeconomic characteristics.

As shown in the correlations in Table 1, the findings point to a significant positive correlation between the four social groups in Israel and the party/parties deemed to represent them. As expected, this confirms that the party representing the social group, the tribe, receives more votes from the people affiliated with the group than from other social groups. It also shows that each one of the social groups has a shared political behavior that is distinguished from the others. This was found true for all four tribes: Arabs, Secular Jewish, Jewish National-Religious, and Jewish Haredi. However, Table 1 also showed positive correlations between socioeconomic characteristics and voting behavior in Israel. Nevertheless, in Table 2, we show that these correlations are affected by social affiliation at the first level (tribe), and those correlations become insignificant or are less significant than social affiliation at the first level for most parties.

For United Torah Judaism (TR4, b = 0.932, p ≤ 0.001), Joint List (TR1, 0.923, p ≤ 0.001), Yamina (TR3, 0.855, p ≤ 0.001), Yisrael Beiteinu (TR2, 0.528, p ≤ 0.001), Likud (TR2, 1.187, p ≤ 0.001), and Shas (TR4, 0.727, p ≤ 0.001) the positive Beta for the associated tribal affiliation was stronger than all other socioeconomic characteristics. Only for two parties was this the opposite. Only for two parties does a socioeconomic variable better explain voting than social affiliation. For Labour-Gesher-Meretz, the percentage of degree holders (ED1, 0.792, p ≤ 0.001) better explains the vote than tribal affiliation at the first level (TR2, =0.742, p ≤ 0.001). Similarly, for Blue and White, the average monthly salary (EC1, 0.963, p ≤ 0.001), followed by the percentage of degree holders (ED1, 0.912, p ≤ 0.001), better explains the vote than tribal affiliation at the first level (TR2, 0.227, p ≤ 0.001).

The model found that the coefficient of the correlation between tribal affiliation and voting behavior trumps the correlation between socioeconomic characteristics and voting behavior. Hence, we accept H1. Voting behavior is explained by social affiliation to the tribe. Affiliation to one of the four tribes in Israel is a better predictor of the vote than the economic or education levels or their geographic location. It helps explain the voting behavior in all parties. In other words, voting in Israel is explained by political in-group favoritism, by belonging to the tribe and voting for the party deemed to represent the tribe—that is, regardless of socioeconomic factors such as district, population size, wages, or the level of education—except for Labour-Gesher-Meretz and Blue and White.

4.2. Findings Related to H2: In Deeply Divided Societies, the More the Residents in a Locality Belong to a Social Group at the First (Tribe) and the Second Level (Subtribe), the More the Party That Represents That Social Group (Subtribe) Receives Votes, Regardless of Socioeconomic Characteristics

At the second level of social affiliation (the “subtribe”), Muslim-majority localities, thus belonging to the Muslim subgroup (ST1), are positively correlated with voting for the “Joint List”. The higher the percentage of Muslims in a locality, the higher the vote for the Joint List (0.935, p < 0.001). The vote in these Muslim-majority localities is negatively correlated with all the other parties. The higher the percentage of Muslims in a locality, the lower the vote for parties other than the Joint List (r ranges from −0.211 to −0.723, p < 0.001). Furthermore, the higher the percentage of Christians in a locality (ST2), the higher the vote for the Joint List (0.237, p < 0.001). Two items distinguish Christians from Muslims: (1) a lower positive significance level of Christians voting for the Joint List and (2) a lower negative significance level of voting for the other parties. The fourth subgroup of the Arabs is the Druze, a group also distinguished by its political behavior. Druze-majority localities (ST3) are not significantly correlated with voting for the “Joint List”. The higher the percentage of the Druze in a locality, the higher the vote for the Labour-Gesher-Meretz, Yisrael Beiteinu, and Blue and White (0.144, p < 0.05, 0.293 p < 0.001, and 0.228, p < 0.001). The vote in these Druze-majority localities is insignificantly correlated with all the other parties.

These findings were not only confirmed in the four subgroups of the Arabs. They were also confirmed in Jewish subgroups, where data were available. For Russian Jews (ST5), we found a significant correlation with the vote for Yisrael Beiteinu, the Likud, and Shas. The higher the percentage of Russian Jews in a locality, the higher the vote for the Yisrael Beiteinu, the Likud, and Shas (0.710, 0.654, and 0.2226, p < 0.001). For Ethiopian Jews (ST6), we found a significant positive correlation with the vote for Yisrael Beiteinu, the Likud, and Blue and White. The higher the percentage of Ethiopian Jews in a locality, the higher the vote for Yisrael Beiteinu, the Likud, and Blue and White (0.288, 0.458, and 0.293, p < 0.001). For Ashkenazi Jews (ST7), we found a significant positive correlation with the vote for Yisrael Beiteinu, the Likud, Blue and White, and Shas. The higher the percentage of Ashkenazi Jews in a locality, the higher the vote for Yisrael Beiteinu, the Likud, Blue and White, and Shas (0.580, 0.596, 0.228, and 0.233, p < 0.001). For Mizrahi Jews (ST8), we found a significant positive correlation with the vote for the Likud, Blue and White, and Shas. The higher the percentage of Mizrahi Jews in a locality, the higher the vote for the Likud, Blue and White, and Shas (0.743, 0.196, and 0.423, p < 0.001). Among the Mizrahi Jews (ST8), there was a negative correlation with the vote for the Joint List (−0.567, p < 0.001), which was higher than the other Jewish subgroups in this study (ST7 and ST6).

Following the findings above, which focused on the Pearson correlations of all variables in this study, the second step was to examine the hypotheses with a regression analysis. This will help determine which is more significantly correlated to voting behavior (H2): is it social affiliation at the second level (subtribe), or is it the other socioeconomic characteristics?

In H2, we expected social cleavages at the first level (tribe) and second level (subtribe) in deeply divided societies like Israel to affect voting behavior by dictating political in-group favoritism, regardless of socioeconomic characteristics. The more residents in a locality belong to a social group (tribe or subtribe), the more the party that represents that social cleavage (subtribe) receives votes. We expected that such a relationship would be more significant than the relationship between the four socioeconomic characteristics (similar to H1): Geography, Population, Economy, and Education. Thus, belonging to the social group at the first (tribe) and the second level (subtribe) trumps socioeconomic characteristics affecting voting behavior.

Table 3. Correlations with voting behavior: social affiliation at the first (tribe) and second level (the sub-tribe) versus socioeconomic characteristics (H2).

Dependent Variables	V1 Labour-Gesher-Meretz	V2 United Torah Judaism	V3 Joint List	V4 Yamina	V5 Yisrael Beiteinu	V6 Likud	V7 Blue and White	V8 Shas
Constant (B)	−0.498 *** (−3.625)	0.955 *** (10.976)	−0.045 * (−7.94)	2.257 *** (17.528)	0.088 * (0.847)	0.261 * (2.961)	−0.185 * (−2.167)	0.870 *** (6.8941)
GE1: District					−0.081 * (−2.061)
PO2: Percentage of residents under the age of 18 in 2018				0.164 *** (3.399)				0.060 * (2.408)
EC1: Average monthly salary of employees				0.159 * (2.027)	0.222 * (2.302)	0.529 *** (6.052)	0.582 *** (7.268)	−0.247 * (−2.580)
ED1: Percentage of degree holders	0.531 *** (4.551)					−0.528 *** (−6.891)	0.520 *** (7.259)	−0.277 * (−3.091)
ED2: Percentage of students in higher education				0.224 *** (5.831)		0.211 *** (4.643)	−0.173 *** (−4.088)
TR1, TR2, TR3, or TR4 (the tribe that is associated with the party)	0.621 *** (5.699)	0.869 *** (23.361)	0.962 *** (32.135)	0.711 *** (20.63)	0.163 * (2.021)	0.726 *** (6.352)	0.526 *** (7.865)	0.626 *** (11.495)
ST1: Muslims		−0.140 * (−2.136)	0.785 *** (19.608)		−0.214 * (−2.942)	−0.365 *** (−5.648)		−0.222 * (−2.424)
ST2: Christians			0.225 *** (10.525)		−0.239 *** (−5.950)	−0.138 *** (−3.865)	−0.117 *** (−3.523)
ST3: Druze	0.200 * (2.886)		−0.463 *** (−32.544)		0.568 *** (5.950)		0.357 *** (8.397)
ST5: Russian Jews					0.903 *** (11.324)	0.312 *** (4.398)	−0.288 *** (−4.349)
ST6: Ethiopian Jews					0.092 * (2.217)
ST7: Ashkenazi Jews				0.187 * (2.544)		−0.299 *** (−3.918)		−0.344 *** (−3.848)
ST8: Mizrahi Jews	−0.385 *** (−4.007)	−0.178 *** (−3.482)	−0.083 * (−1.994)			0.551 *** (8.731)	−0.296 *** (−5.026)	0.471 *** (6.360)
Adjusted R2	0.574	0.886	0.918	0.841	0.768	0.817	0.840	0.754
F	23.840	124.002	204.841	90.651	57.001	76.533	89.801	53.029

Notes: Presenting standardized coefficients (Beta) for independent variables; t statistics are in parentheses; N = 276; significance: * p ≤ 0.05, *** p ≤ 0.001. For the TR, the model included the tribe that was associated with the party, based on Table 1 and Table 2: TR1 for Joint List; TR2 for Labour-Gesher-Meretz, Yisrael Beiteinu, Likud, and Blue and White; TR3 for Yamina; and TR4 for United Torah Judaism and Shas. Variables that were found to be insignificant in this model with all parties are: GE2: Municipal Status, PO1: Density per square meter, and EC2: Percentage of employees with wages below the average salary.

presents the standardized coefficients of the correlations on voting behavior: social affiliation at the first level (tribe) and second level (subtribe) versus socioeconomic characteristics.

As mentioned above, Table 1 shows positive correlations between socioeconomic characteristics and voting behavior in Israel. However, in Table 3, we show that these correlations are affected by social affiliation at the first and second levels (tribe and subtribe), and those correlations become insignificant or are less significant than social affiliation for most parties.

For United Torah Judaism (TR4, b = 0.869, p ≤ 0.001), Joint List (TR1, 0.962, p ≤ 0.001), Yamina (TR3, 0.711, p ≤ 0.001), Likud (TR2, =0.726, p ≤ 0.001), and Shas (TR4, 0.626, p ≤ 0.001) the positive Beta for the associated tribal affiliation was stronger than all other socioeconomic characteristics. This was the opposite for only one party. For Blue and White, the average monthly salary (EC1, 0.582, p ≤ 0.001) remained a better explanatory variable of the vote than tribal affiliation at the first and second levels. For Labour-Gesher-Meretz, adding the subtribe affiliation made the tribal affiliation more significant in explaining voting (TR1, 0.621, p ≤ 0.001), surpassing the percentage of degree holders (ED1, =0.531, p ≤ 0.001).

In addition, the subtribal groups included in our model directly explain the vote for several of the following parties, while making the socioeconomic characteristics insignificantly correlated or by surpassing their correlation: for the Joint List, the Muslim and Christian subtribes (ST1 and ST2, 0.785 and 0.225, p ≤ 0.001), for Yisrael Beiteinu, the Russian Jews and Druze subtribes (ST5 and ST3, 0.903 and 0.568, p ≤ 0.001), and the Mizrahi Jews subtribe for the Likud and Shas (ST8, 0.551 and 0.471, p ≤ 0.001).

Hence, we accept H2. Voting behavior is explained by social affiliation with the tribe and subtribe. Affiliation to a tribe or subtribes in Israel is a better predictor of the vote than economic or education levels or geographic location. It helps explain the voting behavior for all parties. In other words, voting in Israel is explained by political in-group favoritism, which means belonging to a tribe or subtribe and voting for a party deemed to represent the tribe and the subtribe—that is, regardless of socioeconomic factors such as district, population size, wages, or the level of education.

4.3. Findings Related to H3: In Deeply Divided Societies, the More the Residents in a Locality with a Hometown Candidate Belong to a Social Group at the First (Tribe) and the Second Level (Subtribe), the More the Party That Represents That Social Group (Subtribe) Receives Votes, Regardless of Socioeconomic Characteristics

At the third level of social affiliation (the hometown candidate), Table 1 reports a positive correlation between the native hometown candidate (NC1) and the party that includes the candidate, except for Labour-Gesher-Meretz and Yisrael Beiteinu. For the following parties, including a hometown candidate was positively correlated with higher votes for that party: United Torah Judaism (0.256, p < 0.001), Joint List (0.186, p < 0.01), Yamina (0.254, p < 0.001), Likud (0.131, p < 0.05), Blue and White (0.206, p < 0.001), and Shas (0.145, p < 0.05). However, when we used the coefficients versus the socioeconomic characteristics, the direct correlation of the hometown candidate with voting was reduced or became insignificant.

Table 4. Correlations with voting behavior: social affiliation (first, second, and third levels) versus socioeconomic characteristics (H3).

Dependent Variables	V1 Labour-Gesher-Meretz	V2 United Torah Judaism	V3 Joint List	V4 Yamina	V5 Yisrael Beiteinu	V6 Likud	V7 Blue and White	V8 Shas
Constant (B)	0.111 * (2.150)	0.939 *** (1.564)	−0.011 * (−0.023)	2.393 *** (19.451)	0.045 * (0.889)	0.609 *** (7.869)	0.057 * (1.786)	1.338 *** (18.475)
GE2: Municipal status	0.452 *** (6.665)
PO2: Percentage of residents under the age of 18 in 2018					0.285 *** (4.064)			0.266 *** (3.440)
EC1: Average monthly salary of employees	0.471 *** (4.294)					0.432 *** (5.075)	0.772 *** (11.878)
EC2: Percentage of employees with wages below the average salary							−0.262 *** (−3.309)
ED1: Percentage of degree holders	0.646 *** (5.001)					−0.580 *** (−7.074)	0.765 *** (11.952)	−0.237 *** (−3.481)
ED2: Percentage of students in higher education out of all population in 2017				0.171 * (3.005)		0.444 *** (7.529)	−0.224 *** (−4.854)
Tribe: TR1, TR2, TR3, or TR4	0.386 *** (6.198)	0.937 *** (36.074)	0.392 *** (9.590)	0.723 *** (19.536)	0.210 * (2.527)	0.520 *** (8.888)	0.459 *** (12.003)	0.662 *** (17.250)
Subtribe: ST1, ST2, ST3, ST5, ST6, ST7, or ST8.	0.222 *** (4.765)	−0.104 * (−3.171)	0.609 *** (17.158)	0.140 * (2.873)	0.703 *** (11.531)	0.592 *** (9.875)	0.410 *** (14.294)	0.508 *** (10.471)
NC Hometown native candidate NC1 Number of candidates who live in this locality and are likely to be elected in the list- all Israel		0.065 * (2.904)		0.067 * (2.175)
Adjusted R2	0.524	0.879	0.914	0.760	0.586	0.707	0.820	0.733
F	31.227	200.177	292.245	88.165	36.903	67.295	126.201	76.515

Notes. Presenting standardized coefficients (Beta) for independent variables; t statistics are in parentheses; N = 276; significance: * p ≤ 0.05, *** p ≤ 0.001. For the TR, the model included the tribe that had the highest correlation with the vote for the party, based on Table 1 and Table 2: TR1 for Joint List; TR2 for Labour-Gesher-Meretz, Yisrael Beiteinu, Likud, and Blue and White; TR3 for Yamina; and TR4 for United Torah Judaism and Shas; For the ST, the model included the subtribe that had the highest correlation with the vote for the party, based on Table 3: ST1 for Joint List, ST3 for Labour-Gesher-Meretz and Blue and White, ST5 for Yisrael Beiteinu, ST7 for Yamina, ST8 for United Torah Judaism, Likud, and Shas. The following variables were found to be insignificant in this model with all parties: GE1: district and PO1: density per square meter.

presents the standardized coefficients of the correlations on voting behavior—social affiliation (first, second, and third levels) versus socioeconomic characteristics. This model was used to test hypothesis H3. In H3, we expected social cleavages at the third level—native candidates in deeply divided societies like Israel affect voting behavior by dictating political in-group favoritism. The more residents in a locality with a hometown candidate belong to a social group at the first (tribe) and the second level (subtribe), the more the party that represents that social group (subtribe) receives votes. We hypothesize that this effect is regardless of socioeconomic characteristics.

While the results in Table 1 indicated a positive correlation between the hometown candidate and the vote in their hometown for the party where the candidate was running for office, as shown in Table 4, for all parties, the native hometown candidate had no direct significant correlation with voting versus the correlation of the socioeconomic variables with voting.

Still, in this model, adding the hometown candidate (NC) reinforced the correlation between the tribe or the subtribe and voting, thus confirming H3. Those correlations remained better predictors of voting than socioeconomic characteristics, except in the case of Labour-Gesher-Meretz and Blue and White. The percentage of degree holders (ED1) better explains voting for Labour-Gesher-Meretz than social affiliation does on the three levels (0.645, p ≤ 0.001). The average monthly salary (EC1) better explains voting for Blue and White than social affiliation does on the three levels (0.772, p ≤ 0.001). For the remaining parties, the social affiliation at the first level (tribe) or the second level (subtribe) better explains voting than social affiliation does on the three levels. Tribal affiliation better explains the vote for United Torah Judaism, on the first level (TR4, 0.937), as well as for the Joint List (ST1, 0.609), Yamina (TR3, 0.723), Yisrael Beiteinu (ST5, 0.723), Likud (ST8, 0.592), and Shas (TR4, 0.662) (all at p ≤ 0.001).

5. Discussion

This study aimed to shed light on political behavior in deeply divided societies by examining the case of the Israeli society. Using data from the 2020 election, this study investigated the correlation of social affiliation and socioeconomic characteristics with voting behavior. This research examined whether social affiliation better explains voting behavior in the deeply divided Israeli society than socioeconomic characteristics.

This study confirmed that social cleavages affect voting behavior by dictating political in-group favoritism. Our results confirmed the research hypotheses on two levels of social affiliation: social division at the first level (H1) and social division at the second level (H2). The results showed that H1 and H2 were confirmed, as the findings point to a significant positive correlation between social affiliation and voting behavior in Israel. The findings helped attach one or more social groups or subgroups to a specific party deemed to represent them.

The findings show that the correlation of social affiliation with voting behavior in Israel trumps the correlation between socioeconomic characteristics and voting behavior. This was found to be true for the social group at the first level—the tribe (H1)—and the second level subgroup affiliation—the subtribe (H2)—but not for the native hometown candidate. H3 was accepted since the addition of the native candidate showed that the more that residents in a locality with a hometown candidate belong to a social group at the first level (tribe) and second level (subtribe), the more the party that represents that social group (subtribe) receives votes, regardless of socioeconomic characteristics.

Our findings did not apply to two parties: Labour-Gesher-Meretz and Blue and White. For these two parties, socioeconomic variables, namely education and economy, better explained voting behavior. Interestingly, the subtribe associated with these two parties, the Ashkenazi subtribe, was the only one of the subtribes in this study that was found to be an insignificant factor in influencing voting behavior.

One of the contributions of this study is to provide findings that reaffirm the designation of Israel as a deeply divided society based on these divides: the national, ethnic, religious, and political divide over Zionism and socioeconomic divides. The study confirms that Israel’s society can be described as tribal, with four distinct social groups at the first level or “tribes”—“Jewish Secular”, “Jewish national-religious”, “Jewish Haredi”, and “Arab”. These social groups include thirteen coherent subgroups or subtribes; nine are Jewish subgroups, and four are Arab subgroups. Policies and social practices, including endogamy marriages, segregated housing, and education, maintain this deep division. Our study confirmed their distinct political behavior.

Furthermore, a second major contribution of our study is the illumination of how social affiliations translate into political in-group favoritism. Our findings demonstrate a preference for voting in elections for a party associated with one’s social group or subgroup. Notably, this political in-group favoritism, shaped by social affiliations, emerges as a more potent factor influencing voting behavior than other socioeconomic characteristics.

These findings confirm the conclusions of previous scholars regarding Israel’s deeply divided society, such as Al-Haj [49], Mesch and Talmud [51], and Smooha [52]. We further confirmed the existence of political in-group favoritism by demonstrating vote preference by social group affiliation, as suggested by Sani and Bennett [1] and Balliet et al. [2]. We confirmed that such group affiliation extends to race, ethnicity, religion, culture, or geographic location, as suggested by Walton et al. [9], Verkuyten [10], Johnson et al. [11], Perry et al. [12], and Falk and Zehnder [13]. These findings extends Barth’s [14] suggestion of a connection between social affiliation and politics to include sub-societal groups, not just political groups per se. Our conclusion on preferring to vote for an in-group candidate because of their race, ethnicity, religious or cultural affiliation, and geographic location supports the previous works of Ansolabehere and Puy [17], Rand et al. [18], Peele and Morse [19], Kotler-Berkowitz [20], Özdemir and Özkan [21], Walton [22], and Richardson [23]. Our findings confirmed previous scholarship findings on the political in-group favoritism in Israel, as shown by Kimmerling [46] and Arian and Shamir [39].

For the Israeli case, our findings align with Ben-Bassat and Dahan [93], who already found that social affiliation among Arabs in Israel affects political behavior. However, our findings extend this to the social affiliation among Jews in Israel, which aligns with the findings of Arian and Shamir [39]. We extended these findings to show that social affiliation does affect political behavior more than socioeconomic variables, which is our study’s major contribution.

On the other hand, our findings contradict scholars who showed other explanations for political behavior. Scholars have found evidence that the impact of social structure on voting patterns was declining. For example, Van der Eijk et al. [94] showed that political behavior was explained by social characteristics, such as education, income, and religiosity, while we found social affiliation to be more influential. Others, such as Simon [95], included group loyalty to explain political behavior within the framework of human rationality. However, the reference was to groups such as “business corporations, military units, political parties, government agencies, universities”. We extend this approach to include fundamental group loyalty based on social structures, away from the economy, agenda, or institutions. Additionally, we contradict the findings from studies on Israel by other scholars who focused on the economy as the main predictor of political behavior [40,90,91].

Finally, although this work focused on the Israeli case, it is suggested that this approach be tested in other countries. The importance of social affiliation in explaining voting behavior should be examined as a phenomenon that corresponds with recent trends to reject globalization [96], recent populist movements [97], and the re-emergence of nationalism [98]. The findings could further the discussion on improving vote predictions and sampling methods.