Research Project: Voter Behaviour and the Political Dominance of the ANC. Introducing the Research Problem

Transcription

1 Introducing the Research Problem The problem of this research project concerns the political dominance of the African National Congress (ANC), and what drives the majority of the South African electorate to continue to vote for this major South African political party. A dominant political party is one that has seen consistent success in elections. Its defeat either cannot be envisaged or is unlikely foreseeable in its future (Suttner, 2006:227). This is because its influence trumps that of all other political parties for at least a generation so that its rule becomes recognised with a particular epoch (Langfield, 2014: 1). While one-party dominance is not necessarily detrimental to democracy, as for example India - an established democracy has endured lengthy periods of such dominance, it is not necessarily conducive to developing democracies, as it could cause declines in government accountability (Human Sciences Researchers Council, 2005: 3). Since South Africa s first national general democratic election in 1994, the country has been widely acknowledged as a clear-cut and prime example of one-party dominance. The ANC has been prominent in every sphere of South Africa s politics. Apart from seeing victory in the country s five national general elections with nothing less than sixty two percent of the vote, the ANC has also managed to govern all nine of South Africa s provinces, as well as large sections of municipal governments, for lengthy periods of time (Langfield, 2014: 1). Even though the ANC s organisational failures and intraparty factionalism have caused its dominance to deteriorate slightly from the mid-2000s, no drastic changes in South Africa s national general election results have been seen (Langfield, 2014: 1; 7). Furthermore, despite there being evidence of widespread dissatisfaction with the ANC amongst South Africa s electorate, such as the party s failure to deliver basic services and its internal corruption, this has also not translated into any major transformational shift in politics (Langfield, 2014: 8-9; Booysen, 2012: 295). Rather, the struggle of the opposition parties to 1

2 gain support has continued; and, as of yet, they have only been able to produce a few significant advances (Langfield, 2014: 1-2). The explanations that have been provided for the ANC s everlasting dominance, since South Africa s early days of democracy, have been grounded upon the way in which the party manages its extensive coalition (Langfield, 2014: 2). After years of the country s population segregation, brought about by the colonial, Union and apartheid eras, the ANC s appeal to national unity and integration captured the attention of the majority of the electorate (Langfield, 2014: 1). The party s popular support has been reinforced by its former coalitions with the South African Communist Party and the Congress of South African Trade Unions (Langfield, 2014: 2). Moreover, following the period of the Government of National Unity, in which there was a formal institutionalising of inter-party cooperation, the ANC successfully weakened its opponents (Langfield, 2014: 2). In fear of South Africa (SA) becoming a one-party state, the ANC s continued domination raises the pertinent question: why do the majority of South African voters choose the ANC? In turn, this question is informed by the broader question concerning voter behaviour in SA: why do the majority of South African voters vote in the ways that they do? Although there has been some agreement amongst authors in the widely cited research, regarding the theories underpinning voter behaviour since the country s 1994 election, there seems to be hardly any evolving consensus on which thesis is (or combination of theses are) the most appropriate for explaining such behaviour in democratic SA, and, more particularly, for elucidating the ANC s unceasing predominance. Research Question The question of this research project is as follows: What determines voting behaviour in SA for the ANC to have been able to gain the majority of the vote in every South African national general election since 1994? 2

3 State of Knowledge Habib and Naidu (2006: 81), and in a separate study, Naidu (2006: 49) challenge the racial identity thesis on the basis of South African national general election data and attitudinal surveys, which, they claim, portray racial identity as not being the sole determinant or explanation of voter behaviour amongst the South African electorate. Instead, they allege that the data indicates that large proportions of the electorate do make rational choices before casting their votes for the relevant political parties. Moreover, they contend that the racial identity thesis is unsound on two significant bases. Firstly, the explanation of the thesis is flawed because it focuses on race as the independent objective variable, without considering that there are strong overlays between racial and other identities. Secondly, the thesis merely presumes that race is definitely an independent causal factor in the behaviour of voters, wholly disregarding the possibility that other identities, either alone or as a mixture including race, could also constitute such independent causal factors (Habib and Naidu, 2006: 83). They conclude that the racial identity thesis, as a theory for explaining the voting behaviour of the South African electorate is highly problematic, as the findings of their study show that there are indeed large proportions of the electorate who do make rational choices when electing political party candidates. However, they maintain that this does not mean that the consideration of racial identity should be completely precluded from explaining voting behaviour in South African national general elections. Such preclusion would be unreasonable because racial identity certainly does play a role here. Yet, since the correlation between racial identity and electoral behaviour is not robust enough, racial identity cannot be the only objective variable and causal factor in determining and explaining the voting behaviour of the South African electorate (Habib and Naidu, 2006: 91). Likewise, Bratton, Bhavani and Chen (2012: 27) argue that neither racial identity voting nor policy and performance, especially economic, voting, alone, adequately accounts for the voting behaviour in South African national general elections. They claim that racial identities and economic interests both matter when contemplating the voting behaviour of the 3

4 South African electorate. In essence, they allege that the dissimilarity between racial identity and economic voting is overstated, as each of these types of voting behaviours cannot be understood as one-dimensional constructs (Bratton et al, 2006: 30). The outcomes of their study illustrate that while South African voters do vote on the basis of racial identity, they consider their economic interests as being uppermost when electing political party candidates. Further, they calculate their economic interests with more reference to the state of South Africa s economy as a whole, than to the condition of their private finances (Bratton et al, 2006: 46). This is due to the fact that it is difficult to isolate government responsibility for one s personal fortunes (or misfortunes) distinctly from one s knowledge of all other relevant factors (Mattes, Taylor and Africa, 1999: 238). Furthermore, their study shows that where racial identity is factored into the equation of determining the voting behaviour of the South African electorate, its primary role is seen as shaping the economic interests of voters (Bratton et al, 2006: 47). Thus, this confirms their aforementioned allegation regarding the dissimilarity between racial and economic voting being overstated. Mattes, Taylor and Africa (1999: 235), along similar lines as Bratton et al, emphasise the role of political party performance in voting behaviour at South Africa s national general elections. They contend that while the racial identity thesis can be useful to describe elections in divided societies, it cannot sufficiently explicate individual voter behaviour. Like Habib and Naidu, they claim that this thesis suffers from inherent problems, but also do not deny that there has been a correlation between race and voting behaviour in South Africa s national general elections (Mattes, Taylor and Africa, 1999: 236). Focussing on interpreting this correlation, they explicate that voters utilise the information about political parties that is available to them to inform their political party preferences. Where no such information is readily available to voters, they will either make assumptions from any pieces of information that they do have, or will turn to consider the opinions of other individuals or sources, whom they think might be able to provide them with such information. Mattes, Taylor and Africa (2006: 237) term the process of gleaning information where no such information is readily 4

5 available low information reasoning. The past performance of political parties is a key source of such information, since this would provide a somewhat more reliable guide as to how such parties will perform in the future than the promises that are made by such parties regarding what they will do in the future (Mattes, Taylor and Africa, 1999: 237). 1 Where voters are left without any information about the way in which political parties will act in the future, Mattes, Taylor and Africa (2006: 238) go further and argue that voters will look to other aspects, such as race, to infer whether political parties would promote their interests or the interests of their community. Hence, group-based voting differences, for example race, come about or remain because voters views are moulded by the social milieu in which they are exposed to low information reasoning (Mattes, Taylor and Africa, 1999: ). Therefore, Mattes, Taylor and Africa see race as a secondary influence on voter behaviour, but as a primary factor in shaping the ways in which voters view political parties, their policies and their performance. These arguments are made definitive when looking at the findings of their study (Mattes, Taylor and Africa, 1999: 245). An almost identical study conducted by Mattes and Piombo (2001: 102; 122) concluded the same findings about the racial identity thesis and race having a more indirect impact on the voting behaviour of the South African electorate than many political analysts speculate. Similarly, Ramutsindela (2002: 54) challenges the racial identity thesis, alleging that the practical effect of this thesis is that it neither provides an adequate insight of more recent South African voter behaviour, nor highlights an accurate account of the continuities and changes in the electoral choices made by individuals in racial groups. Ramutsindela further contends that this thesis props race up on a pedestal to be viewed in isolation. Yet, ideas of race in SA have always been associated with, or reinforced by, various mechanisms and conditions. Therefore, it is logical to infer that certain issues and interests exist alongside race. However, Ramutsindela (2002: 55) cautions that this does not mean that race ought to 1 This can be contrasted with Bratton et al (2012: 30) who explicate that, regarding economic voting, voters tend to place more value on the promises that political parties make of future economic prosperity, than on information about past economic performance. 5

6 be entirely disregarded when determining the causes of voter behaviour, but that the assumptions that underpin the use of race as a determinant of voter behaviour ought to be questioned. Apart from finding that a majority of South Africans base their electoral preferences upon the policies and performance of political parties, interestingly, the outcome of Ramutsindela s study also unveils that a majority of South Africans will identify themselves as being South African, in other words as having a general identification with the country, before identifying themselves as members of particular racial groups (Ramutsindela, 2002: 55-57). Ferree (2006) takes a different approach to elucidating voting behaviour in multiracial or ethnic societies. She claims that racial identity voting, policy and performance voting and racial or ethnic heuristic voting are all able to justify the electoral outcomes of a country that merely reveal racial or ethnic censuses (Ferree, 2006: 803). She argues that if racial identity voting justifies such an outcome, it simply means that race plays a direct role in electing political party candidates and thus voters vote with the aim of expressing their racial identity (Ferree, 2006: 804). If policy and performance voting justifies such an outcome, it means that race plays an indirect role in electing such candidates. In this instance, although voters appear to be voting for the purpose of expressing their identity, they are instead grounding their candidate preferences on which candidate s policies and performance best serve their interests as, for example, black or white voters. In other words, Ferree is alleging that, in essence, black voters vote for black political parties because they are closest to them regarding policy and because they are generally happy with their overall performance and the same applies for all other races (Ferree, 2006: 805). Finally, if racial or ethnic heuristic voting justifies such an outcome, it also means that race or ethnicity plays an indirect role in electing political party candidates. Here, once again, while voters appear to be voting for the purpose of expressing their identity, they are really basing their candidate preferences on the racial or ethnic credentials of candidates in order to infer which candidate would best include or treat their racial or ethnic group fairly. With these two latter types of voting, voters do 6

7 inherently make rational choices (Ferree, 2006: 805; 808). For voter behaviour in the South African context, the findings of Ferree s study depict strong evidence that obvious patterns of racial identity voting, in which race plays a direct role, have no linkage to the voting behaviour of the majority of the South African electorate. Furthermore, between policy and performance voting and racial or ethnic heuristic voting, both in which race plays an indirect role, racial or ethnic heuristic voting tends to have more of a relationship with most of the South African electorate (Ferree, 2006: 814). Research Hypotheses/Propositions In response to the question of this research project, as stated above, three investigations will be undertaken, each with its own hypothesis. Thus, three competing hypotheses will be tested in this research project. I 1 H 1 The racial identity thesis: In SA, race is an important predictor of voting behaviour. South African voters are generally more likely to vote for political parties on the basis of race. I 2 H 2 The policy and performance thesis: In SA, the policies and performance of a political party are important predictors of voting behaviour. South African voters are generally more likely to vote for political parties that, such voters believe, have the best record, in terms of their polices and performance, to further voters self-interests. I 3 H 3 The ethnic heuristic thesis: In SA, the ethnic credentials of a political party, as an informational shortcut, is an important predictor of voting behaviour. South African voters are generally more likely to vote for political parties on the basis of ethnic credentials, in order to determine whether or not voters ethnic groups would be fairly treated by parties. 7

8 For the first hypothesis (H 1 ), the independent variable is racial identity voting, and the dependant variable is the respondents choices of political parties (which political parties they would likely vote for). There is also a control variable, race, for this hypothesis. For the second hypothesis (H 2 ), the independent variable is policy and performance voting, and the dependant variable is the same as that in H 1. For the third hypothesis (H 3 ), the independent variable is ethnic heuristic voting, and the dependant variable is, once again, the same as that in H 1. Relevance of Potential Findings The findings of this research project are relevant for attaining a deeper understanding of the existing literature of why people in societies that are divided along racial or other types of ethnic lines, like SA, vote in the ways that they do (Mattes and Piombo, 2001: 103). In SA, it has been, and still is being, feared that voters register the identity of political parties as opposed to making rational choices when marking their ballot papers (Mattes, Taylor and Africa, 1999: 236; Langfield, 2014: 2). Essentially then, they choose not to choose but to give their vote predictably on a racial basis to an ethnically defined party (Mattes, Taylor and Africa, 1999: 236). As a result, instead of election results reflecting the preferences of voters, they merely reflect the racially demographic distributions of the country (Mattes, Taylor and Africa, 1999: 236). Yet, irrespective of what actually determines voter behaviour in SA, the tendency of opposition parties has been to base their party strategies and electoral campaigns on crude racial assumptions (Habib and Naidu, 2006: 91). This accounts for their failure to attract cross-sectional electoral support, and, therefore, for their persistent struggle to increase their overall support base, resulting in the ANC s consistent electoral success. As highlighted above, undoubtedly, this would not be conducive to South Africa s democratic consolidation, since the lack of any viable opposition could cause the accountability relationship between the ANC and the citizens of SA to wane (Habib and Naidu, 2006: 91). And, as mentioned above, there is already evidence of the South African 8

9 electorate being dissatisfied with the ANC for various reasons (Langfield, 2014: 8-9; Booysen, 2012: 295). Furthermore, it might be unlikely that SA would be able to be prevented from becoming a fully-fledged one-party state. Thus, the findings of this research project could also be useful in guiding opposition parties away from racially centred party strategies and electoral campaigns towards those that are, for instance, more centred on peoples common interests in general. Research Design Since the problem of this research project pertains to the political dominance of the ANC, SA is the single case that will be used in this research project. In attempting to determine the influences of individual voter behaviour in SA, the level of analysis for this case will be on the micro level, considering the views of 2399 South African respondents (the respondents from the Global Barometer Survey: Afrobarometer). The case of this research project will also be analysed longitudinally, in accordance with the time periods of the final three survey rounds of Afrobarometer ( , and ). 2 Conceptualisation and Operationalization For the investigations that will be undertaken in this research project, the variables have been conceptualised through the predominate literature and have been operationalized through the Global Barometer Survey: Afrobarometer. The data from the three final Afrobarometer survey rounds, mentioned above, will be used for the measurement and analysis of this research project. The fact that other studies testing the behaviour of voters in 2 The final three rounds were selected because they were the only rounds that had data available for the dependent variable that will be used across the investigations in this project. Unfortunately, not all of the three final survey rounds had been completed for the independent variable of I 2 H 2. Nevertheless, this will not drastically effect the investigations of this research project because the averages of the rounds that are used will in any case be calculated, and will be considered when analysing the evidence for this variable. 9

10 SA, for instance Ferree (2006) and Bratton, Bhavani and Chen (2012), have also operationalized their variables through the surveys of Afrobarometer, lends a degree of credibility to the operationalization of the variables in this research project. Investigation 1 Independent variable Racial identity voting Racial identity voting is conceptualised as voting that is viewed as a means to, and for the purpose of, affirming self-worth as group members through racial identities (Ferree, 2006: 804; Bratton et al, 2012: 27). When voter behaviour is characterised by racial identity voting, voters affiliate themselves with and vote for the political party whose members have the same distinct physical characteristics as they do, particularly skin colour (Trimble and Dickson, 2010). Therefore, instead of voters voting for political parties by weighing up alternatives and then making rational choices in order to further their self-interests, they vote for such parties by expressing their membership of the particular racial group that they identify themselves with, in order to attain psychic benefits (Ferree, 2006: 804). The operationalization of this variable will be through the implementation of the control variable explained below. Dependent variable Choice of political parties (which political parties the respondents would likely vote for) Asking the following Afrobarometer survey question will operationalize this variable categorically: If a presidential election were held tomorrow, which party s candidate would you vote for? Owing to the voluminous number of political parties present in Afrobarometer s three final rounds of survey results, for practical sake and since the ANC is the majority party in SA, the respondents responses to the above question will be divided up into the ANC versus opposition parties (a single grouping of all opposition parties). 3 3 It should be noted that the reliability of this variable might not always be sound, since the respondents might not always provide honest answers to this survey question (Bratton et al, 2012: 31). This is principally because they feel that they are being exposed to surveillance and coercion (Bratton et al, 2012: 32). 10

11 Control variable Race For each of the three final survey rounds, Afrobarometer predominantly categorises five racial groups for SA, namely, Black/African, White/European, Coloured/Mixed Race, South Asian (which includes, inter alia, Indians and Pakistanis) and East Asian (which includes, amongst others, Chinese, Koreans and Indonesians), according to which respondents have identified themselves in other surveys carried out by Afrobarometer. 4 Investigation 2 Independent variable Policy and performance voting Policy and performance voting is conceptualised as voters voting for the political party that they believe has the most effectively and efficiently planned, or implemented, courses of action on certain issues, for example crime and unemployment, that will take place, or that are taking place, within a particular country (Ferree, 2006: 808). It is also conceptualised as voters voting for the political party that has most effectively and efficiently managed to deal with certain areas of governance that are especially significant to voters, such as the economy (Ferree, 2006: 809). Hence, unlike racial identity voting, when voter behaviour is characterised by policy and performance voting, voters vote for political parties by weighing up alternatives and then making rational choices in order to further their selfinterests (Ferree, 2006: 804). Asking the following Afrobarometer survey questions will operationalize this variable categorically 5 : How well or badly would you say the current government is handling the following matters, or haven t you heard enough to say: managing the economy, improving the living standards of the poor, creating jobs, keeping prices down, narrowing [income] gaps 4 Although Afrobarometer also mentions other as a race category, such category, unlike the ones stated above, has not been used throughout the three final survey rounds for operationalizing this dependent variable. This, together with the fact that it would not be favourable to I 1 H 1 to have an unspecified race category are the reasons why other is being excluded here. 5 While this independent variable is categorical, some of the responses that it will engender will be inherently scalar (ranked). 11

12 between the rich and the poor, reducing crime, improving basic health services, addressing educational needs, providing water and sanitation services, ensuring everyone has enough to eat, fighting corruption in government, resolving violent conflict between communities, combatting HIV/AIDS, maintaining roads and bridges, providing a reliable supply of electricity and empowering women? In accordance with Afrobarometer s survey, the respondents response options for this question are as follows: very badly, fairly badly, fairly well, very well or don t know. Dependent variable Choice of political parties (the political parties that the respondents would likely vote for) See I 1 for the conceptualisation and operationalization of this variable. Investigation 3 Independent variable Ethnic heuristic voting Ethnic heuristic voting, to a certain extent, can be conceptualised as combining identity and policy and performance voting. Yet, in the case of ethnic heuristic voting, the correlation between interests and behaviour is more indirect than in the case of policy and performance voting. This is due to the existence of an information gap, which encourages voters to make use of cognitive shortcuts. 6 Ethnic divisions can affect the cognitive shortcuts that voters choose to privilege, particularly if group membership is something that they choose to privilege (Ferree, 2006: 805). In this instance, before voters cast their votes, they will assess the ethnic credentials or labels of political parties, and then make inferences as to whether or not such parties would exclude or treat their ethnic group unfairly. They would subsequently vote for the political party that would not exclude or treat their ethnic group unfairly (Ferree, 2006: 809). Thus, in the context of ethnic heuristic voting, the ethnicvoting relationship depicts ethnicity as having an instrumental use rather than being a mere emotional attachment to ethnically exclusive political parties (Ferree, 2006: 805). 6 Cognitive shortcuts refer to easily acquired information that voters use to assist them in deciding which political party would most likely benefit them in the future (Ferree, 2006: 805). 12

13 Number of Respondents (%) Democratic Theory and Practice (POL3043F) Asking the following Afrobarometer survey question, this variable will be operationalized categorically 7 : How often is your ethnic group treated unfairly by the government? In accordance with Afrobarometer s survey, the respondents response options for this question are as follows: never, sometimes, often, always or don t know. Dependent variable Choice of political parties (the political parties that the respondents would likely vote for) See I 1 for the conceptualisation and operationalization of this variable. Evidence Graph 1: Number of Respondents (%) Who Would Vote for the ANC and Who Would Vote for Opposition Parties I 1 H 1, I 2 H 2 and I 3 H 3 Dependant variable; no control variable ANC Opposition N/A Refused Don't Know Would not vote If a presidential election were held tomorrow, which party s candidate would you vote for? (Source: Afrobarometer) 7 See note 3 above. 13

14 If a presidential election were held tomorrow, which party s candidate would you vote for? Black/Africa White/European Coloured/Mixed Race South Asian East Asian Democratic Theory and Practice (POL3043F) Graph 2: Number of Respondents (%) Who Would Vote For the ANC and Who Would Vote For Opposition Parties in terms of race I 1 H 1 Dependant variable; with control variable Would not vote 4 Don't know 31 Refused 32 N/A Opposition ANC Would not vote 7 16 Don't know Refused N/A 3 Opposition ANC Would not vote Don't know Refused N/A Opposition ANC Would not vote 28 Don't know Refused N/A 2 Opposition ANC Would not vote 7 6 Don't know Refused N/A 1 7 Opposition 114 ANC Number of Respondents (%) (Source: Afrobarometer) 14

15 Number of Respondents (%) Number of Respondents (%) Democratic Theory and Practice (POL3043F) Graph 3: How Well Respondents Believe the Government to be Handling Certain Matters ( ) I 2 H 2 Independent Variable Very badly Fairly badly Fairly well Very well Don't know How Well Respondents Believe the Government to be Handling: Graph 4: How Well Respondents Believe the Government to be Handling Certain Matters ( ) I 2 H 2 Independent Variable Very badly Fairly badly Fairly well Very well Don't know How Well Respondents Believe the Government to be Handling: 15

16 Number of Respondents (%) Number of Respondents (%) Democratic Theory and Practice (POL3043F) Graph 5: How Well Respondents Believe the Government to be Handling Certain Matters ( ) I 2 H 2 Independent Variable Very badly Fairly badly Fairly well Very well Don't know * How Well Respondents Believe the Government to be Handling: Graphs 3-5 (*Source: Afrobarometer) Graph 6: How Well Respondents Believe the Government to be Handling Certain Matters (Averages of Rounds for Overall Performance) I 2 H 2 Independent Variable Very badly Fairly badly Fairly well Very well Don't know (Source: Own calculations) How Well Respondents Believe the Government to be Handling Certain Matters 16

17 Number of Respondents (%) Democratic Theory and Practice (POL3043F) Graph 7: Number of Respondents (%) Who Feel that Their Ethnic Group is Unfairly Treated by the Current Government I 3 H 3 Independent variable How Often Respondents Believe that the Government Treats Their Ethnic Group Unfairly (Source: Afrobarometer) Analysis Table 1: I 1 H 1 - Considering Whether There Are Any Correlations ANC Opposition ANC Opposition ANC Opposition Black/African 67% 4% 56% 11% 69% 7% White/European 12% 38% 2% 21% 3% 61% Coloured/Mixed 43% *58 27% 16% *45 39% *14 24% *55 47% Race *9% % % % % South Asian 30% 15% 16% 35% 12% 31% *16 % East Asian 0% *Without control variable. (Source: Afrobarometer) 100 % 0% 33% 78% 22% 17

18 Looking at graphs 1 and 2, as well as table 1, it is clear that the dominance of the ANC, as a choice of political party that the respondents said that they would likely vote for, was mostly due to the Black/African respondents who said that they would likely vote for it, except in the last survey round ( ), in which the dominance of the ANC was obviously due to the East Asian respondents who said that they would likely vote for it. However, it should be noted that in comparison to the Black/African respondents who said that they would likely vote for the ANC, the East Asian respondents who said the same were very few (Afrobarometer). Hence, it seems that their majority support for the party in this round was not a true reflection of its dominance. Without the inclusion of the control variable, the percentages of respondents who said that they would likely vote for the ANC was 58% between 2005 and 2006, 45% between 2008 and 2009 and 55% between 2011 and With the inclusion of the control variable, the percentage of Black/African respondents who said that they would likely vote for the ANC was 67% between 2005 and 2006, 56% between 2008 and 2009 and 69% between 2011 and The White/European respondents who said that they would likely vote for the ANC was 12% between 2005 and 2006, 2% between 2008 and 2009 and 3% between 2011 and The Coloured/Mixed Race respondents who said that they would likely vote for the ANC was 43% between 2005 and 2005, 16% between 2008 and 2009 and 24% between 2011 and The South Asian respondents who said that they would likely vote for the ANC was 30% between 2005 and 2006, 16% between 2008 and 2009 and 12% between 2011 and The East Asian respondents who said that they would likely vote for the ANC was 0% between 2005 and 2006, 0% between 2008 and 2009 and 78% between 2011 and Regarding the Black/African respondents and the opposition parties, it can be seen that 4% of these respondents said that they would likely vote for opposition parties between 2005 and 2006, 11% between 2008 and 2009 and 7% between 2011 and On the other hand, 38% of the White/European respondents said that they would likely vote for opposition parties between 2005 and 2006, 21% between 2008 and 2009 and 61% between 2011 and 18

19 2012. The percentage of Coloured/Mixed Race respondents who said that they would likely vote for opposition parties was 27% between 2005 and 2006, 39% between 2008 and 2009 and 47% between 2011 and The percentage of South Asian respondents who said that they would likely to vote for opposition parties was 15% between 2005 and 2006, 35% between 2008 and 2009 and 31% between 2011 and The percentage of East Asian respondents who said that they would likely vote for opposition parties was 100% between 2005 and 2006, 33% between 2008 and 2009 and 22% between 2011 and Considering that the dominance of the ANC in the survey rounds has mostly been attributed to the Black/African respondents who said that they would likely vote for it; and since the ANC has been construed as the black party (Booysen, 2012: 295), it could be said that a relationship exists between racial identity and the behaviour of voters in SA, at least as far as the Black/African respondents are concerned. This relationship can be bolstered by the fact that very few of the White/European respondents said that they would likely vote for this dominant political party. Similarly, below 50 percent of the Coloured/Mixed Race and South Asian respondents said that they would likely vote for the ANC. Further, besides for the last survey round ( ), none of the East Asian respondents said that they would likely vote for this party. 9 The evidence, however, does not convey this racial-identity-voting relationship as definitively for the respondents of the other four racial groups who said that they would likely vote for opposition parties, than it does for the Black/African respondents who said that they would likely vote for the ANC. 8 Once again it must be remembered that very few East Asian respondents partook in these survey rounds. 9 See note 8. 19

20 Table 2: I 2 H 2 Considering Whether There Are Any Correlations *Overall Performance A N C Opposition *Overall Performance A N C Opposition *Overall Performance A N C Opposition Very badly 21.5% 29.43% 26.64% Fairly badly Fairly well 23.44% % % 8 9% 5 14% 35.94% % 28.93% % 32.55% 5 5 % 16% Very well 14.38% 11.36% 11.27% (Source: Afrobarometer and *Own Calculations) Considering graphs 1, 3, 4, 5 and 6, as well as table 2, it can be seen that between 2005 and 2006, a majority of respondents (35.94%) believed that the ANC performed fairly well across the various matters that it had to handle, 23.44% believed that it performed fairly badly, 21.5% believed that it performed very badly and 14.38% believed that it performed very well. Between 2008 and 2009, a majority of respondents (29.43%) believed that the ANC performed very badly, 28.93% believed that it performed fairly well, 27.86% believed that it performed fairly badly and 11.36% believed that it performed very well. Thus, between the first two rounds, the number of respondents who believed that the ANC performed very badly, as well as fairly badly, increased; and the number of respondents who believed that it performed fairly well, as well as very well, decreased. It can further be seen that between 2005 and 2006, the percentage of respondents who said that they would likely vote for the ANC was 58% and the percentage of respondents who said that they would likely vote for 20

21 opposition parties was 9%; and between 2008 and 2009, the percentage of respondents who said that they would likely vote for the ANC was 45% and the percentage of respondents who said that they would likely vote for opposition parties was 14%. Therefore, between the first two survey rounds support for the ANC decreased, while support for opposition parties increased. Between 2011 and 2012, it can be seen that the majority of respondents (32.55%) believed that the ANC performed fairly well across the various matters that it had to handle, 29.09% believed that it performed fairly badly, 26.64% believed that it performed very badly and 11.27% believed that it performed very well. Hence, between the last two rounds, the number of respondents who believed that the ANC performed very badly, as well as very well, decreased; and the number of respondents who believed that it performed fairly badly, as well as fairly well, increased. Furthermore, between 2011 and 2012, the percentage of respondents who said that they would likely vote for the ANC was 55% and the percentage of respondents who said that they would likely vote for opposition parties was 16%. Hence, between the last two survey rounds, both support for the ANC and opposition parties increased. A strong relationship between the policies and performance of the ANC and the voter behaviour of the South African electorate can clearly be seen for the first two survey rounds, since support for the ANC decreased while support for opposition parties increased as respondents believed the ANC to be performing less well. Regarding the last two survey rounds, the same relationship exists, although, it is not as clear-cut because both support for the ANC and opposition parties increased. The relationship is evidenced by the decrease in the number of respondents who believed that the ANC performed very badly, as well as by the increase in the number of respondents who believed that it performed fairly well between the last two survey rounds. However, it should not be ignored that support for opposition parties continued to increase even though more respondents conveyed that they were satisfied 21

22 with the ANC s performance, as this shows that while a majority of South African voters vote in accordance with the policy and performance thesis, there are definitely some outliers. Table 3: I 3 H 3 Considering Whether There Are Any Correlations Unfairly Treated ANC Opposition Unfairly Treated ANC Opposition Unfairly Treated ANC Opposition Never 43% 26% 55% Sometimes 26% 28% 23% 58% 9% 45% 14% Often 10% 15% 10% 55% 16% Always 8% 9% 7% (Source: Afrobarometer) Observing graphs 1 and 7, as well as table 3, it is clear that between 2005 and 2006, a majority of respondents (43%) felt that the ANC never treated the ethnic groups with which they identified themselves unfairly, 26% felt that it sometimes treated such groups unfairly, 10% felt that it often treated such groups unfairly and 8% felt that it always treated such groups unfairly. Between 2008 and 2009, a majority of respondents (28%) felt that the ANC sometimes treated their ethnic groups unfairly, 26% felt that it never treated such groups unfairly, 15% felt that it often treated such groups unfairly and 9% felt that it always treated such groups unfairly. Hence, between the first two survey rounds, the percentage of respondents who felt that the ANC never treated their ethnic groups unfairly decreased, while the percentage of respondents who felt that it sometimes, often, as well as always, treated such groups unfairly increased. It can further be seen that between 2005 and 2006, the percentage of respondents who said that they would likely vote for the ANC was 58%, while the percentage of respondents who said that they would likely vote for opposition parties was 9%. Between 2008 and 2009, the percentage of respondents who said that they would likely vote for the ANC was 45%, while the percentage of respondents who said that they would likely 22

23 vote for opposition parties was 14%. Therefore, between the first two survey rounds, while support for the ANC decreased, it increased for opposition parties. Between 2011 and 2012, it can be seen that a majority of respondents (55%) felt that the ANC never treated their ethnic groups unfairly, 23% felt that it sometimes treated such groups unfairly, 10% felt that it often treated such groups unfairly and 7% felt that it always treated such groups unfairly. Thus, between the last two survey rounds, the percentage of respondents who felt that the ANC never treated their ethnic groups unfairly increased, while the percentage of respondents who felt that it sometimes, often, as well as always, treated such groups unfairly decreased. Furthermore, between 2011 and 2012, the percentage of respondents who said that they would likely vote for the ANC was 55% and the percentage of respondents who said that they would likely vote for opposition parties was 16%. Hence, between the last two survey rounds, both support for the ANC and opposition parties increased. A strong relationship between the ethnic heuristics of the ANC and voter behaviour in SA can clearly be seen for the first two survey rounds, since support for the ANC decreased while that for opposition parties increased as respondents felt the ANC to be treating their ethnic groups less fairly. The same relationship also exists for the last two survey rounds, yet, it is not as easily seen because there was an increase in support for the ANC, as well as for opposition parties. The relationship is demonstrated by the increase in the percentage of respondents who felt that the ANC never treated such groups unfairly and by the decrease in the percentage of respondents who felt that the ANC sometimes, often, as well as always, treated such groups unfairly between the last two survey rounds. Nonetheless, the increase in support for opposition parties, even though more respondents felt that the ANC never treated their ethnic groups unfairly, cannot be disregarded, since it portrays that while most of the South African electorate vote in accordance with the ethnic heuristic thesis, outliers certainly do exist. 23

24 Conclusions The three investigations that were undertaken in this research project sought out to determine the behaviour of the South African electorate when voting in the country s national general elections, in the hopes that this would be able to shed some light on the everlasting one-party dominance of the ANC. Through an empirical analysis of qualitative public opinion survey data, this research project tested three competing hypotheses, namely, the racial identity thesis: race is a significant predictor of South African voter behaviour, as voters in SA have a general tendency to base their votes for political party candidates on race; the policy and performance thesis: a political party s policies and performance are necessary predictors of South African voter behaviour, since voters in SA are generally likely to vote for political parties that they anticipate have the most appropriate policies and record of performance to advance their self-interests; and the ethnic heuristic thesis: a political party s ethnic credentials is an important predictor of South African voter behaviour because voters in SA have a usual tendency to base their votes for political party candidates on their ethnic credentials, in order to establish whether or not such candidates would treat voters ethnic groups fairly. Although there is an abundance of literature on voting behaviour in SA, there is hardly any evolving consensus. The state of knowledge on this topic commonly rejects the racial identity thesis on the bases that it is an inherently flawed approach to understanding voter behaviour and that it, alone, is insufficient to predict such behaviour in the South African context. There is, however, some general agreement that race plays an indirect role in South African voter behaviour in the sense that it is capable of influencing voters interests, which impacts on their political party preferences, especially where there is a lack of information about a particular party. The disagreement then comes into being when considering which other thesis (or combination of theses) is most suitable for characterising voting behaviour in SA. The authors cited in this research project hypothesise either a racial or ethnic heuristic thesis approach, or a policy and performance thesis and an indirect racial 24

25 identity thesis approach; and their studies have proven their arguments to be correct. With these few viable options, the question about South African voter behaviour, in light of the consistent fear of SA becoming a one-party state, warranted further inquiry. Upon answering this question, this research project found that it is difficult to establish consensus on this topic, since the evidence conveyed that, generally, each of the theses tended to have some sort of relationship with the outcome of who eligible voters would likely vote for. The racial identity thesis seemed to have a relatively persuasive relationship with voting behaviour in all three of the Afrobarometer survey rounds, but chiefly amongst Africans. The policy and performance thesis had an incredibly strong and significant relationship with voting behaviour between the first two survey rounds, becoming slightly weaker and less significant between the last two survey rounds, due to the presence of a very small group of outliers. Similarly, the ethnic heuristic thesis had a very strong and significant relationship with voting behaviour between the first two survey rounds, becoming slightly weaker and less significant between the last two survey rounds, due to the existence of a tiny cluster of outliers. Therefore, the ANC s dominance could be attributed to any one (or a combination) of the aforementioned theses. Since the ANC s dominance could have implications for South Africa s democracy, it is recommended that further, more in depth inquiries of the existing data on South African voter behaviour in the country s elections be undertaken. 25

26 Bibliography Afrobarometer. Online Data Analysis Tool. [Accessed: 22 May 2015]. Booysen, S The Ballot and the Brick: Enduring Under Duress. In Local Elections in South Africa: Parties, People, Politics. S. Booysen, Ed. Bloemfontein, South Africa: Sun Media: Bratton, M., Bhavani, R. and Chen, Tse-Hsin Voting intentions in Africa: ethnic, economic or partisan? Commonwealth and Comparative Politics. 50(1): Ferree, K Explaining South Africa s Racial Census. Journal of Politics. 68(4): Habib, A. and Naidu, S Race, Class and Voting Patterns in South Africa s Electoral System: Ten Years of Democracy. Africa Development. XXXI(3): Human Sciences Research Council Survey on South African voter participation in elections. Available: [Accessed: 22 May 2015]. Langfield, D Managing Factionalism and Overcoming Marginalization: South Africa under ANC Dominance. Available: a%20michmay2014.pdf [Accessed: 22 May 2015]. 26

27 Mattes, R. and Piombo, J Opposition Parties and the Voters in South Africa s General Election of Democratization. 8(3): Mattes, R., Taylor, H. and Africa, C Judgement and choice in the 1999 South African election. Politikon: South African Journal of Political Studies. 26(2): Naidu, S Voting behaviour and attitudes in a post-apartheid South Africa. In South African Social Attitudes Changing Times, Diverse Voices. U. Pillay, B. Roberts, and S. Rule. Cape Town, South Africa: Human Sciences Research Council: Ramutsindela, M Identity and voting trends in South Africa. In Public Attitudes in Contemporary South Africa: Insights from an HSRC survey. Cape Town, South Africa: Human Sciences Research Council: Suttner, R Party Dominance Theory : of What Value? Politikon: South African Journal of Political Studies. 33(3) Trimble, J. E. and Dickson, R Ethnic Identity. Available: [Accessed: 22 May 2015]. 27