British economists believe that they can use Google search data to predict future election results and perhaps even the outcome of May’s general election.
Professor Ronald MacDonald from Glasgow University has already used the same methodology to accurately forecast the vote in last year’s Scottish independence referendum.
Professor MacDonald and his team analysed Google search trends in combination with commercial polling data in order to successfully predict that the yes vote would peak at 45 per cent. The economy and currency expert added that the same big data approach is also being used to forecast the rise and fall of share prices.
According to the Guardian, MacDonald revealed that forecasting the general election was a distinct possibility.
“Potentially, it’s a very useful tool but it’s complementary to the main pollsters,” he said. “It would be feasible to do it on the day [of an election],” he said.
However, due to the increase in scale and complexity of a general election compared to a referendum, the tracking would need to consider a greater number of factors, including the increased number of political parties and voting options. MacDonald also added that their methodology is not able to show how one particular topic in isolation affects voting patterns.
That being said, their prediction of the Scottish referendum result was strikingly accurate. With five days to go until the voting occurred, the group’s modelling predicted the yes vote would lie between 44.8 and 45.2 per cent. In the end, the actual result was 44.7 per cent.
While big data has the potential to positively impact many industries ranging from healthcare to banking, analysts must ensure that their methods are viable and appropriate.
Following the Scottish referendum vote, for example, it was revealed that the Yes campaign believed that they would win by 54 per cent to 46 per cent, after they had used social media data that turned out to be erroneous.