Two major news agencies have released their pre-poll surveys ahead of Tamil Nadu’s assembly elections this year. While the tradition of pre-poll analyses has been long present, phones in the EC’s complaint cell have been ringing for the past few days. Many are lodging their protest against the opinion polls fearing it might sway the electorate.
They're also highly localised, considering that the polls are statewide and the projections are seat-wise.
Dinamalar-News 7’s poll survey released on Sunday has claimed to have covered 2.34 lakh respondents with 1,000 each per constituency in the state.
But they weren’t going purely by the numbers either. Factors such as party history, candidate strength were also taken into account, adding to the "data-driven" results in the process.
“First of all to say they’re doing an opinion poll without a random sampling – the success of a poll depends on randomization. You have to pick a sample in equal proportion to get a full representation of the constituency. You can’t just randomly distribute fliers,” says a prominent analyst - and this is exactly what News 7 did, conduct the survey like an online poll.
News7 admits to the survey not being a scientific one. “We did not have a clear caste, gender or age breakup. We sent it through our News 7, Dinamalar reporters and college students. Our questionnaire had just one question – Which party do you support,” says Thillai, Chief Editor, News 7. “We were particular that they should not go to bus stands and railway stations to avoid a non-representative population. We have tried to ascertain the pulse of the state.”
“These polls have not followed the standard methodology in terms of sampling. When you don’t have a scientific method, how do you interpret the data?" another observer pointed out.
In Pennagram, the survey claimed that Ramadoss would be in third place, an analyst said, quoting Dinamalar News7’s poll. “Putting in perspective Ramadoss’ caste stronghold in the area, it seems that the sample is very unscientific and picked only a group that was far from representative. In another case where a constituency such as Tirupur has a huge migrant population that may not conform to the traditional voting pattern- it is not accounted for if the respondent is a permanent resident.”
Handing out a batch of questionnaires to a coterie of people is also likely to throw up flawed numbers. “They won’t vote as individuals, what one person chooses to write will influence the others,” the analyst adds.
An opinion poll commissioned by Thanthi TV in association with Krish Infomedia had a sample population of around 150 respondents per constituency (state-wide respondents of about 35,600).
Thanthi followed the random sampling process and tried to make it as representative as possible. “We went out in the field with a 4-part questionnaire, provided respondents with sufficient reaction time and did not factor in mob reactions in our data,” says Arun Krishnamurthy, psephologist who carried out the survey. Krishnamurthy also clarifies that the survey is not necessarily a quest to find out the winning or losing candidates, but to track the trend of voting.
Transparency is important at a time when there are multiple agencies that are beginning to do these surveys. “What is the credibility of these surveys? The motive is naturally questionable,” says a senior journalist. In the case of the Dinamalar-News7 survey, in Tiruchendur, it was reported that the covers carrying the questionnaires were not sealed and so they were rejected and a press survey was taken instead.
Analysts also called for the EC to ask agencies to make their poll survey methodology open besides making their caste gender and age demographic breakup transparent. “While a certain section of the electorate is bound to take this with a pinch of salt, another section could be easily swayed,” adds the journalist.
Krishnamurthy is open to the prospect. “We are happy to reveal our methodology and data to the EC.”
News, views and interviews- Follow our election coverage.
Click TN Election Special
Click Kerala Election Special