2009/6/25 Matthew Paul Thomas <m...@canonical.com>: >>... >> If just 1% of Ubuntu users tick the box, that gives us enough data to >> improve Ubuntu by justifying our decisions with evidence. >>... > > The absolute size of a sample is more important, statistically, than its > relative size. In other words, 1136581 popcon submissions is a large > enough sample regardless of how many Ubuntu users there are in total. > What is more important now is reducing bias -- bias towards current > users against potential users, towards users who fiddle with settings > against users who don't, and so on.
What if the install program randomly invited 1% of users to join popcon? Something like: "We want to make sure that Ubuntu suits your purposes. To do this it would help us to know what software you use. Do you agree to automatically provide this information to Ubuntu? We will not store personally-identifying information as we only use this information to determine the popularity of different software packages supported by Ubuntu. Only 1% of users are invited to this survey" The obvious disadvantage would be that it introduces non-determinism into the mix, which may mess up some forms of scripting and users that insist on following step by step instructions exactly. The obvious advantage is that it gives us sample as unbiased as possible without mandatory popcon, while still adding only 0.01 mouse-clicks to the average install. Perhaps the submissions to popcon that come from the specifically invited people could be marked as such so that we could keep the relatively unbiased sample pure, and compare it to the sample of people who actively hunted popcon down. -- John C. McCabe-Dansted PhD Student University of Western Australia -- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
