The methodology section of India Today – CICERO poll on upcoming Delhi Assembly elections reports that the sample size was 4273 and the sample was spread across 210 polling stations covering all the 70 segments of the Delhi Assembly. In the recent General Election, The Election Commission of India has reported that there were about 1.27 crore voters in NCT Delhi. Assuming the sample is representative, the results from the poll using the sample size of 4273 from 1.27 crore voters will have a Margin Of Error (MoE) of 1.5%. MoE adds a range to the actual results by setting the lower limit at “Result (minus) MoE” and the upper limit at “Result (plus) MoE”. In this case, the poll reported that BJP is going to get 39% and AAP is going to get 36% of the total vote. What the poll actually saying is BJP is going to get anywhere between 37.5% (39% -1.5%) and 40.5% (39+1.5%) vote. Similarly, AAP is going to get anywhere between 34.5% (36% -1.5%) and 37.5% (36%+ 1.5%). In order to write-off a clear lead, the difference between the measured results must be more than twice the Margin of Error, in this case it is 3% (1.5% X 2). Here, the difference between the projected vote share of BJP and AAP is exactly 3%. That means, the difference could fall within in the MoE as the probability of both BJP and AAP getting 37.5% is same as BJP getting 39.5% and AAP getting 34.5% vote. As a result, the survey outcome is a statistical tie, which means that victory for BJP in Delhi elections is as likely as it is to AAP.
References:
http://indiatoday.intoday.in/story/bjp-modi-arvind-kejriwal-delhi-assembly-polls-india-today-cicero-poll/1/407623.html