Research firms and pollsters are using a variety of methods these days. At Ipsos, social media, big data and machine learning techniques have become mainstream in some parts of our activities. The art is to combine these approaches with surveys in ways which are effective, rather than trying to force a certain solution if we believe it is not adequate and ready.
While electoral polls are not perfect predictors, they remain an excellent method, based on sound theory. Claiming that one could replace them, for example by simplistic methods derived from social media, is just a fallacy.
At the same time, the polling industry should not be complacent; it needs to continue to invest in scientific progress and rigorous practice. The focus needs to be on implementing the right methods, combined with a responsible attitude toward citizens and media.