The polls were showing 34% to Tories and 34% to Labour. How could they have got it so wrong?
The polls are done by telephone survey, and they ignore the "Do not call" list, because political surveys are one of the exemptions.
The problems are these.
1. They try to choose a random sample, but they can't, because the sample is, to some extent, self-selected. If you're busy, you aren't going to talk to them. If you're about to go out, or if you're already out, you won't talk to them. If you can't be bothered, you won't talk to them. So, the people that are willing to talk to a pollster, are the ones that are actually surveyed, and they won't be a random sample. The people they'll talk to, will be biased towards lonely people, and people with nothing better to do.
2. People like me. I signed up to the "do not call" list for a reason - namely, I don't want to be called by strangers who want a slice of my time for their purposes. I might, for example, be underneath my bike when the phone rings, and I'm really annoyed if I dropped the bike and rushed to the phone for no good reason. So I explain to these survey callers, that because they've called me despite my being on the "do not call" list, every question that they ask me, I'll answer with a wrong response, do you still want to talk to me?
Strangely, they do. The caller has a quota to meet, and they really don't care whether I answer correctly or misleadingly. So my wrong, and somewhat random, answers are added to their poll. I would hope that many other people do the same thng, and eventually, pollsters will realise that, even though they're allowed to call people on the "do not call" list, it's actually a very bad idea to do so.