,
3 tweets,
1 min read
Read on Twitter
An absolutely must read and bookmark thread about how differences between polling methodologies affect Labour's vote share in polls (and also the Greens and the BP). Lots of little things that can add up to big swings between polling firms.
It also suggests to me that the best way to treat polls is as a series of directional indicators on different graphs. So if YouGov are consistent, then the direction of travel *between successive YouGov polls* is instructive. Similarly for Ipsos Mori, ComRes etc.
What doesn't work is to treat them all as lines on the *same* graph.
Note that often the media - and politicians - do just that. But that's because the poll results suit their agenda (it doesn't even have to be a political one - "get more clicks and readers" is also an agenda).
Note that often the media - and politicians - do just that. But that's because the poll results suit their agenda (it doesn't even have to be a political one - "get more clicks and readers" is also an agenda).
