L’analyse pré-électorale proposée par deux chercheurs attachés à l’Université de Colombo, qui avait fait l’objet de fuites dans la presse, est la seule à ce jour à avoir été publiée et à avoir exposé les méthodes employées. C’est à ce titre que nous la proposons ci-dessous à nos lecteurs.
January 01, 2014
We, the undersigned, wish to put the record straight regarding a research we two jointly have undertaken to develop an empirical model to analyse the Sri Lankan voter behavior at Presidential elections over the past 10 years with the objective of contributing to the body of social research.
1. This research team was jointly led by us; T L Gunaruwan, Senior Lecturer attached to the University of Colombo and D S Jayaweera, an independent policy analyst, currently the Director General of the Sri Lanka Tourism Development Authority.
2. There is no formal involvement of the University of Colombo or SLTDA in this research, as it is the case with regard to most of our research activities, and any involvement one could imagine is limited to the extent of us being employed at these organisations, and the research team being comprised of several junior researchers and research students attached to the University of Colombo, and our academic and research contact points at several other Universities.
3. Our interests were purely academic and research in nature, and thus, we never released any of the interim outcomes of our still-ongoing research to media. However, we have recently observed several news items in media pertaining to a “Colombo University survey” on presidential elections, some of which even contained our names, but with some contents we cannot claim ownership.
4. Even though we cannot exclude the possibility of there being other surveys conducted by other researchers attached to the University of Colombo and other academic and research institutions on this subject, it would be unfair if such are quoted or published without the explicit consent of any such researchers, including us, and more so if the contents are not properly reported so that misleading opinions could be developed in the minds of the readers.
5. Since we realise the possibility of these numerous reports quoting out of context our research and their various outputs, we thought of putting the record straight by releasing to media the outcomes of our research as it stood at 27th December 2014, while requesting all concerned to avoid using these for any marketing or propaganda purposes. We, however, have no objection of anyone or organisation using these for further research work or for their internal planning and strategic decision making purposes.
T Lalithasiri Gunaruwan, Senior Lecturer (Economics), University of Colombo, and
D S Jayaweera, Policy Analyst (Currently the DG/SLTDA)
January 01, 2015
Attached: Outcomes of our Research as at Dec 27, 2014.
SURVEY ON VOTING PATTERNS – Presidential Election 2015
(Survey conducted by a research team of the University of Colombo)
(a) Relatively smaller sample (little less than 1000) compared to total population of little over 15 million. Survey forms were collected at convenience, and therefore, district-wise or electorate-wise representative sampling could not be realised. The results have to be perceived having proper understanding of this aspect.
(b) Over 80% of the sample were Sinhala Buddhists, and thus this sample has that bias, given that the overall share of Sinhala Buddhist voters is approximately 70% of the total votes.
(c) No survey responses were obtained from North (except one respondent from Jaffna, and two from Mullativu), while much larger number of responses were obtained from East, North Central, Western and Southern provinces. Only a few responses were obtained from Central, Uva, Wayamba and Sabaragamuwa Provinces.
(d) Sample characteristics indicate that it is overwhelmingly biased towards Mahinda Rajapaksa, as out of those who have cast their votes in the respective elections, over 75% had voted for MR in 2010 and 2005 elections (when his national average stood around 58% in 2010 and just above 50% in 2005).
1. Around 20% for the first time voters and nearly 13%-15% of the total respondents are still undecided. This indecision is still substantially high, and means that a lot could still change, particularly through the events and behavior of parties over the next two weeks.
2. Only 1% of the respondents in the aggregate sample pronounced preference to vote for a third candidate, while nearly 3% would vote for no one. These ratios are 2% and 7%, respectively, with regard to fresh voters. This indicates the likelihood of relatively lesser share of votes polled by any candidate other than the two main contenders, while the cancellation rate also is likely to be relatively low (please note that this percentage is compatible with what was observed at past national elections).
3. In the face of it, 44% to 39% split between MR and MS in a highly “MR-biased” (as explained in “d” above) sample points at the highly competitive nature of the forthcoming election, where the possibility of MR’s share going below 50% is substantial.
4. The present survey thus adopted a different approach to examine the trends. The 2015 vote base was divided into “new votes” and “non-new votes”; and the share of votes polled by individual candidates were considered with regard to “new votes”, while the “shifts” away from MR (in 2010) to MS (in 2015) and vice-versa were estimated and applied on to the population with regard to “non-new votes”.
5. As per the survey results, MR would obtain 38% of the “new-votes” while MS would obtain 33%. The balance would be accounted under votes obtained by other candidates, non-voters, and those who are still undecided.
6. With regard to “non-new votes”, nearly 28% of those who voted for MR in 2010 would shift towards MS this time and 11% would still be undecided, while 4% of those who voted for Sarath Fonseka in 2010 would shift towards MR in 2015 while 11% would be undecided. It is interesting to note that nearly 30% of those who voted for MR in 2005 Presidential election would deflect towards MS in 2015, and around 9% of those voted for Ranil Wickramasinghe in 2005 would deflect towards MR in the coming election.
7. Applying the basic survey results to the overall population, and using the 2010 election results as the basis for projections, it could be estimated that the chances are greater for MS to win this election with around 53% of the total valid votes, assuming that the absentee voter ratio stays unchanged at around 25%.
8. This pattern was cross examined through an “ethnicity based” analysis. This was attempted because the survey sample appeared overwhelmingly “Sinhalese”, and any shift calculated based on such a biased sample would only apply to that particular ethnic electorate. The results indicated the likelihood of MS securing a lead of over 2 lakhs of votes over MR in this electoral block, paving the way for an MS victory with a likely preference for him in Tamil and Muslim electorates. The survey results indicate that MS could secure the required 50% of total valid votes if he manages to get around half of the Tamil and Muslim votes, while MR would require over 60% of the Tamil and Muslim votes if he is to surpass MS in the overall competition and to cross over the crucial 50% threshold.
9. The results thus indicate the importance for both candidates of their margins secured in the Sinhalese electorate, which accounts for over 112 lakhs or three-fourth of the total number of registered voters. For instance, one lakh of lead lost for MS in the Sinhala electorate than the above indicated survey estimates would necessitate him to secure over 57% of minority votes (compared to nearly 50% required in the case of two lakhs of lead in the Sinhala electorate) for overall election victory.
10. Based on the survey estimates and a hypothetical 35:65 ratio of minority votes split between MR and MS respectively, the ethnicity specific analysis would yield an overall outcome closer to that projected under (7) above, with nearly 60 lakhs of votes (53%) polled by MS, 50 lakhs (44%) by MR, and 3% going to others.
11. However, it must be emphasised that nearly 20% of the first time voters, and around 13% of the overall electorate, are still undecided, which could go into either camp over the next two weeks, and could either reinforce or reverse the overall estimates made in this analysis. It is interesting to note that only less than 1% of the overall respondents pronounced interest in constitutional reforms or economic hardships such as high cost of living or unemployment. An overwhelming majority expressing preference for MR was for his war effort, strong leadership against external intervention, national unity and not reverting back to war or separatism, and those preferring MS were largely owing to accountability, transparency and governance issues in the MR rule. It is therefore very clear that the Sinhalese vote base, largely reflected in the sample surveyed, is sensitive to the national issue, and those still undecided among them, even though a majority has not given reasons for their indecision, are prone to tilt towards the camp which assures them of peace and national security. Strategic sail through the next two weeks of campaigning is therefore likely to be crucial for both candidates.
This analysis therefore tends to indicate a probable victory for MS at the forthcoming Presidential elections. However, there is a significant share of undecided voters, which makes the competition still wide open for both parties; and the estimates made in this analysis are likely to be influenced by the decision that will be made by these undecided voters over the next two weeks.
T L Gunaruwan (University of Colombo) and D S Jayaweera (SLTDA)
Joint Survey Analysts
December 27, 2014