My Amazing Internet design company claim "most people have their screens set to 800x600" Should I believe them at face value? Poll It's all about size...
Your poll is answered by the people that choose to answer it. Those people are a subset of the people who have the opportunity to answer it; i.e the people who read your LJ - essentially the people who have friended you, although it's an open post so theoretically anyone with web access could stumble across it and respond. There's probably not enough of those (very likely zero, in fact) to influence the outcome though.
Assuming there's no bias in those who choose to respond vs those who don't (by no means a given, although I'd guess it's likely to be a fair assumption here), then your poll is representative of those who had the chance to answer it. It does not follow that it's representative of the population as a whole, nor of the people who will be using your company's website. In fact, it's quite unlikely that your friends list is representative of either of those two groups.
The Sun Phone-in polls suffer from the same problem - they are representative of Sun readers, not the population as a whole (in fact it's probably worse than that since the responders are limited to those that feel strongly enough about the issue to make the effort (more likely to be a source of bias in this case), and also they are quite likely to have had their opinion influenced by the Sun's editorial policy).
Accurate national polling relies on random selection of large numbers (1000+) of respondants, or something like YouGov which has the demographic data on the people it polls to enable it to select a representaive population and/or weight their responses accordingly
Sounded plausible, but a quick Google suggests it was actually Punt and Dennis on the Now Show (plagiarismindependant inspiration is possible, of course, but I'd expect Radio 4 audiences to spot that sort of thing)
Assuming there's no bias in those who choose to respond vs those who don't (by no means a given, although I'd guess it's likely to be a fair assumption here), then your poll is representative of those who had the chance to answer it. It does not follow that it's representative of the population as a whole, nor of the people who will be using your company's website. In fact, it's quite unlikely that your friends list is representative of either of those two groups.
The Sun Phone-in polls suffer from the same problem - they are representative of Sun readers, not the population as a whole (in fact it's probably worse than that since the responders are limited to those that feel strongly enough about the issue to make the effort (more likely to be a source of bias in this case), and also they are quite likely to have had their opinion influenced by the Sun's editorial policy).
Accurate national polling relies on random selection of large numbers (1000+) of respondants, or something like YouGov which has the demographic data on the people it polls to enable it to select a representaive population and/or weight their responses accordingly
Reply
Reply
Not that sample bias and misrepresentation of statistics is one of my (many) pet subjects. Oh, wait...
Reply
Reply
Reply
I listen to pretty much all the comedy Radio 4 transmit upload - so I can't tell the difference between two shows.
Reply
Reply
Leave a comment