The Gujarat electoral calculus

Part 3: The predictive model

This election analysis trilogy is dedicated to Offstumped (Shashi Shekhar), one of the finest political brains that the Indian right has produced in the digital era.

Accordingly, the poet should prefer probable impossibilities to improbable possibilities

– Aristotle

Modern day democratic process has a unique feature associated with it, that of psephological prophesising. Not only do we want voters to vote and elect their rulers/representatives but also we want to be able to predict the voter’s preference even before the vote is cast. In fact, many of the western democracies have such advanced psephological methodologies that at times it becomes difficult to distinguish between the actual votes and the exit-polls. India of course is still evolving its predictive techniques and still has a long way to go before achieving any degree of accuracy.

The upcoming Gujarat assembly elections have already seen 5 major opinion poll surveys till date and quite a few other psephological calculations by arm-chair pundits, political analysts and electoral experts. Apart from these widely publicised surveys, there are at least half a dozen internal surveys conducted by various political parties. Going ahead, we are promised a host of exit-polls by various organizations in the next 2 days before the actual counting takes place on the 20th. Let us try and analyse the survey data and also extrapolate the past electoral data & factorial impact analysis of Gujarat 2012 on to the opinion poll surveys.

All the 5 big boys of Indian psephology and market research have dipped themselves in the Gujarat elections of 2012, which in itself could be unprecedented for a state election. In the past so many players conducting a pre-poll survey was only for national elections; not even UP, the most populous state was given such importance; which only proves how seriously the establishments in Delhi have taken this particular Gujarat election.

Past performance of market-research firms – the PPP index

In-order to analyse Gujarat elections of 2012, there are two important parameters about psephologists/market-researchers that needs to be looked at – this is the past psephological performance index or the PPP index; total weightage = 60. Which can be further classified into specific performance Index and generic performance index.

  1. A market research firms success/failure rate in the state of Gujarat: barometric choice – 2002 and 2007 assembly elections of Gujarat {Weightage = 20}


[Data Source: Election Commission of India & Past opinion/exit-poll database of the author; all percentages have been rounded off & seats are means of projected ranges] {In 2002 AC-Nielson-ORG-MARG conducted a joint survey}

  1. General past record of the market-research firm in Indian elections: barometric random choices – 2009 general elections, 2012 UP assembly elections and 2008 Karnataka assembly elections – first after delimitation came into vogue {Weightage = 25}

A]


[Data Source: Election Commission of India & past opinion/exit-poll database of the author; all percentages have been rounded off & seats are means of projected ranges]

B]

[Data Source: Election Commission of India & past opinion/exit-poll database of the author; all percentages have been rounded off & seats are means of projected ranges]

C]

[Data Source: Election Commission of India & past opinion/exit-poll database of the author; all percentages have been rounded off & seats are means of projected ranges]

Chronology of sample sizes and geographic spreads for Gujarat 2012 – the triple SI {Weightage = 15}

[Data Source: Lens-on-News, CNN-IBN, ABP-News, India-Today, India TV]

{*C-Voter has not published either the sample size or the no: of seats of the sample survey}

The Sample, Spread & Samay (time-frame) Index – Triple SI, awards 5 points each to the three aspects based on the latest polls, sample sizes and geographic spread of the pre-poll surveys conducted in Gujarat 2012.

Projected vote-shares for Gujarat 2012

[Data Source: Lens-On-News, CNN-IBN, ABP-News, India-Today & India TV]

  1. The gap between the two principal players Congress & BJP is projected to be between 7% & 14%
  2. At 7% gap, BJP is expected to retain its majority similar to that of 2007, whereas at 14% gap (as suggested by CSDS), BJP could potentially sweep the state, while Congress would be totally wiped out of the electoral map
  3. Projected Congress vote-share never crosses beyond 40%, while that of BJP never goes below 47%
  4. This time no pre-poll survey has suggested either a Congress victory or even a close-contest, which could mean a possible landslide for the BJP
  5. The “others” vote-share is projected to be between 12%-15% – suggesting marginal impact of regional parties like GPP

The Factorial Impact Index for 2012 Gujarat

Apart from the pre-poll surveys, it is important to make an assessment of factors that impact a particular election in India. For instance, there are 5 major factors that have had a big impact on Gujarat assembly elections in 2012 and we need to quantify them in order to be able to predict the outcome (for a detailed analysis of these factors, please refer to part 2 of the Gujarat Electoral Calculus).

[Derived from Factorial impact analysis in part 2 of this series]

High Turnout: has always been pro-BJP in the context of Gujarat elections, as we have clearly demonstrated in the previous two parts. The only qualifier being that the turnout has been unusually high in phase 1 (70.75%) – more than 10% increase from the previous elections of 2007 and a whopping 23% increase from the 2009 LS polls. Arguably, at least half of this increase is due to correction of voter rolls (5%), so in reality the actual polling percentage in comparison to 2007 could be around 65%. Yet, due to this unusually high turnout, BJP doesn’t get full 10 points and instead shares 2 points with the Congress.

Delimitation: Increased urbanization and re-organization has been extremely beneficial to the BJP, therefore a 80:20 score has been factored in (refer part 1 & 2)

Development Vote: The NaMO factorial impact has been analysed in the previous part and a general 70:30 ratio is the widely accepted weightage for 2012 elections.

Drought: The drought scenario also factors in the “Patel vote” and gives a big 30:70 advantage to the Congress party, especially in Saurashtra (refer part 2)

Other Parties: Presence of other parties like the GPP should have, in the classical Indian electoral scene, been detrimental to the chances of the incumbent, but in Gujarat (especially Saurashtra) it is increasingly being seen as denting the anti-incumbency vote. Thus BJP and Congress share the honours and 1 point is allotted to the neutral field due to the unpredictable nature of the “others” vote-share.

MADE – Modified Albatross Electoral Doppler

MADE is constituted by three different indices; the PPP Index – Past Psephological Performance Index, the Triple SI – Sample Spread & Samay Index and the FII – Factorial Impact Index. While the PPP index & Triple SI get a combined weightage of 60%, the FII gets an independent weightage of 40%. The first two indices; PPP index & Triple SI; are primary indicators, whereas FII is used for secondary ratification of the trend.

MADE = (PPP + Triple SI) + FII

Success rate of market research firms in Gujarat and India is 50% and less. AC-Nielsen has by far the best track-record with a success rate of about 53%, whereas CSDS & Lens-on-News have a success rate of 38% and 40% respectively. ORG has a success rate of 25%, while C-Voter has almost never been successful in predicting Indian elections. Thus C-Voter can be discarded for all practical purposes in analysing/predicting Gujarat 2012.

The FII is going with the trend of the pre-poll survey data and has a positive score of 9 in favour of a BJP victory in Gujarat. Thus the FII enhances the scores of any pre-poll survey by 9 basis points if it is going along the trend-line. For instance, if an X poll survey is predicting a BJP victory in the elections, then the FII increases the possibility of success of X by a factor of 9. On the other hand if a Y poll survey is predicting a Congress victory, then FII reduces the possibility of success of Y by a factor of 9 (there are no anti-trend survey results published in the context of Gujarat assembly elections of 2012).

The three projected scenarios for 2012 Gujarat

Scenario 1: The most likely scenario with a 52% possibility of success rate, projects 49% vote-share for the BJP and 37% vote-share for the Congress, maintaining a gap of 12%.

Methodology:

  • Mean success rate of the top 3 market research firms; AC-Nielsen + Lens-on-News + CSDS = 43% + FII score of +9 = 52%
  • Mean of the projected BJP vote-share of the top three surveys = 49%
  • Mean of the projected Congress vote-share of the top three surveys = 37%

Scenario 2:
The extreme pro-trend scenario comes with a possibility quotient of 34% and projects a BJP vote-share of 53%, while that of Congress is at 33%, maintaining a whopping gap of 20%

Methodology:

  • The lowest possible success rate of a market research firm in the trend-line is 25% (with an overall PPP + Triple SI score of 15/60), by adding the FII score of +9 = 34%
  • +3% as error margin for the highest projected BJP vote-share = 53% extreme
  • -3% as error margin for the lowest projected Congress vote-share = 33% extreme

Scenario 3:
The extreme anti-trend scenario comes with a possibility quotient of the remaining 14%. Where BJP vote-share is projected at 44%, while that of Congress is 43%, thus maintaining a miniscule gap of 1%.

Methodology: The reverse of scenario 2.

Seat Projections for 2012 Gujarat

Converting vote-shares to seats is probably the most difficult task of an analyst/psephologist in the Indian electoral scenario. In the past we have seen many psephologists making huge errors in converting to seats the projected vote-shares. For instance, Yogendra Yadav who is widely touted to have predicted the UP assembly elections of 2012 accurately was in reality way off the mark. His vote-share projection for the Samajwadis was 34% – a whopping 5% points above the actual vote-share, which when converted to seats could have potentially added another 60 seats to the SP tally, taking its total close to 300!

Thus instead of relying on the methodologies of other market research firms, we have formulated our own seat projections based on various parameters;

  1. The vote-share gap between the principal players
  2. Micro-analysis of different districts and assembly seats
  3. Winability criteria of Congress v/s BJP candidates
  4. Over-all caste matrix (as demonstrated in part 1)
  5. Impact of “others” in close fights
  6. Co-relation between high turnout and assembly seat victories

Epilogue:
Is a BJP defeat in Gujarat a statistical impossibility? That seems to be the general consensus among all pollsters/psephologists/market research firms. The reasons for this unassailable lead by the BJP are many, and all of them have been elucidated in the first 2 parts of the GEC series. Let us look at some contrarian findings as a guide for the future.

An exit-poll conducted in Gujarat by a non-profit trust, as a pre-cursor for the next general elections did come up with some startling results in rural Saurashtra, where Congress was shown to be literally sweeping many constituencies. In 22 polling booths of rural Saurashtra, there was almost a 70:30 vote-share division in favour of the Congress (while GPP was almost non-existent in many areas). Although, this exercise was of only academic interest with a potential to guide future policy decisions of Indian right, targeted towards rural areas, it was revealing enough to the extent that it almost corroborated the socialist theory of doles winning elections in rural India.

There are 5 important pointers for the future of BJP/Indian right in the rural political landscape;

  1. The capacity of NREGS/DCT/FSB and loan-waiver type dole-schemes in winning electoral battles should never be under-estimated and a creative socialistic policy narrative should be initiated at least 1 year/6 months before the next general elections – competitive populism should never be ruled out purely from an electoral stand point
  2. Agrarian issues must be addressed aggressively, even if it means adhering to populism. For instance, the MSP for cotton in Saurashtra is about 850 Rs, whereas input costs during a drought year could be as high as 700 to 750 Rs – which is extremely dis-heartening for a farmer who has invested 6 to 9 months of his time in growing cotton
  3. Rural caste-divisions, especially, the middle castes who are in leadership positions in rural areas should always be given importance – Patels in Gujarat, Lingayats in Karnataka, certain OBCs like Yadavs/Kurmis/Lodhs in the heartland etc. – for their capacity to electorally hurt BJP is immense.
  4. A far-right electoral entity (like GPP in Gujarat) to capture disgruntled core-votes/anti-incumbent votes is a failed experiment, due to its inability to gain traction unlike a left-wing entity like Chiranjeevi’s PRP in AP – all the more reason to discourage a KJP in Karnataka.
  5. Although rural voters have a longer memory than their urban counterparts, they can be easily swayed by immediate economic conditions like drought, inability to get proper price for their hard work or doles and incentives – targeted short-term policy initiatives/schemes can go a long way in winning votes.
The following two tabs change content below.
Praveen Patil

Praveen Patil

Analyst of Indian electoral politics and associated economics with a right-of-centre perspective.