I would caution you to take this survey with a heavy dose of salt.
Several flaws as a result of the survey sampling frame.
Each company was measured through telephonic and web-survey methods at a statistically reliable sample of at least 270 respondents per company.
For telephonic interviews, usually it is the more affluent who have access to telephones. Further, the definition of telephonic needs to be made clear because it could mean only landlines, which is what I am guessing it was. These days, more households have cell phones than landlines, with landlines being acquired by only those who can afford to pay R150/mth even before they lift the handset up to make a tollfree call. The number of Waya-Waya (R120/annum line) owners is negligable and dwindling, since Telkom has since discontinued it if I am not mistaken.
Secondly, web-survey. The first question is, how many were web surveys of each of the 270 and how many were telephonic?
Assuming that web-surveys were in the majority since it is the easier and more convinient method, once again the bias would go towards the brand that appeals more to the affluent and well informed.
It is therefore not a surprise that iPhone and Samsung (Galaxy brand no doubt) were the more satisfied.
The statistics of cell phone ownership in South Africa are more in favor of BlackBerry and Nokia and most owners are in the lower LSMs. But, an analysis of the sampling methods seems to indicate that these were the same groups that were left out of the survey.
Naturally the bias would be in favor of Apple and Samsung as current BlackBerry and Nokia owners in higher LSMs would aspire to purchase the more affluent brands and would be disatisfied with beleaguered brands more due to perception than reality. In the United States for instances, Nokia Lumia owners were reporting satisfaction rates that were on par or higher than iPhone owners.
Link
The survey is only statistically representative of people with telephones (whatever that actually means) and those with access to the Internet. An entire population was left out of the study.
This study therefore, cannot be generalised to the entire South African population.
P.S. It would also be interesting to find out where and how they got their telephone numbers and web-survey respondents. In the research I quoted above, they seem to have used a quota system, taking representative numbers of each brand from the different mobile networks where the brands had a presence.
