A landslide for the BJP, but a mudslide for Pollsters

India has voted and the NDA has had a sweeping victory across the country with 339 seats. Of these, the BJP has taken 284, giving it a majority regardless of the alliances it made with other parties. The UPA, led by Congress, has had a shocking defeat, one of its worst in history, with only 58 seats, of which 44 went to the Congress. Ironically, it will now need to seek alliances with other parties to form an opposition in Parliament. The ADMK has 37 seats, the largest party categorised in the OTHER alliance. Whereas Tamil Nadu has typically played a role in forming the government through alliances, it will now have to decide whether to form an alliance with Congress to form the official opposition. This is the first time, since 1984, that a single party has won a parliamentary majority.

For the pollsters, this win shows their methodologies failed. As an aggregator of the pollsters, our own analysis was wholly off from what the actual result came out to be. Interestingly, the PEW Research, in its survey, was the closest to the result. It predicted 63% supporting the BJP to lead the next government. This translated to 342 seats. However, its prediction for the UPA and OTHER was not as accurate.

Returning to our own predictions and how they were wholly wrong, we feel the number of quality pollsters lacks in India. Whereas the USA elections, for example, have hundreds of pollsters, we were presenting our analyses on a handful of pollsters. We firmly believe more transparent quality pollsters are required, working at the state level.

Over the next few weeks, there will be much analysis of the results and parsing of the data to see how the election was won. Some of the more interesting questions will include the role played by the youth, caste, social media, and by the campaign style itself. Did a presidential-like campaign (ie the Modi Wave) play an important role? Did Amit Shahs' strategy in UP, focusing on local caste affiliations and the like for nominating members play a decisive role in the massive win the BJP had in the State? What role did the youth vote and social media play in the election?

John Oliver on Indian Elections

A great and funny piece on the US media ignoring the Indian election, as well as the present state of India's own media.

 

Overheard on the campaign trail

"Uday was under the influence of alcohol and lost his balance while selecting the button. Holding the machine in right hand, he fell down with the machine"
Nelamangala rural police describe their arrest of an inebriated voter.

“It is wrong to award death penalty for rape. Boys commit mistakes sometimes. If SP is voted to power, we will change the laws”
Mulayam Singh Yadav, Samajwadi Party, promises to lessen penalties for rape!

“Solution is this: any woman if, whether married or unmarried, goes along with a man, with or without her consent, should be hanged. Both should be hanged. It shouldn’t be allowed even if a woman goes by consent.
Abu Azmi, Samajwadi Party, expands on his partys’ promise for dealing with rape!

“They treated us to really good biryani and a quarter of whiskey” 
Rampal Singh, on his local partys’ hospitality

"My parents are traditionally loyal to a political party, but I voted for an individual I think can provide me security and job opportunities." 
Ritika, first time-voter, displaying newfound confidence in younger voters.

 "We have polling on different dates in western Maharashtra and Mumbai. Polling in Satara is on April 17 and in Mumbai on April 24. I advise Marathi workers to first cast vote in Satara on April 17, come back to Mumbai and again exercise franchise on April 24 … Of course, please do not forget to wipe the (electoral) ink before casting the vote the second time."
Sharad Pawar, NCP, providing logistical advice for double-voting!

‘We will chop him into tiny pieces’
Imran Masood, Congress, describing what he will do to PM-hopeful Modi

“Before, if voting was on a long weekend, we would have stepped out of the city earlier to make the most of our holiday…Now we realize the value of our votes. We will head out after.”
Aditi Rao, Software Professional, for whom the vote outweighs the long weekend.

"A couple of elections ago it was not such a game of money…Now you have business people in politics, whereas earlier they were involved in managing their empires."
P.K. Dash, Election Commission, complaining about the cash for vote problem

The Last of the Polls

While we should be awaiting exit poll predictions, NDTV released their last opinion poll, giving the NDA 275 seats - past the required 272 for a majority. Of course, opposing parties have claimed the poll doctored. Performed by Hansa Research, details on methodology have been brief, at best.

It is not inconceivable, if one views the other opinion polls as well as our aggregated analysis, to see the NDA get a majority. However, we still firmly believe it will be through newly formed alliances with parties in States such as Tamil Nadu.

Not right to give death penalty for committing rape: Mulayam

Samajwadi Party supremo Mulayam Singh Yadav today said, "It is not right to give death penalty for committing rape, boys make mistakes. If we come to power, we will check misuse of anti-rape law".

Bahu dilao-Vote pao! A different kind of Bribery

One for the weird files.... A group of men in Bibpur, Haryana, are spearheading a campaign to bribe politicians to ensure wives for villagers in return for their vote in the upcoming general elections.

Could Social Media make a difference?

With all opinion polls now banned while voting continues across India, we look at other interesting topics in relation to the Lok Sabha elections. Some will be serious and some will be fun!

This will unquestionably be considered the first social media election. Many Politicians, such as Modi and Shashi Tharoor, have millions of twitter followers each. With the electorate being increasingly young, every parties presence online will become increasingly important. Recently, the Internet and Mobile Association of India estimated that a well-executed social media campaign could swing 3-4% of votes.

Unfortunately, for those of us who like a little fun, it may mean officials will become more versed in social media and topics such as cloud computing. We will see less of the likes of Vishwa Bandhu!

Final Pre-Poll Aggregation Results: NDA (242), UPA (130) OTHER (171)

Using the approach described here, our final analysis of the pre-poll predictions gives the NDA 242, UPA 130 and Other parties 171. We restricted our pre-poll data to the polls conducted in the last two months only. We thank everyone for visiting the blog and will continue posting as the results come in. Clearly, from our analysis no party comes to the magical 272 and there will clearly be a great deal of jostling to establish a majority government. Good luck to all during the next few weeks of voting!

Methodology Updated

Here, we present our final methodology used for our analysis in relation to the 2014 Lok Sabha Elections.

1. Our first exercise was to database the election results of the previous four election cycles. This allowed us to perform a conditional logic analysis that predicted the odds of winning for each party and alliance within each constituency. As part of this analysis we also made predictions and updates to the data to ensure the delimitation of 2008 was taken into account. Ultimately, we had a database with seat safety for each constituency as well as the average winning margin for that party. The report based on this analysis can be seen in this blog post.

2. Now that we had the historic data captured, we focused on the pollster predictions coming through the media. We did not use any national predictions, but rather focused on the state predictions. As has been seen in previous predictions, we found that a number of leading pollsters had been relatively accurate in their by-state predictions. The best of these pollsters was the CSDS. Of course, 2004 had seen pollsters perform poorly, but even then many of their state predictions were fairly accurate. There were, obviously a few exceptions. Our blog post that goes into further detail on our analysis of the pollsters can be seen here. During this 2014 election cycle we databased the predictions made by CVoter, CSDS, Hansa, IPSOS and AC Nielsen. For each State and Union Territory we databased their seat predictions.

3. For each seat prediction, we used the historic data to determine what the poll margin win would be within each constituency or state. From these margins, we then determined the Median and SEM of winning the seat.

4. We finally created a dataset for each of the three alliances containing the Median and SEM for their winning margin within each of the 543 seats. We used this data to determine the exact number of winners, covering all ways of reaching that number given the win probability. We adapted the approach presented here

Of Cricket, Bookies and the Lok Sabha

Earlier this week, the Supreme Court asked BCCI president N Srinivasan to step down to enable an impartial inquiry into allegations of spot-fixing in IPL matches. His son-in-law, Gurunath Meiyappan, was indicted in February 2014 of illegal betting and passing sensitive information in relation to matches to bookies during IPL 2013. While Srinivasan is an industrialist, the controversy has stoked the question of why so many business men and politicians are heading cricket associations across India. As some claim, has cricket been hijacked? Sportsmen seem to be singularly absent in many of these senior roles. For example, in Himachal Pradesh, Anurag Thakur, also the present Lok Sabha member for Hamirpur constituency, became the youngest selector, at 26 years of age, in 2001 to select the Indian Junior Cricket team. He is the president of Himachal Pradesh Cricket association. Presently, the state vigilance bureau is seeking prosecuction sanctions against him and others in relation to irregularities in the Himachal Pradesh Cricket Association.

Illegal betting and corruption in Indian sports is at epidemic levels. Two former presidents of the BCCI had themselves said IPL games were routinely fixed. More attention should be paid to the bookie network. In May 2013 arrests were made in a major crackdown on bookies within Delhi and Mumbai. Evidence from these arrests and subsequent investigation suggested an active transnational bookie network that stretched across India, Pakistan and Dubai. One could only wonder why so many politicians head these organizations, when it is widely known how manifest the bookies are within this arena. One could only hope these bookies are not illegal funding sources for politicians and business men alike.

Even so, it seems Cricket is not the only thing taking illegal betting action nowadays. Despite IPL 2014 occurring at the same time, the Lok Sabha election seems to be gaining more heat among bookies. Punters have set their rates and the odds are in favour of the NDA coming into power with Modi as PM. The odds for Modi being 6-5 against. His odds have narrowed from around 8-5 against three months ago. For Rahul Gandhi, the signs are worse. His odds of winning have widened from 9-5 against three months ago to about 2-1 against today. The rate being offered by punters for the NDA bagging 220 seats is 20 paisa to a rupee. Punters have suggested that the elections have overtaken Cricket in illegal bets placed across the country.


It seems nothing is sacred when it comes to placing a bet. Unfortunately, it seems behind every bet, whether it is cricket or the elections, business men and politicians are not far behind. 

Could China lose from India's abstinence?

India decided today to abstain from voting for the US-backed resolution relating to Sri Lanka at the 25th session of the UNHRC in Geneva . The resolution demanded an independent international investigation against the alleged war crimes and human rights excesses in Sri Lanka. India gave several reasons for rejecting the resolution, despite having voted in favor of resolutions in 2009, 2012 and 2013. India claims the resolution imposes an 'intrusive approach' of international investigative mechanism which was counterproductive apart from being 'inconsistent and impractical'.

By not voting against Sri Lanka, India gains some mileage with Sri Lanka. Clearly, this has been reflected in the Sri Lankan media. China, which has been heavily investing in Sri Lanka for things such as the maritime silk road, rejected the resolution outright. With the Congress-UPA alliance projected to gain very little in Tamil Nadu during the forthcoming elections, it had little to gain from voting for the resolution. On the other hand, it gains in potentially better relations with Sri Lanka. We suspect all of the Tamil Nadu parties, including the DMK which walked away from the UPA alliance over Sri Lanka, will voice their criticism over the abstinence. We suspect the BJP will gain little momentum from this abstinence, considering its own stance on a 'Tamil Eelam State' within Sri Lanka. However, this being an election year, they will likely criticize the move in Geneva.

Perhaps the biggest gain for the Congress-UPA alliance in the forthcoming elections will be what voters outside of Tamil Nadu think of the abstinence and whether the upward trajectory of China-Sri Lankan relations dampens in favor of India. Finally, whether the rise of Sino-Sri Lankan relations is a concern for the voter has yet to be seen.


The patience of king makers

The complexity of Indias' process for electing its Lok Sabha members in a nation of 1 Billion is only matched by the diversity of it electorate and the plethora of regional parties.

In the previous election cycle there were 8070 candidates contesting 543 seats from 364 different political parties. From all accounts, these numbers have increased for 2014. It's estimated there is an additional 100 million eligible to vote this year, compared to the last election only 4 years ago. This is easily explained by the fact that the median age in India is 25.1 years. The percentage of 0-14 year olds in 2009 was 30% of India's population.

To reach out to these electorates, The Centre for Media Studies estimates that US$5 billion will be spent by the contesting parties. This is three times the amount spent in 2009 and is second only to the US$7 billion spent in the US during their 2012 US elections. Even so, this amounts to a paltry US$6 spent per electorate in India, which is much lower than the US$36 spent per electorate in the US.

While this election spending may be a great temporary boost for the economy, especially with India's advertising industry expecting to see an $800 Million injection, one wonders what it will do for democracy in the future. If the spending per electorate in the US is any sign, there is still room for spending more in India. Will India see its election process dominated by corporations, as has happened elsewhere? 'Money Power' will become an important issue in both this election and in succeeding ones for India.

However, for now, it does not seem that corporations will be able to dictate who will rule. Regional parties maintain the mantel of king maker. Of these, Tamil Nadu will likely play an important part. Neither of the two national parties have strong positions in the State, and it will be the regional parties there that will slog it out to win as many of the 39 seats available as possible. Our own analysis of seat safety, based on the previous four election cycles, noted that 8 constituencies within Tamil Nadu presented bellwether traits, in that they had been part of the ruling alliance frequently over the last two decades. These were Arakkonam, Chidambaram, Cuddalore, Dharmapuri, Krishnagiri, Perambalur, Sriperumbudur, and Vellore. The NDA has its rainbow alliance fielding candidates in each of these constituencies, but the prevailing predictions are that the real battle will be between the DMK and AIADMK in the state. Neither of these parties has joined any alliance. Clearly, they will see what they can gain post-election, when they will inevitably think about alliance-forming. This will be especially true if either of the national parties is unable to form an outright majority. This reminds us of an old Tamil Proverb,

'பொறுத்தார் பூமி் ஆள்வார்'
'The patient will rule the World'


Poll Predictions based on differing scenarios

Herein we present some differing predictions, based on different scenarios in our database.

1. Our first prediction is based on the sole use of All Pollsters in the 2014 election cycle.

2. Considering the concerns following Operation Prime Minister, what would happen if we predominantly focussed on the CSDS predictions?

3. Our research has found a number of seats which are considered either 'Safe' or 'Somewhat Safe' based on the previous four election cycles. What would happen if we included these with the pollster predictions?

4. What if we focussed our poll aggregation solely on the historically 'Safe' and 'Somewhat Safe' seat predictions combinded with the CSDS predictions?

The table below presents our results based on the above four scenarios for aggregating the poll predictions.

Data used
NDA
UPA
OTHER
All Pollsters
231
140
172
CSDS only*
213
156
174
Historically  Safe + Pollsters
227
145
171
Historically Safe +CSDS
246
140
157

* Other pollsters were used, in a limited manner, wherever the CSDS had provided no State or UT prediction

Based on these scenarios, the variation in seat number seems the most for the NDA. In all cases they are still shy of the 272 needed for an outright victory. Interestingly, the UPA and OTHER alliance predictions have a tighter range in the prediction variation.

Methodology

Here, we present the methodology used in our analyses to date.

1. Our first exercise was to database the election results of the previous four election cycles. This allowed us to perform a conditional logic analysis that predicted the odds of winning for each party and alliance within each constituency. As part of this analysis we also made predictions and updates to the data to ensure the delimitation of 2008 was taken into account. Ultimately, we had a database with seat safety for each constituency as well as the average winning margin for that party. The report based on this analysis can be seen in this blog post.

2. Now that we had the historic data captured, we now focused on the pollster predictions coming through the media. We did not use any national predictions, but rather focused on the state predictions. As has been seen in previous predictions, we found that a number of leading pollsters had been relatively accurate in their by-state predictions. The best of these pollsters was the CSDS. Of course, 2004 had seen pollster perform poorly, but even then many of their state predictions were fairly accurate. There were, obviously a few exceptions. Our blog post that goes into further detail on our analysis of the pollsters can be seen here. During this 2014 election cycle we databased the predictions made by CVoter, CSDS, Hansa, IPSOS and AC Nielsen. For each State and Union Territory we databased their seat predictions.

3. For each seat prediction, we used the historic data to determine what the poll margin win would be within each constituency or state. From these margins, we then determined the Median and SEM. Whereas other pollsters in the US took this data further, we decided to stop here and rank the constituencies in order to determine the seat predictions for each alliance. This then resulted in our poll predictions. Although it may not be possible for this election cycle, we are still investigating simulation models that could be used easily in a multiparty election system. Many of the poll aggregation models used to date have been in the USA, where there is typically a two party election and a winner takes all in the state electoral votes. Within a parliamentary system, this is not the case. Seats are won at the constituency level and not a winner takes all at the State level.

In a succeeding blog post we will present some additional predictions based on different scenarios.

Election Forecasts over Time

As part of our poll aggregation analysis, we present the pollster predictions made over time since January 2013.

As the number of polls done is limited in India, for any given month, we used the poll predictions at the timepoint, but filled the gaps for States not predicted with preceding month predictions. For January 2013, we used the poll results presented at that month along with historical seat safety predictions from the four previous election cycles (1998-2004).

As can be seen, the initial polls from over a year ago gave the NDA a very large number of seats. However, over time, this lead has reduced as the number of polls taken increased on a monthly basis. As the regional parties had become more non-committal in their choice of Alliance party, the number in the OTHER alliance has increased in seat prediction numbers.

As an aside, Operation Prime Minister, the pollster sting operation, has shown the need for more regulation within Indias' polling industry. However, as we can see in our results presented in this blog, choosing those Pollsters who have presented seemingly accurately in the past, and aggregating their results into one snapshot, will hopefully give a more accurate prediction for the seats won by each alliance.

FiveFortyThree.in projects NDA to win 231, UPA to win 140 and OTHERS to win 172 seats.


FiveFortyThree.in projects that the NDA will win 231 seats in the forthcoming Lok Sabha election. The UPA is projected to win 140 seats, with Other parties winning 172.

These forecasts are based on a poll aggregation of opinion polls presented from January 2013 until March 2014 by leading pollsters such as CSDS, CVoter, Hansa, IPSOS and AC Nielsen.

Only those pollsters with a reputation gained from having presented in previous elections were included. Pollsters who have only presented for the first time in this election cycle or who presented no details on sample size and methodology were excluded.

Further, our poll aggregation was based on the state level only. Pollsters who only provided national seat projections were excluded. Based on our analysis of previous election forecasts, we knew there was a higher level of accuracy based on state level forecasts. For any gaps in the state level forecasts, our seat safety predictions based on the last four election cycles were utilized. In fact, this was only required for smaller States and union territories which nearly always tend to vote in a predictable manner.
The table below presents these forecasts at the state and union territory level.

State
NDA
UPA
OTHERS
Andaman and Nicobar Islands
1
0
0
Andhra Pradesh
8
13
21
Arunachal Pradesh
1
1
0
Assam
3
9
2
Bihar
24
1
15
Chandigarh
0
1
0
Chhattisgarh
7
4
0
Dadra and Nagar Haveli
1
0
0
Daman and Diu
1
0
0
Goa
1
1
0
Gujarat
21
5
0
Haryana
4
5
1
Himachal Pradesh
3
1
0
Jammu and Kashmir
1
3
2
Jharkhand
5
5
4
Karnataka
12
13
3
Kerala
2
8
10
Lakshadweep
0
1
0
Madhya Pradesh
24
5
0
Maharashtra
29
16
3
Manipur
0
2
0
Meghalaya
0
1
1
Mizoram
0
0
1
Nagaland
0
0
1
National Capital Territory of Delhi
3
1
3
Odisha
7
6
8
Puducherry
0
1
0
Punjab
9
4
0
Rajasthan
17
7
1
Sikkim
0
0
1
Tamil Nadu
8
3
28
Tripura
0
0
2
Uttar Pradesh
38
11
31
Uttarakhand
0
5
0
West Bengal
1
7
34


Based on these forecasts, it is clear that a hung parliament is predicted and that the NDA will need to reach out to regional parties presently grouped in the OTHERS alliance if these forecasts are correct.

More details will be provided on our methodologies in forthcoming blog posts.

Know your Constituency!

Our newly released report titled '2014 Lok Sabha Seat Safety Predictions' provides a comprehensive account for each of the 543 constituencies contesting the Lok Sabha elections. Below is an example of the details provided for Nizamabad (Andhra Pradesh).



Get to know similar details for your constituency. You can purchase our 558-page report using paypal below. For more details, feel free to contact us at jsingh@fivefortythree.in.


Lok Sabha Seat Safety Predictions

As part of our analysis of the 2014 Lok Sabha elections, we’ve prepared a comprehensive analysis of election results in the Lok Sabha from 1998 through to 2009. These previous elections are of particular interest as they saw the government change hands between the NDA and UPA alliances. Through this analysis, predictions have been provided on seat safety for each constituency with respect to the 2014 elections.

The Lok Sabha, the lower house of Indias' Parliament, is made up of 543 constituencies. Each one has a single representative MP. An additional two constituencies are nominated by the President of India from the Ango-Indian community, if it is felt that community is not adequately represented in the house.Constituency level results were databased for the four Lok Sabha elections from 1998 through to 2009. All candidates within each constituency were included. Additionally, the total number of votes, the percentage won for the winning party/alliance as well as the percentage winning margin between the winner and the candidate who came in second was determined.

Some constituencies came into existence or were adjusted in 2009 following the implementation of delimitation of parliamentary constituencies based on the recommendations of the Delimitation Commission of India constituted in 2002. In these cases, predictions were made for the 1998-2004 election winners based on averaging state results or by mapping to previous constituencies which were geographically similar to the new constituency.

A conditional logic model was run with factors for voter choice. Each candidate was included in the election, with weighting for the total number of voters per candidate. The model then produced the odds of winning for each candidate within each constituency.

For each candidate, the percentage of elections won from 1998-2009 as well as the odds of winning were used to determine a category for each constituencies seat safety. The categories used were:
This constituency is considered safe for XXXX
This constituency is considered somewhat safe for XXXX
This constituency is considered marginally safe for XXXX
This constituency is considered a toss-up

Table 1 provides seat safety predictions based on the conditional logic model.
Table 1: Seat Safety Prediction per Alliance

NDA
UPA
OTHER
N/A
OVERALL
Safe
39
16
-
-
55
Somewhat Safe
10
81
1
-
92
Marginally Safe
89
-
2
-
91
Toss-up
-
-
-
305
305
Total
138
97
3
305
543
 
As can be seen, based on elections since 1998, the NDA has marginally more safe seats than the UPA. However, of most interest is that the vast majority of seats (305, 56.2%) are considered a toss-up with respect to the alliance winner.

Our comprehensive 558-page report, which also presents a detailed account for each of the 543 constituencies, delves further into the toss-up states and also highlights 8 bellwether constituencies which have been part of the ruling alliance over the last two decades. We believe our report will be of particular interest to journalists, political analysts and citizens. We believe this is the first time a rigorous statistical analysis of historic results has been presented for each of the 543 constituencies in India’s Lok Sabha. 
To purchase the report, please email jsingh@fivefortythree.in. Alternatively, you're welcome to pay via paypal as per below.