Quote:
Originally Posted by Sephiroth
The mathematics of a static situation (30 people in a room) seems to me to have little/no relevance to an survey/poll conducted with a population sample.
The survey companies have a very large pool of people who represent the demographic shape of the country and then use a proportion of that pool, I believe picked at random, to answer whatever the survey is about. Any shortfall in the demographic profile arising from the random selection is adjusted by the "normalisation" process when the surveys are in.
Whereas the birthday probability calculation is entirely mathematical, survey/polls have a significant subjective element applied to them.
Hugh's comparison is of very limited value, though the example he gave is interesting.
|
This ^^
Although this part of the conversation was triggered by my indifference specifically to YouGov.
They don’t randomly ask 2000 or 1000 people, they ask 2000 or 1000 people “specifically chosen” from a database of “recruited” registered YouGov respondents.
Quote:
|
Panel members are recruited from a host of different sources, including via standard advertising, and strategic partnerships with a broad range of websites.
|
Nothing “random” about it, so you can throw that Gaussian Curve out the window.
I’m not saying it is……but you could quite easily skew a result by selecting your respondents.
You may read the attached and think that’s all fine, and if properly managed and overseen it may be.
https://yougov.co.uk/about/panel-methodology/
But……importantly…..it’s not random. It’s like getting a lucky dip in the lottery, those numbers are computer generated and anything computer generated is not random.