The mainstreaming of Islam in party politics followed by decades-long Islamisation is one of the most significant consequences of the democratisation process in post-Suharto Indonesia.Footnote 1 The Prosperous Justice Party (Partai Keadilan Sejahtera, PKS) has been regarded as a prominent representative of this trend, because it appears to be virtually the only Indonesian party that stably expands its support base through a puritanical Islamic profile rather than relying on money politics and the leaders’ personal appeal. Formed in 1998 by activists of Tarbiyah, an underground campus movement for Islamic proselytisation during President Suharto's New Order regime, PKS is now the largest Islamist party and the sixth largest party in Indonesian politics by steadily increasing its vote share from 1.3 per cent in 1999 to 8.2 per cent in 2019.Footnote 2 PKS’ electoral growth contrasts sharply with the other major parties, which have suffered from fluctuations in their votes owing to a heavy dependence upon ephemeral money politics and clientelistic networks.Footnote 3 In addition to noting its unique origins, strong internal organisation, and commitment to Islam, scholars suggest a range of factors to explain the party's electoral growth, including Muslim economic grievances, growing Islamic piety among voters, the emergence of an educated and urban Muslim middle-class, and service provisions by PKS grassroots activists.Footnote 4
Nevertheless, limitations persist in analyses of PKS’ emergence as a viable party. First, quantitative and qualitative studies of PKS (and other Islamist and/or Islamic parties) fail to choose representative cases. For instance, studies based on quantitative data, mostly raw data from public surveys, have little choice but to disregard non-Muslims, who comprise about 13 per cent of the Indonesian voting public; those respondents are excluded from questions regarding Islamic piety, which is regarded as a key factor in the Muslim electorate's choices of parties and candidates.Footnote 5 Given PKS’ attempts since the late 2000s to reach out to constituencies from non-pious Islamic backgrounds or even non-Muslim voters by developing a pragmatic approach to Islamic proselytisation,Footnote 6 such exclusion undermines a full accounting of the patterns in the party's political mobilisation. Meanwhile, qualitative research on the topic tends to suffer from methodological problems caused by a lack of inferential leverage and ad hoc case selection. This shortcoming in turn implies that a more sophisticated methodological strategy is needed to supplement additional analytic leverage to existing explanations for determinants of support for PKS.
A second shortcoming in the existing literature is the discrepancy between national and local-level perspectives regarding the primary factors that mobilise Indonesian political parties’ electoral bases. Observers focusing on national politics note that PKS’ original constituencies share a specific demographic: the urban and educated Muslim middle class.Footnote 7 In this vein, PKS’ electoral rise can be understood as a successful broadening of the party's appeal to various constituencies, especially those who have benefited from social services provided by grassroots party members.Footnote 8 These studies assert that partisanship and the party's programmes strongly influence voters to vote for PKS. However, such assertions contrast with the findings of studies on the dynamics of local party politics in Indonesia. Typically at the sub-provincial level, researchers have found that clientelistic and personal networks led by local notables play a pivotal role in mobilising voters who expect material benefits to accrue by supporting individual candidates instead of parties.Footnote 9 According to these scholars, institutional reforms—particularly decentralisation, which aimed at giving discretion to local elected representatives mostly affiliated with political parties—paradoxically have enfeebled party accountability and institutionalisation. Parties, including PKS, are now increasingly dependent upon wealthy figures, due to the lack of campaign funds for the many elections Indonesia now holds at both national and local levels.Footnote 10 To my knowledge, these two explanations for voter mobilisation by PKS and other Indonesian parties, grassroots activism and clitentelism, have not been tested simultaneously. This is precisely what this article intends to accomplish.
I seek to better understand how an Indonesian Islamist party like PKS mobilises its voters given the context of localisation of electoral politics. Drawing on Evan Lieberman's three-stage nested analytical framework,Footnote 11 I first show that there is a significant variation in the explanatory power of existing theories among the sub-provincial units with respect to voting patterns for PKS. The findings from a cross-regional comparative analysis of three sub-provincial units demonstrate that the variation is largely attributed to the relative strength of two contrasting historical legacies, that is, sociocultural cleavage structures and former authoritarian single-party institutions, in the pattern of electoral competition. In regions where cleavage-based politics remain salient, voters tend to have relatively strong ties to parties in accordance with their social milieus. Under these conditions, PKS has typically opted for narrow categories of Indonesian voters who are more likely to be mobilised by the party's programmes and ideological orientations. In contrast, in regions where Golkar, the electoral machine of Suharto's authoritarian rule (1966–98), has remained a dominant party at the expense of pre-existing social cleavages, PKS’ voter mobilisation is likely to be ineffectual and dependent on whether the party nominates members of the local political elite. Lastly, in regions where the two legacies disappeared and elections function merely as a selection of (opportunistic) personalities, the voters’ motives for supporting PKS are exclusively instrumental, meaning that neither cleavages nor grassroots mobilisation affects the party's voter mobilisation. Instead, the party's electoral performance is heavily affected by whether its candidates have an ability to capture volatile voters using their personal and clientelistic ties. These findings are confirmed by an additional statistical test.
The article proceeds as follows. First, I conduct a preliminary cross-regional analysis to test the existing determinants of PKS voter mobilisation and to justify selecting the three case studies for further qualitative exploration. Second, based on fieldwork in these three sites, I outline the varied patterns of interparty competition, particularly regarding the relative strength of the personal vote and clientelism. Third, I assess the case studies’ findings using additional statistical tests. The conclusion discusses the article's implications for future research and its contributions to the broader literature on political Islam and party politics.
Statistical observations for case selection
Existing explanations for PKS’ electoral support can generally be classified into two broad categories, Islamic piety and Socioeconomic conditions. While the former emphasises that the electoral outcomes of Islamist parties like PKS depend primarily on the mobilisation of strongly pious Muslim constituencies,Footnote 12 the latter is supported by scholars who have understood the electoral rise of Indonesian political Islam as the outcome of the Muslim community's diversification into multiple variants, which stems from a range of socioeconomic transformations.Footnote 13 More specifically, proponents of the socioeconomic conditions model emphasise that PKS’ sustained electoral growth is mainly attributed to its broadening appeal to specific segments of the Muslim population, namely the urban poor and the educated middle class, who have felt excluded from the politico-economic status quo.Footnote 14
Given the number of causal factors suggested by the literature and the primary goal of this research to provide a new perspective on Islamist political mobilisation in Indonesia, an empirical analysis based on a single procedure may be unsatisfactory. I thus adopt Lieberman's nested analysis as a viable suggestion for a mixed-methods design. Compared with earlier attempts to combine small-n (or qualitative) and large-n (or quantitative) methodological approaches, nested analysis has three unique features. First, its initial step requires statistical analysis with a large number of cases, or a preliminary large-n analysis (LNA). According to Lieberman, preliminary LNA allows the researcher not only to estimate the effects of rival explanations simultaneously, but also to identify space for comparing cases with significant variations in their key explanatory variables.Footnote 15 The second step, small-n analysis, depends on the results of the preliminary LNA. Concretely, when the specified model's robustness is confirmed by visualised inspection of the residual differences between the observed and the expected scores, the subsequent small-n analysis, or a model testing small-n analysis (Mt-SNA), examines whether findings from within-case observations support the results produced by the preliminary LNA. If the preliminary LNA finds the model to be weak, one then proceeds to a model-building small-n analysis (Mb-SNA), which is aimed at developing a new theoretical framework that explains cross-case variation in the dependent variable. Lastly, the general plausibility of new empirical evidence, drawn from both the Mt-SNA and Mb-SNA, is tested using an additional statistical analysis called model-testing large-n analysis (Mt-LNA), the final component of the nested analysis.
The preliminary LNA should be conducted at either the cross-regional or cross-national level to select cases for the in-depth SNA. In other words, an individual-level analysis using survey data, which is the primary analytical tool of quantitative research on political Islam as well as other political forces in the Indonesian electoral arena, is virtually impossible. I thus collected data on the vote shares of every party in the 2014 national parliamentary (Dewan Perwakilan Rakyat, DPR) election,Footnote 16 the most recent election at the time of data collection, in all districts (kabupaten) and cities (kota), except those in Papua province.Footnote 17 The dataset also includes a wide range of demographic and socioeconomic indicators in 2012 or 2013, sourced primarily from the Indonesia Data for Policy and Economic Research (INDO-DAPOER) and Annual Statistical Reports (Dalam Angka) provided by every sub-provincial unit.
I then categorise the explanatory variables according to the existing explanations introduced above: Islamic piety and Socioeconomic conditions. The former has three variables, which represent aggregate-level Muslim religiosity: proportion of Muslims in the population,Footnote 18 number of Islamic religious facilities (per 1,000 persons),Footnote 19 and proportion of Islamic school students among the total number of primary and secondary school students.Footnote 20 Again, this group of variables is used to test to see whether a favourable climate for promoting Islamic piety has yielded electoral benefits for PKS. Regarding Socioeconomic conditions, population density measures the level of urbanisation in each region. In addition, regional educational level is measured as the percentage of the adult population graduating from colleges or a higher educational institution. I also include the unemployment rate, poverty rate, and per capita gross regional domestic product (GRDP) to test arguments on the effects of economic development on Islamist political mobilisation in a regional context. The socioeconomic indicators are thus expected to measure the impacts of varied socioeconomic conditions on electoral support for PKS. Lastly, two regional control variables for Java and Sumatra are included. Unlike ordinary dummy variables that may unnecessarily ‘soak up’ cross-case variance with little theoretical basis,Footnote 21 differences in Islamic religiosity between religiously syncretic Java and the pious ‘Outer Islands’, best represented by Sumatra, are well established in the literature as an influential factor affecting political affiliation.Footnote 22 Appendix A summarises descriptive statistics for all variables in the analyses.
Based on the results of regression analysis introduced in Appendix B, I selected three sub-provincial units for case analysis—Gorontalo City, Ngada district, and Mojokerto district (see fig. 1)—by calculating the differences between the actual and predicted scores on the dependent variable.Footnote 23 The data displayed in table 1 and fig. 2 represents the outputs that justify the case selection. As recommended by Lieberman, for structured comparison, it is desirable to combine more than one ‘off-the-line’ case with one of the best fits, with reference to the regression line derived from the preliminary LNA.Footnote 24 Indeed, Ngada and Gorontalo, where PKS unexpectedly achieved far larger and smaller shares of votes than predicted, respectively, were selected as the ‘most different’ cases. Among the possible cases, Ngada was selected because it is the only region where the overwhelming majority of the population is non-Muslim.Footnote 25 Its inclusion may thus highlight certain political conditions that weaken the cleavage across religions. Lastly, Mojokerto district, where almost no difference was found between the actual and predicted vote shares of PKS, constitutes a case that reflects the statistical patterns of Islamist mobilisation in the electoral arena.
Locations of the three sub-provincial regions selected by the preliminary LNA
Three cases for small-N comparative analysis
Visualising the cases reference to a regression line
The next section presents the sub-provincial case findings, drawing on an analysis of the political process and interviews with politicians, intellectuals, and ordinary voters in the three regions. Using the findings, I then develop a framework that accounts for the variation in PKS’ electoral performance across the cases.
Diverging contours of local Islamist mobilisation: Qualitative evidence
PKS in Gorontalo City: Isolation from elite-driven politics
Gorontalo City is the only urban administrative unit in Gorontalo province, which is located on the northern tip of Sulawesi Island. Historically, it has been the lynchpin of Gorontalo (also known as Hulontalo), which refers not only to a geographic territory maintained since the sixteenth century but also an ethnic group based on linguistic and religious homogeneity.Footnote 26 In 2001, Gorontalo City was designated as the capital of the newly established Gorontalo province, which was separated from the predominantly Protestant North Sulawesi province. Almost all government offices, tertiary educational facilities, and commercial markets in the province are concentrated in Gorontalo City.
Consistent with its nickname, the ‘Verandah of Medina’ (Serambi Madinah), Gorontalo City exemplifies strong Islamic piety in both the public and private sphere.Footnote 27 The proportions of Muslims in Gorontalo City and Gorontalo province are 95.7 per cent and 96.9 per cent, respectively, which are considerably higher than the national average (87.2 per cent). Such demographic superiority of Muslims over other religious minorities can be seen in only a few regions, such as Aceh and West Sumatra, where Islam has played a decisive role in creating the regional identity and in rendering the local communities more puritanical through the implementation of sharia bylaws.Footnote 28 In Gorontalo, four sharia bylaws have been passed since the formation of the province that criminalise adultery and alcohol consumption and that require Muslim students to obtain proficiency in reading and reciting the Quran. Evidence of Islamisation in Gorontalo City can also be found in the rapid increase in Islamic infrastructure. For example, at least 27.2 per cent of the total number of mosques in the city (82 out of 301) have been built since the 2010s. The proportion of students enrolling in Islamic schools also increased from 12.2 per cent in 2010 to 17.7 per cent in 2016.Footnote 29
At first glance, based on the literature, Gorontalo City's high degree of urbanisation with growing Islamic piety should be advantageous to PKS’ mobilisation of voters. In fact, in regions with similar socioeconomic conditions to those of Gorontalo exist, PKS’ electoral outcomes have been consistently better than the national averages.Footnote 30 However, as table 2 shows, PKS has performed poorly in both Gorontalo City and its namesake province in the five most recent legislative elections, earning the party a smaller vote share than its national averages. Moreover, PKS has performed no better in Gorontalo City than in the rest of Gorontalo province, which consists of a rural and far less educated Muslim population. Indeed, the puzzle lies in PKS’ unexpectedly poor performance in Gorontalo City.
PKS’ recent vote shares in Gorontalo (DPR, 1999–2019)
Source: Komisi Pemilihan Umum (General Elections Commission, KPU).
As in other parts of the archipelago, the nationalist movement influenced by pro-independence sentiment penetrated Gorontalo in the early twentieth century. In 1923, a group of young elites led by Nani Wartabone, the prince of a small sultanate in Gorontalo, established Jong Gorontalo, a regional branch of a nationalist youth organisation.Footnote 31 Wartabone also patronised H. Oemar Said Tjokroaminoto, the charismatic leader of the first Islamic political party, Sarekat Islam (Islamic Union, SI), by inviting him to visit Gorontalo. In 1928, SI officially opened a branch in Gorontalo City. A year later, Muhammadiyah, the largest moderate Muslim organisation with its base in Central Java, also opened a branch in Gorontalo to protect the indigenous Muslim community from Christian missionary activity.Footnote 32
From that point until the 1950s, political mobilisation in Gorontalo was largely driven by SI and Muhammadiyah. On the one hand, despite decades-long internal conflicts and splits within SI, which were crucial in the party's loss of nationwide support,Footnote 33 Gorontalo remained enthusiastically supportive of SI and its successor, Partai Sarekat Islam Indonesia (Indonesian Islamic Union Party or PSII, est. 1929). This was mainly because the personal charisma of Tjokroamimoto, who was even venerated as a reincarnation of Ratu Adil (the ‘Just King’) by his followers,Footnote 34 resonated well with Gorontalo's own deep-rooted Sufism, a strand of Islam that accentuates the spiritual authority and prestige of divines.Footnote 35 Furthermore, a subset of local notables who had lineage to small ancient kingdoms and were capable of maintaining their political clout due to the indirect rule of the Dutch, became PSII key cadres.Footnote 36 On the other hand, in addition to active participation by local aristocrats, Muhammadiyah was able to expand throughout Gorontalo by providing a range of social services, including education, healthcare, and Islamic teaching.Footnote 37 PSII and Muhammadiyah's leading role in local politics in Gorontalo continued through the early years of Indonesia's independence from 1945 onwards. As table 3 indicates, in the first legislative election in 1955, Masyumi, a political party led by Muhammadiyah and small Islamist groups, and PSII shared an almost even split of more than 70 per cent of the vote in Gorontalo district.Footnote 38 In particular, PSII's vote share was more than ten times its national average. This contrasted with the contours of national electoral competition, in which strands of Islam generally aligned with four ideologically different mass political parties: that is, the nationalist PNI, the Communist PKI, the traditionalist Nahdlatul Ulama (NU), and Masyumi.Footnote 39
1955 National legislative election in Gorontalo district
Source: Alfian, Hasil Pemilihan Umum 1955: Untuk Dewan Perwakilan Rakjat (Djakarta: LEKNAS, 1971).
Political turbulence then dramatically altered the initial local political dynamics in Gorontalo. In 1959, founding president Sukarno disbanded several political parties and arrested their leaders, including Masyumi's Mohammad Natsir and Burhanuddin Harahap, for their alleged involvement in the rebellions in Sumatra and Sulawesi. These opposition leaders remained banned from political activities after the takeover in 1965 by General Suharto, who regarded Masyumi as the other potential threat to his New Order regime in addition to PKI, which had already been brutally liquidated by the Indonesian Army.Footnote 40 Thus, a significant portion of Masyumi politicians joined Golkar, Suharto's party machine created in the late 1960s, while in 1973 others participated in the establishment of Partai Persatuan Pembangunan (United Development Party, PPP), a forced fusion of Islamic parties by Suharto's New Order.Footnote 41 Similarly, PSII was plagued by intra-party conflict over the choice to join either Golkar or PPP. It eventually joined PPP but subsequently became the party's pro-government faction, owing to a close personal relationship between the factional leader Thayeb Gobel, a Gorontalo-born businessman who won the party chairmanship in the late 1970s, and New Order strongmen.Footnote 42 The decline of Masyumi and PSII thus gave Golkar an opportunity to cultivate a base of popular support in Gorontalo.
The New Order in turn made great efforts to secure Golkar's votes in Gorontalo by patronising local elites, who were descendants of sultans and aristocrats in the pre-colonial period and who maintained a firm grip on local society by exerting control over adat (customary laws).Footnote 43 First, the New Order appointed promising descendants of local aristocratic families to key administrative positions in Gorontalo. In particular, the positions of mayor of Gorontalo City and regent of Gorontalo district, the region's top two offices, were dominated by military officers who were born in Gorontalo and strong supporters of Golkar.Footnote 44 Second, Golkar's local apparatus also promoted adat and Islam as the region's foundational principle. In 1971, for example, Kuno Kalaku, the former regional head of the government's intelligence agency, convened local notables to stipulate that adat and Islam are mutually reinforcing and constitute the central components of Gorontalo society.Footnote 45 Furthermore, in 1985, the regent of Gorontalo district, Col. Martin Liputo, reconstituted Bantayo Poboide (‘Meeting Building’ in the Gorontalo language) as a context for holding regular meetings with notables to discuss issues regarding the two pillars. Such attempts consequently contributed to Golkar's sweeping victories in the legislative elections during the New Order years (see Appendix C).
The close ties between Golkar and Gorontalo remained intact, even after the collapse of Suharto's New Order in 1998, for two reasons. First, Golkar's top figures who were personally grounded in Gorontalo, such as President B.J. Habibie, Suharto's successor, and Rachmat Gobel, a business tycoon and the son of Thayeb Gobel, played a key role in mobilising public support for creating a new province for the ethnic Gorontalo.Footnote 46 Second, although the New Order's political manoeuvres ended abruptly, individual Golkar politicians in Gorontalo were still able to continue to mobilise local voters using the clientelistic networks they had built. In particular, through financial autonomy and the rapid increase in the sub-provincial budget allocations, Golkar and its local governments have been able to distribute benefits to potential voters through donations, subsidies, and projects that include mosques, schools, and government complexes.Footnote 47 In addition to such official resource distribution, the party also used illicit means of garnering its support in Gorontalo, as manifested in a series of corruption allegations against Golkar's elected officials.Footnote 48 Table 4 shows that Golkar's efforts to secure its electoral clout have been successful in Gorontalo City, where its vote share has far exceeded the national average.
Election results in Gorontalo City during the New Order (DPR, 1982–97)
Sources: General Election Board (LPM) (1983; 1988; 1993; 1997).
The evidence thus suggests that the prevalence of personal voting and clientelism in Gorontalo handcuffed PKS in the mobilisation of potential voters. Despite the fact that all of the party's local cadres were born in Gorontalo City and its neighbouring regions, most did not come from notable families that have successfully translated their moral authority into electoral advantage.Footnote 49 Furthermore, as an influential member admitted, PKS’ regional branch had a range of problems, such as a shortage of funds, a thin electoral base and the stereotype of being opposed to adat, all of which rendered the party poorly rooted in Gorontalo society.Footnote 50 Even so, PKS’ unexpectedly weak political mobilisation is at least in part attributable to its marginalisation from the local political dynamics of clientelism and personal networks.
PKS in Ngada district: Figur besar works
Ngada district is located in west-central Flores, eastern Indonesia. Unlike Gorontalo City, where a singular ethnic identity has been maintained through linguistic and religious homogeneity, Ngada traditionally comprised dozens of clans that maintained their own animistic ceremonies and dialects. Several clans were loosely governed by a head of village, who had the authority to implement adat of marriage and communal rituals.Footnote 51 Depending on geographic proximity and linguistic similarity, these clans and villages were classified by the Dutch into four ethnic groups (Ngahda, Riung, Nage and Keo), which were then recognised as ‘kingdoms’. After Indonesia's independence, the central government integrated these kingdoms into its administrative system by establishing Ngada district. The cohabitation of the four ethnic groups ended in 2007, when the Nagekeo district was approved, comprising the Nage and Keo peoples. Since then, Ngada has been ethnically underpinned by the majority Ngahda, who largely reside in mountainous areas centred in Bajawa, the district's capital, and the minority Riung, who live along the northern coast.
Viewed from the context of long-standing identity politics in Indonesia, PKS has overwhelming disadvantages in mobilising voters in Ngada district. First and foremost, more than 90 per cent of Ngada's population is Catholic. As elsewhere in Flores, Catholicism grew rapidly in the early twentieth century as Dutch missionaries provided a range of educational services to Ngada's local populations.Footnote 52 By contrast, Islam, which once flourished in coastal areas through trade with neighbouring sultanates, only retained a presence on the northern littoral area governed by an ethnic Riung Muslim ruler.Footnote 53 Given the empirical observation that ‘[t]he Islamic parties were practically never chosen by the non-Muslims, because the Islamic parties were clearly for the Muslims’,Footnote 54 it is clear that successfully appealing to non-Muslim voters would be a massive challenge for Islamist parties. Second, like its neighbours in Nusa Tenggara Timur (NTT) province, Ngada is one of Indonesia's most rural and most socioeconomically underprivileged regions; the proportion of primary sectors reliance in the district's gross regional income is about 36 per cent, which is nearly three times the national average of 13.7 per cent.Footnote 55 Furthermore, at present, no tertiary education is provided by the local educational institutions in Ngada.
Notwithstanding these barriers to voter mobilisation, PKS has achieved remarkable electoral gains in Ngada: table 5 shows that PKS increased its vote share in the region, from 0.01 per cent in 1999 to 8.52 per cent in 2014, albeit with a notable subsequent drop in 2019. In particular, the party's performance in the 2014 election was above its provincial and national averages. These unexpectedly dynamic past election outcomes in Ngada were enough to call into question the factors that positively (and negatively) affect PKS’ vote share in a predominantly non-Muslim region. As discussed below, this it largely attributable to its capture of the strong tendency of the personal vote.
PKS’ recent vote shares in Ngada district (DPR, 1999–2019)
Note: aResults of the Electoral District (Dapil) I of East Java province, which includes Ngada district.
Source: General Election Commission (KPU).
As in Gorontalo, early political dynamics in Ngada were driven by religion. This is exemplified by the Partai Politik Katolik Indonesia (Indonesian Catholic Political Union, PPKI), established by Ignatius Josef Kasimo Endrawahyono, a young Javanese intellectual, in 1923. PPKI represented the voice of Indonesian Catholics, who cooperated with the Dutch and were wary of the anti-denominational, communist PKI.Footnote 56 Such political distinctiveness continued after the Second World War. In 1945, Flores Catholic elites formed the Persatuan Politik Katolik Flores (Political Union for Florenese Catholics, Perpokaf) to form the State of East Indonesia, a Dutch-sponsored autonomous state.Footnote 57 Some local rulers in Flores, including Muslim village heads in Riung, hoped to establish a predominantly Catholic ‘United Kingdom of Flores’.Footnote 58 However, after the State of East Indonesia was dissolved by an agreement between the Dutch and the newly independent government in Jakarta in 1949, Perpokaf and Catholic elites in Flores joined the Partai Katolik (Catholic Party) founded by Kasimo in 1945 in Jakarta. Table 6 shows that the Catholic Party won the 1955 parliamentary election in Ngada by an overwhelming majority. Its mobilising power, evidenced by its national performance, was exclusively concentrated in predominantly Catholic regions.
1955 National legislative election in Ngada district
Source: Alfian, Hasil Pemilihan Umum 1955: Untuk Dewan Perwakilan Rakjat (Djakarta: LEKNAS, 1971).
Unlike other parties, the Catholic Party survived the upheavals of the 1950s and early 1960s by having relatively good relations with Indonesia's first two presidents. Notably, it became an important ally to the New Order regime, actively participating in the anti-Communist massacres during and after General Suharto's takeover in 1965–66.Footnote 59 Nevertheless, the party's electoral clout was eclipsed in eastern Indonesia, and especially in Flores, by the rise of Golkar and accompanying institutional changes. Like its colonial predecessor, the New Order appointed local notables, including descendants of former rulers, as regents in Flores.Footnote 60 These appointees soon became campaigners for Golkar, because bureaucrats of all levels were forced to pledge ‘monoloyalty’ to the regime by joining the Korps Pegawai Republik Indonesia (Civil Service Crops, KORPRI), Golkar's umbrella association.Footnote 61 The impact of this bureaucratic mechanism in mobilising Golkar voters was especially profound in rural areas such as Ngada, where the political activities of opposition parties were banned.Footnote 62 Golkar enjoyed additional access to the local population in Flores through government-sponsored, supposedly non-political organisations.Footnote 63 Indeed, Golkar greatly benefited from these advantages by securing the overwhelming majority of the vote in Ngada and Flores. By contrast, the Partai Demokrat Indonesia (Indonesian Democratic Party, PDI), which was created through a forceful amalgamation of the ‘nationalist parties’, including the Catholic Party, helplessly lost its grip over the masses (see table 7).
Election results in Ngada district during the New Order (DPR, 1982–97)
Sources: General Election Board (LPM) (1983; 1988; 1993; 1997).
Golkar's electoral advantages structured by the New Order's institutional mechanisms, however, have gradually declined in Ngada and other Flores districts since the New Order's collapse. Despite the party still maintaining dense organisational networks in every village, Golkar lost its role as a patron that provides goods and services guaranteed by the state to the local notables and the people. Furthermore, unlike in the case of Gorontalo, where the party leadership played a key role in establishing a province for the ethnic Gorontalo, Golkar's apparatus in Flores and NTT province could not manage the demands of a range of ethnic groups eager to have their own districts. In turn, instead of maintaining their loyalty to the former regime's party, many local notables—known as figur (figure) or figur besar (big figures)—switched their political affiliations to other parties that seemed to be able to ‘drive’ them to the local parliaments.Footnote 64 As a result, Golkar's clout in mobilising the masses in Ngada dissipated significantly; its vote share dropped from 31.5 per cent in 1999 to 8.8 per cent in 2019. While post-New Order Golkar has lost its grip over the masses across the archipelago more generally, the magnitude of its declining support base in Ngada was exceptionally large (see Appendix D).
It is also worth noting that local politics in Ngada has been plagued by chronic instability in electoral competition. This was initially caused by the rise of small but locally rooted parties led by figur besar. Partai Merdeka (Independence Party, PM) and Partai Penegak Demokrasi Indonesia (Indonesian Democracy Upholders Party, PPDI) in the 2004 legislative election are noteworthy examples. Notwithstanding their political insignificance at the national level, PM and PPDI, both of which were founded by Ngada-born leaders with decades-long political careers, won the third and the fourth largest shares in Ngada, respectively. As requirements for party formation and maintenance became more burdensome, the former leadership and cadres of small parties in Ngada, mostly Catholics, chose Islamic parties, such as Partai Amanat Nasional (National Mandate Party, PAN), Partai Kebangkitan Bangsa (PKB, National Awakening Party), or even the Islamist PKS.Footnote 65 One Catholic district parliamentarian from PAN commented that switching political affiliations to Islamic parties in Ngada was a ‘strategic and probable’ choice for local politicians who seek opportunities for greater candidate nomination odds than they might find with other major nationalist parties with larger candidate pools.Footnote 66 As a result, the vote shares of Islamic parties in Ngada, which were originally lower than their national averages, increased after local non-Muslim notables affiliated themselves with the parties in either local or national elections (see table 8).
Trends among non-Muslim candidates and election results of Islamic/Islamist parties in Ngada (District Parliament, 2004–19)
In the case of PKS in Ngada, the nomination of figur besar also played a decisive role in shaping the party's electoral success. Beginning in 2004, the party appointed Zainuddin Paru, a descendant of the Riung royal family and a successful lawyer in Jakarta, as one of its candidates for three successive legislative elections. He held wider popular appeal in coastal areas than any other PKS candidates using his personal networks with the local community. Despite his Islamic piety, Zainuddin also prioritised pragmatic issues, such as the creation of a new district for the ethnic Riung and improvements in regional economic conditions, over religious agendas during his campaign.Footnote 67 Zainuddin's personal appeal combined with pragmatism was crucial in improving PKS’ electoral performance in Ngada. For instance, in the 2014 legislative election, he received 5,582 votes in Ngada, representing almost 90 per cent of PKS’ total vote share in the district. Interestingly, a significant portion of his votes came from non-Muslims in northern sub-districts.Footnote 68 During the 2019 election, however, when Zainuddin's retirement from his candidacy triggered a shift in the local party leadership's campaign strategy toward grassroots mobilisation linked to dakwah (Islamic proselytisation), PKS lost its previous support base in Ngada almost entirely.
In sum, like the case of Gorontalo, Ngada's political history underscores the importance of the personal vote in Indonesia, where long-standing political cleavages diminished with the rise of hegemonic Golkar. However, in contrast to Gorontalo where PKS has been continuously sidelined because of an absence of links with local elites who maintain their affiliations with authoritarian legacies, PKS in Ngada overcame its religious disadvantages at the ballot box when it nominated candidates who had roots in local society.
PKS in Mojokerto district: Destined for stalemate
Mojokerto district is located in the centre of East Java province. In contrast to Gorontalo and Ngada, Mojokerto underwent a massive socioeconomic transformation beginning in the mid-nineteenth century, due to the early institutionalisation of colonial direct rule. Specifically, hereditary local rulers and aristocrats, or priyayi, were included in the salaried and centralised bureaucratic system and the introduction of an export-dependent plantation economy greatly increased the scope of human interactions and financial transactions.Footnote 69 The socioeconomic transformation rendered Mojokerto part of a new subculture called arek, an amalgamation of pre-existing cultural elements from Hindu-Buddhist practice, syncretised as well as an orthodox versions of Islam, Chinese immigrants, and even Western modernity.Footnote 70
The increase in socioeconomic heterogeneity also triggered polarisation within Muslim society in Mojokerto and Java. Beginning in the early nineteenth century, the local version of Javanese Islam, which highlighted the mystical synthesis of spiritual forces and antinomianism, faced challenges to its hegemonic status. A group of Islamic intellectuals describing themselves as putihan (the white ones) attempted to purify Islam by teaching Quranic interpretation and Arabic to their students, or santri. The putihan's Islamic teaching spread rapidly through the establishment of their own religious boarding schools (pesantren) across East Java. In Mojokerto, for example, since the founding of the first pesantren in 1830, at least six more were established by the early twentieth century.Footnote 71 Regarding pre-existing beliefs and Quranic interpretation, the santri was further divided into a traditionalist group based in East and Central Java and a modernist group based in West Java and Sumatra. Amid the rise and division of santri Muslims, the priyayi also maintained their own Islamic culture influenced by Hindu-Buddhist meditation practices.Footnote 72
The polarisation of Javanese society led to political cleavages by the early twentieth century. In 1926, several traditionalist kyai (the principals of pesantrens), led by Hasyim Asy'ari founded Nahdlatul Ulama (NU) to respond to the establishments of modernist mass organisations, such as Sarekat Islam and Muhammadiyah. Using the kyai's nearly absolute authority over the santri community and role ‘not merely as a mediator of law and doctrine, but of holy power itself’ in rural Java,Footnote 73 NU emerged as the largest Muslim organisation in colonial Indonesia. Meanwhile, the secular-nationalist Partai Nasional Indonesia (Indonesian National Party, PNI) led by Surabaya-born Sukarno also successfully took root in rural East Javanese society by securing support from priyayi and their clients. Lastly, the Partai Komunis Indonesia (Indonesian Communist Party, PKI) had its stronghold among peasants throughout East Java. Mojokerto itself represented one of the bastions of the leftist movement in the late 1940s.Footnote 74 The competition among political parties representing specific variants of Javanese Muslims continued in the first legislative election in 1955 when the ‘big four’ parties earned approximately 86 per cent of the total vote in Mojokerto district. However, compared with the national results, PNI and NU performed better locally, indicating that the two elite groups in Mojokerto society, the kyai and the priyayi, played key roles in mobilising voters (see table 9).Footnote 75
1955 National Legislative Election in Mojokerto district
Source: Alfian, Hasil Pemilihan Umum 1955: Untuk Dewan Perwakilan Rakjat (Djakarta: LEKNAS, 1971).
The cleavage-driven contours of electoral competition, or aliran (stream) politics, in Mojokerto district survived the New Order. In the 1971 election, which was held after the dissolutions of Masyumi and PKI, NU came in second, with 18.7 per cent of the national vote. Its relative success was largely a function of the kyai's active grassroots campaigning through religious lectures and preaching, to protect votes of santri Muslims in East Java from Golkar, which had absorbed a substantial portion of secular-oriented Muslims and priyayi votes.Footnote 76 Furthermore, in contrast to Gorontalo and Ngada, where Golkar continuously scored overwhelming victories, the Islamic PPP had performed relatively well in Mojokerto, securing at least 20 per cent of the vote in Mojokerto in each of the New Order's legislative elections (see table 10). Considering that the kyai's attitude toward Golkar and PPP became less enthusiastic beginning in 1984, when NU withdrew from the formal political arena,Footnote 77 PPP's stable performance may be attributed not only to the kyai's guidance but also to the santri voters’ strong allegiance to the sole Islamic party during the New Order.Footnote 78 In addition, the secular-nationalist PDI gradually recovered some of its electoral support. The successful involvement of Sukarno's daughter, Megawati Sukarnoputri, in PDI's campaigning for the 1992 election was particularly threatening to the regime. Indeed, the political rise of Megawati led to New Order machinations that paralysed PDI before the 1997 election.Footnote 79
Election results in Mojokerto district during the New Order (DPR, 1982–97)
Sources: General Election Board (LPM) (1983; 1988; 1993; 1997).
In Mojokerto, the post-New Order elections marked a return of aliran politics. In the 1999 legislative election, for example, a near balance emerged between voters supporting the secular-nationalist Partai Demokrasi Indonesia-Perjuangan (Indonesian Democratic Party for Struggle, PDI-P) led by Megawati and the Partai Kebangkitan Bangsa (National Awakening Party) founded by Abdurrahman Wahid, the charismatic former leader of NU. Golkar and Islamic parties ideologically inheriting Masyumi drew about 10 per cent and 8 per cent of the vote, respectively. Considering Dwight King's finding highlighting a close association between Golkar's support base and the former stronghold of PKI in East and Central Java,Footnote 80 the electoral outcomes in the 1999 legislative election in Mojokerto were almost identical to those in 1955. Since then, except for the 2009 legislative election in which PKB suffered from a party split caused by internal disputes and Partai Demokrat (Democrat Party, PD) benefited from the nationwide popularity of its founder President Yudhoyono,Footnote 81 electoral mobilisation in contemporary Mojokerto has been largely driven by competition among parties that represent specific elements within the local Muslim community (see Appendix E).
Likewise, PKS’ electoral base in Mojokerto has socioeconomically definitive characteristics divided into two categories: (1) well-educated Muslims unaffiliated with NU; and (2) economically underprivileged Muslims. As mentioned, the majority of Muslims in Mojokerto are traditionalists who accommodate pre-Islamic (Javanese) culture in their understanding of Quranic texts. Traditionalists also highlight the intellectual authority of kyai (or ulama). Such elements of traditionalist Islam are generally incompatible with modernist and Islamist Muslims who maintain scripturalist Islamic views. Thus, in the words of a notable kyai in Mojokerto, ‘the local masses do not tend to understand PKS because of the different interpretations’.Footnote 82 Given that rural kyai in particular exert uncommon electoral influence over their alumni and residents by ‘recommending a proper party or candidate’,Footnote 83 PKS has suffered in appealing to voters of the traditionalist camp in Mojokerto. Instead, the party has focused on mobilising the educated Muslims already familiar with the party and more apt to be concerned with its policies.Footnote 84 In addition, PKS in Mojokerto also paid heed to the electorate in need of economic and welfare assistance. According to the PKS’ regional secretary of Mojokerto district, the party apparatus has regularly offered a range of welfare services, including humanitarian aid, free housing and education, and female enrichment programmes through a Women's Justice Station.Footnote 85
Thus, owing to its reliance on specific milieus rather than individual candidates as its support base, PKS’ electoral performance in Mojokerto has been more stable and predictable than its outcomes in Gorontalo and Ngada. However, this does not indicate that the party's local electoral mobilisation was successful. Rather, as indicated in table 11, PKS has achieved about 4 per cent of total votes in Mojokerto since the 2004 election, which is consistently lower than the party's national average. In other words, PKS’ electoral stalemate in Mojokerto is caused by long-standing and institutionalised voting patterns linked to existing local sociopolitical cleavages.
PKS’ recent vote shares in Mojokerto district (DPR, 1999–2019)
Note: aResults of the Dapil VIII of East Java Province, which includes Mojokerto district.
Source: General Election Commission (KPU)
Statistical observations for model testing
The findings from the qualitative, case-based analysis suggest that the primary factor associated with the political mobilisation of PKS varies with local political dynamics. More concretely, PKS’ national electoral performance outside Java, where Golkar's political dominance during the New Order overwhelmed the pre-existing political cleavages, tended to depend on whether the party recruited high-profile candidates who were also capable of using their own personal networks. By contrast, where a region's long-standing political cleavages survived the New Order's political engineering, PKS’ electoral base was based primarily on specific milieus more likely to be influenced by the party's programmes and ideological underpinnings. The findings thus resonate with several recent studies highlighting that the endurance of religious structures in the Indonesian party system is dependent on subnationally varying degrees of authoritarian legacies, particularly of Golkar and its elites’ control over state resources.Footnote 86 Figure 3 schematically summarises the findings from the Mb-SNA.
Schematic summary of the Mb-SNA
How generalisable are these findings? Do they have the potential to produce a theory capable of explaining other cases at least across Indonesia? To examine the generalisability of the case evidence, I employ an additional statistical assessment by adding new variable(s) found by the Mb-SNA to the model used for the preliminary LNA.Footnote 87 Accordingly, I use two revised versions of the Vote Concentration Index (VCI), which was originally devised by Giora Goldberg,Footnote 88 based on the following formulas:
$${\rm VCI}1\_{\rm PKS}( {\rm a} ) = {\rm \;}\displaystyle{{{\rm no}.{\rm \;of\;votes\;won\;by\;PKS^{\prime}\;top\;candidate\;in\;a\;district}/{\rm city\;}( {\rm a} ) } \over {{\rm total\;no}.{\rm \;of\;votes\;won\;by\;PKS\;in\;a\;district}/{\rm city\;}( {\rm a} ) }} \times 100$$and,
$${\rm VCI}2\_{\rm PKS}( {\rm a} ) = {\rm \;}\displaystyle{{{\rm no}.{\rm \;of\;votes\;cast\;for\;PK}{\rm S}^{\rm ^{\prime}}{\rm \;label\;in\;a\;district}/{\rm city\;}( {\rm a} ) } \over {{\rm total\;no}.\;{\rm of\;votes\;won\;by\;PKS\;in\;a\;district}/{\rm city\;}( {\rm a} ) }} \times 100$$VCI1 measures the relative electoral strength of PKS’ top candidate over either the party's other candidates or the party itself on the list.Footnote 89 Thus, when the VCI1 is close to 1, it refers to a single candidate collecting almost all votes won by PKS. Further, VCI2 is introduced to measure the salience of the party vote, which is frequently regarded as opposite the personal vote in developing democracies.Footnote 90 In other words, a higher VCI2 score means PKS’ ideology and policies are more important vis-à-vis the electoral appeal of specific candidate(s). In addition to the two VCI variables for PKS, I included VCI1_Golkar to examine the influence of the decades-long political dominance of Golkar, which was maintained through a strong patronage network, on PKS’ electoral performance. To minimise the possibility of endogeneity, these three VCIs were calculated from the vote shares in the 2004 legislative election at the district level.Footnote 91
Table 12 reports the results of the regression analysis using the revised set of variables. The results generally support the findings from the Mb-SNA based on the three localities. The first column of the table shows that an increase of one standard deviation in VCI1_PKS correlates with a 0.7 per cent increase in vote share for PKS, the same increase in VCI1_Golkar correlates with a 0.3 per cent decrease in PKS’ electoral performance, other things being equal. This result suggests that PKS generally performed better where it fielded candidates with strong roots (for example, in Ngada district) and poorer in places where high profile local figures remained loyal to Golkar (for example, in Gorontalo City). By contrast, VCI2_PKS, which represents the proportion of the party vote of PKS, is not statistically significant, indicating that PKS’ strategy of mobilising voters through a strong commitment to dakwah and well-organised programmes, which has been recognised as a crucial factor in the party's electoral growth, has only limited influence on the party's electoral performance.Footnote 92 This outcome must be interpreted cautiously, however, given the fact that VCI indices are based on the electoral results in 2004, only six years after the formation of PKS.
Model-testing aggregate analysis of PKS’ electoral performance
Note: 
${\rm p}^{\rm \ast } < {\rm \;}.10, \;{\rm \;\;}{\rm p}^{{\rm \ast{\ast} }} < {\rm \;}.05, \;{\rm \;}{\rm p}^{{\rm \ast{\ast} \ast }} < {\rm \;}.01$.
As shown in the table 12's second column, the pattern of political mobilisation differed in regional Java. While VCI1_PKS and VCI2_PKS are statistically insignificant, VCI1_Golkar correlates with about a 0.9 per cent increase in PKS’ vote share, other things being equal. Furthermore, aside from population density, all the variables that represent socioeconomic indicators significantly affected the electoral performance of PKS, meaning that PKS performed better in Javanese regions with lower poverty rates, higher levels of education, lower levels of economic prosperity, and higher unemployment rates. This largely resonates with research based on a class perspective,Footnote 93 which argues that PKS’ political rise resulted from the mobilisation of economically vulnerable sectors in the urban areas, such as the petty bourgeoisie, the educated unemployed, and low-income workers. In turn, the findings likely reflect the electoral stalemate in Mojokerto district, where Golkar's electoral clout rapidly eroded as cleavage-based voting patterns re-emerged. Unlike urban areas in Java, where PKS has effectively exploited the religious aspirations of middle-class Muslims from diverse backgrounds,Footnote 94 Mojokerto's stable and rural economic landscape hinders the growth of PKS’ support base and has reinforced support for parties and candidates affiliated with the traditionalist NU.
Conclusion
This study attempts to develop a theoretical argument for the political mobilisation of the Islamist PKS in Indonesia. Findings from the nested analysis suggests that variations in PKS’ electoral outcomes across subnational units were the result of local political dynamics. In other words, the party's voter mobilisation depended largely on whether it was able to manage the primary source of electoral appeal in a local electorate. In Gorontalo City and Ngada district, located outside Java, it was personalities and clientelistic networks that determined the electoral clout of each party. By contrast, as shown in the case of Mojokerto district, long-standing cleavages within the Muslim community were more influential in determining voter choice than other factors.
PKS has been often understood as an exception in Indonesian politics because of its unique institutional characteristics, including its focus on good (Islamic) governance, commitment to anti-corruption, organisational cohesion, and grassroots activities, which are frequently absent among other parties in Indonesia.Footnote 95 Scholars thus point out that PKS’ high levels of institutionalisation have been crucial in enhancing the party's electoral appeal and sustainability.Footnote 96 On the contrary, other Islamic/Islamist parties based on factional interests and clientelistic networks were assumed to be inherently fragile and prone to underperformance in elections.
However, the 2014 legislative election confounded the institutionalist framework. Among the parties grouped under the label of political Islam, only PKS experienced a decline in its vote share. In the meantime, other parties, such as PKB, PAN, and PPP, which suffered from chronic internal disputes and a dearth of policy packages, successfully achieved electoral gains compared with the previous election. This study explains this divergence by noting that the political dynamics of local politics do not always welcome an institutionalised party like PKS. The findings can be also applied to the 2019 legislative election. Although PKS successfully increased its vote share by 1.5 per cent, in many regions, the party's electoral growth is largely attributable not to its organisational coherence and institutionalised activism, but rather to strategic manoeuvres that aroused the antagonism of conservative Muslims toward the incumbent president Joko Widodo by joining an electoral coalition led by Prabowo Subianto, Jokowi's political rival.Footnote 97
In the broader context of political Islam, this study contributes to a better understanding of how Islamist parties’ adaptation to pre-existing and localised voting patterns matter in terms of mobilising supporters in the electoral arena. Like PKS, due to its long history of isolation from the political mainstream, almost all Islamist parties in the Muslim world started off as peripheral parties with limited autonomy to gain ground in electoral competitions.Footnote 98 Furthermore, even if the participation of Islamists in the electoral arena reflects popular discontent with existing parties, there is no guarantee that such disappointment automatically translates into voter support.Footnote 99 Under these circumstances, at least to some extent, Islamist parties have to act strategically by adapting to the prevalent patterns of voting and interparty competition. Recent studies have shown that the electoral performances of Islamist parties in Middle Eastern states, such as Morocco, Tunisia, and Turkey, are highly dependent on whether they are able to accommodate and absorb non-religious voter demands in varied localities across a country, where the old regime elites still politically prevailed.Footnote 100 Using the case of PKS, this study confirms that similar patterns of Islamist political mobilisation are found in Indonesia, the world's largest Muslim democracy.
Appendix A. Descriptive statistics of preliminary LNA and model-testing LNA
Appendix B. Preliminary aggregate analysis of PKS’ electoral performance
The table below reports the results of Ordinary Least Squares (OLS) multiple regressions to examine the competing theories regarding PKS’ voter mobilisation. The analyses suggest two important takeaways. First, among the Islamic Piety variables, only the proportion of Muslims is statistically significant with a positive coefficient. This indicates that only the numerical strength of the Muslim community over other religious minorities, instead of other institutional factors favouring the promotion of Islam, positively affect PKS’ electoral outcomes. Second, at least in the case of the 2014 election, higher levels of urbanisation, education, and unemployment are associated with better PKS performance. More importantly, the coefficients and levels of significance for these variables remain stable even with the inclusion of the poverty rate and per capita GRDP, which were statistically insignificant. This result thus contradicts the view expressed by some specialists in Indonesian politics who highlight a direct relationship between absolute levels of personal wealth and the vote for Islamists.Footnote 101 Rather, similar to evidence from the Middle East and North Africa, this suggests that well-educated urban Muslims with employment-related grievances against the current political and economic system form one of the central bases of PKS’ political mobilisation.Footnote 102 The overall findings from the preliminary LNA also leave room for a fine-grained analysis of the determinants of PKS’ vote share using an Mb-SNA.
Appendix C. Golkar's vote shares in Gorontalo City in DPR elections, 1999–2019
Appendix D. Post-New Order election results in Ngada district and Indonesia overall (DPR elections, by largest parties)
Appendix E. Post-New Order election results in Mojokerto district and Indonesia overall (DPR elections, by largest parties)
Appendix F. Robustness checks: Model-testing LNA (Mt-LNA)
I address concerns about multicollinearity in order to assess the robustness of the results from the regressions used in Model-testing LNA (Mt-LNA). Multicollinearity arises when ‘a combination of variables makes one or more of the variables largely or completely redundant’. It is problematic since it may not provide valid information about which independent variables are important with respect to others as predictors.Footnote 103
One simple way to check for a multicollinearity problem is by looking at the variance inflation factor (VIF). The table display presents the VIF of independent variables included in the Mt-LNA introduced in table 12.
The table indicates that none of the VIF scores for each independent variable is higher than 5 or 10, which statisticians consider to be the threshold for multicollinearity. Furthermore, the overall VIF score for Mode 1 is only 1.85, which suggests that multicollinearity is highly unlikely to affect the results of the regression analysis.
