Polling gone wrong: Windows 8 “disinterest” sample heavily skewed

Nov 16, 2012
46

A poll showing supposed Windows 8 disinterest earlier this week has been cut into pieces by a fellow by the name of Ed Bott. Mister Bott takes what's reported by AVAST in a report of a collection of their users and takes it apart systematically, starting with the sample size and demographic of those polled. As it turns out, things aren't as objective as AVAST makes it all out to be.

The poll was conducted on October 25th, one day before Windows 8 was launched. The poll was sent out to 1.6 million users of AVAST's free PC anti-virus product, and they got back a total of 350k responses - of those, 135,329 of them were Windows users from the USA. Bott made it clear this week that "people who have deliberately chosen a free antivirus program, in other words, cheapskates" are probably not the best set of people to be basing the opinions of the general population on.

The key here comes in with the amount of people polled that actually intend on purchasing a new PC soon - just 16% of the total. This makes a lot of sense considering the cross-section of users polled, those being people who, again, have opted for a free service. The number break-down from there goes like this:

12% plan on purchasing a Mac
30% plan on purchasing an iPad
68% plan on purchasing a new Windows 8 PC

These numbers are all part of the 16% of users polled that plan on getting a new computer at all, while the much more massive set, 84% of all users polled, said they didn't plan on buying anything at all. So here's what that really means, for the total polled:

2% plan on purchasing a Mac
5% plan on purchasing an iPad
11% plan on purchasing a new Windows 8 PC
84% have no plan of purchasing anything

Meanwhile Bott also notes that it might not make any sense that the division between those planning on picking something up equaled out to be 110% in the end - we're guessing, on the other hand, that some people answered with more than one option. Like purchasing an iPad and a Mac, or an iPad and a new Windows 8 PC. Still, the end result is this: don't trust a poll without checking the sample size.

Also note this fun find from another AVAST poll which got the Presidential Election wrong (just about as wrong as it can get, in fact), saying Romney would win with 290 electoral votes in the end:

Various factors should be considered. For example, as all those that AVAST polled were users of the firm’s free software in the U.S., there could be economic correlations. Or, the general demographics of avast! users could skew results one way or another.

Enough said - and said by AVAST themselves. Keep this all in mind the next time anyone - even us here at SlashGear - report findings from pollsters. Keep your mind tuned!


Must Read Bits & Bytes