|
An extensive nation-wide survey conducted by the Human Sciences Research Council (HSRC), called the South African Social Attitudes Survey (SASAS), has predicted that the ANC will secure just over a two-thirds majority in next year’s parliamentary election. “The HSRC projection is that the ANC will win 67,8% of the votes next year”, said Dr Udesh Pillay, Executive Director of Surveys, Analyses, Modelling and Mapping (SAMM) in the HSRC. This is fractionally above the 66,7% required for a two-thirds majority. The ANC won 62,6% of the vote in South Africa’s first democratic election in 1994, and 66,4% in the last election in 1999. The Democratic Alliance was predicted to come second, with 10,5% of the vote, followed by the New National Party (8,7%), the Inkatha Freedom Party (7,1%), and the UDM (1,7%). Of the smaller parties, the Pan Africanist Congress (PAC) and the Azanian People’s Organisation (Azapo) are predicted to poll 0,51% and 0,49% respectively. The new Independent Democrats of Patricia de Lille are expected to win 0,44% of the votes. The predicted result would give the ANC 271 parliamentary seats, five more than the 266 it won in 1999. The DA would go up from 38 seats to 42, the NNP picks up 7 seats taking it from 28 to 35 seats, the IFP loses 6 seats and drops from 34 to 28 seats, while the UDM’s representation would be halved, from 14 seats in 1999 to 7 in 2004. Of the smaller parties, the African Christian Democratic Party (ACDP) would drop from 6 seats to 2 and the PAC would drop from 3 to 2. The new Independent Democrats are predicted to win 2 seats. The prediction is based on a massive new survey of South African social attitudes conducted by the HSRC, with a sample of 7 501 respondents of voting age drawn from a random selection across the country. “This is one of the largest sample bases of any attitudinal survey conducted in South Africa, and three times as large as the typical samples for commercial opinion surveys”, said Dr Mark Orkin, the HSRC’s CEO. Nearly a third of the respondents (32,6%) did not indicate which party they will vote for in next year’s election. Early preferences were statistically imputed for this group, using a well-known statistical technique called discriminant analysis. This technique assigns undeclared voters to a political party by establishing predictors of party choice among those who had declared their voting intentions, and applying them to those whose voting intentions were not declared. An additional 7,8% of respondents said that they would not vote in next year’s election, and were not included in the analysis. The results also predict that the ANC will retain control of the Western Cape, with the support of the NNP. In KwaZulu-Natal, the DA-IFP alliance falls just short of a majority over the ANC. The ANC has overwhelming support in all other provinces. The survey’s national results were presented to politicians and the media in parliament today by an HSRC team led by Dr Orkin. Provincial results, including voting intentions in various areas of each province, will be presented later this month to the provincial governments. Dr Pillay noted that “The survey was completed between August and October this year. Clearly the fact that a third of voters are undecided at this stage means that a lot of votes are still potentially up for grabs. Surveys always contain a number of respondents who decline to answer. “We have tried to overcome this by statistically imputing a vote to respondents who did not say or did not know how they would vote. “The breakdowns we have – nationally, provincially, and with divisions by race, gender and age group – will help parties focus on their election challenges and campaign strategies.” The various breakdowns provided by the HSRC revealed that the ANC was the most popular party among all age groups, with its support highest among the 25-34 age bracket. The ANC and the IFP have an overwhelming majority of black supporters (94% and 98% respectively), while the DA support is largely white (75%) and the largest proportion of NNP support is coloured (42%). When voting intentions were analysed by gender, the supporters of the ANC and DA were roughly equally divided between men and women, while women were dominant among supporters of the IFP and the NNP. The South African Social Attitudes Survey is a pioneering survey in South Africa, and one of the few in the world, designed so that repeat surveys can be done, returning to the same, or closely similar, cross-section of respondents each year to track shifts in attitudes, locally, regionally and nationally. “Voting intentions are one part of a substantial survey programme tracking attitudes on a number of issues, including politics. We will monitor party preferences in each annual survey, and political parties and the public will be able to follow the shifts in attitudes as we record them”, said Dr Orkin. Early next year the HSRC will release results, distilled from the same survey, of responses on topics such as household poverty, national identities and patriotism, racism and xenophobia, crime, gender issues, moral issues such as abortion and the death penalty, interpersonal violence and inter-generational attitudes. Responses will be reported by province, gender, race, age group and other demographic variables. Dr Orkin said the new format of the HSRC’s attitudinal survey (SASAS) would provide South Africa with a unique long-term account of shifts in culture, social attitudes and values among the country’s population. “SASAS 2003 is the start of these annual measurements on the big political, social and moral issues of the day.” Dr Stephen Rule, Director of Surveys at the HSRC, said the HSRC had previously conducted annual attitudinal surveys, including surveys of voting intentions. These had been based, however, on a once-off series of questions, only occasionally repeated in subsequent years. The new SASAS series would not only be more accurate, as it was based on a much larger sample, but would be on a regular annual basis and provide detailed breakdowns. “The survey uses a master sample based on 1 000 census enumerator areas spread across all provinces. The sample has been stratified so that we can analyse and report on the responses nationally or regionally, by race and by geographic sub-type (urban, informal, rural or tribal). It is a very powerful research tool,” Dr Rule said. Dr Rule stressed that there are important differences between an election prediction based on an attitude survey, and the actual result of an election several months later. Variations may arise from whether voters register in advance, which voters actually turn out on the day, and important events that may occur between the poll and the election. These variations can differ appreciably for different parties. Full results and analysis of all modules of the survey, including political preferences, will be published by mid-2004. Full presentation and tables
|
