Votes were cast for the Delhi assembly election on Saturday and the fate of the candidates was decided in the electronic voting machine (EVM) units. According to most of the exit polls broadcast after voting, the AAP would continue to rule in Delhi. The ruling party might get 50 seats out of 70.
The BJP campaigned very aggressively in Delhi, and yet it seems to get only 14 seats. The INC may barely get just one seat.
But how reliable are the results of exit polls? According to Delhi BJP chief Manoj Tiwari, exit polls will prove to be wrong. The BJP can get 48 seats, he says.
Looking at the previous record of exit polls, none could predict the performance of the AAP in either 2013 or 2015. In 2013, some exit polls did not show anyone getting a majority. The BJP was shown getting a majority in some exit polls. No one got an absolute majority when the result was out.
No party attained 35. Although the BJP emerged as the largest party, the AAP, with 28 seats, formed a government that lasted for a mere 49 days.
In the four 2013 exit polls, India Today-ORG, Times Now-CVoter, ABP Nielsen and Chanakya polls were but quite right about the AAP’s performance. The private survey agencies had estimated that the AAP could win around 31 seats in Delhi. In the 2013 exit poll, the INC was seen getting 15 to 24 seats, but it managed just eight seats.
In 2015, all exit polls showed the AAP getting a majority, but it got 67 seats, which no one had predicted. All exit polls gave 35 to 45 seats to the AAP, but it got 67 seats.
In most exit polls, BJP seats were seen crossing the double-digit figure. India TV-C voters had shown BJP getting 33 seats. But when the results were out, the BJP won only three seats. The INC could not open get even one.
A reason for the unreliability of exit polls is the small sample sizes. The India Today-Axis exit poll, for example, involved a declared sample size of 14,011 voters whereas there are 1.47 crore eligible voters in Delhi out of which roughly 61% (86,65,860 voters) have exercised their franchise this year. It would take a psephologist tremendous experience and an astonishing intuition to pick a representative sample of 14,000 odd voters from a total of nearly 87 lakh voters.
The other exit polls had smaller sample sizes. The ABP News-CVoter interviewed 11,839 voters. The IPSOS survey had a sample size of 7,321 respondents. Others did not declare their respective sample sizes in the broadcasts.