I'm reading this article on the Telegraph website with a sense of bewilderment. 'Week of bungled plots boosts Labour in poll' is the headline, except it isn't actually true.
Sure, it's true in the sense of what newspapers do when they're bored, stupid, and/or want to play around with a narrative for a while. But it's not the actual truth.
What they've done is compare the ICM poll done over the last couple of days, which had the split at 40/30/18, with another ICM done for the paper at the beginning of December showing 40/29/19.
Except there was another ICM done over a week after that one - for the Guardian - which had the split at 40/31/18. It's one of the most important rules of polling to compare polls with the previous one from that particular organisation; not the last one for that particular paper. It's sloppy, wrong, and it makes the Telegraph look rather simple.
Local By-Election Review March 2017
7 hours ago