On April 27, 1997, Muhammad Zabara stood outside a polling station in the old city of Sanaa. In a neatly pressed suit and tie, his short hair and mustache freshly trimmed, he greeted voters who had turned out for Yemen’s second post-unification parliamentary elections. A team of Western election monitors approached him and asked whether he was a candidate. In English, he answered that he was the district’s candidate from the Yemeni Reform Group, a conservative party with an Islamist agenda. “But Ahmad Raqihi is the Islamist candidate for this district,” said one of the monitors, referring to Zabara’s main rival, who dons a turban and beard. “You don’t even look like an Islamist.” 
Like many Western observers, the monitors presumed that Islamist candidates would be Taliban-esque figures, shouting anti-Western slogans.  These superficial perceptions fuel debates about the so-called paradox of democracy in the Middle East: the fear that democratic elections might bring to power an anti-democratic regime.  Scholars, journalists, regimes and their foreign backers, as well as a range of democratic activists in the region, have expressed concerns about Islamists’ participation in elections. How can one know which groups are truly committed to pluralist politics? Might the exclusion of certain groups, though undesirable in principle, be necessary to ensure that democratic stirrings are not immediately crushed? Should Islamists be singled out as most likely to harbor hidden anti-democratic agendas? Although these concerns are legitimate, evidence suggests that the powerful images of Islam and Islamists that have driven the debate about the paradox of democracy may actually distract attention from regimes’ efforts to manipulate electoral outcomes, rather than illuminate obstacles to democratic reform in the Middle East. An examination of a decade of electoral results and representative trends will enable us to separate fact from fiction concerning Islamists’ participation in elections.
The Representation of Threat
The 1979 revolution in Iran brought to power a conservative and repressive Islamist regime. The June 1989 military coup in the Sudan brought to power a government supported by the National Islamic Front, led by Hasan al-Turabi. Although Algeria’s electoral experiment in the early 1990s ended in an anti-democratic military coup, global collective memory recalls not the putsch but the near-victory of the Islamic Salvation Front (FIS), which won 48 percent of the vote. What stands out about these oft-cited examples of the Islamic “threat” is that not one illustrates the paradox of democracy, i.e., an authoritarian regime coming to power through elections. Algeria’s FIS never had the chance to take power, and although the governments of Sudan and Iran are patently intolerant, neither emerged through elections. Yet these cases fuel generalizations about Islamist political participation. Since these Islamist groups seized or might have seized power by force, Islamists participating in elections should be expected to do the same. Faced with such ominous depictions, even dedicated democratic reformers accept the conclusion that all Islamist participation should be viewed with caution. Meanwhile, these threatening images serve the maintenance of existing non-democratic regimes. A more subtle aspect of the idea of an Islamist “threat” is that not all Muslim electoral participation is dangerous. The ubiquitous media image of a veiled woman casting her ballot, for example, has become a key symbol of the enigma of democratization in Middle Eastern societies. In a sharp dichotomy between desirable and undesirable political behavior, the voting female, a relatively passive individual participant, is juxtaposed with threatening mobs of bearded Islamist activists. Because these dual images are so pervasive, large numbers of Islamists like Zabara — activists but not militants — are routinely overlooked.
Scholars as well as policymakers now distinguish between moderate and radical Islamists, though “moderation” is surprisingly difficult to define. A group may hold moderate and radical views simultaneously, for example with regard to domestic versus foreign policy, or economic versus social issues. One definition hinges on domestic political behavior:
“Moderation” will denote those Islamic groups and activists who formally declare their respect for, and commitment to, pluralism and the democratic principle and renounce the use of violence in achieving their objectives. 
Radicals aim to seize absolute power by force; moderates abide by electoral results. Historical examples of radicals, then, cannot be used to predict what moderates will do.
Yet while this approach promises more nuanced analyses, at least two problems remain. First, the dichotomy between desirable and undesirable participation remains unresolved. “Desirable” is simply redefined to include Islamist activism that does not threaten the existing regime’s grip on power. This leaves no room for groups that accept the limits of the system but seek to change existing hierarchies of power, perhaps by deposing an incumbent elite. Ruling regimes use the image of an Islamist threat to inspire fear and slow the pace of democratization. This obscures the extent to which real obstacles to democratization stem from entrenched political elites’ reluctance to loosen their grip on power. Second, the moderate-radical construct does not address concerns that certain groups may adopt “moderate” behavior in order to hide radical agendas. Before addressing this last issue, we must take a closer look at election results.
Assessing Election Results
In the past decade, Jordan, Kuwait, Lebanon, Morocco, Yemen and the Palestinian Authority have all initiated or renewed political participation based on nationally elected assemblies. Egypt, Pakistan and Turkey have well-established electoral traditions, though longevity hardly entails greater transparency. Each country has seen Islamists declare their commitment to working within a democratic, pluralist system.
In no case has an Islamist group won elections only to establish an authoritarian Islamist state, nor are any poised to do so. On the contrary, most of these states continue to be ruled by entrenched regimes. Islamist groups have won sizable blocs, but seldom if ever a majority.  Most saw only modest changes in their share of seats in subsequent elections, and many saw declines. In Lebanon, Hizballah took seven of 128 seats (6 percent) in 1992, and eight (7 percent) in 1996.  Jordan’s Muslim Brotherhood won 22 of 80 seats (27.5 percent) in 1989, compared to 17 (21.3 percent) in 1993 (when it competed as the new Islamic Action Front). The Yemeni Reform Group also saw its share decrease from 62 of 301 (20.6 percent) in 1993 to 53 (17.5 percent) in 1997.
The extent of a particular group’s success is not always reflected in election results. Where Islamist parties are illegal, blocs are sometimes still identifiable. In Tunisia’s 1989 race, Rachid al-Ghannouchi’s Islamic Tendency Movement ran candidates as independents and claimed 13 percent of the total vote and as much as 30 percent in some districts. 
However, election results may be misleading in terms of the actual size of assembly blocs, in 1993 as many as 30 supporters of the Yemeni Reform Group won by running as candidates for the ruling General People’s Congress party.  In Jordan’s 1989 elections, the Muslim Brotherhood’s 22 seats (of 80) were augmented by 12 independents, giving Islamists the largest bloc with 42 percent. In 1993 the same group won only 17 seats, largely due to changes in the electoral law.  Blocs may also be smaller than election results suggest. Jordan’s seven Islamist deputies elected in 1997 are sharply divided, and only two are members of the Action Front. In Kuwait’s October 1992 elections, candidates representing three Islamist groups won a combined ten seats (of 50), with another ten bringing their number to 20 (40 percent).  In 1996, they won 18 seats (36 percent).  Although these deputies try to function as a bloc, in practice the salafi-oriented Islamic Popular Alliance deputies are frequently at odds with other Islamists in the assembly. The Yemeni Reform Group has no relations with the single deputy from the Islamist Hizb al-Haqq. Lebanon’s Hizballah, Amal and Jama‘at Islamiyya, though divided on a range of issues, forged electoral coalitions to increase their success at the polls. As the case of Pakistan illustrates, the proliferation of Islamist parties may actually serve more to divide the Islamist vote than to increase their influence. 
Finally, alliances are sometimes forged across ideological lines. In Egypt’s 1984 elections, the coalition that competed under the Wafd Party won 58 of 390 seats, eight of which went to Muslim Brotherhood members. In 1987, the new Alliance formed by Egypt’s Muslim Brotherhood and the Labor and Liberal parties won 56 of 400 seats, with 35 going to Muslim Brotherhood members. Jordan’s Islamic Action Front worked with several smaller parties and two former prime ministers to organize a boycott of the 1997 elections. Thus, while Islamists sometimes coordinate their efforts, just as often they fight among themselves or cooperate with non-Islamists.
Prevailing trends in Islamist electoral participation cannot be fully understood without assessing each government’s efforts to influence electoral outcomes. Mechanisms for maintaining a political monopoly include restrictions on voter and candidate eligibility, gerrymandering, choice of electoral system and legal constraints on political parties. Many governments adopt or change laws between elections in order to produce different results. Jordan adopted a law just months before the November 1993 elections which gave each voter only one vote, rather than one for each seat in her district. Political parties and less-prominent independents were immediately weakened, as voters could no longer support both their tribe and a favorite party or independent candidate. Islamic Action Front candidates, who won nearly the same percentage of votes in 1989 and 1993, saw their share of seats decline from 22 to 17. When the law was not substantially revised prior to the 1997 elections, the Front led a boycott.
In Tunisia, President Zine El Abidine Ben Ali rejected the opposition’s demands for a proportional representation system for the 1989 elections. His party won every seat, despite the strong share of votes won by members of the Islamist Hizb al-Nahda.  In Lebanon, the 1996 electoral law — virtually identical to the 1992 law — was applied differently in various electoral districts in order to protect the positions of certain militia leaders.  The new law was found unconstitutional by a high court, but after a few cosmetic touches, Parliament approved it just days before polling began. Although pro-government parties won 95 percent of the seats, opposition candidates won 30-40 percent of the vote in many districts. 
Governments do not always succeed in their attempts to engineer election results. In Algeria, for example, the strong FIS showing in late 1991 was due in part to electoral design.  A new law increased the number of seats from 295 to 430, adding districts primarily in rural areas assumed to be strongholds of the ruling National Liberation Front. The government miscalculated the extent of their support in these regions, however, and won only 16 seats with 1.61 million votes. FIS took 188 seats with only 3.26 million votes. 
In Jordan, the distribution of seats among districts also disadvantages certain groups. Karak, a town of predominantly East Bank Jordanians, is represented by nine seats, one for every 10,000 citizens. Amman, by comparison, has 18 seats, one for every 20,000 citizens. Most marginalized are parties with strong followings among the Palestinian communities of Amman — notably the Islamic Action Front.
Electoral systems also affect the compositions of national assemblies. Most states in the region do not use proportional representation, as it tends to insure that smaller parties are better represented by distributing seats according to the percentage of votes received. Yet because parties under a proportional system must receive a minimum percentage of the vote to be represented in the assembly, governments can increase the threshold in order to exert further control. In Turkey’s proportional representation system, for example, the government sought to exclude Kurdish candidates by setting a 10 percent minimum at both the national and district level, which they were unlikely to reach. Ironically, the strategy had the unintended effect of increasing the success of another disfavored group, the Islamist Refah Party, which many disgruntled Kurds chose on polling day. Governments also influence electoral outcomes indirectly. In Jordan, the Ministry of the Interior refused to process requests submitted by the Islamic Action Front to hold electoral rallies prior to the 1993 elections. Muslim Brotherhood leader ‘Abd al-Majid Thunaybat, a lawyer by training, won a legal case against the ministry, but with only weeks left before polling, Action Front candidates had already been hurt. Egypt’s government undermined the chances of Muslim Brotherhood candidates in 1995 by imprisoning more than 100 members in the months before polling. Only 100 of them eventually stood for the elections, along with another 120 candidates from the Islamist-oriented Labor party; not a single Islamist succeeded in winning a seat. 
Democracy Beyond the Ballot Box
The past decade of elections in the Middle East indicates that Islamist parties are not poised to take over governments through success at the polls, even assuming that less electoral engineering would increase their shares of seats. Greater Islamist participation might even lead to a decrease in these groups’ popularity, as they fail to deliver on campaign promises. Moreover, increased participation might moderate political actors as they face the challenges of maneuvering within a pluralist system.
Islamist parties may even incorporate pluralist principles into their doctrines, as reflected in emerging cooperative arrangements that cross ideological lines. The Yemeni Reform Group explored cooperation with leftist opposition parties when it saw its own share of power decreasing, Jordan’s diverse opposition bloc meets at the Islamic Action Front headquarters in Amman for the simple reason that it houses a large conference room, and Egypt’s Muslim Brotherhood and Lebanon’s Hizballah have both formed electoral alliances with non-Islamists. Such cooperation does not prove that these groups are committed to democratic principles. It does suggest that while regimes are seeking to preserve their power by manipulating electoral processes, a range of political groups, including Islamists, are gaining experience in pluralist political practices.
Internal party structures and practices may also offer insights into levels of democratic commitment. Some Islamist parties have been under the same leadership for decades, while others have adopted democratic structures. Although one may question the extent to which internal elections have produced changes in these parties’ leadership, it is nonetheless significant when groups adopt such practices. Jordan’s Muslim Brotherhood recently saw an upset in its consultative council elections when the leaders of the Islamic Action Front failed to win seats.  By comparison, Egypt’s Muslim Brotherhood has been extremely undemocratic in terms of its rank and file. Recently, young members at odds with the group’s leadership joined with non-Islamists in a failed attempt to establish the Wasat Party. Muslim Brotherhood leaders worked against these younger members, even presenting damaging material to the committee that reviews applications.
Yet few parties in the region are internally democratic. Egypt’s Wafd party has been a champion of democracy for decades, but has never incorporated democratic mechanisms into its internal structure. In fact, Egypt’s Labor, Nasserist, and Wafd parties all remain under the control of pre-1952 generations. Jordan’s new National Constitution Party fell apart when its leaders, hailing from eight smaller parties, could not agree on a mechanism for power-sharing. Yemen’s ruling General People’s Congress, like many ruling parties in the region, is a thin guise for an autocratic military regime. For those concerned about potential threats to democracy, internal practices may hold important lessons for assessing Islamist and non-Islamist parties alike.
As the evidence shows, the so-called paradox of democracy is perhaps the least of the obstacles to pluralist politics in the Middle East. While one may fairly question the sincerity of any group’s commitment to democratic principles, the real puzzle is why Islamists alone are at the center of such debates. What of the blatant manipulation of election outcomes by existing regimes? Why are other internally non-democratic parties not considered dangerous? Regimes point to the Islamist “threat” to justify foot dragging in the expansion of democratic processes, but even many leftist opponents of authoritarianism are surprisingly quiet with regard to attacks against Islamists. The real question is not whether Islamists pose a threat, but what political agendas are served by continuing to paint Islamists as a monolithic, anti-democratic mob.
Perhaps the final irony concerning Islam and democracy is that of Iran, where revolution brought to power an unquestionably repressive government, but which has gradually become more democratic than under the Shah, in large part through the functioning of elected representative bodies.  To be sure, Iran’s political system is far from democratic. Yet this suggests that in promoting pluralist political regimes based on the rule of law and accountability, “Islam” is perhaps not the determining variable of regime character. Indeed, the inclusion of a wide range of voices, Islamist and non-Islamist, may provide the best mechanism to achieve such goals.
 This article draws on two years of field research supported by the Social Science Research Council, the Fulbright Commission, and the American Institute for Yemeni Studies. I wish to thank those who provided comments on earlier drafts.
 “Islamist” refers to the subset of Muslims (followers of the Islamic faith) who seek to promote an Islamic agenda for social, political and economic reform. All Islamists are therefore Muslims; not all Muslims are Islamists.
 For example, Judith Miller argues that “free elections seem more likely than any other route to produce militant Islamic regimes that are, in fact, inherently anti-democratic.” “The Challenge of Radical Islam,” Foreign Affairs 72/2 (Spring 1993), p. 52.
 Gudrun Krämer, “Cross-Links and Double Talk? Islamist Movements in the Political Process,” in Laura Guazzone, ed., The Islamist Dilemma: The Political Role of Islamist Movements in the Contemporary Arab World (Ithaca, NY: Ithaca Press, 1995), p. 42.
 Algeria’s FIS did not even win a majority, though it may have if the electoral process had not been suspended.
 Paul Salem, “Skirting Democracy: Lebanon’s 1996 Elections and Beyond,“ Middle East Report 203 (Spring 1997), p. 28. Augustus Richard Norton reports that Hizballah won seven seats in the 1996 contest. See his “Hizballah: From Radicalism to Pragmatism?” Middle East Policy 5/4 (January 1998), p. 156.
 Krämer, p. 45.
 Sheila Carapico, “Campaign Politics and Coalition Building: The 1993 Parliamentary Elections,” Yemen Update 33 (Summer/Fall 1993), p. 38.
 Muslim Brotherhood candidates in the 1993 contest participated as the Islamic Action Front, which was established in 1992 when Jordan legalized parties.
 The Constitutional Islamist Movement and the Islamic Popular Alliance are both Sunni; the National Islamic Alliance is predominantly Shi‘i. Krämer, p. 45.
 Shafeeq N. Ghabra, “Balancing State and Soociety: The Islamic Movement in Kuwait,” Middle East Policy 5/2 (May 1997), p. 67.
 S. V. R. Nasr, “Islamic Opposition in the Political Process: Lessons from Pakistan,” in John L. Esposito, ed., Political Islam: Revolution, Radicalism or Reform? (Boulder: Lynne Rienner, 1997), p. 153.
 See Christopher Alexander, “Authoritarianism and Civil Society in Tunisia,” Middle East Report 205 (Fall 1997), p. 35. Hizb al-Nahda candidates ran as independents.
 Salem, p. 27.
 Ibid, p. 28.
 Following the strong FIS victory in local elections in June 1990 (winning 55.42 percent, 853 of 1,539 municipalities), the government sought to weaken FIS’s popularity by “stripping the Islamist mayors of their power to sell plots of land to homeless citizens at a symbolic price. This…backfired, as it prompted the FIS to stage mass demonstrations and demand early parliamentary and presidential elections.” Hamou Amirouche, “Algeria’s Islamist Revolution: The People Versus Democracy?” Middle East Policy 5/4 (January 1998), p. 90.
 John P. Entelis, “Civil Society and the Authoritarian Temptation,” in A. R. Norton, ed., Civil Society in the Middle East, vol. 2 (Leiden: E.J. Brill, 1996), p. 85.
 Mona Makram-Ebeid, “Egyptian Elections,” Middle East Policy 4/3 (March 1996), p. 131.
 Action Front leaders Ishaq al-Farhan and ‘Abd al-Latif ‘Arabiyyat both failed to win seats in the Muslim Brotherhood’s 120-seat consultative council elections in June 1998.
 See Mohsen M. Milani, “Political Participation in Revolutionary Iran,” in Esposito, pp. 77-93.