Colombo, Elections, Politics and Governance, Polls

Surveys with conflicting outcomes

There was a survey on the next presidential election conducted by an Indian firm-Viplav Communications Pvt Ltd during January 6-13.  They say the sample size is 10,225 and forecast “The poll shows Rajapaksa (MR) leading in all provinces other than the Tamil-dominated north and multi-racial east and enjoying a 12 percent lead over his opponent in the island as a whole.”

I got two feedbacks from two prominent personalities in Sri Lanka on this survey;

Response 1: “Cannot go by Indian survey as it is, because most people who have changed their minds after PC elections do not reveal the change unless the one who is interviewed is personally known to the officer carrying out the survey, due to sheer fear. Environment is so bad, the govt. sources take revenge at slightest knowledge that someone is in favour of SF. Therefore my understanding is that SF will win provided election/polling is carried out fairly without ridging votes, which hardly can be believed.”

Response 2: “That Indian outfit is well known and reputed. They will lose customers in India if they fudge data to support one side in tiny Sri Lanka.  I would take non Lankan survey more seriously than any Lankan survey. Specially one by an outfit with a reputation to lose!” and he said MR will win.

Therefore a group of people here in Auckland decided to carry out a telephone survey during the last weekend (16-17 January) using a very small sample in 6 districts; Colombo, Gampaha, Matara, Kalutara, Puttlam and Moneragala. According to that survey General (Rtd.) Sarath Fonseka (SF) is going to win the next presidential election. We have to wait until the results are out to verify which sample is more representative and to see whether the sample size really matters. We captured 100 responses.  We assessed four critical sensitive factors (CSF) and the outcome is as follows;

CSF 1: Percentage of valid votes in the whole country except for Northern  Province (62.5% in 2009 PC election; 77.7% in 2005 presidential election).

Projection for 2010 is 78.12%; apx. 11,000,000 votes.

CSF 2: How do those who have not voted in 2009 PC election would vote on 26th Jan 2010? This is the most sensitive factor as it counts nearly to 2,500,000 votes.

Many respondents of the view that those who did not vote in 2009 PC election are mostly hardcore UNP votes and the survey outcome suggests 3 – 4 out of five of such votes would be for SF; 1,500,000 – 2,000,000 votes.

CSF 3: Is there any change of voting pattern from last PC election?

Survey suggests that 0-2 out of 5 (0% – 40%), who voted to UPFA in the last 2009 PC election, would vote against UPFA. Colombo reported 40% while Moneragala reported 0% and other four districts reported 20% swing to SF. However, giving the benefit of the doubt to MR the whole island average swing is projected as moderate 13.6% (5,162,000 of UPFA votes except Northern Province in 2009 PC election) which is apx. 700,000 votes.

CSF 4: Northern Province polling rate and percentage to SF.

Projection is that 420,000 votes would be polled and 60%-70% of valid votes would be for SF; Apx. 250,000 votes.

FORECAST

MR SF Others
Approximate vote-bases according to 2009 PC election without NP 5,162,000

(UPFA+)

2,900,000

(UNP+JVP+MC)

18,000
CSF 2 – Out of 2,500,000 not polled in 2009 PC election 700,000 1,700,000 100,000
CSF 3 – 13.5% of 5,162,000 polled in 2009 to UPFA+ (700,000) 700,000
CSF 4 – Northern Province; 420,000 expected 150,000 250,000 20,000
Total Votes; 11,000,000 out of 14,080,000 = 78.12% 5,312,000 5,550,000 138,000
Percentage 48.29% 50.45% 1.26%

Viplav’s forecast of 12% lead should result in that MR polling at least 55.5%, so that SF would be polled 43.5% leaving 1% to others. But if we project CSF-2 towards more favourable to MR, that is only 1,500,000 vote-swing to SF, then MR would get apx. 50.4%. Therefore, according to our survey it is very unlikely that MR would lead by 12% to post 55.5% as published by Viplav pvt Ltd, instead he would be polled just between 48.3% and 50.4% as on 17 Jan.

C A Saliya

Auckland

18 Jan 2010

  • Sarath Nanda

    What do you think about this?

    ඡන්දය හා අපේක්ෂිත ප්‍රතිඵලය (අනාවැකිය) | Taboo’s vote and the result forecast

    http://taboosubjects.wordpress.com/2010/01/18/2821

  • Paths

    Sarath is likely to win with Tamil votes; douglas will be ignored by Tamils as douglas is interested in himself.
    It is in the larger interest of the Island Sarath wins as it is now ruled by the 5 Rajapakse bros; 3 of them were not living in this country for over 20 yrs.
    Thugery & violence in plenty will be unleashed by the ruling clic not to mention rigging.
    If Sarath wins as likely & HONOURS the promises to the TNA, then the Island can be developed on a fast track basis with the combined efforts of the 3 ethnic groups.
    Historical facts are Tamils were always cheated with broken promises( by DS) broken pacts by SWRD, Srimavo, & Chandrika. I hope Sarath will not follow suit &
    I hope he will bring devolved power to a merged N&E under a federal system of govt. A new secular federal const is a MUST with Singala, Tamil, & English as official languages with Singala prevailing to overcome legal loopholes.
    Paths

  • No frames

    MR 50.5-52.5%
    SF 46.0-49.5%

    Rest calculate for your self.

  • Groundtruth

    Neither.

  • Dayan Jayatilleka

    Dear Saliya,

    There is a qualititive diffence between your lower and uppoer projections, right? If MR poll 48.3% he would/could lose , depending on the figure his opponent has polled but if he gets 50.4% he wins. The upper figure of your survey, namely 50.4%, is , if I remember rightly, approximately Mr Premadasa ‘s winning percentage in Dec1988, so if MR polls that, he’s home and dry.

    As one of the two personalities you have kindly quoted, may I say on the record that my projection hovers in the middle space between your upper figure and the Viplav one, but tends to be closer yours.

  • Dushantha Kurera

    Telephone surveys will not be accurate in SL as most in the rural sector do not have. This is the area where MR is strong. The indian survey employed Sri Lankans and it would have been more accurate. However, since 5th Jan SF has gained ground and I would say It will be around 51-52% for MR and 47-48% for SF. The most accurate is the Kelaniya University one which gave a12% lead to MR in end dec2009. They were spot on in 2005 elections.
    All districts in the North and the East and Colombo,Kandy, N’eliya will vote SF due to the Pact or Agreement or favourable response to TNA demands. This will increase the MR vote in other areas as there is a definite shift due to this among the Sinhala Voters ……………… In 2005 Tamils did not vote for Ranil and this time Tamil vote will be the loosing point again for Ranil &Co !!!!!!!!!!!

  • SomewhatDisgusted

    Dear Saliya,

    While I don’t know anything about how the Indian survey was conducted, and the prediction doesn’t quite sound right, I have a few questions on the methodology used in your own survey.

    While sample size might not matter too much, I believe a key to getting a correct result is proper representation of various stakeholders. Can you explain how the 100 candidates were picked? Did you use a telephone directory to pick users? Also, did you include/exclude the far greater number of mobile phone users? The poorer segments of the population are more likely to use mobile phones than landlines.

    I mainly want to determine how well a reasonably representative cross-section of the populace was captured and what the margin of error is likely to be.

    Thanks!

  • WeSriLanka

    You are making an estimation based on something in the past. Nonsense. Indian Telephone survey is more valid.
    Wait another week to learn that you are absolutely wrong. It is hard to convince you now.

  • Saliya

    Dear Dayan,

    Yes, you are right. My point in that para is to dispute Viplv’s claim of 12% lead by MR. Whoever wins the margin is likely to be capped with 2%-4% and, the survey outcome favours SF as on 17 Jan.

  • Shamed Lankan

    Does any of this matter, really? Is there anyone who believes that the Rajapakses will allow themselves to be voted out of office? Doesn’t matter who casts the votes. What matters is who counts the votes and the Rajapakses would have already taken care of that. They were never going to let the people decide this election. This election was over the day Mahinda called for it.

  • statistics are likely to err on various grounds.there are likely to be last minute swings.there is a possibility that due to the Tamils being given some special concsideration it would antagonise Singhalese voters.so swings are possible for both.it could result in a hung vote .Neither candidate getting the required 51 % . Your calculation is based oin the Tamils voting Some Tamils may refuse to vote to any candidate.If some onecan takeallthesefactors into account which will require higher mathematics they may get a statistically significant result.Many Tamil voters are not having their ID card.The situation may simulate the condiitons in Iran or Afghanistan or both. But who is to mediate between the two leading candiidates? Will at the last minute Dr.W’bahu give his votes to S.F ? It will be decisive. Sam

  • Indians have always been having their vested political interest in SL. What they are expressing is manipulative statistics to fool people into supporting MR.

    Or has MR planned election rigging to get the results and India is aware of it?

  • Travelling Academic

    With that small a sample size, the error bars in your predictions will be much higher than the 2% difference in outcome you project (speaking as an amature statistician).

  • StuckInTheMiddle

    Very happy to see that though you are are oceans away stiil keep a keen interest on happennings in srilanka. I would like to know whether

    1. Do you still think whether the total poll will stand same with the present happenings here
    2. If there any swing of votes after your projections.
    3. What will be your prediction if North/East poll is very low

  • Sarath Gamage

    Shame Lankans comment is shameful. He says MR will win the election by rigging or otherwise. Being a retired govt. servant , I know how many precautionery measures have been taken by the Commissioner of Elections to conduct fair & free election.

  • Saliya C A

    Thanks, Stuckinthe Middle,

    1. I have the same feelings and wish people will be brave enough to safeguard their franchise
    2. The signals so far do not indicate a significant change
    3. Yes, it’s critical. OUtcome might be different as it was in 2005. That’s why we have identify it as the 4th Critical Sensitive Factor.

    Dear Somewhat disgusted.

    I can tell you this 100 responses represent lower and upper middle class. Yes there were mobile users but I do not think they represent the bottom of the populace. 64 of them declaired that they voted for MR in 2005. Therefore the sample is somewhat bias towards MR.

    Dear We Srilanka,

    Me too very anxious about the actual results on 27 Jan. I am not trying to convince anybody but just did a research as I did in the past and possessed a credible track record for correct predictions. But as you wish to remain ananimous I donno how did you try to convince me and realised it was hard? Also, You will be vanished if SF wins.

  • niranjan

    I feel that MR has the edge due to the Sinhala rural vote. The rural vote will go to MR because he has ended the war. The people in those areas are happy that the war is over because their sons and daughters who are in the army will not get killed and they themselves can live in peace without the LTTE. He has also carpeted the roads in many rural areas(even though such activities might have had its origins during the UNF time) and development activities are taking place on a daily basis.

    The Tamil vote is split even though the majority will go to SF. The Muslim vote is also split. Not all Muslims will vote SLMC.

    I also believe the Indian poll is more accurate. MR is likely to get between 50%-53% of the vote. There are over 14, 000, 000 registered voters in this country now. However, not everyone will vote and there will be spoilt votes as well as valid votes that will go to candidates other than the main two. Therefore lets assume that there are about 10,000,000 valid votes for the two main candidates. That would mean the the winning candidate has to get 5 million votes to win this election. Last time MR got 4, 887,152 compared to RW’s 4, 706, 366. Therefore, MR has to increase his vote marginally whereas SF has to get over 200,000 + votes to reach the required number of 5 million votes. The question is has SF convinced the voter that he is the better choice in order to get that number of votes?

  • StuckInTheMiddle

    Dear Saliya,

    I have heard one of your predictions some years ago, that one of the Sri Lankan conglomate is finanacially mismanaged and that will collapse unless the chairperson listens to professional advice rather than getting carried away with what people around him say for personal benefits.It happened as you predicted.

    That shows your ability to analyse financial statments. SO LET’S TEST HOW GOOD YOU ARE IN POLITICS.

  • Saliya C A

    Dear Nieranjan,

    My article is not based on what I believed or my perceptions, it’s an analysis based on a survey.
    I am giving below my concerns which I have not incorporated into my predictions, which are purely based on survey results.
    Those concerns are;
    • unleashing violence to fear the general public and avoid people using their franchise
    • unbelievably escalation of violation of election laws
    • spreading racism when the need of the day is reconciliation
    • State media has became an obvious prostitute which offers free service to the government
    .

  • c a saliya

    Dear Niranjan,

    Thank you for letting us know your feeling and beliefs. Everyone has his/her own feelings, beliefs and perceptions about how people would vote.
    My article is mainly based on a survey on the intention of voting towards and past voting patterns. I tried my best not to contaminate the predictions with not only my own feelings and beliefs but even with my observations/concerns such as;
    • Violation of election laws; illegal/unlawful publicity and promotions by cut-outs, banners/posters etc
    • Use of state media for unlawful/illegal publicity, to spread baseless allegations and intense mudslinging.
    • Unleashing terror towards (assaulting / killing) opposition supporters
    • Election day (26th January) rigging
    • Provoking racism in the rural Sinhala dominated areas etc

  • NAVIDI SL

    Recently, we pampered 3 wonderful weeks (22/12 – 12/1) in the Paradise travelling around but for North and East. As an unrelated casual and somewhat ‘get to know people’ chatter, we randomly queried over 400 people on who they think the next president-elect would be. Of the 237 people responded, 122 endorsed the General showing he is nearly 3% ahead. However, we neither asked for the reasons nor their voting intentions.

  • Namal Jayasuriya

    Dear Dr Dayan Jayatilleke,

    I like the sound of your survey, as its from someone we can trust. Could you please, time permitting, let us know what your sample size was and some more information regarding the survey.

    Thanks

    Namal

  • niranjan

    CA Saliya,

    Both the leading candidates in the present Presidential election are not fit to govern the country. Therefore, your concerns and feelings are valid.
    I am in agreement with you on them. My biggest worry is the whipping up of racism in the south by both leading parties. Perhaps the only non-racist candidate in the contest in Bahu. I feel it is better to vote for a third party rather than two candidates who are similar in attitudes and beliefs at least where race is concerned. In my opinion race should not be an issue in a civilised country. Whether Sri Lanka is a civilised country is another debate.

    My predictions are not based solely on beliefs and my feelings. My predictions are based on discussions I have had with some rural people and from what I have observed in rural areas in the very recent past. My predictions are also based on the Presidential results of 2005.
    Another thing that should be kept in mind is that all Presidents of Sri Lanka have got a second term in office. The only exception was Premadasa.

  • Kanchana

    Now that your prediction has gone wrong, and also those 100 persons living in NZ had no feedback except e mail, sms, newspapers in the web,news from their friends and relatives.
    If you had the opportunity to travel around the villages, you should have come out with a different prediction Saliya. Biut i appreciate your efforts, and invite you to visit your former motherlasnd before the April elections and offer us an acceptable prediction.

  • Saliya

    Thanks, Kanchana,

    In fact those 100 persons are from Sri Lanka; urban as well as rural. I believe the perceptions changed in the last 10 days with the 3 of above conditions I have doubted in the answer to Niranjan, they are;

    Violation of election laws; illegal/unlawful publicity and promotions by cut-outs, banners/posters etc
    • Use of state media for unlawful/illegal publicity, to spread baseless allegations and intense mudslinging.
    • Unleashing terror towards (assaulting / killing) opposition supporters
    • Provoking racism in the rural Sinhala dominated areas etc

    However, now MR has won and I wish he will attend to rectify the division among ethnic groups and make a genuine effort for reconciliation and economic/socila development of Sri Lanka.

    Thanks again for reading my article and making comments