ICM, UKIP and the BBC

Nigel Farage is still sticking to his guns following last night leaders' debate and that audience, going after both the BBC and opinion pollsters ICM:

Sometimes these things go wrong, sometimes you get groups who apply to be on programmes who perhaps aren't as truthful on their applications as they could be.

In this case, the BBC gave the job to a polling company called ICM who are famous for getting everything about Ukip wrong and that I think was the mistake.

Am I going to make a complaint? I've got an election to fight. What matters isn't the 200 people in the room, what really matters are the millions of people watching on television.

Is Nigel right that there's a problem with ICM and UKIP? 

Well, in a fascinating piece about the different results achieved by phone and online polling at UK Polling Report by YouGov's Anthony Wells, it appears he could have a point:

Only one polling company gives UKIP a single figure average this year - ICM. 

And that's not the only result that seems out-of-step with other pollsters. The same post at UK Polling Report also features a graph monitoring the average Conservative lead in 2015 and, again, ICM is at the furthest extreme from the average:

Is there something about ICM's methodology which produces results that - at least in these two respects - seem very much like statistical outliers?

Could it be something to do with their reliance on telephone polling? Do people lie more when polled by telephone (I know I do!)? 

Could that help explain why the audience seemed so unbalanced? Could it be that it was unbalanced because of ICM's methodology?

And was the BBC aware of any of this when commissioning ICM?

All of this may well back up Nigel Farage, but it still doesn't explain the loud applause for the many attacks on David Cameron and the Conservatives or the whoops for the idea of Labour and the SNP getting together to kick the Tories out. ICM's polling doesn't seem at all biased against the Conservatives, does it?