The Hot, the Cold and the Lukewarm: Exploring the Depth and Determinants of Public Closeness to the African National Congress

ABSTRACT

South African politics has been dominated by the African National Congress (ANC) since the country’s successful transition to democracy. But what is the degree of ANC support in South Africa at the micro-level? This paper will examine individual party closeness to the ANC for the period 2008–2014. We will use data from the nationally representative South African Social Attitudes Survey. Multivariate tests will be employed to identify the determinants of party closeness at the start of this period and at the end. Of those factors that our models revealed as statistically significant predictors of ANC support, the strongest was race group. Class identity, by contrast, was not strongly correlated with support. Interestingly, perceptions of electoral instrumentality were positively associated with higher levels of support for the ruling party. The findings of this paper suggest the need for future research on party closeness in the country and we conclude by outlining potential avenues of study.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Steven Lawrence Gordon http://orcid.org/0000-0002-6393-2118

Notes

1 Consider an interesting example of how political partisans may be persuaded to switch their vote during an election. In the 1972 general elections in the United States, the candidate for the Democrat Party was George McGovern. A significant segment of the Democratic Party had a poor opinion of McGovern and major players in the party, like the American Federation of Labour and Congress of Industrial Organizations, refused to support him. Some within the Democratic Party even organised ‘Democrats for Nixon’ and said they planned to vote for McGovern’s opponent. During the 1972 general election many people who identified as Democrats did not vote for McGovern (for a more detailed examination of voting behaviour in that election, see Miller et al. 1976).

2 Research has shown that an individual’s reaction to a wide variety of information is mediated by their prior beliefs through different biases. In an interesting study, partisanship was found to affect how people perceived the utility of government schools or clinics in a new African democracy. In Uganda Carlson (2016) found that supporters of the incumbent president systematically overestimated what they have received from the government, while opposition supporters systematically underestimated. In another example, research in Brazil by Winters and Weitz-Shapiro (2015) has shown that party partisanship increases tolerance of corruption cases that involve their own party. These studies demonstrate the power of partisan schema to influence judgements about governance in new democracies.

3 During the SASAS 2014 round, which took place in October 2014, respondents were asked who they had voted for in the 2014 national elections which had occurred in May of that year. Of those who felt close to the ANC and were age eligible to vote, 18% said that they had not voted for the ruling party in the 2014 elections. This suggests that a significant number of potential voters who felt attached to the ANC did not vote for that party in the 2014 general elections – this finding is even more significant given that people tend to over report their turnout during elections.

4 We looked at which groups were most likely to refuse to answer this question. Members of the Indian minority were more inclined to refuse to answer a question on party closeness in both 2008 and 2014. Black African respondents in KwaZulu-Natal and the Free State were also more likely to refuse to answer the party closeness question than their counterparts in other provinces.

5 The 2014 result showed that 45% of the adult public said they were lower class, a quarter reported that they were working class, 26% indicated that they were middle class and a 20th that they were upper class. The 2008 results were more or less similar although a smaller share identified with the upper class or upper middle class in that year.

6 The models help explain differences in the partisan loyalties to the ruling party in the different provinces. In both 2008 and 2014, provincial residence was a statistically significant determinant of attachment to the ANC. If provincial was removed then the share of variance explained by the model would drop by about 2% in each model. This suggests that the dynamics of each province explain some share of variance in ANC partisan loyalty in South Africa but that the explanatory power of the provincial dynamics should not be overstated.

7 We used a multinomial regression analysis to identify the determinants of our subjective class variable. We found that 17% of the variance in subjective class status selection could be explained by labour market status, geographic location, marital status, asset ownership and years of education. This result suggests that there are other factors that are driving class identification in South Africa. Burger et al. (2014) contend that class identification in the country is complicated by perceived tensions between racial and cultural identity and persistent confusion around what terms like ‘middle class’ represents.

8 In SASAS respondents were asked a series of questions about their household’s standard of living (i.e. egocentric conditions). Respondents were asked if the following were inadequate, just adequate or more than adequate for their household’s needs: (i) housing, (ii) access to transport, (iii) health care, (iv) children’s schooling, (v) clothing and (vi) the amount of food your household had over the past month. At the bivariate level, egocentric conditions were associated with partisanship with the ANC. Those who did not feel close to the ruling party had worse self-reported conditions than those who felt very close or quite close to the ANC. However, when we include these variables in our multinomial model, egocentric conditions were not found to be statistically significant determinants of our dependent.

9 Existing research suggests that objective macroeconomic changes impact on the relationship between partisan identity and general evaluations of the economy. In his analysis, Stanig (2013) found that during economic downturns, citizens of different partisan identifications were more likely to agree that their nation’s general economic situation was grim. Conversely, general economic evaluations are polarised by political affiliations during economic recoveries (also see Lewis-Beck, Nadeau, and Elias 2008).