¹ Michels, Robert, Political Parties: A Sociological Study of the Oligarchical Tendencies of Modern Democracy (London: Jarold and Sons, 1915)Google Scholar

² Field, Bonnie N. and Siavelis, Peter M., ‘Candidate Selection Procedures in Transitional Polities’, Party Politics, 14 (2008), 620–639CrossRefGoogle Scholar

³ Carey, John M. and Polga-Hecimovich, John, ‘Primary Elections and Candidate Strength in Latin America’, Journal of Politics, 68 (2006), 530–543CrossRefGoogle Scholar

Öhman, Magnus, ‘The Heart and Soul of the Party: Candidate Selection in Ghana and Africa’ (doctoral thesis, Uppsala University, Sweden, 2004)Google Scholar

⁴ Carey and Polga-Hecimovich, ‘Primary Elections and Candidate Strength in Latin America’.

⁵ Poiré, Alejandro, ‘Biased Ambitions: Party Nominations, Discipline, and Defection. Mexico's PRI in Comparative Perspective’ (doctoral thesis, Department of Government, Harvard University, 2002)Google Scholar

Kemahlioglu, OzgeWeitz-Shapiro, Rebecca and Hirano, Shigeo, ‘Why Primaries in Latin American Presidential Elections?’ Journal of Politics, 71 (2009), 339–352CrossRefGoogle Scholar

⁶ Serra, Gilles, ‘Why Primaries? The Party's Tradeoff between Policy and Valence’, Journal of Theoretical Politics, 23 (2011), 21–51CrossRefGoogle Scholar

Adams, James and III, Samuel Merrill, ‘Candidate and Party Strategies in Two-Stage Elections Beginning with a Primary’, American Journal of Political Science, 52 (2008), 344–359CrossRefGoogle Scholar

Jackson, Matthew O.Mathevet, Laurent and Mattes, Kyle, ‘Nomination Processes and Policy Outcomes’, Quarterly Journal of Political Science, 2 (2007), 67–92CrossRefGoogle Scholar

Oak, Mandar P., ‘On the Role of the Primary System in Candidate Selection’, Economics and Politics, 18 (2006), 169–190CrossRefGoogle Scholar

⁷ Kemahlioglu et al., ‘Why Primaries in Latin American Presidential Elections?’

⁸ In Botswana, for example, all party members of the ruling Botswana Democratic Party were allowed to vote in primaries before the 2004 elections. Elsewhere, leaders have veto power over the results, as in Cameroon's ruling party, or a committee selects nominees without local participation, as in Mozambique's ruling party. The system used for Senate nominations in Nigeria by the ruling People's Democratic Party, with elections among local delegates coupled with elite-level interference on a case-by-case basis, seems to most closely mirror that of Ghana (Öhman, ‘The Heart and Soul of the Party’).

⁹ Key, V. O., Politics, Parties, and Pressure Groups (New York: Thomas Y. Crowell, 1942)Google Scholar

¹⁰ Miguel De, LucaJones, Mark and Tula, Maria, ‘Back Rooms or Ballot Boxes? Candidate Nomination in Argentina’, Comparative Political Studies, 35 (2002), 413–436Google Scholar

¹¹ Langston, Joy, ‘The Changing Party of the Institutional Revolution’, Party Politics, 12 (2006), 395–413CrossRefGoogle Scholar

¹² Rahat, Gideon and Hazan, Reuven Y., ‘Candidate Selection Methods: An Analytical Framework’, Party Politics, 7 (2001), 297–322CrossRefGoogle Scholar

¹³ Field and Siavelis ‘Candidate Selection Procedures in Transitional Polities’.

¹⁴ Lundell, Krister, ‘Determinants of Candidate Selection: The Degree of Centralization in Comparative Perspective’, Party Politics, 10 (2004), 25–47CrossRefGoogle Scholar

¹⁵ Meinke, Scott R.Staton, Jeffrey K. and Wuhs, Steven T., ‘State Delegate Selection Rules for Presidential Nominations, 1972–2000,’ Journal of Politics, 68 (2006), 180–193CrossRefGoogle Scholar

¹⁶ Serra, ‘Why Primaries?’, p. 23Google Scholar

¹⁷ Meinke et al., ‘State Delegate Selection Rules for Presidential Nominations, 1972–2000’.

¹⁸ James Adams and Samuel Merrill III, ‘Candidate and Party Strategies in Two-Stage Elections Beginning with a Primary’.

¹⁹ Poiré, ‘Biased Ambitions’.

²⁰ Kemahlioglu et al., ‘Why Primaries in Latin American Presidential Elections?’ p. 343. The analysis summarized in their Figure 1, p. 348, indicates that the likelihood of a primary increases in the previous vote share of a party or coalition and this relationship appears strongest for non-coalition centrist parties. This finding appears to depend on a very small number of cases, however, as only 3 per cent of all non-coalition parties in their dataset had primaries.

²¹ Fish, M. Steven and Kroenig, Matthew, The Handbook of National Legislatures: A Global Survey (New York: Cambridge University Press, 2009)CrossRefGoogle Scholar

Barkan, Joel D., ‘African Legislatures and “The Third Wave” of Democratization’, in Joel D. Barkan, ed., Legislative Power in Emerging African Democracies (Boulder, Colo.: Lynne Rienner, 2009), pp. 1–32CrossRefGoogle Scholar

²² Carey and Polga-Hecimovich, ‘Primary Elections and Candidate Strength in Latin America’; Kemahlioglu et al., ‘Why Primaries in Latin American Presidential Elections?’

²³ Lindberg, Staffan I., ‘What Accountability Pressures Do MPs in Africa Face and How Do They Respond? Evidence from Ghana’, Journal of Modern African Studies, 48 (2010), 117–142CrossRefGoogle Scholar

²⁴ A senior official in the ruling New Patriotic Party of Ghana argued that just because someone can use money to influence the primary does not mean they have the resources to do the same in the general election, when the electorate is larger by two orders of magnitude. ‘It has become a very sore problem for us.’ (Author's interview with former NPP Communications Director, Accra, 28 January 2010.)

²⁵ Calvo, Ernesto and Murillo, M. Victoria, ‘Who Delivers? Partisan Clients in the Argentine Electoral Market’, American Journal of Political Science, 48 (2004), 742–757CrossRefGoogle Scholar

²⁶ Wantchekon, Leonard, ‘Clientelism and Voting Behavior: Evidence from a Field Experiment in Benin’, World Politics, 55 (2003), 399–422CrossRefGoogle Scholar

²⁷ McDonald, Michael and Samples, Johneds., The Marketplace of Democracy (Washington, D.C.: Brookings Press, 2006)Google Scholar

²⁸ An even better option for the party leader would be for the party leader's favoured aspirant to make the payments to the other aspirants in exchange for their withdrawal. But inducing the favoured aspirant to do so would have its own costs to the party leader and would appear the same to local party members, so we do not consider it as a separate option here.

²⁹ Kemahlioglu et al., ‘Why Primaries in Latin American Presidential Elections?’

³⁰ Poiré, ‘Biased Ambitions’; Langston, ‘The Changing Party of the Institutional Revolution’; De Luca et al., ‘Back Rooms or Ballot Boxes?’

³¹ The number of constituencies was increased from 200 to 230 just prior to the 2004 elections.

³² Constituency development funds like those in Ghana exist in Kenya, India, the Philippines, Uganda, Malawi and Zambia, among others. See Philip Keefer and Stuti Khemani, ‘When Do Legislators Pass on “Pork”? The Determinants of Legislator Utilization of a Constituency Development Fund in India’ (The World Bank, Working Paper, 2009).

³³ Lindberg, Staffan I., ‘It's Our Time to “Chop”: Do Elections in Africa Feed Neo-Patrimonialism rather than Counteract It?’, Democratization, 10 (2003)CrossRefGoogle Scholar

³⁴ Vieta, Kojo T., Know Your Ministers (Accra: Flagbearers, 2005)Google Scholar

³⁵ Nicolas van de, Walle, ‘Presidentialism and Clientelism in Africa's Emerging Party Systems’, Journal of Modern African Studies, 41 (2003), 297–321CrossRefGoogle Scholar

³⁶ Barkan, Joeled., Legislative Power in Emerging African Democracies (Boulder, Colo.: Lynne Rienner, 2009)CrossRefGoogle Scholar

³⁷ For an example, see Halifax Ansah-Addo, ‘NPP to Expel Appiah-Ofori’, Daily Guide, 8 July 2009.

³⁸ For example, see ‘Voting progresses at NPP Conference’, Ghana News Agency, 17 December 2005, http://www.ghanaweb.com/GhanaHomePage/politics/artikel.php?ID=96234

³⁹ Party supporters can become ‘card-carrying’ members by applying at a local branch office of the party and paying their annual dues, which are about US$1 (author's interview with NPP Regional Secretary, Brong Ahafo Region, Sunyani, 6 October 2009).

⁴⁰ Article 11 of the Constitution of the New Patriotic Party, Accra, Ghana (1998).

⁴¹ Öhman, ‘The Heart and Soul of the Party’.

⁴² The former General Secretary (1998–2005) and current Vice-Chairman of the NDC said of his party's approach to candidate selection: ‘In 2000 we decided to go a bit more bureaucratic … and we paid for it because we lost a number of seats … So in 2004 we didn't make the mistake. We opened the flood gates so that everybody’ could run (Author's interview with former NDC General Secretary (1998–2005) and current Vice-Chairman, Accra, 6 May 2010. Also author's interview with NDC General Secretary, Accra, 5 March 2010). Although this decision suggests the adoption of primaries by the weaker party, in line with Serra, this view is not supported by our data analysis (Serra, ‘Why Primaries?’). Though the NDC had more primaries in 2004 than in 2000, the party leadership continued to prevent some primaries in both 2004 and 2008. The NPP was still more likely to hold primaries than the NDC in both 2004 and 2008.

⁴³ Article 11 of the Constitution of the New Patriotic Party, Accra, Ghana (1998); Article 42 of the Constitution of the National Democratic Congress, Accra, Ghana (2002).

⁴⁴ Generally, to be eligible for the primary in 2004 and 2008, aspirants had to be dues-paying members in the constituency, hold only Ghanaian citizenship, pay a filing fee, reported to be roughly US$350 in 2008, and also demonstrate support from party members in the constituency.

⁴⁵ Lindberg, ‘What Accountability Pressures Do MPs in Africa Face and How Do They Respond?’, p. 125Google Scholar

⁴⁶ Author interview with NDC General Secretary, Accra, 5 March 2010; Author interview with NPP Chairman (2005–10), Accra, 26 April 2010.

⁴⁷ For example, see Collins Agyekum-Gyasi, ‘Baffour de Graft elected parliamentary candidate’, Daily Graphic, 29 January 2008.

⁴⁸ For example, see Simmons Yusuf Kewura, ‘Tears Flow at Minister's House’, Ghanaian Chronicle, 12 August 2004; Kwame Asare Boadu, ‘Boakye is NPP Candidate for Asunafo South’, Daily Graphic, 7 July 2007.

⁴⁹ Author interview with NPP Ayawaso East constituency nominee (2008 election), Accra, 12 May 2010.

⁵⁰ For a typical example, see Moses Dotsey Aklorbortu, ‘Shama NPP Supporters Protest Against MP’ Daily Graphic, 28 May 2008. Also see Victor Kwawukume, ‘NDC Group Threatens to Vote “Skirt and Blouse” ’, Daily Graphic, 22 April 2008.

⁵¹ For example, in the Effutu constituency in the NPP in 2004. See ‘Confusion Rocks Effutu NPP’, modernghana.com, 18 October 2004, http://www.modernghana.com/news/64919/1/confusion-rock-effutu-npp.html.

⁵² See Enoch Darfah Frimpong, ‘NPP Fomena Appeals to National Executive’, Daily Graphic, 14 March 2008.

⁵³ See Samuel Kyei-Boateng, ‘Protest Over NDC Plan for Wenchi East’, Daily Graphic, 25 September 2004.

⁵⁴ Our main sources were the AllAfrica.com database search engine, which contains articles from three Ghanaian daily newspapers (Ghanaian Chronicle, Accra Daily Mail, and Daily Searchlight) and microfilm copies of all issues of Ghana's ‘paper of record’, the Daily Graphic, from January 2003 through June 2008. Microfilm of the Daily Graphic from July 2008 to 2012 is not yet available, although most primaries for the 2008 elections had already been conducted by this point. Newspapers in Ghana often have political affiliations – the Chronicle devoted far more coverage to developments in the NPP, for example – but the Daily Graphic had more balanced coverage.

⁵⁵ Ansolabehere et al., ‘The Decline of Competition in US Primary Elections, 1908–2004’.

⁵⁶ Having a primary election is partly mechanically related to the number of aspirants who come forward to seek a party's nomination for parliament, since having only one aspirant always means no primary. Having more than one aspirant is necessary for, but does not mechanically imply a contested primary, since party leaders can still interfere to prevent primaries from taking place.

⁵⁷ Outcomes for both election years can be viewed as largely independent decisions, in that party-constituencies that held primaries in 2004 were not more likely to have primaries in 2008 than those that did not have primaries in 2004. In our data, 65 per cent of the constituencies that did not have primaries in 2004 had primaries in 2008, while 69 per cent of constituencies that did have primaries in 2004 had them again in 2008.

⁵⁸ For example, in Ellembelle constituency in the Western Region of Ghana, the NPP presidential candidate won 61 per cent of the vote in the 2004 election, but the NPP deferred to the CPP, a minor party, and did not field a parliamentary candidate. As the NPP prepared for the 2008 elections, a more realistic expectation of the potential support for a hypothetical future NPP parliamentary candidate, should the party have chosen to end the alliance, would have been something closer to the 61 per cent the party received in the 2004 presidential election than the 0 per cent it received by default in the 2004 parliamentary election.

⁵⁹ Posner, Daniel N., Institutions and Ethnic Politics in Africa (New York: Cambridge University Press, 2005)CrossRefGoogle Scholar

⁶⁰ We also estimate models including higher order terms and interaction terms between ethnic fractionalization and the previous vote share and find that these terms were not statistically significantly different from zero (not shown), so they are excluded from our favoured specification. We also estimate a generalized additive model, allowing for an interaction between ethnic fractionalization and the previous vote share. With the vis.gam function in the mgcv package in R, we visually confirm that the parametric form of the equation displayed above, including the squared term on past vote share and the assumption of no interaction between ethnic fractionalization and past vote share, is reasonable (not shown).

⁶¹ Results using the parliamentary vote share rather than the presidential vote share are very similar (not shown).

⁶² Brambor, ThomasClark, William R. and Golder, Matt, ‘Understanding Interaction Models’, Political Analysis, 14 (2006), 63–82CrossRefGoogle Scholar

⁶³ Serra, ‘Why Primaries?’; Adams and Merrill, ‘Candidate and Party Strategies in Two-Stage Elections Beginning with a Primary’.

⁶⁴ To check, we have estimated this model with the quadratic term (not shown) and find that the probability of a primary election predicted by the models with and without the squared term are quite similar, is generally increasing with previous vote share, and always falls within the 95% confidence interval of the other model. The percentage of observations correctly predicted also hardly differs across the models. Since a simpler model is generally preferred and our theory for the restricted sample is that the probability of a primary would increase in the previous vote share, we prefer to exclude the squared term.

⁶⁵ We created an index for the development level of a constituency created by conducting a factor analysis with a principal-component factor model of three extremely highly correlated measures of living conditions recorded in the 2000 Census data – percentage of households using electricity, percentage of households using modern sanitation facilities (toilet), and percentage of households with running water (either pipe-borne or from a tanker).

⁶⁶ The proportions are statistically indistinguishable at higher numbers of aspirants because they are very close to 1 and the sample sizes become much smaller. We ignore the difference in proportions between observations with only one aspirant and those with additional aspirants because it is mechanically not possible to have a contested primary election with only one aspirant.

⁶⁷ For example, Agyekum-Gyasi, ‘Baffuor deGraft Elected Parliamentary Candidate’; Magdalene Sey, ‘NPP MP aspirant goes independent’, Ghanaian Chronicle, 6 August 2008.

⁶⁸ For example, see Simmons Yusuf Kewura, ‘Tears flow at Minister's house’, Ghanaian Chronicle, 12 August 2004; Kwame Asare Boadu, ‘Boakye is NPP candidate for Asunafo South’, Daily Graphic, 7 July 2007.

⁶⁹ Author interview with former NDC General Secretary (1998–2005) and current Vice-Chairman, Accra, 6 May 2010.

⁷⁰ Serra, ‘Why Primaries?’; Adams and Merrill, ‘Candidate and Party Strategies in Two-Stage Elections Beginning with a Primary’.

⁷¹ Rather than quality in campaign skills, the party leader may be interested in education, experience, or other qualities that would indicate competence in the legislature. But we find that incumbents with less education or without ministerial experience are no less likely to seek re-election than their more educated or more experienced counterparts. Including these variables together in the model does not affect any of our estimates (not shown).

⁷² Montalvo, José and Reynal-Querol, Marta, ‘Ethnic Polarization, Potential Conflict, and Civil Wars’, American Economic Review, 95 (2005), 796–816CrossRefGoogle Scholar

⁷³ Serra, ‘Why Primaries?’; Adams and Merrill, ‘Candidate and Party Strategies in Two-Stage Elections Beginning with a Primary’; Carey and Polga-Hecimovich, ‘Primary Elections and Candidate Strength in Latin America’.

⁷⁴ Levitsky, Steven and Helmke, Gretchen, ‘Informal Institutions and Comparative Politics: A Research Agenda’, Perspectives on Politics, 2 (2004), 725–740Google Scholar

⁷⁵ Aldrich, John H., Why Parties?: The Origin and transformation of Political Parties in America (Chicago: The University of Chicago Press, 1995)CrossRefGoogle Scholar

Przeworski, Adam and Sprague, John, Paper Stones: A History of Electoral Socialism (Chicago: University of Chicago Press, 1986)Google Scholar

Kalyvas, Stathis, The Rise of Christian Democracy in Europe (Ithaca, N.Y.: Cornell University Press, 1996)CrossRefGoogle Scholar