What do you mean by pollster bias in an election? I came across harvards election poll data science question. I which they ask:
Is there a pollster bias in presidential election polls? What exactly is a pollster bias? can some one explain it to me in layman terms?
 A: Usual estimators in sampling are unbiased, but this only holds under certain conditions that are unlikely to be met in case of election polls (no non-response, probabilistic sampling, etc.).
If estimators used by pollsters were truly unbiased, errors would be equally distributed around 0, but it's clearly not what happens: see for example http://fivethirtyeight.blogs.nytimes.com/2012/11/10/which-polls-fared-best-and-worst-in-the-2012-presidential-race/
I guess the models and methods used by pollsters tend to be re-used from one election to another, which means "pollster bias" can somehow be
measured and predicted (which is also probably why Nate Silver used a pollster rating to compute his famous election predictions: http://fivethirtyeight.com/features/how-fivethirtyeight-calculates-pollster-ratings/)
In layman terms, I would sum this up as : Part of the estimation depends on mahtematical models. Models used are specific to each pollster, and are built upon view or experience. Modeling errors lead to biases, and pollsters are likely to re-use models they've already built for previous elections.
