An analysis on how Delhiites will vote in the upcoming elections.


Abrupt political experiments that erupt out of nowhere are not new in India, although the media would want us to believe so. Electoral history has taught us that such neo-political experiments always begin with a sudden big bang that induces a temporary belief that politics has changed forever, but then end up fizzing out in a long whimper. There have been at least two major national experiments and three state-level experiments that fall into this category.


On 14th October 1985, in Golaghat, many agitating student’s unions merged to form the Assom Gana Parishad (AGP) which decided to contest the then upcoming assembly elections. Within 2 months, on 16th December 1985, the party won a thumping mandate with 60% vote-share in the state assembly elections.


In fact, there wasn’t enough time for the Election Commission to recognize AGP as a regional party, so all its candidates contested as Independents and still managed to win. Just 2 years before that, in January 1983, in the southern state of Andhra Pradesh, a Telugu Movie Superstar Nandamuri Taraka Rama Rao, popularly known as NTR, emerged victorious in the Andhra Pradesh assembly elections by winning a historic 199 seats out of 294.


His party TDP was formed just months before the elections and once again the EC did not have enough time or data to recognize the party so all candidates had contested as Independents and yet won.


What happened to both these parties in subsequent elections, tells us how politics in India operates. The AGP experiment fizzled out in 5 years and lost the 1991 elections badly, whereas the TDP experiment lasted slightly longer for about 6 years until 1989.


There were also two national experiments of similar nature. The first one was the JP experiment of 1977 which lasted for just over 2 years until July-August 1979 and the second one was the V.P. Singh experiment which lasted for about a year.



The latest entrant in this list is the AAP experiment which began in Nov-Dec 2013 and is trying hard to stay relevant just a year later. As we have seen, historically, the voter in India doesn’t believe in giving a second chance to such neo-political experiments (although TDP and AGP did return back later, they were both normal political parties by then).


To be sure, AAP had fizzled out just 4 months later in April because of its own foolishness of giving up after a 49-day run as the government in Delhi with nothing to show to the voter in terms of tangible achievements. Yet, today, it is once again being propped up as the sole answer to Modi’s leadership by the media-intellectual class. Let us try and see how much of that is actually true.


The December 2013 Delhi elections will be recorded as an election with never-before seen levels of freshness. One of the underlying themes that was widely propagated after the elections was the presence of so many rookie professionals within the AAP ranks, while the other political parties – mainly Congress and BJP – had used the same old political formulae.


Even BJP’s internal assessment was that at least 4-5 close defeats of the party had come about due to wrong ticket distribution choices of pitting old outdated war horses in the fray. The impact of this “freshness” quotient on all political parties was so deep that none of them came forward to form a government after the elections lest they anger this “fresh new India” which was supposed to be loath to all other forms of time-tested politicking!


Freshness can only go thus far in the Indian milieu. Even the “new India” that was supposed to hate all politicking, needs every day governance. Thus, a year later, after the fervor of the neo-revolutionaries of Delhi assembly, the Somnath Bharatis and the Rakhi Birlas has already gone cold, Delhi seems to have become that much wiser in her choices.


We asked Delhi-ites what matters more to them when it comes to their vote-preference, “local candidate” or “party/leadership” and not so surprisingly 18% chose the former and 31% chose the latter, while 25% chose the please it all option of “both matter equally”. Therefore, essentially, the Delhi battle is likely to be one dictated by leadership styles rather than any deep localized undercurrent.


It is in this backdrop that the overall vote-share percentage becomes even more crucial. Indian electoral college can be essentially divided into two groups – bipolar contests and multipolar contests. First, let us consider the latter.


One overriding feature of Indian elections in recent times has been the crucial 35thpercentile. Whenever political parties achieve this 33 to 35% range, in multi-cornered fights, they tend to achieve clear majorities. For instance, let us consider some very recent examples.


In Jharkhand, BJP+ secured 35.5% and in Haryana the party got 33.3% and in both the states BJP has a clear mandate, whereas in Maharashtra BJP+ was stemmed at 31% and consequently just fell short of a clear majority. Similarly, in the Telangana assembly election of 2014 (another multi-cornered contest), TRS received a vote-share of 34.3% and a clear mandate.


Obviously, this 35th percentile phenomenon doesn’t hold true in bipolar contests. Two very recent examples of such bipolar contests can be found in Andhra Pradesh and Madhya Pradesh. Congress secured 37% vote-share in MP and yet came a distant second because BJP had a much higher 46% vote-share. Even more curiously, in AP, Jagan’s YSRC secured a whopping 44.6% of the popular vote and yet lost with a big difference in the number of seats to the TDP-BJP combine that had polled 47.1% votes.


How does one reconcile Delhi with this vote-share mathematical construct of Indian electoral politics? Delhi is an anomaly to this 35th percentile logic. For instance, last time in 2013, BJP+ had secured 34.3% vote-share and yet fell short of a clear majority despite a multi-cornered contest. There are two ways in which we can try and understand this anomaly of Delhi – first is the statistical way and the second is the more practical method.


The 35th percentile rule fails in very large geographies, simply because the sheer breadth of geography and the number of seats make it redundant, therefore there are two crucial anomaly electoral colleges to the 35th percentile rule – the first is obviously India on the whole in terms of parliamentary elections where BJP could win a clear mandate by securing just above 31% votes, while the second anomaly is what we term as a “country within a country”, Uttar Pradesh, where again parties win mandates with under 30% vote-shares.


Delhi is the statistical counter to this very large-geographies theory by virtue of being a glorified municipality of a state which makes it a counter-anomaly of needing more than 35% vote-share. Of course, there is a simpler practical explanation too.


Although a multi-cornered contest at the outset, Delhi has essentially been a bipolar state in reality so the 35th percentile doesn’t hold true – as was witnessed in the LS polls of 2014 when it was a direct contest between BJP and AAP (or in the past when BJP v/s Congress battle ensued).


This is why Congress’s performance in the present elections of Delhi becomes crucial. If Congress repeats its LS poll performance of being much below the 20% mark, then either of the main contestants will need to cross the 35th percentile in a big way and move towards the 45th percentile range for a bipolar majority.


Apart from its inability to win, Congress is also in a crucial transformative electoral phase as the party is now leveraging its micro-electoral presence. Yes, today Congress is essentially a national party by nomenclature but a micro-sub-regional party structurally.


This is why pollsters are unable to capture Congress party’s performance metrics with accuracy. For instance, the party gets much lower overall vote-share and yet manages to win disproportionate number of seats unlike what happened in the summer LS elections where Congress received a whopping figure of 10 Cr votes but only 44 seats.


The one general truth about Congress party is that it has a thinly spread vote-share which results in far fewer seats as compared to its votes, but now after the party has lost its base in a big way, the reverse is slowly becoming true. As we saw in Maharashtra or J&K, Congress party is winning seats because of its sub-regional presence despite receiving low overall vote-shares.


Thus, the one worry for pollsters in Delhi is that the Congress party may deceive with its lower overall vote-share but better sub-regional performance. Probably what we are observing in Delhi is the exact reverse of what we saw in 2013 – this is now possibly a multipolar contest in the disguise of a bipolar contest.



The year-long instability has taken its toll on average Delhi voters, who are probably now looking for a stable government. The biggest reason given by 43% of those, who are voting for BJP is not Modi or Bedi, mind you, it is a “stable government”; this is in contrast to the biggest reason among those supporting AAP, which at 37% is “Arvind Kejriwal”.


One gets a sense that the average voter in Delhi is intelligent enough to know that AAP is unlikely to provide a stable government. Thus, as it usually happens closer to polling date, more and more voters may tend to lean towards a stable government since India in the 21st century essentially votes for majority governments.


To understand this 9% gap between BJP and AAP and the next step of the 12% gap between AAP and Congress (as per our projections), we must try and classify Delhi in two ways. The first classification is the class division, while the second is the ethnicity division. In that sense, Delhi is a rare state where electoral analysis is not just a caste-vote matrix but a slightly more complicated phenomenon. Caste does still play a role, but it is more intermingled with ethnicity than in any other state.


Class division of population in India is a nightmarish exercise as the criteria vary vastly as per different definitions. In order to keep it simple, we have classified our target respondents into two simple categories – those with a family income below 80k per annum as poor and those with family incomes above 80k as “affluent”. In our survey, roughly 55% of the respondents belonged to the affluent category and 45% to the poor category.


The difference between BJP and AAP is a yawning 19% among the middle classes, whereas both parties are almost on equal footing among the poorer sections of Delhi voters. Congress, which is still shunned by middle classes, shares its vote-base with AAP which is hurting Kejriwal’s party more than what most political pundits have admitted.



Yet, this classification on its own doesn’t tell the full story as it encompasses a wide range of subgroups into just two categories. In order to understand how Delhi is actually voting, we must divide this city into her various ethnic vagaries.


Again, there are no hard and fast rules for dividing Delhi, so we (at 5Forty3 datalabs) have our own way of classifying Delhi socio-geographically. On a broader level, the eastern districts of Delhi comprising mainly of North-East Delhi and East Delhi can be termed as what we call as “Chhat Puja” territory, while the western part comprising mainly of North-West Delhi, West Delhi and some parts of South Delhi is termed as “Punjabi” territory (which is now more of a Haryanvi-Punjabi territory to be precise).


The central part of Delhi, comprising mainly Chandni Chowk, New Delhi and some parts of South Delhi areas, is essentially the “urban” territory. Of course, there are many overlaps and intermingling in this broad classification, but it more or less holds true demographically.


Ethnically, Delhi is comprised of seven major sub-groups – Poorvanchalis, Punjabis, Paharis, Dalits, Muslims, Baniyas and others. While Poorvanchalis dominate the Chhat puja territory, Punjabis dominate the Punjabi territory (for our own convenience, Punjabis also include Sikhs, Gujjars and Jats). Baniyas and Muslims are present in big numbers in urban Delhi (especially in Chandni Chowk area).


Not surprisingly, Muslims have picked on AAP as the best bet to defeat BJP as their first choice. So AAP is deriving indirect benefit which should augur well for the party if it wants to reinvent itself as a secularism messiah in the future.


Dalits are still favoring BJP but not by much as AAP comes in a close second, while Mayawati still stands tall among some 10% of Dalit voters. Baniyas seem to be suffering from a split mentality as their loyalties are divided between BJP and AAP, but it is the Punjabis and Poorvanchalis, who are powering the BJP in Delhi.



Considering all these findings, we can safely say that the main contest in Delhi is between BJP and AAP, but that the former enjoys a clear advantage with almost a double-digit margin of difference. Congress, although a distant third, may be the crucial player, who will eventually decide how many seats the two main contenders win.


There are a few caveats though. Smaller parties do tend to get under-represented in surveys while ruling parties tend to get over-represented. Although we have taken enough care to under-weight BJP and over-weight Congress (which is essentially a small party these days), there could yet be some small aberrations.


As for AAP, we think it has a core vote-base of around 22% (which is the number of voluntary responses prior to multiple-choice options) and a 10% additional vote. BJP’s core-vote stands at 29% and additional vote is 12%. The standard error margins of 3% apply to all our survey findings.


We also conducted a unique experiment of asking the voters if they could be changing their voting preference in the next two weeks and some 7% respondents answered in the positive which essentially means 93% (or at least a rounded off figure of 90%) voters in Delhi have made up their minds.


Very interestingly, a big chunk of almost 44% of these 7% “shifting” voters belonged to Congress while only 28% and 21% belonged to AAP and BJP respectively. We will be conducting a second pre-poll survey in the last days leading up to the polling date on the 7th which should give us a far better measure of any shift in voting patterns.


[Tomorrow in the third and concluding part we will try to statistically project vote-shares to seats and also understand Delhi at a deeper micro-level.]


Note: Our poll survey was conducted in 381 locations of carefully chosen 121 swing polling stations spread across 42 representative assembly segments. We had a target sample-size of 3260, but were able to achieve 2945 pre-determined respondents derived from Delhi voter-rolls using our path-breaking filtering mechanisms.


Of these, 1360 respondents were female and 1585 respondents were male. Adequate representation was given to all castes/religions and different economic classes of the Delhi society. While deriving the final numbers from our survey findings, proper statistical modelling techniques were followed giving requisite weightage to modified census data based on our own methodologies.


All our interviewees went to the homes/places of dwelling of the respondents to conduct the interview in a clear language that all Delhi citizens could understand. We incurred a cost-per-response rate of 90 rupees including all the overheads. Part of the capital was raised through crowd-sourcing, but mostly was met through self-funding by Swarajya-5Forty3.