The Real Story behind Exit Poll Results

The exit poll results have been declared on 16th May and there is a surprise consensus across states among agencies except for Tamil Nadu. While all incumbent governments are expected to be thrown out, Mamata in Bengal is expected to be back as CM denying a Congress-Left alliance to make a comeback. There is a bit of a suspense over Tamil Nadu with C-Voter predicting an AIADMK win while others predict Karunanidhi to make a comeback in line with historic trend.

The range of seat estimates is also big especially in case of Bengal with pollsters predicting a simple majority (News Nation 153 seats) to a landslide (India Today 243 seats). This is essentially because a seat by seat exit poll is not conducted and state wide vote share is extrapolated to arrive at seat share. Exit polls point to a direction which way the results will be and they have been successful in this aspect.

While bigger names like C-Voter and Today’s Chanakya have come out with their numbers, India Today has partnered with Axis My India, the same agency which got Bihar bang on but its results were not aired by CNN-IBN as they refused to believe them. Axis in the end had the last laugh.

Axis and Chanakya have come up with detailed studies – caste, sex, age, issues, popularity ratings etc. While both point towards the same winner in states the story they narrate is entirely different. Let us analyze the story behind these results.

1. Kerala

This South Indian state has always thrown out incumbent governments for past three decades and this time too trend was maintained. BJP entry made the contest three cornered and threatened the real Hindu party Left in the state. We had mentioned in our article in DailyO that Left is expected to sweep the state and it won’t be a close elections like last time. While Chanakya predicts vote share of 42%-37%-15%, Axis predicts 43%-35%-11% in favour of Left.

Overall, the BJP is gaining between 5-9% versus last assembly election with bulk of the damage to the Congress which appears to be losing between 9-11% points while the Left is losing 2-3% points. The scale of anti-incumbency against the Congress is to do with the series of corruption scandals, the Jisha case and the gradual shift of Hindu voters towards BJP.

Both polls show that BJP won more upper caste vote than Congress in Kerala. Chanakya poll also claims that BJP won nearly 30% of the Ezhava vote versus Left at 49%. This could be due to its alliance with Ezhava party BDJS. Further, BJP’s best performance came amongst the 18-25 age group. These results could be the first step towards consolidation of Hindu votes in favor of the BJP in Kerala and indicate a bright future for BJP in Kerala.  

Chanakya Theme – Left losing vote share across caste groups, but Congress losing more benefitting LDF

As per Chanakya, Left is seen losing Nair (-10%) and Ezhava (-16%) votes to BJP while Congress loss is bigger amongst Nairs (-19%) and Ezhavas (-14%). Congress is seen losing Muslim (-8%) and Christian (-9%) votes as well but this is not going to Left.

Axis Theme – Left losing vote share among Hindus, gaining amongst minorities because of fear of BJP

As per Axis, Left is losing Ezhava (-18%) and SC votes (-16%) which is going to BJP. But this is getting negated by increase in minority community votes from Congress (Muslims +8% and Christians (+12%) which is in line with our prediction for DailyO that even minorities are turning away from Congress. If these trends continue over the next few elections, the two major parties could be Left and the BJP. 

2. West Bengal

Mamata is seen getting 200+ seats in both polls however they differ significantly in terms of vote shares. While Chanakya gives vote share of 42%-33%-13%, Axis gives vote share of 51%-29%-10% in favour of TMC.

Chanakya Theme – Left losing vote share across caste groups, Mamata the gainer

Left-Congress combine is seen losing votes from even staunchest of supporters, Muslims (-13%) and STs (-10%) compared to LS polls. Obvious gainer is Mamata. BJP seen losing its upper caste and OBC voters obviously due to absence of Modi factor. Others at 12% is much higher than historic values and may definitely be an overstatement. .

Axis Theme – Left losing vote share among Hindus, Mamata gaining amongst minorities because of fear of BJP

Left-Congress combine is seen losing massive votes from Muslims (-22%), STs (-16%) and SCs (-9%). This is massive having implications for a Congress rout in its citadel Malda and Murshidabad. BJP loses Obvious gainer is Mamata. BJP seen losing half of its Hindu support to TMC. Others at 11% is very high.

As we had opined in our article in DailyO, Muslims hold the keys to victory in the state with influence in 102 seats (35%). A swing in favour of Mamata from the community ensures her victory. However, 45 seats and 70 seats as projected by these two agencies for Left-Congress combine is too low in our opinion as they have historically won 60 / 102 Muslim dominated seats. So their number is expected to be much higher.

3. Assam

BJP looks set to mark its entry in North East with a sweep here. Gogoi suffered from a 15 year anti-incumbency and his failure to form an alliance. Only C-Voter is a dissenting voice here as it predicts a hung assembly with BJP as single largest party. We however opine that this is unlikely to happen. The division of Muslim votes as we had opined in our article in Yahoo has led to this likely defeat for Congress.

As per Chanakya difference between BJP and Congress vote share is 11% which is explained by AIUDF vote share. As per Axis, the difference is 15% out of which 60% is explained by AIUDF which is expected to get 9%.

Both the surveys do not give separate numbers for Assamese & Bengali Hindus, Assamese & Bengali Muslims, Tea garden workers groups. However, AIUDF garners 23%-25% Muslim votes in these surveys accounting for 8% vote share which will likely be the decider eventually

In the Axis Poll, the gap between BJP and the Congress is the lowest amongst lower income voters (8-10%), while it is massive (19% to 23%) amongst middle and high income voters. This trend is worrisome for Congress as more and more voters are likely to move to middle income in the next 5-10 years.

4. Tamil Nadu

DMK led alliance is expected to make a comeback is a setback of sorts as there was no visible anti-incumbency against Jaya govt. However, last mile smart campaigning led by Stalin and voters penchant for change in line with historic trends sealed the deal for DMK. C-Voter and a few regional channels predict  AIADMK victory and hence the suspense continues as C-Voter has got Tamil Nadu right for the past 1.5 decades as claimed by them.

Axis predicts vote shares of 39%-37%-24%. This effectively means that DMK has not gained vs 2011 elections (39.4%). AIADMK also has not lost significant vote share (37% vs 38.4% in 2011). However, her erstwhile alliance partners (DMDK and Left) have taken away their chunk of vote share and hit her hard (13%). Axis doesn’t predict major gains for DMK across caste groups, however there is a swing from AIADMK to Others across caste groups with highest amongst SC (-15%), BC and MBC (-17%) voters. 

Chanakya predicts vote shares of 44%-38%-18%. They estimate that DMK will gain a vote share of 5% from across caste groups with a minor decline of 1% from Muslims. AIADMK vote share is expected to be at same levels as of 2011 with allies taking away almost all of their votes from her.

Jayalalithaa also performed well amongst lower income, lower middle class and women voters. But DMK alliance notched up big leads amongst middle and upper income voters taking an overall lead (Axis Poll). The big issue is what kind of turnout model was considered in the exit poll analysis? If the model wrongly estimated lower number of low/income and women voters than the actual numbers, the final results could even be in favor of AIADMK.This probably explains wide variance in vote share forecasts between varous exit polls. 

To sum up, each State has a different story to tell. However, the exit polls in many ways  confirmed who will win in Kerala, Bengal and to some extent Assam. Only Tamil Nadu looks a little dicey because of a variety of factors including differing strengths for the DMK and AIADMK. As far as seat shares are concerned, exit polls have a terrible record of estimating them correctly (less than 40% accuracy) so one can be very confident that most exit polls WOULD definitely get the seat forecasts wrong. It is best to sit back and wait until counting day to know the exact seats picture. An entertaining day indeed on the 19th of May 2016.