Opinion Poll or Opinion Building?

An opinion poll by Cicero – India Today on upcoming Delhi elections was made public yesterday. The opinion poll predicted 34 to 40 seats for BJP and 25 to 31 seats for AAP. Congress has been predicted to shrink further to 3 – 5 seats while 2 seats might go to others.

The actual fate of Delhi would be decided by 1.25 crore voters from over 12000 polling stations. The results predicted by Cicero India Today are based on inputs received from 4459 people representing 210 polling stations from 70 constituencies.

Far from the raw data, details of the methodology used are not disclosed anywhere. The raw data can be disclosed without compromising with confidentiality of participants. The motive behind not disclosing the raw data would always be questioned till the same is made public.

Is the sample representative of population?

Though the number of participants from each assembly constituency is not disclosed, it translates to around 65 for each, presuming equal distribution. Each constituency has around 200 polling stations, while the total number of polling stations represented in this poll is 210, averaging to 3 per constituency. Elections in India have time and again proven importance of focusing on polling booth. Voters from polling booth often have inclination for a particular candidate. A sample of any 3 polling stations even if choosen randomly cannot give a clear idea about the whole constituency. Voters from 3 polling stations would be of no use in guessing the mood of a constituency consisting of 200 polling stations. This methodology may well be in practice in west but Indian voting pattern demands representation of each polling booth to have a fair idea.

Sampling method – does it matter?

The sampling method not being disclosed, we would not be sure whether it was a random or stratified sampling. Factors like age,  genders, religion, cast, socioeconomic status, educational background, profession, largely determine the inclination of voters, at least in India. Till the time we are sure of inclusion of these factors in the sample, we may not get the true representation. Broadly the people who agree to participate in such opinion polls, are less likely to go for voting while those who religiously go for voting often are not approached or don’t participate in opinion polls.

Reliability of data

People responsible for collecting the data are tasked to collect the response forms. I am not sure if there is any check to confirm the genuineness of data presented for analysis.

Statistical methodology?

I often call statistics as a tool to fool people. If you have some results in your mind, statistics offers you the leverage to get expected results. There are a number of methods each giving different results. Adding weight is a leverage that is often misused to skew the results.

I might sound skeptical putting all these questions. But then I have all the rights to do so, unless the motive behind not disclosing raw data is made clear.

What could the motive be?

Let’s go back to 2014. Various opinion polls predicted AAP to get 2-8 seats in Delhi Assembly elections 2014. This surely had an influence on a large number of voters who did not want to see congress in power again. They were made to believe that AAP is nowhere in the race. A good number of people would have voted for BJP to keep congress out  of power. AAP registered victory in 28 assemblies in spite of such opinion polls being telecasted on major news channels just before the polling date. Imagine what the results could have been if these opinion polls were not telecasted.

In 2014 the motive was to make people believe that AAP is not a serious contender. The same can’t be said this time. The ground reality cannot be ridiculed.

So this time the opinion polls are giving a message that BJP is about to form government. If AAP does a little better, there would be hung assembly again. So if you want a stable government with full majority in Delhi, vote for BJP.

When political parties can spend thousands of crores on elections they would not shy away from cooking up such opinion building polls.

If it sounds too skeptical I would like to remind you of the sting operation carried out last year by News Express. 6 different reputed agencies involved in the business of opinion polls were caught on sting camera agreeing to forge results against cash.

I strongly belive that the opinion polls are opinion building exercises with a clear motive. Having raw data and a clear disclosure of methodology used may be helpful in winning confidence of people.

Comments are welcome.