View Single Post
Old 31-05-2017, 22:44   #856
1andrew1
cf.mega poster
 
Join Date: Dec 2013
Posts: 14,234
1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze
1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze1andrew1 is cast in bronze
Re: June 8th General Election

Quote:
Originally Posted by Damien View Post
I don't understand how YouGov's poll can be right? If it is then we really don't know everything about how voters will behave. Part of Labour's resurgence seems to be Tory missteps and the return to two party politics but even so?
Quote:
Financial Times
What YouGov produced is not a poll in itself, but a model based on polling data, a projection of how seats would fall based on current voting projections, rather than a forecast of what will happen on election day.
A conventional poll attempts to create a representative sample of how the country intends to vote. By contrast, the YouGov model combines survey data from the last election, census data and current polling to try to identify types of voters, assess their number, work out how likely they are to vote and who they intend to vote for...
Where the technique can still go awry — as traditional methods can — is when the underlying data has what statisticians call “in cell” bias. For example — and without knowing what the precise details of the YouGov model are — if there are large numbers of people in the sample who are political junkies, but this is not a category in the model, then the results will be skewed.
https://www.ft.com/content/39b6628c-...7-59b4dd6296b8
1andrew1 is offline