Author's note: I presented earlier versions of this article at a workshop in the Department of Near Eastern Studies, Princeton University, and at a meeting of the Project on Middle East Political Science, George Washington University; I thank the participants for their comments. I also thank Melani Cammett, Amaney Jamal, Nikolay Marinov, Shibley Telhami, and four anonymous reviewers for their additional suggestions on this article. For their generous financial support of the survey, I thank the Center for International Development and Conflict Management, the Department of Government and Politics, and the Designated Research Initiative Fund at the University of Maryland, as well as the MacMillan Center and the Institution for Social and Policy Studies at Yale University.

¹ The “season for money” quip comes from a senior official in the Lebanese Association for Democratic Elections (LADE), http://www.lade.org.lb, Lebanon's premier independent civil society organization focused on election monitoring and electoral reform. Interview, July 2008.

² Among scholars of the Middle East, see Lindsay J. Benstead, “Does Casework Build Support for a Strong Parliament? Legislative Representation and Public Opinion in Morocco and Algeria” (PhD diss., University of Michigan, 2008); Blaydes, Lisa, Elections and Distributive Politics in Mubarak's Egypt (New York: Cambridge University Press, 2011)Google Scholar; Brownlee, Jason, Authoritarianism in an Age of Democratization (New York: Cambridge University Press, 2007)CrossRefGoogle Scholar; Cammett, Melani, “Partisan Activism and Access to Welfare in Lebanon,” Studies in Comparative International Development 46 (2011): 70–97CrossRefGoogle ScholarPubMed; Cammett, Melani and Issar, Sukriti, “Bricks and Mortar Clientelism: Sectarianism and the Logics of Welfare Allocation in Lebanon,” World Politics 62 (2010): 381–421CrossRefGoogle ScholarPubMed; Jamal, Amaney, Barriers to Democracy: The Other Side of Social Capital in Palestine and the Arab World (Princeton, N.J.: Princeton University Press, 2007)Google Scholar; and Lust-Okar, Ellen, Structuring Conflict in the Arab World (New York: Cambridge University Press, 2005)CrossRefGoogle Scholar.

³ Magaloni, Beatriz, Voting for Autocracy (New York: Cambridge University Press, 2006), 19CrossRefGoogle Scholar.

⁴ Harik, Iliya, “Voter Participation and Political Integration in Lebanon, 1943–1974,” Middle Eastern Studies 16 (1980): 30CrossRefGoogle Scholar.

⁵ Interviews, senior officials in the Lebanese Transparency Association, June 2009.

⁶ Corstange, Daniel, “Sensitive Questions, Truthful Answers? Modeling the List Experiment with LISTIT,” Political Analysis 17 (2009): 45–63CrossRefGoogle Scholar; Kuklinski, James H., Cobb, Michael D., and Gilens, Martin, “Racial Attitudes and the ‘New South,’” Journal of Politics 59 (1997): 323–49CrossRefGoogle Scholar.

⁷ On “dual games,” see Cammett and Issar, “Bricks and Mortar Clientelism.”

⁸ The “missing majority” moniker refers to the fact that Michel Aoun's Free Patriotic Movement (FPM), a major player in the anti-Syria demonstrations, broke with the March 14 alliance to contest the 2005 elections. A senior FPM official argued that there was “no such thing as March 14” after the 2005 elections. Interview, July 2008. A member of Hizbullah's politburo bluntly stated that “March 14 doesn't represent the real majority in Lebanon anymore.” Craig S. Smith, “Lebanon's Future: Bending Toward Hezbollah or Leaning to the West?” New York Times, 22 September 2006.

⁹ According to Article 65 of the constitution, the government needs two-thirds support of the cabinet to decide on “national issues” such as constitutional amendments, the budget, and the electoral law.

¹⁰ Interviews with March 14, March 8, and independent politicians, April 2009. For more detail on the electoral campaign and outcomes, see Corstange, Daniel, “The Parliamentary Election in Lebanon, June 2009,” Electoral Studies 29 (2010): 285–89CrossRefGoogle Scholar.

¹¹ The districts most cited as “hot” by members of the two alliances included Zahla, West Bekaa–Rashaya, al-Matn, Kisrawan, Batrun, and the Beirut first district. Interviews, April 2009.

¹² We can define the district's vote Gap between the district's last winner (LW) and first runner-up (FRU) as Gap = (LW – FRU)/(LW + FRU), then multiplying by 100 to express the vote difference between 0 and 100 percent. We can then try to explain district-to-district variation in the vote Gap according to the district Shiʿi and Christian demographic shares, along with a district Diversity index (briefly, the chances that any two people in the district are from different sects). Using the 26 electoral districts as the units of analysis, we estimate via linear regression that the vote Gap = 49_0.00 + 0.61_0.00 (Shiʿi) – 0.25_0.02 (Christian) – 0.33_0.02 (Diversity), with subscripted p values (conventionally, those less than 0.10 mean statistically detectable effects). Substantively, the estimates confirm that heavily Shiʿi districts are less competitive (i.e., have a larger vote gap), while districts that are heavily Christian or multisectarian are more competitive (i.e., have a smaller vote gap). Adding one percentage point to the Shiʿi share of the district population increases the vote gap by nearly two-thirds of a percentage point. Conversely, the same size increase in the Christian share or the amount of diversity closes the gap by a quarter and a third of a point, respectively. Demographic data come from the 2009 voter rolls as reported on http://elections.naharnet.com/locations (accessed 4 June 2009). Election returns data come from official results released by the Ministry of the Interior. All data available upon request.

¹³ March 14 chose not to run lists in Zahrani, Sur, Bint Jbeil, al-Nabatiyya, Marjaʿyun–Hasbiya, and Baalbek–al-Hirmil. Independents formed lists in the last two districts with no hopes of winning (including the Asaʿad list in Marjaʿyun). List composition taken from “Candidates’ lists according to electoral districts,” NOW Lebanon, 1 June 2009.

¹⁴ Again using the 26 electoral districts as the units of analysis, we estimate that Turnout = 55_0.00 + 0.12_0.27 (Shiʿi) – 0.20_0.08 (Gap). Overall, this model fits the data poorly (F = 2.10, p = 0.15) and performs even worse if we adjust it for idiosyncratic districts such as Bsharri or the Beirut second district. Given how poorly this proposed model aligns with the actual outcomes, we can conclude that neither Shiʿi demographics nor district competitiveness can tell us much of anything about turnout. Note that Figure 1b and 1c reports simple scatterplots, and the slopes of their fitted lines come from bivariate regressions which are slightly different from the multivariate estimates reported in this note.

¹⁵ Among many others, see Binder, Leonard, ed., Politics in Lebanon (New York: John Wiley & Sons, 1966)Google Scholar; Cammett, “Partisan Activism”; Cammett and Issar, “Bricks and Mortar Clientelism”; Hamzeh, A. Nizar, “Clientelism, Lebanon: Roots and Trends,” Middle Eastern Studies 37 (2001): 167–78CrossRefGoogle Scholar; Johnson, Michael, Class & Client in Beirut (London: Ithaca Press, 1986)Google Scholar; and Khalaf, Samir, “Changing Forms of Political Patronage in Lebanon,” in Patrons and Clients in Mediterranean Societies, ed. Gellner, Ernest and Waterbury, John (London: Gerald Duckworth and Co., 1977), 185–205Google Scholar.

¹⁶ Rewards are targetable and excludable when parties and patrons have discretion over which voters receive them and which do not; civil service jobs are classic examples. In contrast, a school for a village might be broadly targetable insofar as it is built in Village A rather than Village B, but it is not excludable because the patron cannot prevent nonsupporters in the village from enjoying the benefits of the school.

¹⁷ See Blaydes, Elections and Distributive Politics; Cammett, “Partisan Activism”; Jamal, Barriers to Democracy; Johnson, Class and Client; Kitschelt, Herbert and Wilkinson, Steven I., eds., Patrons, Clients, and Policies (New York: Cambridge University Press, 2007)CrossRefGoogle Scholar; and Schaffer, Frederic Charles, ed., Elections for Sale: The Causes and Consequences of Vote Buying (Boulder, Colo.: Lynne Rienner, 2007)Google Scholar.

¹⁸ Contingency distinguishes clientelistic rewards from pork-barrel projects, which are localized public goods, the benefits of which accrue to all residents regardless of their support. There is, however, some conceptual ambiguity for collective payoffs in discrete localities that patrons can hold collectively responsible for aggregate vote totals. Compare what Hicken calls “collective clientelism” and what Cammett and Issar call “bricks and mortar clientelism.” Hicken, Allen D., “Clientelism,” Annual Review of Political Science 14 (2011): 289–310CrossRefGoogle Scholar.

¹⁹ The quotes are from Bratton, Michael, “Vote Buying and Violence in Nigerian Election Campaigns,” Electoral Studies 27 (2008): 627CrossRefGoogle Scholar; Stokes, Susan C., “Political Clientelism,” in The Oxford Handbook of Comparative Politics, ed. Boix, Carles and Stokes, Susan C. (New York: Oxford University Press, 2007), 606Google Scholar; Schaffer, Elections for Sale, 5–6; and a member of the Boutros Commission for Electoral Reform (interview, Beirut, June 2008). See also Brusco, Valeria, Nazareno, Marcelo, and Stokes, Susan C., “Vote Buying in Argentina,” Latin American Research Review 39 (2004): 66–88CrossRefGoogle Scholar; and Hicken, “Clientelism.”

²⁰ For prewar accounts of vote trafficking, see, for example, Harik, “Voter Participation”; Hudson, Michael C., The Precarious Republic (New York: Random House, 1968)Google Scholar, chap. 6; and Landau, Jacob M., “Elections in Lebanon,” Western Political Quarterly 14 (1961): 120–47CrossRefGoogle Scholar. By most accounts, the phenomenon has increased substantially in scope since the end of civil war, financed by large amounts of money flowing from wealthy candidates and foreign governments. Interview, think-tank director, Beirut, June 2008. See also el Khazen, Farid, Intikhabat Lubnan ma baʿd al-Harb 1992, 1996, 2000: Dimuqratiyya bi-la Khiyar (Beirut: Dar al-Nahar, 2000), 40–41, 202Google Scholar.

²¹ See Bashir, Iskandar, al-Taʾifiyya fi Lubnan ila Mata? (Beirut: University Institute for Studies, 2006)Google Scholar; Binder, Politics in Lebanon; Hudson, The Precarious Republic; el Khazen, Intikhabat Lubnan ma baʿd al-Harb; and Johnson, Class & Client in Beirut.

²² See el Khazen, Intikhabat Lubnan ma baʿd al-Harb; and idem, al-Ahzab al-Siyasiyya fi Lubnan: Hudud al-Dimuqratiyya fi al-Tajriba al-Hizbiyya (Beirut: Lebanese Center for Policy Studies, 2002). The “Hobbesian” quip come from Khalaf, “Changing Forms,” 198, while the “musical chairs” observation comes from Hudson, The Precarious Republic, 148.

²³ See Hudson, The Precarious Republic, 232; el Khazen, Intikhabat Lubnan ma baʿd al-Harb, 126; and Landau, “Elections in Lebanon,” 136.

²⁴ Interview, member of parliament, July 2008.

²⁵ Interview, member of parliament, April 2009.

²⁶ Dhahir, Nasim, ʿAn al-Ahzab wa-l-Dawla fi Lubnan (Beirut: Dar al-Nahar, 2008), 80Google Scholar. Interviews, think-tank director, publisher, and former and current Free Patriotic Movement officials, June and July 2008. See also el Khazen, al-Ahzab al-Siyasiyya fi Lubnan.

²⁷ First-generation studies of clientelism highlighted normative, personalized bonds between patrons and clients; subsequent studies have placed more emphasis on the transactional nature of the exchanges. See Binder, Politics in Lebanon; Hamzeh, “Clientelism, Lebanon”; Hudson, The Precarious Republic; Johnson, Class & Client in Beirut; Khalaf, “Changing Forms”; Kitschelt and Wilkinson, Patrons, Clients, and Policies; and Stokes, “Political Clientelism.”

²⁸ Cammett, “Partisan Activism,” provides evidence that parties direct more (and potentially more lucrative) benefits to party activists than to passive supporters. While the parties may try to service these “core supporters” in between elections, they have strong incentives to focus on “swing voters” during the election campaign itself.

²⁹ See Cox, Gary W. and Kousser, J. Morgan, “Turnout and Rural Corruption: New York as a Test Case,” American Journal of Political Science 25 (1981): 646–63CrossRefGoogle Scholar; Nichter, Simeon, “Vote Buying or Turnout Buying? Machine Politics and the Secret Ballot,” American Political Science Review 102 (2008): 19–31CrossRefGoogle Scholar; and Stokes, Susan C., “Perverse Accountability: A Formal Model of Machine Politics with Evidence from Argentina,” American Political Science Review 99 (2005): 315–25CrossRefGoogle Scholar.

³⁰ Interview, senior Kataʾib Party (March 14) campaign strategist, April 2009.

³¹ LADE officials also noted that both campaigns spent millions of dollars on the seat but that the Aounist campaign outspent Gemayel. Interviews, July 2008. Less charitably, one of Gemayel's confederates noted that “our ally Amin is a bit of a miser. He could have won if he spent more on vote buying or on facilities.” Interview, senior March 14 politician, June 2008.

³² These are all acceptable titles according to a head polling officer. Participant observation, election day 2009.

³³ In 2009, a total of six polling officers worked in each polling station (each of which had one ballot box). Two of the officers were appointed by the Ministry of the Interior. The remaining four were selected at the opening of the polls from among the voters present at the start of the day, who were, unsurprisingly, mostly party representatives. Two were selected by the the ministry's designated officers, and the other two were selected by the voters themselves.

³⁴ Information in this paragraph derives from interviews with LADE and foreign experts, April and June 2009, and confirmed by participant observation on election day.

³⁵ The Boutros Commission had, in fact, recommended that ballot counting be conducted centrally to disrupt the machines’ ability to match individual ballots and voters. Voters, meanwhile, tended to prefer that counting be done locally, ostensibly because it increased the transparency of the process. Although true as far as it goes, one plausible interpretation is that many voters want that transparency precisely so that they can credibly sell their votes for a profit. Interviews, senior LADE official, March 14 and March 8 politicians, election observers, journalists, July 2008, April and June 2009.

³⁶ Interviews, Boutros Commission member and senior LADE officials, July 2008 and June 2009.

³⁷ Interview, Beirut, July 2008.

³⁸ Interviews, LADE officials and nongovernmental organization (NGO) official, July 2008 and June 2009. See also Cammett and Issar, “Bricks and Mortar Clientelism”; and Gaspard, Toufic K., A Political Economy of Lebanon, 1948–2002 (Boston: E. J. Brill, 2004)Google Scholar.

³⁹ Interviews, foreign technocrats, April 2009; interviews, senior LADE officials, July 2008.

⁴⁰ Harik, “Voter Participation,” 30.

⁴¹ Interviews, July 2008 and June 2009. “Baroud: Electoral Money a Disease with More Talk than Evidence,” al-Hayat, 2 June 2010. Note that the interior minister, Ziad Baroud, was formerly the head of LADE.

⁴² Bratton, “Vote Buying and Violence”; Brusco et al., “Vote Buying in Argentina.”

⁴³ See Corstange, “Sensitive Questions, Truthful Answers”; Gonzalez-Ocantos, Ezequiel, de Jonge, Chad Kiewiet, Meléndez, Carlos, Osorio, Javier, and Nickerson, David W., “Vote Buying and Social Desirability Bias: Experimental Evidence from Nicaragua,” American Journal of Political Science 56 (2012): 202–17CrossRefGoogle Scholar; Kuklinski et al., “Racial Attitudes”; and Streb, Matthew J., Burrell, Barbara, Frederick, Brian, and Genovese, Michael A., “Social Desirability Effects and Support for a Female American President,” Public Opinion Quarterly 72 (2008): 76–89CrossRefGoogle Scholar.

⁴⁴ Note that I also ask members of the direct group the yes/no question about vote selling. Although not necessary to estimate the prevalence of vote selling in the list group, this answer allows us to compare what people say when asked directly to what they say when offered the anonymity of the list.

⁴⁵ Information International conducted the interviews after sampling residents of the main town and two randomly selected villages in each district. Sunnis were always interviewed by Sunnis, Shiʿa by Shiʿa, and Druze by Druze. Christians were always interviewed by Christian surveyors, and Armenians interviewed Armenians.

⁴⁶ Although not strictly necessary, I conducted a randomization check to confirm that assignment to the direct and list groups was indeed random; as expected, assignment did not vary systematically with any of a large set of demographic covariates.

⁴⁷ el Khazen, Intikhabat Lubnan ma baʿd al-Harb, 130, observes that “the most basic and prominent reasons [for voting] are the personal services and benefits provided by candidates in exchange for loyalty. In the Lebanese dictionary, this is known as ‘services’ [al-khidamāt].” Numerous elite interviewees identified al-khidamāt al-shakhṣiyya as one of the most common terms used to describe the payoffs.

⁴⁸ One might speculate that “personal services” reveal a candidate's commitment to the citizen's constituency and that commitment sways the voter rather than the specific reward. The payoffs being offered, however, are private goods to individuals rather than the collective goods such as “bricks-and-mortar clientelism” that might convince voters of the candidate's commitment to their constituencies (Cammett and Issar, “Bricks and Mortar Clientelism,” 389). Alternately, voters might view “personal services” as an indicator of candidate quality. Given the obvious lack of programmatic appeal, however, such payoffs would imply only that the candidate would be a competent future patron or source of wāsiṭa. However, interviews with candidates, academics, and various NGO officials emphasize much more straightforward accounts of the payoff itself swaying voters; no interviewees mentioned payoffs of this nature as indicators of competence or commitment.

⁴⁹ See Bratton, “Vote Buying and Violence”; Brusco et al., “Vote Buying in Argentina”; and Schaffer, Elections for Sale.

⁵⁰ More formally, the sample means, standard deviations, and sample sizes are 1.29, 1.11, and 1127 for the control group and 1.84, 0.95, and 1049 for the treatment group, respectively. A t test estimates a difference in means of 0.55 ± 0.09 at the 95 percent confidence level. Note that, given the intentional way the question was worded, the 55 percent figure is inclusive of switch buying, turnout buying, and abstention buying.

⁵¹ Because these data come from a random sample of the Lebanese population rather than a census of the full population itself, we need to account for the natural, random “noise” that represents the luck of the draw and makes repeatedly drawn samples look slightly different from each other. The estimate in the dot indicates our “best guess” of what the value in the full Lebanese population actually is. If we were to draw 100 random samples of that population, 95 of them would produce estimates falling within the 95 percent confidence interval, and 90 would fall within the 90 percent interval.

⁵² I leave the Druze out of the main comparisons because their small subsample size (142 respondents) means that there is too little information in the data about this community to learn much from the difference-in-means estimator. More involved multivariate analyses that also take into account respondents’ socioeconomic status and basic demographics reveal that Druze, Christian, and Sunni voters are virtually indistinguishable from one another, while Shiʿi voters are detectably more likely than their peers to sell their votes. Results available upon request.

⁵³ Although the list-group differences between March 14 and either March 8 or the unaligned (10 and 16 percentage points, respectively) are sizable in magnitude, there is not enough information in the sample to detect these differences reliably. The former is statistically significant at the 69 percent level and the latter at the 75 percent level.

⁵⁴ Daniel Corstange, “Ethnicity on the Sleeve and Class in the Heart,” British Journal of Political Science (forthcoming); Gaspard, A Political Economy of Lebanon; Hudson, The Precarious Republic.

⁵⁵ The median family income in each of the communities was $501 to $1,000 a month; Mann–Whitney tests (roughly, for differences in medians) indicate that Shiʿi respondents are, nonetheless, somewhat less wealthy than their Sunni or Christian counterparts. The median respondent in each community completed secondary school; a Mann–Whitney test detects no difference in education between Shiʿa and Sunnis, although a t test (assuming interval data) finds the Shiʿa to be marginally better educated.

⁵⁶ See n. 52.

⁵⁷ Shiʿi respondents expressed detectably less confidence than Sunnis and Christians in the government, the courts, and civil society associations. Although there were virtually no differences between the communities in their levels of interpersonal trust, Shiʿi respondents were noticeably more likely to place greater trust in members of their sect than in other Lebanese. Note, however, that Shiʿi respondents were always interviewed by Shiʿi interviewers.

⁵⁸ Note that the speaker directed these remarks against Lebanese politicians in general, not just against those in the Shiʿi community. NGOs voiced particular concern about intimidation stemming from Hizbullah's weapons, however. One observer remarked that “we as monitors had some problems . . . where Hizbullah wouldn't let us in. If you don't have weapons I can photograph you, but if you have a weapon I'll think twice about taking that photograph.” Interview, LADE official, July 2008.

⁵⁹ Interviews, March 14 activist, senior March 14 leader, March 14 leaning parliamentary candidate, Beirut, July 2008 and April 2009.

⁶⁰ According to the 2009 voter rolls, 47.6 percent of Shiʿi voters live in the districts of Bint Jbeil, al-Nabatiyya, Sur, and Zahrani, where neither March 14 nor independent figures bothered to field a slate of candidates against the Amal–Hizbullah lists. Another 29.9 percent live in the districts of Baalbek–al-Hirmil and Marjaʿyun–Hasbiya, where March 14 ran no candidates but where independents fielded lists with no hopes of winning. Combined, 75.5 percent of the Shiʿa live in districts with virtually no competition.

⁶¹ Interview, former minister, July 2008. The measure to which he was referring is the Polity index: http://www.systemicpeace.org/polity/polity4.htm (accessed 22 February 2012).

⁶² Interview, Beirut, July 2008. March 14 elites commonly cited Iranian funding for March 8, and their March 8 counterparts likewise cited Saudi funding for March 14. LADE officials, in turn, acknowledged the substantial amount of foreign funds flowing into the country for the campaign. Interviews, Beirut, April and June 2009.

⁶³ My seatmate on the flight into Beirut a few days before the election was such a voter. He saw a photograph of one of the alliance leaders in the newspaper I was reading, leaned over, and said, “That's my guy.” He happily admitted that he had no idea what issues were at stake but that he was using the election as an opportunity to “go lay on the beach for a week.”

⁶⁴ Each alliance claimed that the other was flying in voters to the “hot” districts, where each vote was extremely valuable. Such voters were allegedly given a free plane ticket and $1,000 (some claimed up to $5,000) to spend during a week in Lebanon. Interviews, March 14 and 8 officials, April 2009. Although the magnitude of this phenomenon is difficult to assess, the number of expatriates flown in was likely small relative to the size of the overall electorate (given the expense) but nontrivial in the hot districts into which they were flown and where the margin of victory was often on the order of a few hundred votes.

⁶⁵ Trafficking in votes—and regretting it—are both time-honored traditions in Lebanon. Citing interviews with prominent politicians in the 1960s, for example, Hudson, The Precarious Republic, 252, observes that several such figures “admitted quite freely to allocating certain amounts for vote buying. They regretted the necessity but argued that it was standard practice in their districts.” Part of the problem, they complained, was that the price of votes had been increasing rapidly—“but then so has the income of politicians,” as Hudson wryly concludes.

⁶⁶ See The Civil Campaign for Electoral Reform (al-Hamla al-Muduniyya li-l-Islah al-Intikhabi), a 2007 joint booklet of the Lebanese Transparency Association, the Lebanese Center for Policy Studies, and LADE.

⁶⁷ Compare the cautionary tale in Cox and Kousser, “Turnout and Rural Corruption,” on ballot reforms leading to vote suppression in upstate New York, where parties switched to buying abstentions.