Coming Up Wed 5:00 PM  AEDT
Coming Up Live in 
Hindi radio

Can Indian exit polls be trusted?

A polling officer marks the finger of Indian tribeswomen at a polling station at Pedda Shapur on the outskirts of Hyderabad. Source: Getty

Exit poll predictions ahead of Indian election results have sparked a debate about the credibility of such prognoses about the results.

Almost all exit polls have predicted a clear majority to BJP-led National Democratic Alliance (NDA) with over 300 seats in some cases in a house of 543 seats. Some surveying-agencies have gone as far as giving 350-360 seats to the ruling alliance.

BJP supporters have lauded these polls. Sydney-based Vipul Marwah of Friends of BJP, Australia, believes the exit-polls are a true reflection of actual results.

"The BJP is winning big in West Bengal. They have made a comeback in MP, Gujarat and Rajasthan after losing in state elections. They are not losing much in UP either. They are doing very good in Karnataka and Maharashtra. So, I have no reason to doubt these exit polls," Mr Marwah told SBS Hindi.

However, the opposition parties have rejected the polls as gossip and unreliable.

Manoj Sheoran of Indian Over Congress, Australia, says the digit polls are failing all over the world so we cannot trust them.

"From Brexit to New Zealand, from the US to Australia, the exit polls have been proven wrong time and again. In India, majority of the exit polls failed to predict the results since 1998. So, there is no reason for us to trust these results," says Mr Sheoran.

Exit polls stunned Australian voters last week after Liberal-National coalition managed a comfortable victory despite the surveys predicting a Labor win.

Economist Dr Vinod Mishra of Monash University, Melbourne agrees that the exit polls are failing to predict the electoral sentiments correctly.

Why are surveys failing?

Dr Mishra suggests two reasons behind this trend,  the first being the sample selection bias.

"The success of a poll or survey depends on how well the sample of surveyed individuals is representative of the voting population. The sampling techniques developed a while back, may not be able to pick a representative sample of the current voting population," he says.

"A lot is changing in terms of demography, societal aspirations, the outlets through which people consume news and form perceptions and in the nature of work and workplace. All these factors, if unaccounted for can lead to severe sample selection problems, where a skewed sample leads to skewed and inaccurate predictions."

Many have pointed to the sample size for surveys in India where more than 600 million voters have voted.

Economist and Co-author of Indianomix Ms Rupa Subramanya echoed Dr Mishra in her tweet.

"Take exit polls with a pinch of salt even if they're all saying the same thing. They could all be suffering from the same bias. Apart from giving you sample size which is least important, none of them share their methodology, and neither are surveys externally audited (sic)," Ms Subramanya tweeted.

However, some news TV channels in India have defended their surveys citing a large sample size.

"Polling for India Today – AxisMyIndia exit poll has concluded at a sample size of a whopping 742,187. Not just is the sample size massive; it's also scientifically selected. 95% accuracy record so far," said Mr Rahul Kanwal of Aaj Tak, the channel which has given a massive majority to NDA with more than 350 seats.

Truth vs lies

However, Dr Mishra questions the surveys at their core. He says people might not be telling the truth to a surveyor.

"What I mean is that when asked by a surveyor, they may not express their intentions, for fear of surveyor judging them negatively. Even though most surveys are anonymous, often when talking to another individual, we avoid saying something, because of which the listener may form a negative image of us. For example, a person may actually like Donald Trump, but given the stereotypical image of Trump supporters, he may not want to be associated with them, so he says he is planning to vote Hillary but does the exact opposite at ballot," said Dr Mishra.

India has seen many exit polls predictions going south in the recent times, so people are always in doubt. However, Dr Vinod Mishra believes, this time they might have caught the sentiments.

"It's possible they are accurate this time, in case they are inaccurate, I think the margin of error would be on the plus side for NDA rather than on the minus side, as opposed to last time. A possible overcorrection of the sampling technique; thereby sampling more NDA supporters than the population," he said.

Election results will come on 23 May 2019.

Follow SBS Hindi on Facebook for election results on 23 May.