Introduction

In recent years, the rise of nationalism and populism has posed new challenges for minority groups everywhere. In an age of resurgent nationalism, understanding the factors that affect the nature and degree of the political engagement of minorities is of particular interest and importance. Whether and how minorities choose and are able to engage politically will influence the extent to which divisions in society are reduced or develop into seemingly insurmountable rifts.

This paper examines how the voter turnout of members of minorities and their political party preferences in national elections respond to changes in the economic, institutional, and political climate. We draw insights from the case of Israel, where the relationship between the Jewish majority and the Arab Palestinian minority has long been a particularly difficult one and is likely to be more relevant in a world where majority-minority relations are under growing strain.

While the case of Israel shares strong similarities with other multicultural and multi-ethnic societies in which minority groups face discrimination, one distinctive feature is the process of modernization that has transformed Palestinian Arab society over the past three decades. Over this period, three key socioeconomic and political developments have profoundly affected the Israeli Arab population. The first is the steady increase in their human capital and standard of living (Yashiv and Ksir 2018), the second is a demographic transition following a steady decline in the fertility rate (Winckler 2002; Yashiv and Ksir 2018), and the third is a rise in electoral support for Arab parties or disengagement from voting at the expense of support for Zionist, or Jewish-majorityFootnote 1 parties.

This paper aims to test five theories of voting behavior in examining how the participation of members of the Palestinian Arab minority in national elections responds to changes in the economic, institutional, and political climate: economic voting, political exclusion, substantive representation, instrumental voting, and expressive voting. Our method employs a fixed-effects linear model that examines the effects of a range of socioeconomic indicators on voter turnout and party preferences in several elections held since 1996. Our data includes national election results from seven Knesset elections between 1996 and 2015, which allows us to exploit temporal variation in locality-level variables.

There are two main advantages of using these data. First, since the period includes the second Intifada—an uprising by Palestinians residing in the West Bank and Gaza— it is possible to test, using panel data, whether this major political event affected voter turnout and political party preferences among Palestinian citizens of Israel. We also have an indicator variable to identify those Israeli localities in which there were clashes with the police during the Second Intifada's October 2000 disturbances. To the best of our knowledge, this is the first article examining whether the localities directly exposed to these clashes were affected differently.

The second advantage of this data set is that it allows us to cover a period during which the electoral system in Israel experienced increasing political fragmentation, involving a fall in the share of votes and the number of parliamentary seats gained by each of the largest parties. Given increased polarization in several settings, we believe that our findings on the determinants of the electoral behavior of the Arab Palestinian minority provide invaluable insights that may be relevant when seeking to understand other contexts.

We find evidence for the political exclusion and substantive representation theories in the context of voter turnout and evidence for the instrumental voting and expressive voting theories in the context of party preferences. While voter turnout seems unaffected by socioeconomic variables, it is highly sensitive to political events: for example, voter turnout significantly decreased following the Second Intifada in 2000. After investigating mechanisms for the last result, we find that we cannot ascribe the decline in turnout to exposure to the aftermath of political violence since residents of localities that experienced violence were as likely to turn out as their counterparts. Additionally, we show that the quantity of Arab political representation, measured as the share of Arab politicians represented in Jewish-majority parties, is significantly associated with voter turnout. We also present suggestive evidence that an increase in the quality of Arab representation (due to recent institutional changes that prompted several Arab political parties to join forces to meet new eligibility requirements) led to an increase in voter turnout. Concerning party preferences, we find that improvements in locality-level socioeconomic variables are somewhat associated with increased support of Arab parties, along with reduced support of Left-wing Zionist parties.

The rest of the paper is structured as follows. In the following section, we provide an analytical framework and formulate five hypotheses regarding voter turnout and political-party choice determinants. In "Voting patterns of Arabs in Israel" section, we summarize Arab voting patterns in Israel. In "Data and Methods" section, we explain our data sources and methodology. Section "Results" presents our results. Section "Discussion" discusses and interprets these in light of the empirical and theoretical literature. Section "Concluding remarks" concludes.

Theoretical framework

In this section, we take an exploratory approach to formulate hypotheses about the factors most closely associated with changes in voter turnout and other voting patterns of the Arab minority in Israel. As with previous studies, examining both turnout and the share of votes gained by particular parties (or political blocks), we test our hypotheses by adopting an approach based on the following two stages: (1) analyzing the decision of whether to vote, based on the share of voters as a proportion of eligible voters, and (2) analyzing the votes obtained by a particular party (or block) as a proportion of the total number of votes cast (Ben-Bassat and Dahan 2018).

It is important to note that not voting can be considered a political stance of a politically engaged public. Consider that Israeli Arab society has experienced important political movements encouraging voters to boycott general elections and has seen a marked difference between the high proportion of individuals voting in municipal elections and the low proportion voting in general elections (Ben-Bassat and Dahan 2012).

We begin by reviewing the theoretical and empirical literature on voter turnout, focusing on trends for minority groups, and stating three main hypotheses. We then review the literature on the factors that attract minority voters to various political parties and formulate hypotheses concerning this second question.

Voter turnout

According to the classic economic voting theory, voters tend to reward incumbents in good economic times but punish them when the economy worsens (Lewis-Beck and Paldam 2000). In addition, high-income and well-educated individuals are more likely to vote and engage in politics than those of lower socioeconomic status (Chan 2021; Logan et al. 2012; Verba et al. 1978). In support of the theory, Afriat and Dahan (2010) find that, while turnout decreased across Israeli localities from 1996 to 2006, localities that experienced lower household income growth rates saw a greater drop in turnout.

Another factor that may influence voter turnout is increased mistrust in institutions fueled by feelings of political exclusion, which can lead individuals to abandon the political process (Jones et al. 2017). Thus, numerous studies draw on political exclusion theory to explain how patterns in voter turnout differ between groups in several contexts worldwide. For example, using US county-level data, Williams (2017) finds that counties where individuals were exposed to more intense violence in the past—proxied for by the number of (Black) lynchings per county—currently had lower voter turnout rates among Blacks. While feelings of political exclusion by one group may stem from a history of violence with the majority group, they might also be psychologically motivated if the minority group lacks trust in the political system (Chan 2021; Logan et al. 2012). In fact, Van Londen et al. (2007) show that greater civic engagement in the Netherlands raises voter turnout by fostering social trust.

The aforementioned theories have also been explored in the Israeli case. For example, to examine how politically motivated violence between Palestinians and Israelis shaped political participation, researchers conducted face-to-face interviews immediately after the Second Intifada (January–May 2001). They found that, among Arab citizens of Israel, those who identified as Palestinian were much more likely to engage in legal forms of protest and to boycott national elections than those who identified as Israeli (Lowrance 2005). Similarly, deteriorating political and economic conditions led Palestinians to question whether Israel was a democracy in which the Palestinian minority could hope to achieve any meaningful representation. Many questioned whether Palestinians might do more to improve their lives by engaging in alternative forms of political pressure other than voting (Jamal 2002; Kook 2017; Rekhess 2008). Feelings of political exclusion likely intensified during the 2006 and 2009 elections following the Lebanon War of 2006 and the Gaza War of 2008, which may have contributed to further decreases in voter turnout (Rekhess 2007; Rouhana et al. 2011).Footnote 2 Therefore, not only did Arabs feel socially and politically excluded from Zionist parties, but, as many researchers have noted, they also felt that Arab leaders were powerless to enact real change for the Palestinian minority (Jamal 2002; Rekhess 2008; Rouhana et al. 2005).

Naturally, when group members feel politically excluded, they express a deep mistrust in the political process, making it challenging for the group’s leaders to persuade them to vote. However, when the group is targeted, threatened, or excluded through a particular policy, leaders can mobilize their constituents by appealing to their emotions and creating a sense of urgency to vote to avoid the consequences of the threat. For example, in a recent paper, Weiss et al. (2022) show that in the wake of Trump’s proposed peace plan, Palestinian citizens of Israel were more politically active and increased their turnout in national elections; many had feared that their citizenship was at stake.

Thus, voter turnout may increase when individuals’ policy concerns are directly addressed by their politicians. This leads us to the third theory used to test changes in voter turnout, the substantive-representation theory (Rosenthal et al. 2018). It’s worth noting that substantive representation does not necessarily imply the use of quotas–which may have a limited effect in allocating benefits to disadvantaged groups (Dunning and Nilekani 2013) to attain sufficient political representation for a minority group. Instead, substantive representation implies that politicians represent their constituency by addressing the most palpable policy concerns of the group.

In the Israeli context, the increase in the minimum electoral threshold required to acquire parliamentary seats—from 2 to 3.25%- was viewed as a policy that served to exclude Arab parties from having seats in the Knesset. In response, leaders of Arab political parties united under the Joint List in 2015, after which Arabs increased their turnout (Kook 2017). The reason for the popularity of the Joint List was that, rather than focusing on the Palestinian-Israeli conflict, it focused on issues fundamental to the Arab minority in Israel (Rosenthal et al. 2018). Specifically, the Joint List campaigned to bridge socioeconomic gaps between Arabs and Jews and worked with the government to increase the budget for Arab localities, potentially allowing Arabs to enjoy substantive representation. In this way, the Joint List provided not only the Arab minority with ethnic representation in the Knesset but also quality representation by serving the community's needs. In addition, some Jewish-majority parties have quotas or secured representation in their party lists for Arabs (Kenig and Rahat 2023, pp. 91, 180) designed to attract and better represent Arab voters.

In general, due to data constraints and measurement issues, it is difficult to differentiate between quantitative and qualitative representation. For example, as argued by Teney et al. (2010), ethnic minorities in Belgium are more likely to vote for parties where they are better represented. However, we cannot discern whether voters’ choices are driven by the ethnic origins of leading politicians associated with those parties or party platforms. Either way, in the absence of representation, ethnic voters may rationally decide not to vote at all. These theories allow us to formulate the following testable hypotheses about how minority groups respond to economic, political, and institutional changes:

H1a

Increases in living standards (as proxied for by income per capita, dependency ratio, the share of subminimum wage earners, and the share of the population in receipt of income maintenance) lead to higher voter turnout in the locality (classic economic theory).

H1b

Major political events (for example, the Second Intifada) involving a major clash between the majority and the minority over fundamental rights depress voter turnout in the locality (political exclusion theory).

H1c

Increased opportunities for quantitative and qualitative representation raise the voter turnout rate in the locality (substantive-representation theory).

Which parties to vote for?

In this subsection, we supplement the theories mentioned above with two other major theories (instrumental and expressive voting) to generate hypotheses about the determinants of voting for minority (Arab) versus majority (Zionist or Jewish-majority) parties.

One might vote for a party type (Arab, left-wing Zionist, right-wing Zionist) for instrumental reasons—in pursuit of expected benefits stemming from the election’s outcome—to maximize material welfare (Buchanan and Yoon 2004). There is a wealth of evidence that ethnic minorities in Europe and the United States, especially immigrants, tend to support left-wing parties (Bergh and Bjørklund 2011; de la Garza and Cortina 2007; Marcos-Marne 2017; Teney et al. 2010).

One explanation for why minority groups lean to the left is that left-wing parties usually promote government intervention in the economy to close socioeconomic gaps, benefiting ethnic minorities, who are usually less well-off (Marcos-Marne 2017). Thus, if one assumes that welfare benefits are more appealing to those of lower socioeconomic status, then left-wing (Zionist) parties are less appealing as the socioeconomic status of minority groups improves.

Another explanation is that left-wing parties usually promote more inclusive policies toward ethnic minorities and protect their rights (Sanders et al. 2014).Footnote 3 In the US, having experienced racial discrimination or having dealt with microaggressions is positively associated with identifying as a Democrat and with having negative views of the Republican party (Kuo et al. 2017). Likewise, in the context of Israel, most right-wing Zionist parties’ platforms include politically charged nationalist rhetoric that many view as ethnically discriminatory against Palestinians and, more broadly, as antithetical to Palestinian aspirations for statehood (Adnan and Miaari 2018).

Note that if minority members primarily vote for left-wing parties to avoid greater power allocated to right-wing parties, which are viewed as more discriminatory, the socioeconomic status may be unrelated to the vote share. However, we think this is unlikely in the case of Israel since voters can challenge right-wing parties by voting for left-wing Zionist parties or Arab parties. We contend that in Israel, the Arab minority usually votes for left-wing Zionist parties to gain material benefits.

Expressive voting, in contrast, is motivated by concerns that are not purely economic or do not stem from the election’s outcome (Brennan and Brooks 2013; Brennan and Hamlin 2000; Hamlin and Jennings 2011). One of the main motivations for expressive voting is social identification with a group (Chan and Jasso 2023; Hamlin and Jennings 2011; Kuo et al. 2017). A good example of social identification associated with high mobilization is municipal elections in Arab localities. In this case, extended family (or hamula) identity (Ben-Bassat and Dahan 2012) and religious identity (Hillman et al. 2015) are associated with high turnout.Footnote 4 There are also examples of expressive voting in a Western context.Footnote 5

Weiss et al. (2022) argue that in the face of extremist rhetoric, members of a minority group may be motivated to vote by factors related to either instrumental or expressive voting. If the group is threatened, some may vote to avoid the threat and achieve a tangible outcome (instrumental voting). Others may vote because their social identification with the minority group is enhanced through exclusion (expressive voting).

These theories allow us to generate the following hypotheses about the political-party choice of Palestinian citizens of Israel:

H2a

Higher socioeconomic status is associated with favoring Arab parties and reduced support for left-wing Zionist parties in the locality (instrumental voting theory).

H2b

Socio-economic status does not affect political party affiliation. Instead, a perceived threat to the minority group, such as a major clash between the majority and the minority over fundamental rights (for example, the Second Intifada), is associated with favoring Arab parties and being less inclined to vote for Jewish-majority partiesFootnote 6 in the locality (expressive voting theory).

Voting patterns of Arabs in Israel

In the period immediately after the independence of Israel, Arab turnout in general elections was consistently high, averaging about 80–90% in elections held in the years 1949–1969 (Israel Democracy Institute 2009).Footnote 7 Over this period, Israeli Arab votes were divided between the Arab satellite parties of Mapai, the ruling party in Israeli politics until 1977, and the Israeli Communist Party, Maki (Kenig 2004). The formation of these satellite parties by Mapai was part of a patronage system (Brake 2018) in which localities with a high share of votes for these parties received a better supply of public goods. Maki attracted Arab voters not coopted by Mapai. Along with its later radical left successors, Rakach and Hadash, it remained outside of the Israeli political consensus (Kenig 2004). The satellite parties faded away in the seventies. The patronage and votes moved to Jewish-majority, especially left-wing Zionist parties, such as the Israeli Labor Party, the successor of Mapai (Brake 2018). "Anti-establishment" Arab voters continued voting for Hadash and smaller binational left-wing parties (Kenig 2004).

A significant change took place in the nineties, during which, for the first time, independent Arab parties (who were neither communist nor officially binational, like Hadash) gained representation in the Knesset, the national legislature of Israel. These parties were, like Hadash, outside the Israeli political consensus. Officially, they represented varying ideologies, such as Socialism (Hadash), Islamism (Ra'am), CentrismFootnote 8 (Ta'al), and Arab Nationalism (Balad). Israeli Jewish politicians and the media treated them as a single block of "Arab Parties." With regard to the stance they took on the most topical issues of the day, such as the peace process and the future of the West Bank and Gaza, they were all considered radical left-wing parties. They were not an official part of any ruling coalition but did provide vital support for the Rabin government in votes of confidence and votes to approve state budgets during 1992–1995. In return, these parties gained additional financial resources for infrastructure development and Arab municipalities (Kenig 2004). The formation of these parties was accompanied by many Arab voters leaving the left-wing Jewish-majority parties. The implementation of direct popular election of the Prime Minister in the 1996 and 1999 elections, which weakened the two major big-tent parties, Labor and Likud, and which benefited smaller sectoral parties, also encouraged the formation and favored the rise of independent Arab parties, at the expense of Left-wing Zionist parties.

From September 2000 to the end of 2004, Israel experienced a violent conflict with its Palestinian neighbors, known as the Second Intifada.Footnote 9 While Arab communities in Israel did not participate in this conflict, the wave of demonstrations and protests during October 2000 quickly deteriorated into violent clashes with the Israeli police. The October 2000 clashes had a lasting, negative effect on Jewish–Arab relations in Israel (Miaari et al. 2012). There was a dramatic reduction in Arab voter turnout in elections held in the years that followed the events of October 2000 (Fig. 1). Turnout did not recover until 2015 when the four main Arab parties formed the Joint List. The political appeal of this union is indicated by the sharp decrease in Arab voter turnout and the fall in the share of votes gained by Arab parties in the April 2019 and 2021 elections, when the Joint List split, and by the rise in turnout in September 2019, when the Joint List was re-established. The vote share of left-wing Zionist parties continued to fall throughout this period. Another important feature of the data to note is the difference between the political preferences of Druze voters and the voting patterns of Muslims and Christians. The Druze citizens of Israel have historically been more accepting of the Zionist narrative and political establishment (Nisan 2010). Therefore, the proportion of Druze voting for Jewish-majority parties, both on the Left and the Right, is far greater than is the case for other Arabs: From 1996 to 2015, the average vote share for Jewish-majority parties in the Druze localities included in our dataset was 84.3%, compared to only 25.0% in other Arab localities.

Fig. 1
figure 1

Source Israel's central elections commission. This is the turnout as a proportion of eligible voters in all Arab localities, including Druze localities and smaller localities that are not included in the statistical analysis. Two general elections were held in 2019, in April and September, marked 2019a and 2019b, respectively

Turnout in Arab localities, 1996–2022.

Full size image

Data and methods

We use two data sources in our paper. The first consists of the results of all Israeli general elections, from 1996 to 2015, held for electing a new Knesset.Footnote 10 These results include the number of eligible voters in each voting precinct, the total number of valid and disqualified ballots, and the number of ballots cast for each party or list in each locality. We include localities defined by the Israeli Central Bureau of Statistics (ICBS) as "Arab."Footnote 11

We construct four dependent variables: "Turnout," "Arab Parties," "Jewish-majority Parties," and "Left-wing Zionist Parties." Turnout is the number of valid ballots as a share of eligible voters. "Arab Parties" is the vote share of non-Zionist parties that specifically aim to serve Arab voters and draw most of their electoral support from the Arab population. From 1996 to 2015, they included four major parties participating in several different electoral alliances: The Socialist Hadash,Footnote 12 Islamist Ra'am, Nationalist Balad, and Centrist Ta'al. This variable also included smaller Arab parties, who received less than 1% of the votes in the general elections studied. The full list of parties we consider "Arab" is given in Table A2. We consider the rest of the parties as Jewish-majority Parties who draw the majority of their electoral support from Jewish voters, and the variable "Jewish-majority Parties" is their vote share. Finally, "Left-wing Zionist Parties" represents the vote share of the two parties considered left-wing and Zionist – The Israeli Labor Party and MeretzFootnote 13 (Hazan 2020). All Left-wing Zionist Parties are considered Jewish-majority Parties, but not vice versa.

The second data source consists of locality-level socioeconomic characteristics drawn from ICBS publications.Footnote 14 We focus on five variables that consistently appear in all publications: (1) The median age of a locality’s residents; (2) The dependency ratio, calculated as the size of the dependent population (including children and youths of ages 0–19 plus the elderly, aged 65 and above), divided by the size of the working-age population, (aged 20–64); (3) The average household income per capita; (4) The share of subminimum wage earners; the share of employees and independent workers (as a proportion of all employees and independent workers) whose monthly income is under the statutory minimum wage for a full-time worker; (5) The share of individuals, (out of the total population), receiving an income maintenance payment from the National Insurance Institute.Footnote 15 Descriptive statistics are shown in Table 1, and a correlation matrix for the five main independent variables is presented in Table B1.

Table 1 Descriptive statistics
Full size table

We merge each set of election results with a set of socioeconomic indicators from the closest ICBS survey year, as demonstrated in Table A1. The five elections held in 2019–2022 are not included in the data because no appropriate ICBS survey data is available after 2015. We excluded small localities that were not surveyed separately by the ICBS. We also excluded ten Druze-majority localities due to their different degree of integration with the Jewish majority (Nisan 2010) and voting preferences.Footnote 16 The result is a panel of 63 localities over seven time periods from 1996 to 2015. We provide a list of the localities in Table A3 and a list of excluded and merged localities in the notes accompanying Table A3.

Our main empirical strategy is a fixed-effects linear regression. We focus on two dependent variables (noted as \({Y}_{i,t}\)): (1) voters' turnout in locality \(i\) during election year \(t\), (2) Arab parties' vote share in the locality. The independent variables are the observed socioeconomic characteristics for that locality at that time. Our regression can be written as follows:

$$Y_{i,t} = \alpha + \beta X_{i,t} + \rho \;{\text{Re}} presentation_{t} + \lambda \;Intifada_{t} + \gamma_{i} + \varepsilon_{i,t}$$

where \({X}_{i,t}\) is a vector of our five observed socioeconomic characteristics: median age, dependency ratio, the natural logarithm of real average household income per capita, share of subminimum wage earners, and share of individuals receiving income maintenance payments. \({\gamma }_{i}\) is a locality fixed-effect.Footnote 17

\(Representation\) is an index for the political representation of Arabs in Jewish-majority parties, accounting for the relationship between minority representation in party lists and political preferences (Teney et al. 2010) and aimed at testing the substantive-representation theory in the context of Jewish-majority parties. The index equals the share of Arab candidates included in the top 10 (or 20) individuals in the Jewish-majority party list and can be calculated as follows:

$$Representation_{t} = \frac{{\mathop \sum \nolimits_{{p \in \left\{ {Jewish - majority} \right\}}}^{P} Arabs 1 - 10_{p,t} + \mathop \sum \nolimits_{q}^{2} Arabs 11 - 20_{p,t} }}{10 \cdot P + 10 \cdot 2}$$

\({Arabs 1-10}_{p,t}\) is the number of Arab candidates in the top ten places of the party list of a Jewish-majority party \(p\) at time \(t\). \({Arabs 11-20}_{p,t}\) is the number of Arabs amongst those ranked 11-20th in a party list, and we only consider the latter in the case of the two largest parties in a particular election.,Footnote 18Footnote 19 We use the candidate lists published by Israel's central elections committee. We present the number of Arab candidates in each party for each election in Table A5. Figure 2 gives the trend in representation for the studied period.Footnote 20

Fig. 2
figure 2

Source Authors' calculations based on Israel's central elections commission data. See the Data and Methods section for more details on how the figures are calculated

Representation of Arabs in Jewish-majority party lists, 1996–2015.

Full size image

\(Intifada\) is a dummy variable with a value of zero for elections held before October 2000 (the 1996 and 1999 elections) and to one otherwise. The term \({\gamma }_{i}\) denotes the locality fixed effect and \({\varepsilon }_{i,t}\) the independent error. Our model is identified by temporal variation in voting patterns and economic patterns within a locality over time.

\(Clashes\), is a dummy variable which receives the value of one for localities in which civilians clashed with police forces during the October 2000 disturbancesFootnote 21 and zero otherwise. Because \(Clashes\) is time-invariant, we use it only as part of an interaction term with \(Intifada\). Finally, we check the robustness of our main model using a first-difference fixed effect linear regression.

Threats to the identification strategy

Two main reasons our identification strategy may not hold are reverse causality and omitted variable bias. Reverse causality may stem from political parties rewarding localities that give strong electoral support with benefits that improve economic conditions in these localities. Decisions regarding Israeli local government are under the authority of the Interior Minister (Ordinance of Municipalities, 1964). Therefore, if there is reverse causality, we would expect localities with a greater share of votes cast for the Minister's party to be better off economically. In section B2 of the supplementary Materials, we provide a test of reverse causality and demonstrate that it is not a concern in our context (see test results in Table B2).

Table 2 The relationship between turnout and socioeconomic variables
Full size table

Another concern related to omitted variable bias is that particular unobserved social identities that arise within a given locality might affect economic conditions and voting patterns simultaneously. Arab localities are traditionally built around distinct social groups (or “clans”) called hamulas. Each hamula is a group of families who share common ancestry (Rosenfeld 1974). The strength of the hamula as an identity group varies across localities and affects voting behavior, at least at the level of municipal elections (Ben-Bassat and Dahan 2012). Because there is no evidence to indicate that the relative power of hamulas changes across the studied period,Footnote 22 we assume that this variable is time-invariant and accounted for by the locality fixed-effect.Footnote 23

Results

Voter turnout

Table 2 presents the results for the turnout dependent variable. We find that none of the socioeconomic variables was significantly associated with turnout, but representation of Arabs in Jewish-majority parties contributed to turnout. A one standard deviation increase in the representation index was associated with a 2.0% points (pp) increase in turnout. We also find that in elections following the second Intifada, turnout decreased by 19.6 pp (column 1). The turnout in localities exposed to the 2000 clashes was not significantly different from that in other localities (column 2). The findings were robust to first differencing (Table B3).

Party preferences

Table 3 presents the results for the Arab Parties' vote share dependent variable. We find that both higher median age and dependency ratio were negatively associated with Arab Parties' vote share, while income per capita was positively associated with it. A one standard deviation increase in median age was associated with a − 5.8 pp change in Arab Parties' vote share, while a one standard deviation increase in dependency ratio was associated with a − 12.6 pp change. A 1 pp increase in income per capita was associate with a 1.2 pp increase in Arab Parties' vote share. We also find that in elections following the second Intifada, Arab Parties' vote share increased by 3.8 pp (column 1), and that this increase was not significantly different in localities exposed to the 2000 clashes (column 2). The findings were not robust to first differencing (Table B4).

Table 3 The relationship between Arab parties' vote share and socioeconomic variables
Full size table
Table 4 The relationship between Left-wing Zionist parties' vote share and socioeconomic variables
Full size table

Table 4 presents the results for the Left-wing Zionist Parties' vote share dependent variable. We find that the dependency ratio was positively associated with Left-wing Zionist Parties' vote share, while the share of wage earners on subminimum wages was negatively associated with it. A one standard deviation increase in the dependency ratio was associated with a 4.8 pp increase in Left-wing Zionist Parties' vote share. In comparison, a one standard deviation increase in the share of wage earners on subminimum wages was associated with a − 3.7 pp change. Furthermore, a one standard deviation increase in the representation index was associated with a 2.3 pp increase in these parties' vote share. We also find that in elections following the second Intifada, Left-wing Zionist Parties' vote share decreased by 2.8 pp (column 1) and this decrease was not significantly different in localities exposed to the 2000 clashes (column 2). Excluding the Second Intifada variable, these findings are mostly robust to first differencing (Table B5).

Table B6 presents the results for the Jewish-majority Parties' vote share dependent variable, and Table B7 presents the results for the first differencing. As detailed in the data and methods section, all non-Arab Parties are considered Jewish-majority Parties. Therefore, the same variables positively associated with Arab parties' vote share were negatively associated with Jewish-majority Parties' vote share, and vice versa.

Discussion

Voter turnout

The results of our analysis provide evidence to support some of the hypotheses we posed for voter turnout. We find no evidence for H1a, as none of the socioeconomic variables has a statistically significant association with turnout in our main model (Table 2, column 1).

In contrast, we find strong empirical backing in favor of H1b. Specifically, the Second Intifada marked a major turning point in Arab political participation. The post-intifada variable primarily explains the drastic decrease in turnout in the years following this crisis. In particular, our regressions indicate significant trend breaks when the post-Intifada indicator is included in the regression. However, the change is not necessarily due to the exposure of Arab voters to political violence since these trend breaks are similar in all Arab localities, regardless of whether there were clashes with the police in October 2000 (Table 2).

Previous research on turnout between 1996 and 2006 (Afriat and Dahan 2010) suggests that the post-Intifada elections saw a decrease in turnout for all localities in Israel and, after controlling for socioeconomic variables, the decrease in Arab localities was not significantly different from that seen in Jewish ones. However, since we find a significant effect for the Intifada even after controlling for socioeconomic variables, we cannot disprove H1b. Thus, we interpret this result as one where eligible voters turned away from voting in reaction to political violence, as stated in the political exclusion theory, even if they did not experience the violence directly.

We also find evidence for H1c, as we observe that an increase in the representation index is associated with a significant increase in turnout (Table 2), implying that quotas for Arabs in Jewish-majority Parties might attract Arab voters looking to be represented. In addition, even though our multivariate analysis ended in 2015 when the Joint List first ran, we can also interpret the aggregate voting patterns during 2015–2022 as constituting a preference for substantive representation in the form of the Joint List. During this period, turnout (Fig. 1) and Arab Parties' vote share (Fig. 3) increased whenever all major Arab parties ran together and decreased otherwise. This may also result from voters favoring parties with less risk of falling under the electoral threshold or as some signaling regarding the importance of sectoral unity.

Fig. 3
figure 3

Source Israel's central elections committee. This is the share of voters for each party group as a proportion of the total qualified votes cast in all Arab localities, including Druze localities and small localities that are not included in the statistical analysis. Two general elections were held in 2019, in April and September, marked 2019a and 2019b, respectively. See the Data and Methods section for further information on the composition of each group

Parties' vote shares in Arab localities, 1996–2022.

Full size image

Party preferences

Regarding party preferences, we find mixed evidence for H2a (the instrumental voting theory) and more conclusive evidence for H2b (the expressive voting theory). An increase in the dependency ratio, which corresponds with a lower standard of living, (since there are more dependents per working-age person), is associated with a decrease in Arab Parties' vote share and an increase in Left-wing Zionist Parties' vote share. Furthermore, an increase in income per capita is associated with an increased vote share of Arab Parties. On the other hand, an increase in the share of wage earners on subminimum wages, which also corresponds with a lower standard of living, is associated with a decrease in Left-wing Zionist Parties' vote share, the opposite of what we would expect based on the instrumental voting theory. Our findings are, therefore, not entirely consistent with our hypothesis.

In addition, the second Intifada is associated with an increase in the Arab parties' vote share. This finding is consistent with the expressive voting theory because we would have expected such an event, where a clash between the majority and the minority over fundamental rights can be perceived as a threat, to spur support for Arab parties. However, when taken together with our findings on turnout, this finding might be instead explained by voters who used to vote for Jewish-majority Parties deciding to abstain in higher numbers than those who used to vote for Arab Parties, resulting in the latter's higher vote share.

Thus, our interpretation is that as the socioeconomic status of Arabs improved, they became less dependent on the expected benefits of supporting Left-wing Zionist parties, (or even Jewish-majority parties in general), who lost support to Arab parties. In other words, greater upward mobility in Arab localities allowed constituents to transfer their support to parties that were likely to reflect their political preferences instead of transferring their support to parties that would address their economic concerns. The shift from Jewish to Arab parties after the Second Intifada suggests that sentiments related to expressive voting were important, with this political event pushing voters towards political parties that represented their social/cultural identities.

Concluding remarks

Our analysis provides a coherent narrative that connects the socioeconomic trends in Arab society to important political outcomes. These findings are especially important in the context of the latest political developments: the transition of Arab parties from an external opposition outside the Israeli political consensus to a position of influence. Their growing political power is evidenced by the fact that one of the Palestinian Arab parties, Ra’am, became part of a governing coalition in 2021. This novel development will allow us to observe whether and how its voters reward or punish it based on deeds rather than declarations and aspirations expressed when in opposition. We feel that our findings may be relevant to understanding the experience of minorities in other contexts and perhaps also to policymakers in Israel and beyond.

Our findings raise the question of whether other ethnic minorities experienced the same changes in political preferences over time. Two ethnic groups of immigrant Jews in Israel, Sephardic Jews in the 1950s and Russian-speaking Jews in the 1990s, also changed their voting behavior: initially tending to support ruling left-wing Jewish-majority parties and later switching their support to sectoral parties. However, the political allegiances of members of these groups have become more diverse over time and now include support for left-wing, right-wing, and sectoral parties. It is unclear whether these changes resulted from the same economic and demographic changes affecting Israeli Arabs.

In addition, understanding what affects whether and how Palestinians vote is extremely important, given the wider impact on the region and the world of the state of Israeli-Palestinian relations. Understanding the electoral behavior of Palestinians sheds light on the state of this relationship and the possible means of improving it. We look forward to future research that explores the determinants of voter turnout and political party choice, especially among racial/ethnic/religious minorities.

Research limitations

It should be noted that our results only apply to the locality level, and are not necessarily generalizable to the level of the individual. Nevertheless, since voting is completely anonymous and it is not possible to directly observe an individual's political preferences and socioeconomic status simultaneously, this locality level analysis allowed us to estimate the relationship between them on average.