The last two years have proved pretty tumultuous, with unforeseen political victories arising during the Brexit vote, the American election and the Italian referendum. In all cases, one of the reasons these victories were so unexpected was because exit polls suggested that voters had swayed in the opposite direction – or towards the widely predicted result. It came as a genuine shock when it was announced that Britain had voted to leave the European Union, Donald Trump had beaten Hillary Clinton to become President of the United States and that Italy had voted ‘No’ to constitutional reform, causing Prime Minister Matteo Renzi to step down.

It appears that researchers should not have relied so heavily on voters to accurately report their own preferences in recent political votes. Aside from undecided voters, there are also those who lie to disassociate themselves from a vote they believe will make them seem controversial or unpopular. Often, people give the answer they think the interviewer seeks, even if their opinion is completely different. Society has becoming much more complex and, given this complexity, individuals in some cases are not able to qualify precisely and reliably what their real reactions are. As far as political issues are concerned, in several cases they simply don’t want to tell the truth. This is becoming more relevant, given populist parties and other new and strange phenomena going on in politics, and goes some way in explaining why exit polls during events like Britain’s referendum and the American election did not accurately reflect the way many ultimately voted.

So, how do we make the data we collect about voters’ preferences more reliable? One suggestion is by implementing the use of biometrics. These are tests that interpret signals such as blood pressure, heart rate and microfacial expressions to measure participants’ emotional reactions to political ideas and stimulus, which indicate which way they would really vote. As complex as these tests sound, they only require small sample groups and are not invasive, completed on wearable devices. Through the use these devices to identify someone’s “emotional arousals” – or the physical and biological changes that occur when people answer questions – exit pollsters would not have to rely on the self-selecting accuracy of people’s political preferences. This is likely to address the problem of someone lying to disassociate themselves from a controversial vote.

Although the devices are non-invasive, we must acknowledge that some people might not want to be approached by a stranger standing outside a polling station, asking them to put on a t-shirt full of biometric technology for the purposes of market research. This is why market researchers must be trained with specific techniques to put participants at ease about the process and the use of the results. They must also understand that not everybody wishes to be involved, for privacy concerns. However, this is a small cost to pay for a more reliable output.

Biomarketing as a science is far broader in scope than neuromarketing, which reveals and collects brain signals through electroencephalograms. In cases where biometrics are measured, all parameters come from the body: blood pressure, heart rate, breathing rhythm, sweating and microfacial expressions. The tests simply undermine our natural defences such as shame and reserve, to reveal the true secrets of the mind. These tests are also far more efficient than exit polls, which require tens of thousands of participants. Results can be obtained from small sample groups of roughly 200 – 300 individuals.

The method is supported by a wealth of academic literature on emotional arousal which reinforces its robust and scientific value. Biometrics provide a tried and tested means of predicting which way people will vote, and of understanding the drivers behind why people gravitate towards, or away from, individual political ideas and parties. Not only could using these methods provide a clearer indication of the result of upcoming political votes, but political parties themselves could use the findings to better design manifestos that serve the masses, by focusing in on individual reactions. It is clear that this technology has a place in the political arena and that those who are early adopters of it are likely to reap the rewards of a better understanding of voters’ preferences.

Giuliano Noci is a professor of strategy and marketing at Politecnico di Milano School of Management, Italy.