I'm not convinced this is the question the article is talking about. It says, "One in five black people interviewed gave the regime ... a positive rating - a result which analysts attributed to crime and unemployment." And later: "... as time passes you tend to forget the negative things and emphasise the things that you had then and don't have now, such as law and order and jobs. Apartheid was a harsh, repressive, but seemingly efficient government which made the trains run on time." Is there a question in the survey that directly asks about how efficiently the government was run?

@DavidRicherby: I understand your point - I concede it is possible there was a second closely related question with very, very different results, and it was in the 2002 report, but was dropped in later reports, despite being the most important result for the media. I think it unlikely though. Finding the 2002 report would resolve this. [Edit: Oh, just read the comment that HAS the 2002 report. Will read it now!]

@elmer007: Thanks for pointing out the bit that needed more explanation. Updated.

The article also closes by saying, "The apparent nostalgia for apartheid emerged when people were asked to compare governments." The question you mention doesn't compare governments. It compares the person's living conditions.

@Oddthinking Q53 looks potentially relevant.

@DavidRicherby: After reading all the questions in Klanomath's link, I think it is clear there was no direct question about "better run". There were several questions about specific aspects, and the very raw data in the Codebook would suggest the responses were about the same (e.g. 30%, give or take a big margin, rating the apartheid regime better).

@jpmc26: The issue of giving an absolute positive rating is separate from the relative "is it better now" issue. I think the original Guardian article was rather confusing (and confused) about this. Are you happy with the wording of the answer now?

@Oddthinking I was considering the same possibility until I saw Q53. It has 4 parts (A through D). In part A, 57.4% indicate apartheid was at least as good at law enforcement (40.6% say apartheid was better). In B, 54% say the new government is not better at delivering services (34.1% say apartheid was better). In part C, over 70% of respondents indicate the new government is at least as corrupt (52.6% say apartheid was less corrupt). 60% say the new government is less or equally trustworthy than apartheid (34.2% say the new gov. is less trustworthy than apartheid).

@Oddthinking If we take the article at face value, it also seems like they were looking at break downs from a finer level of detail that this report gives us (e.g., the "one in five black people" giving a "positive" rating). They appear to have spoken to people who analyzed the raw data directly. It also of course depends strongly on what a "positive rating" is considered; it might refer to some aggregate results on a few questions. So the claim is unclear given the data we have, but I don't see enough evidence to simply dismiss the claim as confused.

@jpmc26 The Guardian article makes several different claims. The question used to repeat a couple of them, but I edited it to focus it on one. (Before I thought to actually respond - which was triggered when I noticed that the article was almost 15 years old, so the OP was probably looking in the wrong spot.)

Whether the Guardian actually spoke to the pollsters or, as I suspect, based their article on a press release, isn't important - but they certainly were looking at more than just one result.

I hope I don't come across as dismissing the claim as confused, but go slightly deeper and effectively say "It is more complicated." The questions asked were nuanced, and shouldn't be mixed together.

So in answer to the second part of my query: is 'the 20% of black respondents answered positively' too convoluted to state as a fact?