Political Brand Association: the Tale of Two Political Party Brands in Twenty Years

Title: Political Brand Association: The Tale of Two Political Party Brands in Twenty Years.

Dr. Kobby Mensah1 University of Ghana Business School Department of Marketing and Customer Management P. O. Box LG78, Legon, Accra Ghana Email: [email protected]

Draft Paper. Not for citation © March, 2015

1 My areas of expertise include Political Marketing, Tourism Marketing and Financial Services Marketing.

I use this opportunity to offer my sincere appreciation to Dr. Lord Mensah and Dr Eric Ofosu-Hene of the Department of finance, UGBS; my Graduate and National Service assistants, Edem Amenuvor and Deborah Narh, and all the MPhil students, especially Alex Anlesinya and Bukari Zakari for the various contributions in getting this research completed successfully.

Abstract

The aim of this study is to establish whether changes in party

leadership could result in changes in party behaviour and voter choice

in Ghana. The study also offers insights into who the likely voters are

for the two dominant parties in Ghana, the National Democratic

Congress (NDC) and the New Patriotic Party (NPP), how their voting

patterns have evolved, and the causes of this evolution. In order words,

the study tries to identify the age, gender, and class distribution of

voters for the two parties. It argues that changes in party leadership

over the past 20 years have influenced voter perceptions on party

brands and party choice over the years.

1. Introduction Since 1992 when Ghana was returned to democracy, academics, political commentators, party activists, and indeed the electorates, have held the view that the two dominant parties, National Democratic Congress (NDC) and New Patriotic Party (NPP), appeal to two sections of the society, broadly speaking. The NPP of the right is known to be predominantly upper class, and the left of centre NDC party is predominantly lower class.

For twenty years of multi-party democracy these perceptions on party-voter class associations have held same amongst academics, practitioners, experts in media and politics, despite changes in both parties’ leadership and management styles. Furthermore, the two parties are known to have entrenched regional and ethnocentric appeals, which will be discussed further in the paper. Most electioneering campaign studies in Ghana (Mensah, 2009; Ninsin, 2006; Youde, 2005; Gyimah-Boadi, 2001; Nugent, 2001; Ayee, 2000; 1997; Anebo, 1997) affirm this belief, until Tweneboah-Koduah et al (2010) study on political brand choice in Ghana which alludes to a gradual shift from the previously held class, regional and ethnocentric associations. However, the latter study is also limited in varied ways, and will be discussed. That notwithstanding, these studies have played pivotal role in unearthing the phenomenon of how political parties in Ghana seek to brand themselves, serving as foundation scholarship on which further studies could evolve.

Indeed, scholarship in Ghana could not be the only culprit to blame for failing to track the changes to political brand associations that leadership change brings. The political parties themselves could only hazard a guess about the profile of their core voters, and most undoubtedly, are unsure about the effect of leadership change on voter choice. The leadership of the NPP, for example, have on various platforms denied being a party of predominantly affluent and well educated class (Mensah, 2012). They argue that if anything at all the party’s origin from farming areas of the country such as Ashanti, Brong Ahafo, Western and northern Ghana should suffice in thinking that their political brand has a broader appeal across class. The party again argues that its main opposition, the NDC is equally elitist as it also appeals to those sections of society.

Perhaps the NPPs claim is premised on the assumption that the NDC, over the years, has changed in ‘temperament’ as a result of changes in its frontline political operatives and leadership. The NDC party today can boast of young politicians, political operatives and followers who could be considered elites. Just as the NPP, leadership of the NDC today are predominantly graduates from some of the country’s renowned universities; a variant of its past when it was perceived as a party belonging to ‘those of the streets’. Since 2000, the NDC has been led by a professor and a public relations expert, Atta Mills and John Mahama respectively, both graduates and of elite backgrounds. The scenarios above, on both parties, draw us to the study questions as follows:

 What is the age and gender distribution of votes for the two parties?  What is the perceived class distribution of votes for the two parties?  Would change in party leadership affect party behaviour and voter choice?  How have the voting patterns for these two parties changed, and what are the causes for the change?  Are NPP and NDC still perceived as predominantly upper and lower classes respectively?

Answers to these questions are cardinal to devising a compelling political marketing strategy effective in targeting the voter. This research therefore offers the base information on which these two parties could adequately profile the voter and target them effectively with campaign messages. Thus, the sections are divided into five key areas. The first is to elaborate on the emerging literature of political marketing in Ghana, focusing on political brand association, and how that sits within the global context of political brand association, party behaviour and voter choice, as the next section. The methodology through which the study was carried out will be accounted for in the third section, followed by findings and discussions in the fourth section. The concluding section will draw the curtain on the discussions.

1.1 Brand Association: definition, types and its application in political marketing The concept brand association has been defined and operationalized in many ways (Uggla, 2005; Schneider, 2004; Kapferer, 1997). From party management perspective, Kapferer (1997), for example notes that political parties seek to brand associate by personalizing the political brand through functional and emotional attributes in order that they can reinforce the brand in the minds of the consumer (Kapferer, 1997: 109). Kapferer's views are supported by Schneider (2004), who observes that brands that have the ability to generate recall are those that have

3 established knowledge structures of what they stand for in the minds of consumers over time. Hence, consumers who see a reflection of themselves in these brands would choose them during purchase. For example, the automobile company Volvo has over the years been able to link safety as benefit to its brand, and Mercedes-Benz for prestige. This means, not only do the marques of these automobile brands generate recall of the brand name, but also the benefits they offer the consumer. Therefore, a consumer for safety need would choose Volvo, and that for prestige would choose Mercedes-Benz.

In British politics, the Conservative and Labour parties have long been associated with fiscal discipline and social justice respectively, due to their long standing policy positions on Taxes, the NHS and other socio economic needs of the people. Hence a voter who predominantly seeks social justice through redistribution of wealth would choose the Labour party, and one whose self-concept aligns with fiscal discipline, individual responsibility would choose the Conservatives.

Uggla (2005) also talks of brand association as anything that symbolically represents the brand in the minds of the target customer. Thus defining the concept as ‘the link a brand establishes with its stakeholders through, for example, people, places, institutions that add to (or subtract from) customers knowledge of the brand’ (2005: 789). That means not only do institutions such as political parties use ideology to anchor their brand, but are also able to use "iconic" personalities and issue positions to do same; and such anchors can add or detract value through the action or inaction of the brand, and/or other players in the industry. For example, a consistent on-time delivery of consignments by a postal company (a functional attribute) will naturally enhance the impression held about the company as a dependable partner (emotional attribute), especially if the statement ‘a dependable partner’ happens to be the postal company’s slogan. On the other hand, a customer perspective association could be the emotive personal attribute of the company as a dependable partner that the customer beholds, is most likely to weigh on the costumer’s choice when faced with a similar purchase decision in future.

In developing brand association to manage brand/market relationships, Farquhar and Herr (1993: 265) note a two-dimensional framework of brand-to-associate and associate-to-brand. Within the wider debate on how the political brand is managed, brand-to-associate is market-driven whereas associate-to-brand is market-driving, to use Day’s (1994) market orientation theory. The benefit in operationalizing the two, according to Kapferer, is the opportunity not only to build a brand but also to stretch the brand—leveraging—to other terrains not original to it. In implementing the market-driven brand-to-associate strategy, an organization examines internally held values of the brand, its identity for example, to understand the brand’s basic uniqueness and who it targets (Kapferer, 1995: 30). This first step helps gain insights in product performance features that are suitable for the target market. It also assists in developing the brand’s marketing activities, such as communications to support the brand.

In the Volvo example cited earlier, the brand name is synonymous or associated with the word safety because the company inculcates the idea of safety, not only in their communication but also in product design, according to the managing director and the technical manager, Assar Gabrielsson and Gustav Larson (1927).

On the other hand, operationalizing market-driving associate-to-brand orientation means identifying market expectations (consumer values) and associating the brand’s values to match those expectations. It involves re-examination of the brand’s identity and keeping track of the brand’s image and reputation—opinions about the brand held by consumers. The process of associate-to-brand is crucial to renewing the brand’s relevance to the market, as it is in stretching the brand to other markets not original to the brand. The process involves asking two main questions, according to Kapferer (1997: 169). The first is how do we adapt to the changing conditions whilst keeping the brand’s identity? And the other is what should we adapt and what do we leave untouched? This exercise is thus important as the brand’s clientele and market expectations change overtime.

Associate-to-brand strategy is especially crucial for brand extension, when a decision is taken to advance a product into other markets. In the case of Volvo, for example, when the association ‘safety’ is mentioned, it should evoke brand Volvo. However, with people becoming ‘time poor’; having more to do within a short time, the brand needs to find out other values that are of concern to consumers aside ‘safety’. This may yield results such as less time spent in commuting, not necessarily safety, for automobile buyers, especially the younger generation. This means for Volvo to remain relevant and expand its market share, the brand needs to adapt the association ‘fast’, whilst keeping its original identity of ‘safety’, hence becoming the ‘safest fast’ car in an attempt to tap into the ‘yuppie’ (young urban professional) market of today.

The exposition of brand association, its forms and applicability in politics and in commerce, as noted above, suggest that not only could NDC and NPP use ideology, an inherent value of the political offering, to represent the brand in the minds of voters but also adapt values external to them as their political brand representation in the minds of voters. In this regard references could be made to some few studies in the political marketing literature (Mensah, 2009; Tweneboah- Koduah et al, 2010; Hinson and Tweneboah-Koduah, 2010) that have researched into the concept of political brand association in Ghana and others in political science and sociology that have looked at how parties in Ghana campaign and attract votes (Ninsin, 2006; Youde, 2005; Gyimah-Boadi, 2001; Nugent, 2001; Ayee, 2000; 1997; Anebo, 1997). Of these studies, Tweneboah-Koduah et al (2010) is the closest to the current study.

2. Literature on political brands and voter choice in Ghana In a study conducted in 2007 on voter choice in Ghana from political brand perspectives, Tweneboah-Koduah et al (2010) allude to a confused state of political brand association amongst voters. Their central objective was to understand whether political parties in Ghana were managed as brands, and if that impacts on voter choice. The qualitative study had a total of 82 focus groups discussions conducted in eight regions (42 constituencies out of a total of 230 at the time) in Ghana. The questions that the study operationalized, as noted in the responses, were predominantly party and policy based (Tweneboah-Koduah et al, 2010: 84-86). The focus group of 8-12 members in each group were registered voters with no indication of party membership or any strong attachment to a party; neither were respondents’ level of education important to the study.

The main objective was to gain deep insight of voter behaviour in political choice making amongst Ghanaian voting age population of 20-45. The study was to enable them understand

5 whether voters in Ghana could link certain constructs to the two political brands, NDC and NPP, and if such constructs could influence political choice (2010: 83). The study applied Kapferer’s brand identity prism (2004), measuring voters’ response against six theory frameworks as listed below:

 The Brand Physique  The Brand as a Personality  A Brand has Its Own Culture  A Brand is a Relationship  The Brand is Also a Customer Reflection  A Brand Speaks to the Target’s Self-Image

Per their findings, almost all the constructs, except brand as a personality, had less than half of the respondents identifying, positively or negatively, an association to the party brand. For example, under political party brand’s physique, which is to measure a respondent ability to recall any relevant functional or physical component of a brand when the name is mentioned, the study note that when respondents were asked which policies, programs…would they associate with NPP and NDC, less than half associated NPP with road construction, national health insurance, freedom of speech and human right. More than half however, could recall NPP’s election 2004 campaign slogan “4 more years… so far so good”, without prompting.

For the NDC, respondents could not associate the party with any specific projects and programs, according to the study, and went further to credit NDC initiated projects such as GETFUND to NPP, thus indicating the chaotic nature of the party's communication, the study suggests. On brand as a personality, the study reported more than half of respondents describing NPP’s personality in the negative as “betrayer, boastful...” with the positives as “democratic, peaceful…” The NDC associations were “violence...lack of internal democracy.” The NDC was also not associated with any philosophy, policies and programs.

There is however a significant revelation in Tweneboah-Koduah et al (2010) study that departs from previously held assumptions on voter-party brand associations in Ghana. On brand as customer reflection and voter self-image constructs, which seek to measure the likelihood of a voter choosing a political brand based on the nearness of their characteristics and values to that of the party brand, in the question: who was more likely to be a supporter/ member of NDC/NPP, the study reported that less than half of the focus group respondents claimed someone from Ashanti Region and, or, Eastern region for NPP, and someone from the Volta Region or parts of the North, for the NDC.

This is in contrast with previously held view of regionalisation and ethnocentric pattern of voter- party brand association. Findings in extant literature (Ninsin, 2006; Youde, 2005; Gyimah- Boadi, 2001; Nugent, 2001; Ayee, 2000; 1997; Anebo, 1997) suggest that respondents would overwhelmingly refer to regional and ethnocentric associations for partisan identification when asked similar questions. Again, in Tweneboah-Koduah et al (2010) study, the finding that less than half of the focus group respondents think the “masses” across ethnicity may vote NDC and the “elites” vote NPP is also startling, and contrary to previously held beliefs, as the reverse had been the case in Ghana, it is believed.

Tweneboah-Koduah and colleagues thus concluded that though there were evidence of branding, political parties are however “not properly managed as brands in order to invoke mental associations nor intense emotional effects with voters as power brands do with consumers” (2010: 87). The study findings suggest least existence of political brand association enough to offer the party brand equity, as there is in consumers’ minds about commercial brands. This is contrary to findings in extant political science literature that suggest strong partisan identification amongst Ghanaian electorates (Gyima-Boadi, 2001; Anebo, 2001; CDD, 1999).

Though important, as it indicates a gradual shift from previously held views, and could offer insight to political practitioners, Tweneboah-Koduah et al (2010) study is limited in some ways. It lacks contextual appreciation where the political system, media system and voter dynamics differ from country to country. For example, the difference in the conceptualisation of party- voter relations and product-consumer relations, where the level of information/knowledge structures possessed in the latter is higher than that of the former must be appreciated. In the case of the former, political maturity would determine a respondent’s ability to recognise certain associative constructs, such as a party’s ‘philosophy’ for example, as in the response that: NDC… was not associated with any philosophy, policies and programs.

In this regard, Anebo (2001: 85), in a study on “Ghana election 2000: voter choice and electoral decisions”, notes that low level of political knowledge of the majority of the Ghanaian electorate, coupled with the community based interest politics, and other primordial characteristics of the Ghanaian social structure, citizens have little incentive to evaluate parties and candidates on issues and competency. Secondly, the revelation that less than half of study respondents feel that voter-party brand association is less regional, ethnocentric and class oriented itself demonstrates that the approaches adopted by political parties to wean themselves from these associations in order to appeal to a broader target are working. Hence, the claim that political parties are not being effectively managed as brands is contrary to their own findings and existing knowledge.

To claim that there is least existence of political brand association constructs in the minds of the respondents just because they were unable to articulate detailed policy or party values, as Tweneboah-Koduah et al (2010) suggest, depart from political science theory, and also new studies in political marketing on candidate brands, where voter choice is underpinned by ideology and other factors such as candidate values; meaning the closer a voter’s belief or value is to a party’s ideological position or candidate characteristics, the likely the choice of that party in elections. This approach is market-driven, as parties brand-to-associate with voter needs and characteristics.

It is also noted in other studies (Mensah, 2009; Gyimah-Boadi, 2001; Nugent, 2001) that not only have parties been market-driven, by branding to associate with voters, but have been market driving where parties have associated with other constructs outside them in order to brand in Ghana. For example, Mensah notes of the 2008 elections that NPP, with the elephant as its emblem adopted also the “hopping kangaroo dance” of the national football team, the black stars and a popular gospel song, go high, as their campaign song in order to reinforce their message: “yee – ko ye – nim,” to wit, “we are moving forward.” The NDC with the umbrella as traditional party symbol also adopted the signs that football coaches and fans do at tournaments, “finger

7 wriggling” of both hands lifted upwards to indicate a desire to change a player. A move to reinforce their message of “ye – re se sa – mu,” meaning, “change”. These symbols as introduced in the 2008 elections, and those before them in previous elections, have high recognition amongst voters in Ghana, and are quickly recalled and associated with their respective parties that ‘owned’ them even years after their introduction.

Other studies have also shown that voters rely on cognitive shortcuts as a substitute for detailed information (Needham, 2005; Lupia and McCubbins, 1998; Neuman et al., 1992). They explain that when a political element—the party, candidate or policy—is heavily promoted and gains visibility unequal to the rest, and is used as a representation of the rest in the party political structure, it is most likely that voters will have a ‘remarkable’ knowledge of that element which is mostly promoted. This makes Tweneboah-Koduah et al dominant emphasis on inherent party and policy associations problematic for the respondents in that study. In this direction, Needham (2005) observes that aside ideology voters have impressionistic perceptions about candidates in electoral choice-making.

In Ghana, findings of electoral campaign studies (Mensah, 2009; Gyimah-Boadi, 2001; Ayee, 2000) show that the political candidate is highly promoted in recent Ghanaian elections than the party and the policy. Hence the difficulty in voter recollection of associations linked to the party and the policy brands beyond the broad generalisation of party name, slogans, emblems, colours and campaign themes, such as ‘the agenda for positive change’.

Based on the discussions above, could the results of Tweneboah-Koduah et al (2010) be different should they have broadened the theories of associations in the study? It suffice to say, therefore, that should we accept a lack of association, even as basic as ideological nearness based on which voters make political choice then we are admitting that voters in Ghana are becoming less ideological in their choices. Is it also the case that political parties are becoming so, or both? Can we say that voters know very little about political candidates, and where they belong given that our media-political landscape in the last 20 years have been predominantly presidential – personality driven?

These scenarios, as suggested by Tweneboah-Koduah and colleagues, could be disputed given that the two parties have consistently mastered on or around 40 per cent of the voter share each electoral cycle in the last 20 years. For example, the Centre for Democratic Governance (CDD) 1999 ‘Afrobarometer’ survey of the Ghanaian electorate suggests a substantial partisan identification in Ghana with about 66 per cent of respondents interviewed claiming to identify with a party strongly. In that survey 38 per cent of respondents interviewed identified with the NDC, a mirror image of what the party claims is the size of its membership relative to the total electorate (Ninsin, 2006: 18). On the other hand, 25 per cent identified with the NPP. Both parties boast of having that share of the electoral market as constant and only need to fight a little bit more to scale the 50+1 per cent mark to win elections. These claims have consistently been proven by the two parties’ electoral performance every electoral cycle since 1992. Perhaps it could be argued that Tweneboah-Koduah et al framework with which the study was conducted is not broad enough to deal with the inherent limitations of typical marketing frameworks that we cloak on political phenomenon. The study does not take into account the concepts of political sophistication (Luskin, 1990; Converse, 1964; Campbell et al., 1960), where Luskin, for

8 example, defines political sophistication as ‘the extent to which a person’s political cognitions are numerous, cut across a wide substantive swath, and are highly organized or ‘constrained’ (Luskin, 1990: 332).

According to Luskin, political sophistication suggest that interest and intelligence, representing motivation and ability, have major effects, but that education and media exposure, the big informational variables, do not necessarily provide the impetus for sophistication (1990: 331). This means that voters, just like consumers, are not equal in their understanding of—and interest in—the political product regardless of levels of education and exposure to news media, and how it impacts on their motivation to seek political knowledge, vote or to abstain from voting. This and many other factors that limit the extant literature on this subject have been dealt with in the current study. It could thus be argued that Tweneboah-Koduah et al study is a victim of the caution given by some political marketing experts (Henneberg, 2004; O‘Shaughnessy, 2001; Lock and Harris, 1996) that the conceptualisation of the political phenomenon in the same way as the market place is problematic. That is not to say they entirely negate marketing‘s applicability in politics, and so the study is majorly flawed. Rather, they suggest the need to adapt and develop frameworks capable of responding to the tensions as explained.

O‘Shaughnessy, for example, notes that “we should not assume political contexts are invariably analogous to business to the extent that methods can be imported and used with equal effect.” (2001: 1047). Henneberg (2004) cautions that our increasing quest to demonstrate that marketing works (or worse, do not work) in politics has resulted in stunted growth of political marketing, where there is lack of effort to develop new theories to explain the political marketing subdomain. And Lock and Harris (1996) also caution the direct transfer of the marketing principle and associated theories into politics, identifying seven key differences that make the application of political marketing in politics difficult (Lock and Harris 1996: 14-15). Among these factors is that:

The political party or candidate is a complex intangible product which the voter cannot unbundle. As a consequence most voters have to judge on the overall packaged concept or message (Lock and Harris, 1996, pp. 14-15).

In the above quote, Lock and Harris argue that because voters are unable to unbundle the electoral product offering, the majority choose on the basis of the overall political package, concept, or image. They argue that while there are other complex products or services which consumers are unable to unbundle, the range of concepts and issues in the political bundle distinguish it from such situations noting that, ‘voters’ perceptions of party characteristics can be blurred and their knowledge of specific policies is low’ as against consumer knowledge in commerce where ‘consumers have remarkably homogeneous perceptions of product characteristics’ (Lock and Harris, 1996: 14-15). This gap in product knowledge sophistry explains the inability of respondents in Tweneboah-Koduah et al (2010) study to link policy initiatives such as ‘Getfund’ to the right party brand.

The question of voter choice and association in the Ghanaian political space, especially in the 2000 elections, has also been highlighted by other studies (Anebo, 2001; Gyima-Boadi, 2001; Ayee, 2000; Mensah, 2012), where it is noted that the NPPs electioneering strategy in 2000 made

9 use of policy and candidate attributes as associative elements to differentiate itself from competitors in that elections. According to these studies the NPP campaign in the 2000 general elections used their candidate’s image to project difference between them and their rival party, NDC. Despite this development of voter targeting, using political brand association it is still unclear who the core voter is to the NPP and NDC, and how have they changed over the years? It is thus argued in this study that for twenty (20) years since the inception of multiparty democracy and the increasing intensity and sophistry of political campaigning, parties in Ghana still lack the understanding of voter dynamics; who is the target voter, how are they changing, and what is informing the change?

The blurring of who is likely to vote NPP, NDC or any other party for that matter is leading political parties in Ghana to miss-strategise and miss-target their campaign. A case in point is the 2012 general elections when “free or quality education” became the centre stage of the two parties’ campaigns and the ensued debate as to who got it right. In marketing for example, effective product and service proposition is premised on the understanding of who your target consumer is, and that of your competitor. Without this insight, an organisation such as a political party will not be able to identify rightly the voter’s need and satisfy it. A political party is also likely to mismanage resources in operationalizing the 4Ps because of their inability to appreciate rightly the profile of the voter, leading to an overarching failure in strategy for success. This research therefore offers political parties insight into who their likely voters are, how their voting patterns have evolved, and the causes of this evolution. Based on this information, political parties will be able to position their offerings and target better their political campaign messages in the coming 2016 general elections.

2.1 Does leadership change affect party behaviour, and voter choice? This study is also aimed at understanding whether party leadership changes in Ghana are impacting on party behaviour, brand image and voter choice. Hence, a look into the literature in this direction is paramount. The political brand literature is replete with strategies for political parties to maintain difference from competitors and also to stand identified to voters. In practice, studies (Worcester and Baines, 2006; Needham, 2006; Reeves et al., 2006; Norris, 2001) suggest an increasing use of the brand concept by political parties to differentiate and identify themselves just as commercial firms do. Based on the knowledge that the traditional means of political branding using ideology is limited and obsolete (Lees-Marshment, 2001; Needham, 2006; Reeves et al., 2006; Scammell, 1999), research in political branding propose the idea of using candidate attributes as the basis for developing and managing the political brand, and to influence voter choice (Needham, 2006; Reeves et al., 2006; Scammell, 1999). Hence, the creation of a political brand based on attributes associated with candidates (Needham, 2006; Reeves et al., 2006).

The concept of candidate branding, and its impact on a party’s electoral fortune, is one of the topical issues in political brand literature (Needham, 2006; White and De Chernatony, 2001; Worcester and Baines, 2006). Although policy positions and party ideals continue to play significant roles in electioneering, the process of image-building of political candidates and the public perceptions of attributes considered desirable of a candidate are regarded as some of the factors that impact on election outcomes in most Western democracies, especially in countries of presidential governments like the US (Scammell, 1999; Norris, 2001; Needham, 2006; Worcester

10 and Baines, 2006).

Needham (2006), for example, argues that due to the decline in institutional attachments by consumers, with its reflections in politics, parties are no longer able to persuade voters with their inherent values. Instead, political parties need to identify and orient their appeals around values that voters perceive as desirable to vote for. In this direction, parties aim to choose candidates around these values. The chosen candidate then nurtures and enhances the party’s image through his or her leadership and management style that is manifested during discussions and debates, interaction with voters, the media and other stakeholders all under the spotlight of the interactive media, well-connected with voters. Through this process, the party begins to remould itself in the minds of the electorate. The approach, according to Needham (2006), possesses the potency to create memorable, emotional, and functional pictures in the minds of the electorate. Examples of parties that have remoulded themselves based on candidates’ characteristics and been successful at the polls, according to Needham, are Blair and the New Labour party in Britain, and Clinton and the Democrats in the US (Needham, 2006). These leaders chiefly dictated new ways of organizing their parties and saw them renewed much along the values of the leaders themselves (Needham, 2006).

In Ghana, both the NDC and NPP have had changes in party leadership, with presidential candidates of different characteristics taking charge at different electoral cycles in the past 20 years. We have also seen overwhelming emphasis on candidate personalities in media reporting before, during and after electioneering (Gyima-Boadi, 2001), making Ghanaian elections more presidential, akin to that of the US. The question is, have these changes influenced party behaviour, party brand image and voter choice, eventually? These are the questions the study seeks to address through the methods below.

3. Methodology

3.1 Study design and sampling This study is quantitative and exploratory, using both secondary and primary sources. In sampling, the Electoral Commission’s (EC) data compilation on elections, and the Accra Metropolitan Assembly’s (AMA) area demarcation records were used to determine the sample frame of 23 constituencies of the Greater Accra region. The AMA’s classification uses demographic terms, including income levels of residents, housing characteristic and environmental conditions of the residential suburbs of the city. The classification is stratified into four (4) income zones to enable viable determination of level of poverty, according to the AMA. This classification has been adopted by the AMA and gazetted in the Local Government Bulletin of the Assembly (January, 2002).

The electoral register and voting records from the electoral commission, and the AMA’s area demarcation records were necessary to establish the share of votes for the two parties in constituencies that are predominantly lower and upper classes. The determination of share of votes in the last 20 years was necessary to gain initial insight on the relationship between class and voter choice of party. Out of the 23 constituencies, seven (7) were selected through stratified random sampling technique using the AMA’s four (4) classification, based on which further

11 reclassification was made for the purposes of this study. In this direction, constituencies that are predominantly 1st and 2nd class residential areas were grouped as upper class stratum, and those with predominantly 3rd and 4th class residential areas were classed lower. The two strata were later arranged in alphabetical order and each was assigned numbers to simple randomly select the sample. This exercise resulted in the following constituencies: Ablekuma North, Ablekuma South, Ayawaso West Wuogon, Dade Kotopon, Ledzokuku, Odododiodio and Okaikwei South

In the primary research, the constituencies mentioned above were surveyed in an attempt to explore from the inhabitants perceptions on the political brands and what informs the voting trends. Again, stratified random sampling (disproportionate stratification, where the sample size of each of the stratum is not proportionate to the population size of the same stratum) was applied. In choosing sample communities within the constituencies for the administration of the questionnaire, simple random sample was again used in selecting only one community from each constituency, leading to the following communities….

3.2 Data collection Data were collected using structured questionnaire with a lay out of five (5) themed areas, namely socio-demographic characteristics, voter-party identification, party leadership, party brand personality and candidate brand personality. Questions defining such dimensions related to political party leadership and voter choice, for example, are based on a five-point Likert scale. Every one of the fifteen dimensions of political party and voter choice was clearly represented in differentiated blocks in the questionnaire. For every dimension, a set of items was included in the questionnaire. For every item a Likert scale ranging from 1 (strongly agree) to 5 (strongly disagree) was used to measure the agreement of voter choice and party leadership, for example. The measuring variables were collated from the literature on party leadership, party behaviour and voter choice.

For pretesting and assessment of the variables, students, research assistants and lecturers were requested to check and remark the questions for clarity, consistency, and duplications. These exercise resulted in some syntactic changes but none reported concept misuse or difficulties in understanding. In order to ensure reliability of the study, a pre-test pilot study was also conducted using 50 students as a pilot sample size. A high response rate was recorded with all the students participating. The pilot study was helpful as it facilitated the reformulation of research instrument, refined data collection plans in terms of content and procedure. After all corrections were made, the questionnaires were distributed and collected from participating qualified voters in seven (7) constituencies in the Greater Accra region.

The study surveyed people of voting age, 18 years and above. In order to attract the maximum number of participating respondents, an introductory letter was attached to each questionnaire asking respondents for their participation while stressing the importance of the study and the researcher’s commitment to privacy and protection of data. In this regard, respondents were also requested not to provide any personal data that could make references to their person. By the end of the survey, a usable data of 339 were realized from a total of 400 questionnaire disseminated, which constitute 85%. The questionnaire was administered in English. However, voters who could not speak English were assisted by the interviewers in translating into local languages for the respondents. All questions were answered in the presence of the interviewers administering

12 the questionnaire, and it took two (2) weeks for the exercise to be completed.

3.3 Data Analysis The data obtained was analysed using Statistical Package for Social Science (SPSS) version 20. The SPSS was used to generate descriptive statistics, chi-square test and Spearman correlation for the analysis. The descriptive statistics such as percentage and frequency table were used to present the demographic profile of the research participants, class of voters and their party of choice.

The use of these descriptive statistics allowed the researcher to summarise large quantities of data using measures that can be easily understood. In addition, chi-square was used to test the significance of differences between two categorical variables, such as the association between voters’ age and the party voted for in those elections; the association between voter's gender and the party voted for; and they association between voters' class and the party of choice. In the age variable, the investigation involved a 3×3 design where age was measured on three levels (ie. 18- 39, 40-61 and 62 and above) and party choice also measured on three levels (ie. NPP, NDC and others). In gender and party choice, the investigation involved a 2×3 design where gender was measured on two levels (i.e. male and female) and party choice also measured on three levels (i.e. NPP, NDC and others). In analyzing the data for the third objective, “how have the voting patterns for the two parties changed”, the study makes use of survey respondents’ accounts of how they have voted in the past 20 years, in addition to the secondary data of voter records of the seven (7) constituencies within the same duration. These two sources are compared to understand the voting pattern.

For the fourth objective, “would change in leadership affect party behaviour, brand image and voter choice”, Spearman rank correlation was used to establish whether there is a relationship between candidate’s characteristics and party behaviour under their leadership. In this direction respondents were asked to identify the characteristics of party candidate and that of the parties they lead in order to establish whether there is a relationship between party characteristics and candidate characteristics. In analyzing data for the final objective, voter perceptual mapping was created, using strength of partisan and moderation index, which were gleaned from questions that asked the respondents to indicate which of the voter categories would vote for only NDC, only NPP or both. The count for only NDC and NPP votes is considered as strength of partisanship – that is the extent to which a voter category would vote for these parties, and the count for moderation is the figure indicating the extent to which a voter category would vote cross party.

Table 3.1: Respondents’ characteristics

Registered to vote: Frequency Percentage % N Yes 320 94.4 No 16 4.7 Missing 3 0.9 339 Gender: Male 179 52.8 Female 133 39.2 Missing 27 8 339 Age: 18-39 233 68.7 40-61 85 25.1 62+ 17 5 Missing 4 1.2 339 Education: No schooling received 14 4.1 Basic school certificate 42 12.4 Senior school certificate 75 22.1 Trade/Technical/Vocational certificate 38 11.2 Diploma 45 13.3 Professional certificate 23 6.8 Bachelor's degreeDegree 68 20.1 Master's degree 26 7.7 Doctorate degree 1 0.3 Missing 7 2.1 339 Employment status: Yes 216 63.7 No 107 31.6 Missing 16 4.7 339 Age at first vote: 18-28 273 80.5 29-39 33 9.7 40-50 8 2.4 51-60 3 0.9 62-72 1 0.3 Missing 21 6.2 339

Source: Field survey (2014)

4. Results

Table 1 shows the summary profile of the research participants with respect to their voting status, gender, age, education, employment status, and age at first vote. From the Table, most (94.4%) of the participants said they are registered to vote, while few (4.7%) are not registered voters. Three participants representing 0.9% did not indicate their voting status. Also, more than half (52.8%) of the participants identified themselves as males, and (39.2%) as females. Few (8.0%) did not indicate their gender. In terms of age, majority (68.7%) of the participants were within 18-39 age bracket, followed by 40-41 (25.1%) age bracket, and 62 years and above (5%) age bracket. Few (1.2%) did not provide any answer regarding their age bracket.

With respect to highest educational qualification, majority (22.1%) were holders of senior school certificate, followed by bachelor degree (20.1%) holders, diploma (13.3%) holders, basic education certificate (12.4%) holders, trade or technical or vocational (11.2%) holders, master

14 degree (7.7%) holders, professional certificate (6.8%) holders, no schooling received (4.1%) and doctorate degree (0.3%) holders. 2.1% of the participants did not indicate their level of educational qualification.

Based on this results, it can be said that most (93.8%) of the participants are educated and can make meaning decisions. Furthermore, regarding employment status, majority (63.7%) was employed and a little less than one-third (31.6%) said they were not employed. In terms of participants age at first vote, most (80.5%) were within 18-28 age bracket, followed by 29-39 (9.7%) age bracket, 40-50 (2.4%) age bracket, 51-61 (0.9%) age bracket and 62-72 (0.3%) age bracket. 6.2% of the participants did indicate the age bracket they were when they first voted in a national election.

4. Discussions on study objectives and findings Conducted in seven (7) constituencies in the Greater Accra region using survey, this study was to investigate four objectives, namely the age, gender and the class distribution of voters for the two parties, NPP and NDC; whether the voting patterns for these parties have changed, and the causes of the change; how changes in party leadership have impacted on party behaviour, party brand image and voter choice; and finally whether the NDC and NPP are still predominantly lower and upper class parties. The findings from the study are presented in statistical tables and graphs, and are discussed in the sections that follow under each of the objectives as sub headings.

4.1 What is the age, gender and class distribution of voters for NPP and NDC from 1992- 2012 in the seven (7) constituencies surveyed?

As discussed in the introduction, it is a universal knowledge in Ghana that the lower class of society are likely to vote NDC whereas the upper class vote NPP. This knowledge is based on the ideological categorization of parties, and of society and had increasingly shown in voter studies in Ghana (references here). However same cannot be said of age and gender. It is equally difficult for academics and politicians in Ghana to determine the support base of political parties in these voter categories. Hence, one of the objectives of the study to understand the age and gender distribution of voters for the two parties, NPP and NDC.

The following statistical tables summarise the relationships between age and gender distribution of voters and their party choice from 1992 to 2012. These tables are analysed in the sections that follow, with accompanying graphs for emphasis.

Table 4.1 A simplified contingency table (cross tabulation) of Choice of Political Party by age and gender in 1992-2000 elections in Ghana 1992 1996 2000 % of votes % of votes % of votes Age (years) NPP NDC Others P-value NPP NDC Others P-value NPP NDC Others P-value 18-39 33.9 50 16.1 15.262 0.004 36.5 50 13.5 11.927 0.018 55 38.5 6.4 7.561 0.109 40-61 45.5 49.1 5.5 48.6 44.3 7.1 60.5 32.9 6.6 62+ 31.2 25 43.8 7.6 6.8 28.6 43.8 31.2 25 Gender Male 40.6 44.9 14.5 1.251 0.535 39.3 47.6 13.1 0.752 0.687 59.8 33.9 53.8 0.802 0.669 Female 34 55.3 10.6 40.7 50.8 8.5 53.5 38 8.5 Source: Field survey (2014)

Table 4.2 A simplified contingency table (cross tabulation) of Choice of Political Party by age and gender in 2004-2012 elections in Ghana 2004 2008 2012 % of votes % of votes % of votes Age (years) NPP NDC Others P-value NPP NDC Others P-value NPP NDC Others P-value 18-39 52.7 40.3 7 7.943 0.094 50.6 44 5.4 10.353 0.035 46.1 45.6 8.3 3.299 0.509 40-61 59.7 33.8 6.5 53.8 41 5.1 53.8 38.5 7.7 62+ 37.5 37.5 25 31.2 43.8 25 47.1 35.3 17.6 Gender Male 55.9 36.4 7.6 0.316 0.854 47.6 46.2 6.3 2.434 0.296 47.4 43.5 9.1 0.803 0.669 Female 54.8 39.3 6 56.1 40.8 3.1 48.2 45.6 6.1 Source: Field survey (2014)

4.1a Summary of findings on Age distribution of voters for NPP and NDC parties. From the study findings, as noted in tables 4.1 and 4.2 above, and figures 4.1 and 4.2 below, respondents of all ages in the study have voted more for the NPP than the NDC in the seven (7) study constituencies over the last 20 years. The findings suggest that at the early years of Ghana's fourth republic, younger voters of ages 18-39 in the surveyed constituencies voted more, 50.0% for NDC as against 33.9% for the NPP. They again favoured the NDC in the 1996 elections, but at a reduced gap of 2.6% over the NPP. However, there was a shift in favour of the NPP from the 2000 elections, when the party saw its share of this voter market dramatically increased to 55.0% whereas the NDC saw a drastic reduction to 38.5%. This gains by the NPP has been reducing consistently, with the gap between them and the NDC narrowing to 0.5% in 2012 elections.

In ages 41-61, again the study findings show that the NPP has an overwhelming dominance over the NDC in the seven (7) constituencies surveyed. Apart from 1992, where the NDC recorded 49.1% vote over the NPPs 45.5, the NPP surpassed the former with 48.6% as against the NDC's 44.3%. At the height of the NPPs surge in 2000, the party established an astonishing 27.6% gap over the NDC's 32.9% vote, and by the 2012 elections the former was still leading by 15.3% gap over the latter. Between the ages of 62 and above, NPP has been dominant once again since 1992, with the exception of 2008 when the NDC gained 12.6% lead over the NPP. By the 2012 elections, the NPP has reestablished its dominance with 11.8% gap.

The study findings also establish significant association between the two variables, age and party choice in three electoral cycles, 1992, 1996 and 2008 elections, but no relationship between age and party choice in 2000, 2004 and 2012 elections. The reason behind the association of voter age and party choice in the early years of Ghana’s democracy is documented in the literature (Gyima-Buadi, 2001), where as a result of their experiences of the military regime, out of which came the NDC, voters of upper age brackets are most likely to vote NPP or other parties.

On the other hand, it is noted in literature that significant number of young people in Ghana joined the youth wing of the NDCs predecessor military regime, PNDC as cadres and so the party in the early years of democracy enjoyed their support. Furthermore, majority of young people who were obvious of the military experiences fancied the “youthful exuberance” of the NDC and its leader at the time, Jerry Rawlings as opposed to the “old age, conservative” character traits of the NPP. It is not surprising however that 2000 saw no significant relationship between age and party choice as there was near consensus amongst all category of voters in Ghana for a new party to take charge after years of P/NDC rule. This consensus might have followed the party into 2004. What is however surprising is the 2012 elections, where the study findings suggest no relationship in age and party choice. This is because of the candidature of

John Mahama of the NDC. He is the youngest to have contested for presidential elections in Ghana’s history, and his campaign was built around it in the hope that he would appeal to younger and female voters, yet no relationship was found in the study. Equally surprising is the 2008 elections, where both candidates Atta-Mills of the NDC and Akufo-Addo of the NPP were of similar demographic characteristics and are most likely to appeal to similar age brackets. There is also no evidence that these two parties championed any campaign message or issue that was significantly aimed at age.

Fig. 4.1 Age distribution of voters across parties Source: Field survey (2014)

Fig 4.2 Age distribution of voters across parties Source: Field survey (2014)

4.2 Detailed report on the study findings

4.2a Age and party choice in 1992 elections In figure 4.1 above, a representation of the findings on voter age category of participants who voted in the 1992 elections and their choice of party could be found. In the 18-39 age category, the results revealed that half (50.0%) of the participants voted for NDC compared to 33.90% for NPP and 16.10% for other political parties. Also, in 40-61 age category, majority, 49.10%, voted for NDC compared to 45.5% for NPP, and 5.50% for other parties. Furthermore, in the 62 and above year’s category, majority (43.80%) of the respondents voted for other parties followed by

NPP (31.20%) and NDC (25.00%). This finding suggests that younger and old people voted more for NDC compared to NPP and other parties in the 1992 elections, but older people voted more for other parties against the two major parties. It also suggests that there is an association between voters’ age group and party choice in the 1992 general election. The chi-square test showed a significant association between age of voters and choice of party (Chi-square = 15.262; df =4 p<0.05).

4.2b Age and party choice in 1996 elections As shown in Figure 4.1 above, half (50.00%) of the participants in the 18-39 age category voted for NDC in the 1996 general election as against 36.50% for NPP and 13.50% for other parties. In addition, majority (48.60%) of the participants voted for NPP followed by NDC (44.30%) and other parties (7.10%) in the 40-61 age category, whereas in the 62 and above year age category, majority (37.50%) voted for other political parties compared to NPP (31.20%) and NDC (31.20%). Based on this result, it can be argued that NDC appealed more to younger people compared to its major competitor in the 1996 general elections; however NPP appealed more to old people in the 1996 election than NDC. Also, other political parties appealed more to older people than the two major parties in the 1996 election. This seems to suggest an association between voters’ age category and party choice in the 1996 election. The chi-square test indicated a significant association between age of voters and choice of party (Chi-square = 11.927; df =4 p<0.05).

4.2c Age and party choice in the 2000 elections In figure 4.1 above, it is noted that more than half (55.00%) of the participants in the 18-39 age category voted for NPP compared to the NDCs less than two-fifth (38.50%), and few (6.40%) for other parties in the 2000 elections. Furthermore, in the 40-61 age category, more than half (60.50%) of the participants voted for NPP as against 32.90% for NDC and 6.60% for other parties. Again, majority (43.80%) in the 62 and above age category voted for NPP, followed by NDC (31.20%) and other parties (25.00%). Based on this result, it can be suggested that majority of the voters in the various age categories voted more for NPP against NDC and other parties, and by extension an indication of association between age category of voters and party choice in the 2000 general election. The chi-square test performed indicated a no significant association between age of voters and choice of party in the 2000 election (Chi-square = 11.9; df =4; p >0.05).

4.2d Age and party choice in 2004 elections From Figure 4.2 above, more than half (52.70%) in the 18-39 age category voted for NPP against a little over two-fifth (40.30%) for NDC and few (7.00%) for other parties. Similarly, in the 40- 61 age category, more than half (59.70%) voted for NPP in contrast to 33.80% for NDC and 6.50% for other parties. However, in the 62 and above age category, participants voted equally (37.50%) for the two major parties compared to other parties (25.00%). This results appear to suggest that the NPP party appealed more to both young and old people in the 2004 election compared to NDC. Furthermore, the two major parties also seem to appeal to older people in the

2004 elections compared to other political parties. In addition, this distribution may indicate association between age category of voters and party choice in the 2004 election. The chi-square test showed there is a no significant association between age of voters and choice of party in the 2004 election (Chi-square = 7.943; df =4, p >0.05).

4.2e Age and party choice in the 2008 elections As noted in Figure 4.2 above, more than half (50.60%) of the participants in the 18-39 age group voted for NPP in contrast to 44.00% for NDC and 5.40% for other parties. Likewise, in the 40-61 age group, more than half (53.80%) of the participants voted for NPP compared to 41.00% for NDC, and 5.10% for other parties. However, in the 60 and above age group, majority (43.80%) voted for NDC against NPP (31.20%) and other parties (25.00%). Inferring from this finding, it can be said that NPP in the 2008 election was more attractive to both young and old people than NDC. In the contrary, NDC was more attractive to older people in the 2008 election than NPP and other parties, hence indication of association between age category and party of choice in the 2008 election. The chi-square test indicated there is a significant association between age of voters and choice of party in the 2008 election (Chi-square = 710.353; df =4; p<0.05), but the association is weak (Cramer's V= 0.141).

4.2f Age and party choice in 2012 elections As shown in Figure 2.6.1, majority (46.10) of the participants in the 18-39 age bracket voted for NPP followed by NDC (45.60%) and other parties (8.30%). Also, in the 40-61 age bracket, more than half (53.80%) of the participants voted for NPP compared to a little below two-fifth (38.50%) of the participants for NDC and few (7.70%) for other parties. More so, in the 62 and above age category, majority (47.10%) voted for NPP followed by NDC (35.30%) and other parties (17.60%). Based on this result, it can be concluded that voters in the various age categories voted more for NPP against NDC and other parties in the 2012 general election, and may suggest an association between voters’ age category and party choice in the 2012 election. From the chi-square test performed, there is no significant association between age of voters and choice of party in the 2012 election (Chi-square = 3.299; df =4; p >0.05).

4.3 Gender and party choice in national elections Another objective of the study was to determine whether there is a significant association between gender of voters and party choice in national elections in Ghana. The study employed descriptive statistics (contingency graph and table) and chi-square test to analyse and present the research findings, as below.

4.3a Summary of findings on gender distribution of voters for NPP and NDC parties. The pattern of votes in the gender category was similar to that of age. The findings suggest that both male and female categories voted more for NDC than the NPP in the early years of 1992, and 1996. Since then the NPP had been dominant from 2000 till date in both male and female categories of study respondents in the seven (7) constituencies surveyed.

The departure point from that of the age category is that in each electoral cycle, there was no association established between these two variables – gender and party choice. That potentially suggest that political parties are not doing enough to tempt the votes of a particular gender group to warrant their relationship. Each of the gender group may vote for one party over the other based on other factors other than gender related. The two figures below, 4.3 and 4.4, show graphical representation of the study findings, and are explained in detail below, cycle on cycle.

Fig 4.3 Gender distribution of voters across parties Source: Field survey (2014)

Fig 4.4 Gender distribution of voters across parties Source: Field survey (2014)

4.3.1 Gender and party choice in 1992 elections Evidenced in figure 4.3 above, majority (44.90%) of male participants voted for NDC followed by NPP (40.60%), and other parties (14.50%) in the 1992 elections. In addition, more than half (55.30%) of female participants voted for NDC compared to NPP (34.00%) and other parties

(10.60%). It is thus noted from this results that NDC appealed more to both male and female participants in the study constituencies than NPP in 1992. The chi-square test shows there is no significant association between gender of voters and choice of party in the 1992 election (Chi- square = 1.251; df =2; p >0.05).

4.3.2 Gender and party choice in 1996 elections As shown in figure 4.3 above, more (47.60%) male participants voted for NDC compared to NPP (39.30%) and other parties (13.10%). Similarly, more than half (50.80%) of female participants voted for NDC in contrast to 40.70% for NPP and 8.50% for other parties. It is argued based on this research results that both genders voted more for NDC in the 1996 general election against NPP and other parties that took part in the election. The chi-square test showed there is no significant association between gender of voters and choice of party in the 1996 election (Chi- square = 10.752; df =2; p >0.05).

4.3.3 Gender and party choice in 2000 elections It is noted in figure 4.3 above that more than half (59.80%) of male respondents voted for NPP in contrast to NDC (33.90%), and other parties (6.20%) in the 2000 elections. Similarly, more than half (53.50%) of female participants voted for NPP against NDC (38.00%) and other parties (8.50%). Thus, the findings suggest the NPP appealed more to both male and female voters in the 2000 general elections, but no associations between gender and party choice. The chi-square test showed there is no significant association between gender of voters and choice of party in the 2000 election (Chi-square = 0.802; df =2; p>0.05).

4.3.4 Gender and party choice in 2004 elections From figure 4.4, more than half (55.90%) of male participants voted for NPP compared to 36.40% for NDC, and 7.60% for other parties. Likewise, more than half (54.80%) of female participants voted for NPP against NDC (39.30%), and other parties (6.00%). It can be concluded from this results that both genders voted more for NPP in the 2004 general election compared to NDC and other political parties. The chi-square test showed no significant association between gender of voters and choice of party in the 2004 election (Chi-square = 0.316; df =2; p >0.05).

4.3.5 Gender and party choice in 2008 elections Noted in figure 4.4 above, majority (47.60%) of male respondents voted for NPP compared to NDC (46.20%), and other parties (6.30%). Likewise, more than half (56.10%) of female participants voted for NPP against NDC (40.80%), and other parties (3.10%). From this results it could be said that both genders voted more for NPP compared to NDC and other political parties in the 2008 election. The chi-square test shows there is no significant association between gender of voters and choice of party in the 2008 election (Chi-square = 2.434; df =2; p >0.05).

4.3.6 Gender and party choice in 2012 elections From Figure 4.4, majority (47.40%) of male participants voted for NPP followed by NDC (43.50%) and other parties (9.10%). Similarly, majority of female study participants voted for NPP against NDC (45.60%), and other parties (6.10%). Thus, both genders voted for NPP against NDC and other political parties in the 2012 general elections. The chi-square test shows there is no significant association between gender of voters and choice of party in the 2012 election (Chi-square = 0.803; df =2; p >0.05).

5. What is the perceived class distribution of votes for the two parties? The second objective of the study was to assess respondents’ perception on class and party choice - who they think would vote for which party in the last 20 years. This objective was necessary to examine whether for the last 20 years, there has or has not been any change in the way we perceive class-party association. Here, the prevailing assumption in Ghana is that majority of low income people, and/or who lives in suburbs considered as lower class, or slams are most likely to vote NDC, and those that are in higher class income brackets, and live in upper class neighbourhoods would predominantly vote NPP. Hence the study sought to test this assumption by ascertaining the various classes of voters, according to job category and broad generation of class – rich and poor – and the party they are most likely to vote for. In this categorization, upper income groupings includes lecturers, professors, journalists, lawyers, university students, businessmen, banks/telco professionals and medical doctors. Lower income groupings include: farmers, fisher folks, market women, teachers and nurses.

Other peculiar categories such as religion and geographic location (Muslims, Christians, Zongo) were included in this section as they are usually discussed in relations to class in Ghana. It is believed that poverty is prevalent in the Muslim and Zongo communities than there is in the communities that are predominantly Christian. The results were presented using descriptive statistics as shown in the Table 5.1 below. The electoral cycles were divided into two halves, with 1992 to 1999 being the first, and 2000 to 2012 being the second. This is because the 2000 elections is always considered as the beginning of a new dawn in Ghana’s democracy, where for the first time in the fourth republic one party handed over to another through the ballot box (Gyima-Buadi, 2001).

The two tables below, Table 5.1 and 5.2 show the various classes of voters and the party they are most likely to vote from 1992 t0 1999 and from 2000 till now, assuming only NPP and NDC existed.

Tabe 5.1: Class of voters (in percentage %) and party likely to vot for from 1992- 1999 Class of voters NDC Gap NPP Total partisan vote Gap Cross party vote Not sure Missing Total % Most rich people 13.3 24.5 37.8 51.1 21.6 29.5 12.1 7.4 100 Most poor people 21.8 4.7 26.5 48.3 17 31.3 12.4 8 100 Most farmers 20.1 7 27.1 47.2 13 34.2 10.3 0.3 100 Most fisher folks 21.5 1.7 19.8 41.3 1.2 40.1 11.2 7.4 100 Most lecturers 18.9 16.1 2.8 21.7 17.8 39.5 9.4 7.4 100 Most professors 12.7 18.6 31.3 44 3.6 40.4 8 7.7 100 Most journalists 18.9 3.5 22.4 41.3 0.3 41.6 9.1 8 100 Most lawyers 36.6 20.1 16.5 53.1 20.9 32.2 6.8 8 100 Most university/tertiary students 9.7 38.4 48.1 57.8 30.4 27.4 7.1 7.7 100 Most market women 51 41.3 9.7 60.7 34.7 26 5.3 8 100 Most business owners 21.2 1.8 23 44.2 5 39.2 8.3 8.3 100 Most banks/telecom professionals 20.9 5.3 15.6 36.5 9.2 45.7 9.7 8 100 Most teachers/nurses 18.3 5.6 23.9 42.2 0.3 42.5 0.7 7.7 100 Most medical doctors 17.4 10.3 27.7 45.1 4.1 41 6.2 7.7 100 Most muslims 21.5 2 19.5 41 1.2 39.8 11.8 7.4 100 Most zongo people 39.2 27.1 12.1 51.3 18.9 32.4 8.3 8 100 Most christians 11.5 14.2 25.7 37.2 4.7 41.9 13 8 100 Sample size (n) =339 Source: Field survey (2014)

Tabe 5.2: Class of voters (in percentage %) and party likely to vot for from 2000-2012 Class of voters NDC Gap NPP Total partisan vote Gap Cross party vote Not sure Missing Total % Most rich people 12.4 15 27.4 39.8 0.6 40.4 11.5 8.3 100 Most poor people 28 9.7 18.3 46.3 11.2 35.1 9.7 8.8 100 Most farmers 20.9 4.5 25.4 46.3 8.8 37.5 7.7 8.6 100 Most fisher folks 15 12.7 27.7 42.7 3.5 39.2 9.7 8.3 100 Most lecturers 8.3 22.4 30.7 39 4.4 43.4 9.4 8.3 100 Most professors 7.1 42.2 49.3 56.4 27.5 28.9 6.5 8.3 100 Most journalists 8.8 20.7 29.5 38.3 5.9 44.2 9.4 8 100 Most lawyers 4.4 46.4 50.4 54.8 25.9 28.9 8 8.3 100 Most university/tertiary students 38.9 23 15.9 54.8 24.4 30.4 6.5 8.3 100 Most market women 13.6 14.4 28 41.6 0.9 42.5 8 8 100 Most business owners 4.4 36 40.4 44.8 7 37.8 8.8 8.6 100 Most banks/telecom professionals 15.6 8.6 24.2 39.8 0.9 38.9 10 11.2 100 Most teachers/nurses 17.4 5 22.4 39.8 3.2 36.6 13 10.6 100 Most medical doctors 11.8 4.1 15.9 27.7 6.5 34.2 20.1 18 100 Most muslims 44.8 38 6.8 51.6 21.8 29.8 10.6 8 100 Most zongo people 39.8 33 6.8 46.6 10.3 36.3 9.4 7.7 100 Most christians 13.6 1.8 11.8 25.4 29.8 55.2 11.2 8.3 100 Sample size (n) =339 Source: Field survey (2014)

To satisfy this objective, participants were ask to indicate in their opinion, for which party would most of these voter markets/categories vote, between NDC and NPP from 1992 to 1999, and from 2000 till date.

From the results, as evidenced in the two tables above, from 1992 to 1999, there was a pervasive view that rich people in society would vote almost three times more for NPP than they would for NDC, with the statistics showing 13.3% for NDC, 37.8% and bipartisan votes being 29.5. The voter share for NPP however was reduced by 10% and bipartisan votes increased to 40% and the

NDC votes stayed the same from 2000 till now. This could lead to the inference that the rich people in society are becoming more centrist, but to the NPPs disadvantage.

In terms of lower class category, it is noted in the findings that there was no significant ship to their persuasion in the last 20 years. Both partisan and bipartisan vote have remained almost the same, with the only observable difference being a switched from NPP to NDC in terms of those who would vote partisan. The bipartisanship vote only moved up 4% for the last 20 years.

Same could be said of farmers. The findings indicate very slight movement amongst this voter category. whereas those who would vote NPP had reduced but slightly from 27.1 to 25.4, that of the NDC had remained almost the same with bipartisanship shifting by 3% upwards. What is interesting however, is the notion that NPP seem to be leading the NDC in this voter category. Perhaps, the fall in place of the NPPs claim that the party's tradition is from predominantly farming communities, and that does not make them elitist.

The findings suggest that bipartisan vote amongst fisher folks, just like farmers has remained the same at almost 40%. Again, what is interesting is the NPPs lead over the NDC in this category also. And quite revealing also is the increase of the NPP share of this voter category by almost 8% from their previous 19.8 when NDC was leading with 21.5%

In the direction of most lecturers and professors, there is a yawning gap of between the NDC and NPP perceived association. The NDC is thought to have had the backing of the lecturers in the early years of 1992 to 1999, by 18.9% to NPPs 2.8%. The reverse is the case now, and the respondents interviewed are of the view that there is supported by lecturers by 30.7% against the NDCs 8.3%. However, bipartisanship vote tops partisan votes in both eras, at 39.5% and 43.4% in the first and second eras respectively.

With regards to professors, the first era statistics suggest a whopping 20% gap in favour of the NPP, and bipartisan vote remaining high amongst this group. However, the data shows that unlike the lecturers bipartisan vote has reduced and the NPP has increased its shares tremendously against the NDC with the difference increasing to almost 40%.

Amongst journalist, bipartisan vote is thought to be high at 41.6%, as against 18.9% and 22.4% for NDC and NPP respectively in the first era. In the second, from 2000 till now, both bipartisan and NPP votes appreciated to the expense of the NDC. Bipartisan vote increased to 44.2%, the NPP to 29.5% and NDC reduced to 8.8%.

The most lawyers’ category recorded NDC with 36.6% as against the NPPs 16.5% and perceived bipartisan vote of 32.2%. This changed dramatically, with respondents suggesting that lawyers in the second era would vote the NPP against the NDC almost 5:1, with bipartisan votes reducing to 29%

There is also a substantial shift in how university and other tertiary students would vote. The NPP led the NDC in this category by 48% against the NDCs 9.7% at the early years, this pattern has changed from 2000 till now, with NDC leading the NPP by 23% with bipartisanship moving 3 percentage point up to 30.4%.

There is also a substantial ship in patterns amongst market women. Respondents were of the view that the market women have become very bipartisan, with almost 43% likely to vote both parties, as against 26% in the early years when the NDC controlled this voter market with 51% as against the NPPs 9.7%. From 2000 till now, the surveyed respondents think the NPP however has the upper hand over the NDC with 28% and 13.6% respectively in terms of partisan votes.

NPPs dominant in the business owner category is still entrenched with a sharp rise to 40.4% as against the NDCs 4.4% from previously perceived 23% for the NPP and 21.2% for the NDC. Bipartisanship amongst this category reduced to 37.8% from the previous 39.2%.

There is a reduced bipartisan votes amongst banks and telecom professionals, from the previous 46% to 39% currently. With the partisan vote, the NDC and NPP swapped leadership position with the NDC moving down by almost 5% from their previous 21% and the NPP moving up by 9% from 15.6%

There has been little movement within the teachers and nurses market amongst the two parties but bipartisanship reduced somewhat by about 6% from the previous of 42.5%. In both eras, the NPP has led its rival by just about 5% difference

Medical doctors are perceived to be predominantly bipartisan in both eras, despite the slight decrease in recent times from 41% to 34.2%. Partisan votes amongst the two parties were distributed as 17.4% for the NDC and 27.7 for the NPP in the early years, and 11.8% and 15.9% for the NDC and NPP respectively for recent years.

Bipartisanship amongst Muslims has reduced drastically by almost 10% difference, from 39.8% to the benefit of the NDC, which controls almost 45% from its previous 21.5% as against the NPPs current standing of 6.8% down from 19.5%

Similar findings reflected in the Zongo category, where the gap between the NDC and NPP has remained very wide, almost 40% for the NDC against the NPPs 6.8% currently. However bipartisanship seems to have increased amongst this category by about 5%.

The Christian voter market is perceived as highly cross party, increasing from the previous 41.9% to 55.2%. Leadership in partisan votes amongst the two parties has been interchanged in the two eras, with the NPP having lead the previous and NDC leading the current era.

6. Would change in party leadership affect party behaviour, brand image and voter choice? The political marketing and branding literature have noted extensively the impact that leadership change has on party brand image and voter choice (Less-Marshment, 2001; Needham, 2001 etc). For example, Lees-Marshment notes in her work how the new labour project was very much calved around Blairism, and so was the Clintonian democrats. Other case studies from Austria, Austria etc have all recounted how party leadership change affect the brand, and voter choice. In Ghana, though there is no study that has singled out the effect of party leadership change on party behaviour, Mensah (2009) noted how the leadership of John Kufuour of the NPP party changed party behaviour and attitudes towards new ways of doing things, and went on to win the elections in 2000. Thus, in this study the attempt was made to explore whether changes in party leadership could lead to change in party behaviour. In this direction, survey respondents were asked to map out the characteristics of the party candidate that reflects on the party equally. In other words respondents were asked to match the characteristics that the parties have in common with the candidates under their leadership.

Here, the study findings reveal that survey respondents found significant commonality between candidate traits and the parties under their leadership. This is presented in tables 6.1, 6.2, 6.3 and figure 6.1 below

Table 6.1: Descriptive table on perceived candidate characteristics Five characteristics that study respondents believe most describe the candidates Candidates John J. Rawlings John A. Kufour John E. Atta-Mills Nana A. Addo John Mahama Bundle of characteristics freq % freq % freq % freq % freq % Tribal, Sincere, Diplomatic, Elitist, Competent 38 11.2 27 8 19 5.6 34 10 41 4.1 Inexperience, All Tribes and religion, Modest, Corrupt, Diplomatic 30 8.8 41 12.1 30 8.8 54 15.9 33 9.7 Diplomatic; Corrupt; Modest; All Tribes and Religion; Competent 13 3.8 30 8.8 58 17.1 *63 18.6 *95 28 Modest; Diplomatic; Competent; Elitist; Corrupt *77 22.7 37 10.9 41 12.1 49 14.5 66 19.5 All Tribes and Religion; Accountable; Competent; Modest; Diplomatic 39 11.5 28 8.3 *91 26.8 52 15.3 59 17.4 Arrogant; Accountable; Diplomatic; All Tribes and Religion; Competent 26 7.7 25 7.4 5 1.5 9 9.7 12 3.5 Sincere; Accountable; Modest; Diplomatic; Competent 50 14.7 *87 25.7 44 13 18 5.3 10 2.9 Missing 66 19.5 64 18.9 51 15 60 17.7 50 14.7 Source: Field survey (2014)

Table 6.2 Descriptive table on perceived party characteristics Five characteristics that study respondents believe most describe the political parties under the leadership of the candidates Political Parties National Democratic Congress (NDC) New Patriotic Party (NPP) Under JJR Under JEAM Under JM Under JAK Under NAA Bundle of characteristics freq % freq % freq % freq % freq % Tribal, Sincere, Diplomatic, Elitist, Competent 43 12.7 31 9.1 19 5.6 19 5.6 33 9.7 Inexperience, All Tribes and religion, Modest, Corrupt, Diplomatic 27 8 49 14.5 *74 21.8 *87 25.7 *65 19.2 Diplomatic; Corrupt; Modest; All Tribes and Religion; Competent 26 7.7 *75 22.1 62 18.3 58 17.1 46 13.6 Modest; Diplomatic; Competent; Elitist; Corrupt 19 5.6 27 8 46 13.6 24 7.1 47 13.9 All Tribes and Religion; Accountable; Competent; Modest; Diplomatic 57 16.8 57 16.8 39 11.5 53 15.6 22 6.5 Arrogant; Accountable; Diplomatic; All Tribes and Religion; Competent *68 20.1 6 1.8 12 3.5 6 1.8 36 10.6 Sincere; Accountable; Modest; Diplomatic; Competent 50 14.7 57 16.8 44 13 45 13.3 47 13.9 Missing 49 14.5 37 10.9 43 12.7 47 13.9 43 12.7 Source: Field survey (2014)

Figure 6.1 Shared characteristics between candidates and parties Notes: CC – Common Characteristics; Disparate Characteristics Source: Field survey (2014)

Table 6.3: Spearman Rank Correlation between perceived characteristics of candidates and perceived characteristics of parties under their leadership Measure 1 2 Mean SD 1. Characteristics of NDC under: JJR - 0.437** 4.9 5.7 2. Characteristics of Candidate: JJR 0.437** - 4.2 1.97

1. Characteristics of NPP under: JAK - 0.217** 3.7 1.88 2. Characteristics of Candidate: JAK 0.217** - 4.53 2.15

1. Characteristics of NDC under: JEAM - 0.395** 3.62 2.2 2. Characteristics of Candidate: JEAM 0.395** - 4.03 1.68

1. Characteristics of NPP under: NAA - 0.407** 3.86 2.01 2. Characteristics of Candidate: NAA 0.407** - 3.47 1.65

1. Characteristics of NDC under: JM - 0.298** 3.76 1.84 2. Characteristics of Candidate: JM 0.298** - 3.69 1.34 ** Correlation is significant at 0.05 levels (2-tailed) Source: Field survey (2014)

From the descriptive tables 6.1 and 6.2 and figure 6.1 above, it is noted that survey respondents found significant commonality in character between candidates and the parties under their leadership. In some cases, for example candidates JEAM, NAA and JM, respondents indicated four (4) out of five (5) characteristics as common between the candidates and their respective parties. Similar finding is found in the correlation table 6.3, albeit moderate for candidates

JEAM, NAA and JM and weak in the case of JJR and JAK. It is thus possible to argue that the findings are in support of existing literature that change in party leadership would results in change in party behaviour, and voter choice as a result.

7. How have the voting patterns for these two parties changed, and what are the causes for the change?

Based on the actual vote measurement (see tables 4.1 and 4.2 and figures 4.1 to 4.4 above), as per responses given by the surveyed respondents in the study, significantly the findings suggest that year 2000 elections was a watershed moment for both parties. The early years’ elections saw a clear generational divide between the young and older voters preferring NDC and NPP respectively until that elections when there was a seemingly implied consensus towards the NPP in all ages, and gender groups. The NPP established dominance as a result, until 2008 when there was an incredible support towards the NDC, especially amongst older voters of ages 62+ gaining 12.6% lead over the NPP.

Again the significant association of age and gender that characterized the early years of electioneering had dissipated going forward, with again 2000 elections being the turning point. Since then the study findings suggest no association between ages and gender. This means one does not choose to vote a political party because of their age or gender, as evident in the early years of 1992 and 1996. It thus safe to argue that when it comes to age and gender cohorts, volatility has taking hold of Ghana's electoral market, where these cohorts can decide to choose any of these parties when properly targeted.

Again, it was discovered that there is a significant swing in people's perception about who would vote only NPP, only NDC or either of them at any point in time. Perceptions that certain class of people would vote a certain party is changing dramatically, especially amongst voters considered as economically upper class, with the shift predominantly towards the centre - cross party vote, as opposed to partisanship. However it is also important to note that views on others are entrenching, especially for professors on NPP. The study findings suggest that people perceive lecturers as moving towards the centre, reducing the gap in partisanship versus cross party votes of this cohorts from 17.8% to 4.4% between the years of 1992-1999 and 2000-2012 electoral cycles.

On perceptions of lower income category and how they would vote, a significant swing towards the centre was observed from the market women category, reducing partisanship vote of 60.7% between 1992-1999 to 41.6% between 2000 to date, and increasing cross party vote from 26% in 1992-1999 electoral cycles to 42.5% in 2000-2012 electoral cycles.

8. Are NPP and NDC still perceived as predominantly classed based parties? It is evident in the findings of this study conducted in seven (7) constituencies in the Greater Accra region of Ghana that voters’ perceptions on the political brands of these parties are changing. It is quite clear that given the right policy options voters would vote according to what they think are of value in policy choices. It is thus the responsibility of the parties to respond to

28 the volatility in the Ghanaian electoral market by offering the right mix of policy options for different cohorts of voters.

References Anebo, F. K. G (1997) Voting Pattern and Electoral Alliances in Ghana‘s 1996 Elections. ‘ African Association of Political Science, Vol. 2, No. 2, pp. 38-52. Ayee, J. R. A (1997) The December 1996 General Elections in Ghana.‘ Electoral Studies, Vol. 16, pp. 416–427. Ayee, J. R. A (2000) Deepening Democracy in Ghana: Politics of the 2000 Elections.‘ Freedom Publication, pp. 1-11. Campbell, A., Converse, P., Miller, W., Stokes, D. (1960) The American Voter. Chicago: The University of Chicago Press Center for Democratic Development (CDD) Ghana (2005) ‗Financing Political Parties in Ghana: Policy Guidelines.‘ Accra: CDD Farquhar, P.H., Herr, P.M. (1993) ‗The Dual Structure of Brand Associations.‘ In Aaker, D.A., Biel, A. L. (Eds), Brand Equity & Advertising: Advertising's Role in Building Strong Brands. New Jersey: Lawrence Erlbaum Associates, Inc. Gyimah-Boadi, E. (2001) A Peaceful Turnover in Ghana.‘ Journal of Democracy, Vol. 12, No. 2, pp. 103-117. Hay, C. (1999) The Political Economy Of New Labour: Labouring Under False Pretences: Manchester University Press Henneberg, S. C. M. (2003) ‘Generic Functions of Political Marketing.’ School of Management Working Papers, University of Bath, UK. Henneberg, S. C. (2004) “The Views of an Advocatus Dei: Political Marketing and its Critics”, Journal of Public Affair, Vol. 4, No. 3 Hinson, R and Tweneboah- Koduah, E.Y (2010) Political Marketing Strategies in Africa. Expert opinions of Recent Political Elections in Ghana. Journal of African Business, 11(2). Hogan, E. J., Lemon, N. K., and Rust, T. R (2002) Customer Equity Management-Charting New Directions for the Future of Marketing. Journal of Service Research, Volume 5, No.1, pp 4-12 August 2002, Sage Publications Jain, D and Singh, S (2002) Customer Lifetime Value Research in Marketing: A Review and

Future Directions. Journal of Interactive Marketing, Vol. 16, No.2 pp.34-46 Spring 2002 Kapferer, J. N. (2004). The New Strategic Brand Management: Creating and Sustaining Brand Equity Long Term. London, Kogan Page. Kapferer, J. N. (1997) Strategic Brand Management: Creating and Sustaining Brand Equity Long Term. London: Kogan Page Kapferer, J. N. (1995). Brand confusion: empirical study of a legal concept. Psychology & Marketing, 12(3), pp.551-68. Kumar, V and Reinartz, W (2012). Customer Relationship Management: Concept, Strategy, and Tools, Springer Lees-Marshment, J. (2001) Political Marketing and British Political Parties. Manchester: Manchester University Press Lilleker, D. G., Negrine, R. (2003) ‘Mapping a Market-Orientation: Can We Only Detect Political Marketing through the Lens of Hindsight?’ In Elections on the Horizon: Marketing Politics to the Electorate in the USA and UK, 15 March 2004, London: The British Library. Lock, A., Harris, P. (1996) 'Political Marketing – Vive la Différence!' European Journal of Marketing, Vol. 30 No. 10/11, pp. 14-24. Lupia, A., McCubbins, M. (1998) The Democratic Dilemma: Can Citizens Learn What They Need to Know? ‘Cambridge: Cambridge University Press. Luskin, R. (1990) Explaining Political Sophistication. ‘Political Behaviour, Vol. 12, No 4, pp. 331-361. Mensah, K. (2009). Segmentation and brand development: an Afican perspective. In Political Marketing: Principles and Applications, by J. Lees-Marshment. Oxon and New York: Routledge. Mensah, K (2009) Political Marketing in Ghana, in J. Lees- Marshment, J Stromback and C. Rudd (Eds) ‘Global Poltical Markeing. ‘Routledge UK. October, 2009. Pp 189-201 Mensah, A. (2011). Political Brand Management: Forms and Strategies in Modern Party Politics. Ph.D thesis, The University of Sheffield Needham, C. (2005) Brand Leaders: Clinton, Blair and the Limitations of the Permanent Campaign.‘ Political Studies, Vol. 53, No 2, pp. 343-361. Needham, C. (2006) ‘Brands and Political Loyalty.’ Journal of Brand Management, Vol.13, No. 3, pp. 178‒187 Neuman, R. W., Just, M. R., Crigler, A. N. (1992). Common knowlesge. News and political

30 meaning. Chicago: University of Chicago Press. Ninsin, K. A. (2006) Political Parties and Political Participation in Ghana. Accra, Ghana: Konrad Adenaur Stiftung Norris, P., Gavin, N. T. (1997) Britain Votes, 1997. New York: Oxford University Press Inc. Norris, P. (2001). Digital Divide: Civic Engagement, Information Poverty, and the Internet Worldwide. Cambridge University Press Nugent, P. (2001) ‘Ethnicity and Recent Democratic Experiments in Africa.’ African Issues, Vol. 29, No. 1/2, pp. 2-7 African Studies Association O‘ Shaughnessy, N. (2001) ‗The Marketing of Political Marketing.‘ European Journal of Marketing, Vol. 35, No.9-10, pp. 1047-1057 Reeves, P., L. de Chernatony and M Carrigan (2006) ‘Building a Political Brand: Ideology or Voter Driven Strategy.’ Journal of Brand Management, Vol.13, No.6, pp. 418-428. Robert Hinson, R and Tweneboah-Koduah, E. Y (2010) Political Marketing Strategies in Africa: Expert Opinions of Recent Political Elections in Ghana. Journal of African Business Volume 11, No. 2, 2010 Roslender, R and Wilson, M.S, R (2012) The Marketing / Accounting Interface. Routledge Scammell, M. (1999) ‘Political Marketing: Lessons for Political Science.’ Political Studies, Vol. 47, pp. 718-739 Smith, G., Saunders, J. (1990) ‘The Application of Marketing to British politics.’ Journal of Marketing Management, Vol. 56, pp. 1-17. Schneider, H. (2004) Branding in Politics–Manifestations, Relevance and Identity-Oriented Management. ‘Journal of Political Marketing, Vol. 3, No. 3. Tweneboah-Koduah, Y. E, Akotia, M., Akotia, S. C and Hinson, R (2010) "Political Party Brand and Consumer Choice in Ghana," Journal of Management Policy and Practice, Vol. 11, No. 5, pp.79 - 88 White, J., De Chernatony, L. (2002) ‘New Labour: A Study of the Creation, Development and Demise of a Political Brand.’ Journal of Political Marketing, Vol.1, No.2/3, pp.45-52. Worcester, R. and Baines, P (2006). Voter Reseaerch and Market Positioning: Triangulation and its Implications for Policy Development. In Winning Elections with Political Marketing, (Eds) Philip J. Davies & Bruce I. Newman. New York: Howarth Press Youde, J. (2005) Economics and Government Popularity in Ghana.‘ Electoral Studies, Vol. 24,

31 pp. 1–16.