This weekend the Sunday Times stated that "The African National Congress will be returned to parliament with a greater majority than it achieved in the previous election, despite an internal split which has given birth to a new political party."

The basis for this claim was an opinion poll conducted by Plus 94 Research which, the newspaper reported, found that if "South Africans were to participate in elections now, an overwhelming 74% of decided voters would cast their ballot for the ANC - 5% more than in the 2004 election."

The wire services picked on this story - and generally the story has been reported as being good news for the ANC and extremely bad news for the Congress of the People. But the full results of the poll can be subjected to the opposite interpretation. If COPE realises its full potential the ANC could be in serious trouble electorally.

3500 homes were covered by the researchers, who conducted face-to-face interviews. The respondents were asked, in the voting intention questions, whether they had a "favourite political party...that is to say, one you would definitely vote for in the next general election?" 61% of respondents said yes, and 39% said no. Of those who expressed a definite preference 74% said they would vote for the ANC in 2009, 12% said they supported the DA, and only 0.2% said "Shikota".

It is on the basis of these results that the Sunday Times came to the conclusion that it did. However, there are two reasons for treading carefully before directly extrapolating these results to the elections next year.

The first reason for caution is that compared to other opinion polls 39% is a very high non-response rate. (For instance, the non-response rate for the poll conducted by Markdata earlier this year - see here - was 15,5%.) This could be due to the way the question was asked or inexperienced field work teams. Or, it could possibly reflect an increase in uncertainty within the electorate. It is not impossible that the breakaway has confused existing voter preferences.

Essentially, what the poll found was that 45% of respondents said they would definitely be voting for the ANC next year. The question is whether those who don't have a definite voter preference (yet) will vote in the same proportions for the different parties as those who currently do will.

In this regard the non-responses were higher for minorities - generally more supportive of the opposition - than for the black majority. 35,3% of black respondents said they were undecided, as compared to 45,8% of white, 60,3% of Coloured, and 50,5% of Indian respondents. The 84,2% of black "definites" who said they would be voting ANC, accords with the findings of other polls. But the 15,2% of white respondents who said they would vote ANC is not a credible reflection of voter intentions. Markdata found that of the white respondents who expressed a party preference 85,8% said they would vote DA, and only 1,6% said they would vote ANC. This is unlikely to have changed much in the past few months.

The second ground for caution, especially when it comes to judging the effect of COPE on ANC support, is that the poll was conducted in the last two weeks of October and the first half of November. This was as the split within the ANC was happening, and before the breakaway had acquired its final name and identity. As noted before, as long as the ANC remained united, the supporters of the different warring factions would back the same party.

Generally, opinion polls tend to pick up shifts in party support quite late. One shouldn't therefore expect the voter preference question of an opinion poll - especially one conducted at such an early stage of the process - to provide any kind of reliable indication of the possible electoral consequences of a split.

However, other questions were asked which provide a more accurate guide to COPE's prospects. To begin with the respondents were acutely aware of the divisions within the ruling party. 80% of all respondents, and 78,7% of black respondents, said that they believed the ANC as a party was divided.

The poll also picked up a high level of discontent at the way in which former President Thabo Mbeki was treated. 47,3% of respondents, and 46,4% of black respondents, thought the decision by the ANC NEC to peremptorily recall him was a "bad decision for the country." The answers as to why these respondents thought this suggest a high degree of loyalty to, and sympathy for, the former president. There was also an ethnic/provincial divide with 52,3% of respondents in the Eastern Cape thinking it a bad decision, as opposed to only 27,4% of respondents in KwaZulu-Natal.

In assessing COPE's prospects the crucial question that was asked was: "If the supporters of President Mbeki were to form a new political party, would you vote for this party in the next general election in 2009?"

A stunning 24,9% of all respondents replied "yes" and 75% said "no". Interestingly, 22,3% of male respondents said "yes" as opposed to 27,5% of female respondents. The poll indicated that the breakaway would cut deeply into the ANC's core black support base. 26% of black respondents said they would vote for the breakaway, as compared to 19,8% of whites, 20,1% of coloureds, and 30% of Indians. (See Table 1)

Again there were important provincial divides. 40,4% of respondents in the Eastern Cape said they would back a new party, but only 10,8% in KwaZulu Natal. (See Table 2)

These figures accord with Lawrence Schlemmer's estimate (see here) that the breakaway could draw as much as 20% of the vote from the ANC, something which would bring the ruling party's support down to close to 50%.

Click here to sign up to receive the Politicsweb daily headlines in your email inbox in the morning