Polarization of the Sri Lankan Polity: An Analysis of Presidential Elections (1982 – 2005)

Polarization of the Sri Lankan Polity: An Analysis of Presidential
Elections (1982 – 2005)
Yajni Warnapala

Roger Williams University
DOCS@RWU
Arts & Sciences Faculty Publications Arts and Sciences
5-2008

Roger Williams University, ywarnapala-yehiya@rwu.edu
Zufni Yehiya
Follow this and additional works at: https://docs.rwu.edu/fcas_fp
Part of the Mathematics Commons
Recommended Citation
Warnapala, Yajni and Zufni Yehiya. 2008. “Polarization of Sri Lankan polity: An analysis of presidential
elections (1982-2005).” Journal of Global Intelligence and Policy 2 (1).
This Article is brought to you for free and open access by the Arts and Sciences at DOCS@RWU. It has been
accepted for inclusion in Arts & Sciences Faculty Publications by an authorized administrator of DOCS@RWU. For
more information, please contact mwu@rwu.edu.
1
Polarization of the Sri Lankan Polity:
An Analysis of Presidential Elections (1982 – 2005)
ABSTRACT
Sri Lanka is a multi-ethnic, multi-religious developing country that has
enjoyed continuous universal adult franchise since 1931. Under a new
constitution enacted in 1978, Sri Lanka moved to a presidential system of
government. Since 1982 five presidential elections were conducted. This paper
analyzes voter behavior by looking at all the five presidential elections. This
study shows that all the winners of the presidential elections (except in 2005)
won them by appealing across racial and religious boundaries with a popular
mandate. In 2005, there was a shift; the winner was able to secure victory by
promoting a hard-line pro-Sinhala nationalistic platform. This signals a
departure from the previous elections, as in the past it was understood that
minority support is crucial to win the Presidency. The 2005 election sends a
dangerous signal to a country that is ravaged by ethnic violence for over 20
years. Further, this study looks at the voter behavior in urban vs. rural areas.
Similar to the red vs. blue states divide in the US, in Sri Lanka, there is a
strong urban-rural division in voter behavior. Logistic regression was used to
analyze the results of the elections.
Keywords: Sri Lanka, presidential elections, logistic regression, minority votes
2
Introduction
Introduced in 1978, the second republican constitution of Sri Lanka changed the
governmental structure to a Presidential system, replacing the “British style” parliamentary
form of government. This new constitution was a stark departure from its predecessor and
followed the French model, where the president holds executive powers and is directly elected
by the people. Under this new model, the president is the chief executive of the country,
supreme commander of the military, head of the government and the head of state. He or she
also has authority to appoint the Prime Minister, the Supreme Court, and the Cabinet and also
has the power to convene, suspend and dissolve the parliament.

,Prior to 1978, Sri Lanka enjoyed a fairly stable form of government, where there was a
ceremonial president as the head of state and the head of the military. The Prime Minister
was the head of the government. In the past the members of the parliament were elected by
the “first past the post” electoral system, the new constitution changed the method of election
to Proportional Representation (PR) (Warnapala & Yehiya, 2005). Sri Lanka became the first
democracy in the region to introduce a presidential form of government along with a PR
system. Both were two alien concepts to the region and even after 25 years, no other country
in the region has followed Sri Lanka’s lead and changed their form of government.
Many observers believe that this system was introduced purely because of the personal
ambitions of J. R. Jayawardena (Moore, 1990: 381). Even though there was no public clamour
for constitutional or electoral reforms, in 1977, the United National Party (UNP) led by J. R.
Jayawardena came to power promising that he will change the constitution and move the
country towards a presidential system of government. Jayawardena received a landslide
victory with 5/6 seats in the parliament (Commissioner of Elections). Along with
3
constitutional changes, Jayawardena also used his victory to steer the country away from the
rigid state control economic model (Athulathmudali, 1984: 76-77). He also introduced much
needed social and economic reforms and moved Sri Lanka’s foreign policy away from the
Soviet/Indian axis. On February 4, 1978, Jayawardena became the first executive president of
the country.
Ethnicity is a vital aspect in Sri Lankan politics and it is impossible to separate ethnic
politics from “regular” political discourse. Therefore, it is not surprising that Sri Lanka is the
home for one of the world’s longest and bloodiest civil wars. This conflict began in 1950s and
took a violent turn in the 1980s. The two protagonists are the majority Sinhalese and the
minority Tamils. This crisis that has taken over 70,000 lives is based on ethnicity and not on
religion (Reuters, 2008). The minority Tamils are demanding for a separate state called
“Eleam” in the north and east of the country based on some contested historical claims and
alleged discrimination by successive Sinhala-dominated governments (Gunasinghe, 1983;
Manogaran, 1987; Wilson, 1988).
Ethnically the Sri Lankan population can be categorized into four distinct
communities, Sinhalese (74%), Tamils (12.6%), Muslims (7.1%) and Indian Tamils (5.6%)
(Statistical Pocketbook, 1989: 14)1
. The religious composition of the country is 69.3%
Buddhists, 15.5% Hindus, 7.6% Muslims and 7.5% Christians (including Catholics)
(Statistical Pocketbook, 1989: 12). Sinhalese are predominately Buddhists and the Tamils
and Indian Tamils are predominantly Hindus and all three ethnic groups have a small, but
highly influential Catholic/Christian minority (Warnapala & Yehiya, 2005: 440).
In this study we analyzed the voter behavior in Sri Lanka via two lenses. Firstly we
analyzed each ethnic group’s voter behavior in the five presidential elections. Did the winner
4
appeal to all ethnic groups? Do you need the minority support to win the presidency? To
what extent the cultural terrain need to be understood to craft the campaign message?
Secondly we examined the rural-urban divide. Since 1982, all presidents have had a strong
urban bias, but in 2005 this stranglehold was broken and a rural candidate was able to defeat
an urban candidate.
The Presidential Electoral System
Sri Lanka elects its president every 6 years. Up to date there have been 5 elections
(1982, 1988, 1994, 1999 & 2005). In the Sri Lankan presidential elections one unique feature
is that both the first and the second rounds are conducted simultaneously (Reilly, 2001:
117-119). In most countries where there is a presidential system (ex: France, Liberia, Peru,
Haiti), if the winner of the first round does not get more than 50% of the vote, there is a
second round between the two top candidates.
In order to reduce cost and time, Sri Lankans vote for their candidates based on their
preferences, by marking 1, 2 and 3 (Reilly, 2001: 118). If the top candidate fails to get over
50% of the votes, all the other candidates get eliminated, except the second-place candidate.
Then second preferences of the eliminated candidates will be counted. If their second
choice is for one of the top two candidates, the votes will get re-distributed. It is worth
noting that up to date no election has gone to this second round counting and all the winners
have won in the first round itself.
The Study, Methodology and Data
To examine Sri Lankan voter behaviour, we selected the Logistic Regression technique.
Logistic Regression allows one to predict a discrete outcome from a set of variables that may
be discrete. Generally, the dependent or response variable is dichotomous, such as win/loss.
5
When the dependent variable is dichotomous, then the dependent variable can take the value 1
with a probability of success p, and value 0 with the probability of 1-p. This type of variable
is called a Bernoulli variable. Logistic Regression makes no assumption about the distribution of the independent variables. They do not have to be normally distributed, linearly related or of equal variance within each group. The relationship between the
predictor and the response variables is not a linear function; instead, the logistic regression function is the logit transformation of p.dichotomy, which is pi(1 – pi), allowing us to interpret the net effect of a dichotomous independent variable on the dependent variable’s probability (Bohnstedt & Knoke, 1994: 343). The data for this study was taken from the Department of Elections Web site. The web site had results of all the presidential and the parliamentary elections. The data on the distribution of ethnic groups in each polling division was directly obtained from the Department of Census and Statistics (unpublished data). They are based on the 2001 census. Prior to the introduction of the PR system, Sri Lanka had 160 electorates (polling divisions). Even after the PR system was abolished, political parties use these electorates as their organizing unit to appoint local party organizers and to set-up local political party branches. To this day unofficially this unit continues to function, and the Commissioner of Elections uses these units to announce election results. Therefore, for this study we decided to use these 160 polling divisions to observe the pattern of voter behavior2 .

By using the 6 ethnic distribution data, we rank ordered the ethnic composition of each polling division according to the proportion of each ethnic group. Points were assigned in the following manner: 00.01% – 20.00% = 1; 20.01% – 40.00% = 2; 40.01% – 60.00% = 3; 60.01% – 80.00% = 4; and 80.01% – 100.00% = 5. The logistic regression coefficient will show the change in the predicted log odds of win/loss for one unit change (20% increases in population) in the independent variable. We felt that a 20% interval is desirable, as it will keep provisions for population growth, migration and displacement of people due to the ethnic conflict and due to the devastating Tsunami of December 2004. Although the population distribution for each polling division would have changed between 1982 and 2005 the (period of our study), the available data is reasonably sufficient to capture the ethnic composition of each polling division for the entire duration of our study.The validity of this assumption is consistent with the ethnic composition of the members of the parliament for each area during the period of our study (Goonerathne, W. G & Karunaratne, 1996; Commissioner of Elections, 2006). Therefore, utilizing the 2001 census data to analyze the ethnic composition does not distort the picture of the polling divisions. We also categorized each of the 160 electorates according to the degree of urbanization3 . In Sri Lanka, local government administration is divided according to the degree of urbanization and the size of the population it serves (Baxter et al, 2002: 350-351). In order to divide along the degree of urbanization, we categorized the 160 electorates into municipalities, urban councils or pradeshiya sabha (local councils). Out of 160 polling divisions, there were 24 municipalities and 34 urban councils and the remaining 102 were pradeshiya sabhas. We excluded the pradeshiya sabas from the urban model. We classified 7 the urban council and the municipal councils in the following manner: Urban Councils (medium size cities) = 1, Municipal Councils (large metropolitan areas) = 2.

In the cases where local government councils and parliamentary polling divisions were not properly over lapping with each other, we aligned them according to the population centers. The data for categorizing a polling division whether it belongs to a municipality, urban council or a pradeshiya sabha was taken from the Commissioner of Elections. In the next section, we will examine each presidential election using logistic regression techniques. The results of our study revealed some interesting patterns that are not regularly taken into consideration in Sri Lankan electioneering. 1982 Elections Sri Lanka had its first presidential election in 1982. The election was between a hugely popular incumbent president (J. R. Jayawardena), supported by a strong government and against a very weak opposition. The main opposition leader and former Prime Minister Sirima Bandaranayke was prevented from contesting as she was found to be guilty of alleged corruption and abuse of power by a special presidential commission (De Silva & Wriggins, 1988). As a result, she lost her civic rights for seven years, thus preventing her from running for the presidency. No one was able to mount an effective challenge to Jayawardena and his well-run UNP’s political machinery. For most political observers, the only opposition candidate worth observing was Rohana Wijeweera, the founder-leader of the People’s Liberation Front (Janatha Vimukthi Peramuna) (JVP), a former Marxist terrorist group that had attempted to overthrow the government in 1971.

Jayawardena released the JVP leadership from prison in 1977 under the condition that they would accept democratic principles. The JVP, as a 8 political party deriving support from the younger and less formally educated rural youth has shown its ability to display some originality in its propaganda designed primarily to attract the socially oppressed and the disillusioned (Warnapala, 2001). It was in this 1982 election that the JVP tested its political strength at the ballot boxes. Table 1: 1982 election results Candidate Party Votes % J.R. Jayawardene United National Party 3,450,811 52.91% H.S.R.B. Kobbekaduwa Sri Lanka Freedom Party 2,548,438 39.07% Rohana Wijeweera Janatha Vimukthi Peramuna 273,428 4.19% Kumar Ponnambalam Akila Ilankai Thamil Congress 173,934 2.67% Colvin R. De Silva Lanka Sama Samaja Party 58,531 0.88% Vasudeva Nanayakkara Nawa Sama Samaja Party 17,005 0.26% Total Registered Voters 8,145,015 Total Polled 6,602,617 81.06% Total Valid Votes 6,441,667 Total Rejected Votes 80,470 No. needed for outright victory 3,261,073 No. of votes above the 50% mark 189,738 Date of Poll: Sept 20, 1982 No. of Polling Stations: 6985 Source: Election Commissioner Ethnic Model -1982 When we carefully examined the 1982 election results, we made a few interesting observations. In 1982, Jayawardena won 133 polling divisions (83%, p = 0.863), including all 7 Muslim divisions and the only Indian Tamil majority polling division. The logistic regression model that was used to depict the results of the winner was: L = -0.952Sinhala-
ranked2.095Tamil + 2.459Muslim + 1.027Indian-Tamil + 6.356.

Thus the Sinhala and Tamil polling division coefficients were negative and the Muslim and Indian Tamil polling division coefficients were positive. Therefore Sinhala and Tamil polling divisions had decreased log odds of voting for Jayawardena, while Muslim or Indian Tamil polling divisions had increased log odds of voting for Jayawardena. The effect of the Sinhala polling division coefficient was -11%. Similarly, the effect of the Tamil polling division coefficient was -25%, 9 the effect of the Muslim polling division coefficient was 29% and the effect of the Indian Tamil polling division coefficient was 12%. Thus Muslim and Indian Tamil polling divisions had an increased probability of voting for Jayawardena by 29% and 12% relative to non-Muslim and non-Indian Tamil polling divisions. This indicated that even though Jayawardena was able to secure votes from all segments of the country, he had more support from Muslims and Indian Tamils. This is consistent with UNP’s philosophy and Jayawardena’s voter base. Muslim community has traditionally supported the UNP and they directly benefited from the UNP’s economic policies. In the case of the Indian Tamils, the UNP managed to get the support of the main political party/trade union, the Ceylon Workers Congress (CWC). Its leader S. Thondaman was a member of the Jayawardena’s cabinet (De Silva & Wriggins, 1988). In the 1982 ethnic model, the Sinhala and the Tamil polling division coefficients were negative. We argue that this is because the Tamil community’s vote went to the Tamil candidate (Kumar Ponnambalam), who won 8 out of 25 Tamil majority electorates (Election Commissioner). The majority Sinhala vote was fragmented among the Sinhala or Marxist political parties and the main opposition SLFP, which always had a strong Sinhala voter base. Therefore, Jayawardena having an 11% negative coefficient effect in the Sinhala electorates relative to non-Sinhala electorates is understandable. Urban Model -1988 When we looked at the corresponding urban model, the logistic regression equation was as follows: L = 0.188urban + 1.57. In this model the urban polling division coefficient was positive. Jayawardena won 86.2% (p = 0.862) of the urban polling divisions and the effect of the urban coefficient was 2%. According to this model, an urban polling division had 10 increased the chance of voting for Jayawardena by 2% relative to a non-urban (rural) polling division. From this model, we interpolated that the least urban polling division had an 85% chance of voting for Jayawardena. Figure 1: The more urban a polling division becomes Jayawardena’s probability of winning increases.

1988 Elections
Sri Lanka conducted its 2nd presidential election in 1988. UNP nominated Jayawardena’s Prime Minister R. Premadasa. It was one of the most violent elections in Sri Lankan history (Baxter, et al, 2002: 345). SLFP nominated its leader Sirimavo Bandaranaike, who had re-gained her civic rights after a presidential pardon in January 1986. In the Sinhala heartland, the JVP control was strong and the JVP’s boycott might have robbed 1988 of a Bandaranayake victory. Premadasa, with the help of the state machinery, cheated and his
urban appeal scraped a narrow victory.

Table 2: 1988 election results
Candidate Party Votes %
Ranasinghe Premadasa United National Party 2,569,199 50.43%
Sirimavo Bandaranaike Sri Lanka Freedom Party 2,289,960 44.95%
Osvin Abeygunasekara Sri Lanka Mahajana Pakshaya 235,719 4.63%
Total Registered Voters 9,375,742
Total Polled 5,186,223 55.32%
Total Valid Votes 5,094,778
Total Rejected Votes 91,445
No. needed for outright victory 2,547,389
No. of votes above the 50% mark 21,810
11
Date of Poll: Dec. 19, 1988
No. of Polling Stations: 8060
Source: Election Commissioner
Ethnic Model – 1988
In 1988, Premadasa won 61.3% of the polling divisions (p = 0.613) and the logistic regression model for Premadasa was as follows: L = -0.856Sinhala – 1.104Tamil + 0.847Indian-Tamil + 4.934. Thus the Sinhala and Tamil polling division coefficients were negative and the Indian Tamil polling division coefficient was positive. Therefore Sinhala and Tamil polling divisions had decreased log odds of voting for Premadasa, while Indian Tamil polling divisions had increased log odds of voting for Premadasa. In this case, the effect of the Sinhala polling division coefficient was -20%. Similarly, the effect of the Tamil polling division coefficient was -26% and the effect of the Indian Tamil polling division coefficient was 20%. Thus Indian Tamil polling divisions had an increased chance of voting for Premadasa by 20% relative to non-Indian Tamil polling divisions. Also, the Sinhala and Tamil polling divisions had a decreased probability of voting for Premadasa by 20% and 26% relative to non-Sinhala and non-Tamil polling divisions. For the 1988 elections, the Muslim polling division’s coefficient was dropped from the model because of its high significance. Urban Model – 1988 In the 1988 urban model and the logistic regression equation was as follows :L = 0.887urban – 0.887. Premadasa won 58.6% (p = 0.586) of the urban polling divisions and the urban coefficient was positive. Therefore the effect of the urban coefficient was 21%. Thus in 1988, move an urban polling division had an increased chance of voting for Premadasa by 21% relative
to a non-urban (rural) polling division. From the above model, we interpolated that the least urban polling division (rural) had a 50% chance of voting for Premadasa.
Figure 2: The more urban a polling division becomes Premadasa’s chance of winning increases.
12
It is apparent from our urban model that Premadasa got elected to the presidency due to his urban appeal. Premadasa was a “cross-over” politician who represented the only Muslim-majority Colombo electorate in the parliament. Born to a poor working-class family,
this self-made populist was a street politician with a “common man” appeal. On the May Day of 1992, Premadasa was murdered by an LTTE suicide bomber. Premadasa was succeeded by his Prime Minister D. B. Wijetunge. Wijetunge was able to control the lawlessness in the
country.
1994 Elections
By 1994, after 17 years of UNP rule, the country was fatigued with the UNP and the people were yearning for a change. Bandaranayake was replaced by her charismatic daughter Chandrika Bandaranaike Kumaratunga. She was able to shelf the old socialist philosophy and move SLFP away from Sinhala nationalism. She had solid liberal credentials and promoted cultural pluralism. Kumaratunga was one of the few Sinhala politicians who took personal and political risks in meeting with the LTTE leadership in the 1980s. She spoke of national reconciliation and was very popular in the Tamil heartland. This was a mark of departure for SLFP, as a party that had always portrayed itself as a Sinhala nationalist party.
13
In 1994, first the parliamentary elections were held. SLFP and its grand coalition the People’s Alliance (PA) led by Kumaratunga, won with a narrow victory (113 vs. 112). With this razor-thin majority, Kumaratunga formed a coalition government with the Sri Lanka Muslim Congress (SLMC) and became the Prime Minister. A few months later the Presidential elections were conducted and the UNP nominated Gamini Dissanayake, a charismatic popular leader previously marginalized by Premadasa, who in turn had co-founded another political
party. During the election campaign, Dissanayake was killed by anSLFP-led LTTE suicide bomber. Dissanayake’s candidacy was replaced by his widow, a political novice4
.
Table 3: 1994 election results
Candidate Party Votes %
Chandrika Kumarathunga People’s Alliance 4,709,205 62.28%
Nihal Galappaththi Sri Lanka Progressive Front 22,749 0.30%
Srimathi Dissanayake United National Party 2,715,285 35.91%
A.J. Ranasinghe Independent 22,752 0.30%
Harischandra Wijethunga Sinhalye Mahasammatha
Bhoomiputhra Pakshaya
32,651 0.43%
Hudson Samarasinghe Independent 58,886 0.78%
Total Registered Voters 10,945,065
Total Polled 7,713,232 81.06%
Total Valid Votes 7,561,526
Total Rejected Votes 151,706
No. needed for outright victory 3,780,763
No. of votes above the 50% mark 1,928,442
Date of Poll: Sept 11, 1994
No. of Polling Stations: 9580
Source: Election Commissioner
Dissanayake’s widow was no match for Kumarathunga. Kumarathunga as the incumbent Prime Minister has already consolidated her power. In this environment, Kumaratunga’s victory was a certainty. Kumaratunga as predicted got a landslide victory. She won all but 1 polling division in the country. Therefore we could not use the logistics regression models to analyze the 1994 election, as the independent variable (win/loss) was not dichotomous in this case.
14
1999 Elections
The main contenders for the 1999 elections were the incumbent President Kumarathunga and UNP’s Ranil Wickramasinghe. Between 1994 and 1999, Kumarathunga continued UNP’s neo-liberal economic policies and bridged the gap between the UNP and the SLFP in economic policies and progress. As a result, the 1999 election was not a battle on bread and butter issues, but on how to handle to ethnic crisis (Jayasuriya, 2005). On the eve
of the elections, there was a Tamil Tiger suicide bomb attack. The target was Kumarathunga’s last election rally. She survived the attack, but permanently lost sight in one eye. The public sympathy after the attack, the power of the incumbency, and Wickramasinghe’s lack of charisma, ensured a second-term victory for Kumarathunga. The election was marred with violence and the international election observers found several election violations on both sides (Law and Society Trust, 2000; PAFREL, 1999).
Table 4: 1999 election results

Candidate Party Votes %
Abdul Rasool Sri Lanka Muslim Katchi 17,359 0.21%
Alwis Weerakkody Premawardhana People’s Freedom Front 3,983 0.05%
Ariyawansha Dissanayaka Democratic United National Front 4,039 0.05%
M. D. Nandana Gunathilaka Janatha Vimukthi Peramuna 344,173 4.08%
Kamal Karunadasa People’s Liberation Solidarity Front 11,333 0.13%
Chandrika Kumaratunga People’s Alliance 4,312,157 51.12%
Tennyson Edirisuriya Independent 21,119 0.25%
W. V. M. Ranjith Independent 27,052 0.32%
Ranil Wickramasinghe United National Party 3,602,748 42.71%
Rajiva Wijesinha Liberal Party 25,085 0.30%
Vasudeva Nanayakkara Left and Democratic Alliance 23,668 0.28%
Hudson Samarasinghe Independent 7,184 0.09%
Harishchandra Wijayatunga Sinhalaye Mahasammatha Bhoomiputhra Pakshaya
35,854 0.43%
Total Registered Voters 11,779,200
Total Polled 8,635,290
Total Valid Votes 8,435,754
Total Rejected Votes 199536
No. needed for outright victory 4,217,877
No. of votes above the 50% mark 94,280
Date of Poll: Dec 21, 1999
No. of Polling Stations: 9912
15
Source: Election Commissioner
Ethnic Model – 1999
The SLFP-led People’s Alliance (PA) candidate Kumaratunga won 81.3% of the
polling divisions (p = 0.813). The logistic regression model was :L = 1.196Sinhala + 0.465Tamil + 0.581Muslim – 0.729Indian Tamil -3.655. Thus in this election, the Sinhala, Tamil and Muslim polling division coefficients were positive and the Indian Tamil polling
division coefficient was negative. Therefore Sinhala, Muslim and Tamil polling divisions had increased log odds of voting for Kumaratunga, while Indian Tamil polling divisions had decreased log odds of voting for Kumaratunga. An Indian Tamil polling division had
decreased chance of voting for Kumaratunga by 11% relative to a non-Indian Tamil polling division.
When we carefully examine Kumaratunga’s victory, unlike Jayawardena or Premadasa, it is clear that she was able to get the support of the majority Sinhalese. While Jayawardena and Premadasa in the ethnic models had negative coefficients for the Sinhala and the Tamil polling divisions, and traditional Kumaratunga had a positive coefficient in both elections in which she won the presidency. This goes to show SLFP’s strong Sinhala voter base. It is apparent that Kumaratunga’s attempt to “secularize” the party had no significant impact on the
ballot boxes. Extreme Sinhala voters had not alienated the party and had continued to support the SLFP, in spite the JVP putting forward a candidate with an ultra-Sinhala nationalistic platform. It seems Kumaratunga’s popular appeal assisted the party from any backlash from
the hard-line elements within the Sinhala voter base. To her credit, she moved the party away from communal politics. In the ethnic model, Kumaratunga also had a positive effect on the Muslim polling divisions. Even though traditionally Muslims have supported the UNP,
16
Kumaratunga was able to get their votes in both elections. This is because she was able to get
the support of the SLMC and its splinter groups. The only group that Kumaratunga was
unable to get a positive coefficient in the ethnic model was in the case of the Indian Tamils.
Indian Tamils are considered a “bloc vote” led by their trade union the CWC. CWC aligned
itself with the UNP in all the presidential elections.
Urban Model – 1999
For the 1999 elections, the logistic regression equation for the urban model was as
follows: L = -0.668*urban + 1.846. Even though Kumaratunga won 70.7% (p = 0.707) of the
urban polling divisions, as the urban polling division’s coefficient was negative, the effect of
the urban coefficient was -14%. Thus an urban polling division had a decreased chance of
voting for Kumaratunga by 14% relative to a non urban (rural) polling division. From this
model we interpolated that a least urban polling division (ex: Deniyaya in the Matara district)
had a 76% chance of voting for Kumaratunga.
Figure 3: More urban a polling division becomes, Kumaratunga’s probability of winning slowly diminishes.

But more urban a polling division got the percentage of votes Kumaratunga received
decreased. Therefore in comparison, in the urban model, Kumaratunga was not very
17
successful. This showed that the SLFP is still very much a rural political party with a strong
rural voter base, while the UNP is a strong urban party with a strong urban voter base.
Elections 2005
Next election was held in 2005. The run-up for this election was marred with
controversies and legal battles. An ultra Sinhala-Buddhist party called Jathika Hela Urumaya
(JHU) (National Heritage Party) went to court5
. UNP again nominated Wickramasingha, who
had led the UNP for over 10 electoral defeats (at national and local levels) (Jeyaraj, 2005).
The SLFP led coalition UFPA (United Freedom People’s Alliance) nominated Kumaratunga’s
Prime Minister Mahinda Rajapaksa. Although Rajapaksa came from a rural political dynasty,
his image and approach was very much akin to Premadasa’s image. During the campaign he
portrayed himself as a down to earth, small town lawyer, with a Sinhala-Buddhist and a
socialist political orientation. Rajapakshe was also considered as a non-intellectual and he
even refused to debate Wickramasingha. On the other hand UNP’s Wickramasingha was seen
as non-charismatic, intellectual, aloof and elite. The contrasts were similar to what occurred
in the US during the 2004 presidential elections, between George Bush and John Kerry: Kerry
was the aloof, intellectual, aristocrat, while Bush was perceived as down to earth friendly,
non-intellectual, evangelical Christian. During the campaign, most observers would agree that
Kumaratunga was undercutting Rajapaksa’s campaign to ensure a Wickramasingha victory
(The Sunday Leader, 2005a; 2005b).
Table 5: 2005 election results
Candidate Party Votes %
Mahinda Rajapaksa United Freedom People’s Alliance 4,887,152 50.29%
Ranil Wickramasinghe United National Party 4,706,366 48.43%
Siritunga Jayasuriya United Socialist Party 35,425 0.36%
Ashoka Suraweera Jathika Sangwardhena Peramuna 31,238 0.32%
Victor Hettigoda Eksath Lanka Podujana Pakshaya 14,458 0.15%
Chamil Jayaneththi New Left Front 9,296 0.10%
Aruna De Zoysa Ruhunu Janatha Party 7,685 0.08%
18
Wimal Geeganage Sri Lanka National Front 6,639 0.07%
Anura De Silva United Socialist Party 6,357 0.07%
J.A. Ajith Kumara Democratic Unity Alliance 5,082 0.05%
Wije Dias Socialist Equality Party 3,500 0.04%
P. Nelson Perera Sri Lanka Progressive Front 2,525 0.03%
Shantha Dharmadwaja United National Alternative Front 1,316 0.01%
Total Registered Voters 13,327,160
Total Polled 9,826,908 73.74%
Total Valid Votes 9,717,039
Total Rejected Votes 109,869
No. needed for outright victory 4,858,520
No. of votes above the 50% mark 28,632
Date of Poll: Nov. 17, 2005
No. of Polling Stations: 10,486
Source: Election Commissioner
Ethnic Model – 2005
In a hard fought battle, SLFP’s Mahinda Rajapaksa was the winner. In 2005,
Rakapakse won 58.1% of the polling divisions (p = 0.581). The logistic regression model
was: L = 2.754Sinhala + 1.911Tamil – 1.352Muslim – 1.464Indian Tamil – 11.461.Thus
the Sinhala and Tamil polling division coefficients were positive and the Muslim and the
Indian Tamil polling division coefficients were negative. Therefore a Sinhala or a Tamil
polling division had increased log odds of voting for Rajapaksa, while an Indian Tamil or
Muslim polling division had decreased log odds of voting for Rajapaksa. The effect of the
Sinhala polling division coefficient was 67%. Similarly the effect of the Tamil polling
division coefficient was 46%, the effect of the Muslim polling division coefficient was -33%
and the effect of the Indian Tamil polling division coefficient was -36%. Obtaining a positive
coefficient from the Tamil polling divisions warrants an explanation and it will be discussed
later. Thus an Indian Tamil and Muslim polling divisions had decreased chance of voting for
Rajapaksa by 36% and 33% respectively relative to non-Indian Tamil and non-Muslim
polling divisions. While a Sinhala or a Tamil polling division had increased probability of
voting for Rajapaksa by 67% and 46% relative to a non Sinhala and non Tamil polling
19
division. Rajapakesha’s election victory has created long term implications to Sri Lankan
politics. He was able to break a cardinal rule in Sri Lankan presidential politics, i. e. that a
serious candidate cannot win without minority support.
This sends out a dangerous message to the minorities. It says that Sri Lanka is no
longer a pluralistic multi-cultural, multi-racial country, but a Sinhala-Buddhist country, and
that minority votes do not count, thereby giving ammunition to the groups such as the LTTE
to demand for a separate state. It is in this backdrop Rajapaksa was able to get a positive
coefficient for Sinhala and the Tamil poling divisions. Obtaining a positive coefficient from
the Tamil polling divisions warrants an explanation. Between 2002 and 2008 Sri Lankan
government had a cease-fire agreement (CFA) with the LTTE. The Norwegians negotiated
this ceasefire and until the end of the ceasefire, they served as the monitors and the
intermediaries for the peace talks. It was assumed that the Tamils who live in the North and
East would vote for Wickramasingha, as he was the architect of the CFA. Further, prior to the
presidential elections, LTTE pledged that they will not interfere with the elections and would
give the Tamil population a free hand in voting (The Sunday Leader, 2005c; The Global
Information Gateway, 2005). But days before the elections, LTTE issued a decree, banning
the Tamils living in the north and east from voting (Hindustan Times, 2005; Jayasinghe and
Gunasekera, 2005; Rhode, 2005). As a result of this forced boycott, only .36% of the people
voted in the Tamil heartland (Jaffna District), thus preventing over 700,000 from voting
(Commissioner of Elections, 2006; European Union Report, 2006). Most of the polling
divisions in the Northeast (Trincomalee District) and the East (Batticoloa & Digamadulla
Districts), where the boycott could not be enforced were won by Wickramasingha. In the
Vanni district where the boycott was semi successful, where only 26% voted,
Wickramasingha won 2 out of 3 polling divisions. In the final tally,
1.86% more than Wickramasingha and if not for the boycott, the outcome would have being
vastly different. For most observers, the
casting the most important vote (the boycott). Wickramasingha, as the Prime M
between 2002 and 2004, was able to turn the world opinion against the LTTE and push them
towards a negotiated settlement.
Urban Model – 2005
In the analogues urban model, the logistic regression model was:
1.123. Rajapaksa won 60.3% (p = 0.603) of the urban polling divisions and as the urban
polling coefficient was negative, the effect of the urban coefficient was
urban polling division had a decreased chance of voting for Rajapaksa by 24% relative to a
non urban (rural) polling division. When we changed the logistic model into the analogous
exponential model, the new model was:
From this model we calculated that
voting for Rajapaksa.
Figure 4: More urban a polling division becomes, Rajapaksa’s chance of winning steadily decreases.
It is clear from the figure
percentage of votes decreased.
the urban areas. Urban polling divisions are UNP’s traditional strong holds. In the 2005
Wickramasingha won 2 out of 3 polling divisions. In the final tally, Rajapaksa
86% more than Wickramasingha and if not for the boycott, the outcome would have being
vastly different. For most observers, the Rajapaksa victory was engineered by the LTTE, by
casting the most important vote (the boycott). Wickramasingha, as the Prime M
between 2002 and 2004, was able to turn the world opinion against the LTTE and push them
towards a negotiated settlement.
In the analogues urban model, the logistic regression model was: L= –
won 60.3% (p = 0.603) of the urban polling divisions and as the urban
polling coefficient was negative, the effect of the urban coefficient was -24%
urban polling division had a decreased chance of voting for Rajapaksa by 24% relative to a
on urban (rural) polling division. When we changed the logistic model into the analogous
exponential model, the new model was: P = e-1.005urban+1.123/ (1+ e-1.005urban+1.123
From this model we calculated that a least urban polling division had a 53% c
Figure 4: More urban a polling division becomes, Rajapaksa’s chance of winning steadily decreases.
figure that more urban a polling division become Rajapakse’s
d. It is not surprising that Rajapaksa’s performance was weak in
the urban areas. Urban polling divisions are UNP’s traditional strong holds. In the 2005
20
Rajapaksa got only
86% more than Wickramasingha and if not for the boycott, the outcome would have being
victory was engineered by the LTTE, by
casting the most important vote (the boycott). Wickramasingha, as the Prime Minister
between 2002 and 2004, was able to turn the world opinion against the LTTE and push them
-1.005*urban +
won 60.3% (p = 0.603) of the urban polling divisions and as the urban
%.Therefore an
urban polling division had a decreased chance of voting for Rajapaksa by 24% relative to a
on urban (rural) polling division. When we changed the logistic model into the analogous
1.005urban+1.123).
polling division had a 53% chance of
Figure 4: More urban a polling division becomes, Rajapaksa’s chance of winning steadily decreases.
that more urban a polling division become Rajapakse’s
’s performance was weak in
the urban areas. Urban polling divisions are UNP’s traditional strong holds. In the 2005
21
election the tradition of urban/rural divide has continued. When we carefully dissected the
five presidential elections, the results of this study exposed some facts that were clearly
interesting. What is evident is that the Sri Lankan polity is gradually getting “Balkanized”
along ethnic, religious and regional lines. Today there are political parties that cater to all
these groups. Jayawardena & Kumaratunga were able to appeal across ethnic and racial lines
and also along the rural/urban divide. Even Premadasa, who campaigned during one of the
most violent times was able to get more rural votes and had popular appeal across the
spectrum. This characteristic faded in the 2005 elections. No major candidate could appeal to
all groups.
The Sri Lankan urban/rural divide is similar to the red states/blue states divide in the
US. UNP dominates the urban areas, while the SLFP dominates the rural Sinhala polling
divisions. This urban/rural divide was breached in the 1982 (Jayawardena) and 1994
(Kumaratunga) elections. Both winners were able to encroach into each others territory. It
can be argued that this breach happened not because of the winning candidate, but because of
the weakness of the loosing candidate. In 1982 SLFP (Hector Kobbekaduwa) and in 1994
UNP (Srima Dissanayake) were weak opposition candidates, who did not appeal to their party
or to their base; therefore, because of lack of choice, the public voted by default.
Now let’s re-examine the 2005 elections and its future implications. Because
Rajapaksa was able to win the presidency with a hard-line agenda, this will be seen in the
future as the recipe for winning elections. Long-term implications of the Rajapaksa victory
are profound, especially for the ethnic crisis that is engulfing the northeast region. During the
campaign he pledged to change SLFP’s stated policy of supporting a Federal solution for the
ethnic problem and revert back to a unitary state (Mahinda Chintana, 2005: 32). He also said
22
that he would cancel the post-tsunami aid-sharing program (P-TOM) with the LTTE (Daily
Mirror, 2005); amend the CFA (Mahinda Chintana, 2005: 35) and replace Norway as the
mediator6
.
Rajapaksa ran his campaign similar to a United States presidential candidate in the
primary elections. In the US, during the primaries, “playing to the base” is a standard
practice, so the base will get motivated to vote. But during the presidential elections both
major party candidates try to “appeal to the middle.” In Sri Lanka, the opposite happened;
“playing to the base” was Rajapaksa’s winning strategy, while Wickramasingha’s “appealing
to the middle” back fired as he was seen as a “dove” that would cave in to the LTTE. It is
clear that Sri Lanka is a highly polarized country and the politicians are clearly exploiting
these cleavages for their advantage. Until there is a national consensus on how to resolve the
ethnic question, the crisis will continue with a military stalemate.
History is full of examples, where hard-line hawkish leaders made concessions to their
opposing parties and their fellow citizens trusted the deals made by these leaders. Richard
Nixon with China, Ronald Reagan with Soviet Union, Arieal Sharon on Gaza and Menachim
Begin with Egypt are just few examples. Perhaps Rajapaksa with his “hawkish” agenda can
come out of this deadlock. However, there has to be a partner to negotiate. All the
indications are that the Tamils have not produced one.
Concluding Remarks
In conclusion, the presidential system of government has further polarized an already
fragmented country. This system of government also introduced the possibility of cohabitation between an all-powerful president from one party and a legislature from a different
party. Sri Lankan political culture does not allow for this kind of political co-habitation.
23
From 2002 and 2004, there was a period of cohabitation, and the public did not see a level of
“check and balance” but saw constant “check mate” between the president and the parliament.
The Presidential system has also created an “imperial” presidency. It was President
Jayawardena, who famously said that he had the power to do everything “other than making a
man a woman, or vice versa” (Asian Human Rights Report, 2006). With this kind of an
attitude towards the presidency by the architect of the presidential system, one might view
suspiciously the reasons behind moving the country towards an all-powerful presidential
system, while undermining the authority of the legislature. Further, Sri Lanka enjoyed a
stable governmental structure for over 70 years and there was no real justification to
consolidate power under one person.
The cease-fire which was in existence since 2002 was cancelled in early 2008. After
the 2005 elections, the cease-fire was blatantly violated by both parties. Since November
2005 violence has steadily increased and within 6 months the death toll has exceeded 5000
(Reuters, 2008). Most of the blame for the escalation of violence lies directly at the feet of
the LTTE. After conspiring to elect a hard-line Sinhala president (via an election boycott),
LTTE has steadily increased their terror campaign. LTTE is currently banned in the European
Union, the US, Canada, India and host of other countries. Within the UNP the failure at the
elections have prompted calls for reforms. UNP wants to project that they too can cater to the
needs of the majority Sinhalese and that they too prefer to distance themselves from the
minorities. This phenomenon is similar to the debates within the Democratic Party in the US.
The moderate Democrats are demanding that the party should appeal to the middle, while the
left wing Democrats are pushing the interest of the minorities, such as Blacks, Hispanics,
24
Unions & Women’s groups. It is clear that the UNP is in danger of imploding, not because it
went “soft” on the ethnic issue, but because of lack of leadership.
As indicated at the beginning, the presidential system has contributed to further
polarization in Sri Lankan politics. Once this system was seen as a system that can unify the
nation, but today it has increasingly become a tool for divisive electoral politics. Because the
2005 election was won only with the Sinhala votes, Rajapaksa will not feel he represents the
entire polity, and will not have any obligation to take in to consideration of minority interests
in policy issues. Further, in the future presidential elections the candidates will not see a
reason to cater to the needs of the minority and advocating a hard-line anti minority posture
will be seen as the recipe for victory. This will send out a dangerous signal to the minorities,
which in return will fuel further political alienation and resentment. Further research is needed
on the efficacy of the presidential system in Sri Lankan politics and its long-term
implications. There are several proposed constitutional amendments, among them are: revert
back to the old “British style” parliamentary system, reduce the power of the executive
president, create multiple vice-presidents representing all the ethnic groups, etc. No clear
alternative is in sight.
List of References
Asian Human Rights Commission (2006) Press Release on Sri Lanka. Available:
http://www.ahrchk.net/statements/mainfile.php/2006statements/572/
Athulathmudali, Lalith. (1984) ‘The Elections of 1982’ in James Manor (ed.) Sri Lanka in
Change and Crisis, pp. 76-83. New York: St. Martin’s Press.
Baxter, Craig, et al (2002) Government and Politics in South Asia. (5th Edition). Westview
Press, London.
25
BBC News. (2005) ‘Sri Lanka PM ‘seeks presidency’. Available at:
http://news.bbc.co.uk/2/hi/south_asia/4722369.stm
Bohrnstedt, George W., Knoke, D. (1994) Statistics for Social Data Analysis.
(3rd Edition). Illinois: F.E. Peacock Publishers, Inc.
Commissioner of Elections. (2006) Results of Sri Lankan Elections: 1948 – 2005.
Available at: http://www.slelections.gov.lk/index.html
De Silva, H. L. (1989) Indo-Sri Lanka Peace Accord: An Appraisal, in Shelton Kodikara
(ed), Indo-Sri Lanka Agreement of July 1987. Sri Lanka: Sridevi Printers.
De Silva, K. M. & Wriggins, Howard. (1988) J.R. Jayewardene of Sri Lanka a Political
Biography. Honolulu: University of Hawaii Press.
European Union Report. (2006) The Final Report of the EU-EOM for the 2005 Presidential
Elections. European Union. Available at: http://www.dellka.cec.eu.int/en/index.htm
Goonerathne, W. G. & Karunaratne, R. S. (1996) Tenth Parliament. Sri Lanka:
Associated Newspapers of Ceylon Ltd.
Gunasekara, Tisaranee. (2005) “The Victor, the Vanquished and the Tiger.” The Island.
Available at: http://www.island.lk/2005/11/20/features1.html
Gunasinhe, P. A. T. (1983) The Tamils of Sri Lanka. Sri Lanka: H. W. Cave & Co. Ltd.
Hindustan Times. (2005) LTTE asks Tamils not to take interest in poll. The Hindustan Times.
November 10, 2005. Available at:
http://www.hindustantimes.com/news/7598_1542607,000500020002.htm
Jayasuriya, L. (2005) Changing Face of Electoral Politics in Sri Lanka, (1994 – 2004).
Singapore: Marshall Cavendish
26
Jeyaraj, D. B. S. (2005) Ranil must not bid farewell to politics at this time. Minority Matters.
Available at: http://minormatters.blogspot.com/2005/11/
ranil-must-not-bid-farewell-to.html.
Law and Society Trust. (2000) Presidential Election 1999: People’s Trust. Colombo: Law
and Society Trust.
Mahinda Chintana. (2005) Colombo: Print & Print Graphics.
Manogaran, Chelvadurai (1987) Ethnic Conflict and Reconciliation in Sri Lanka. Honolulu:
University of Hawaii Press.
Mendis, Dulith. (2005) Janabala Meheyuma Final Day Attracts Massive Crowds to
Colombo, Sri Lanka, Lanka Everything. Available at:
http://lankaeverything.com/index.php?subpage=news&pid=20050712123929.php&cat
=archive
Moore, M. (1990) Economic Liberalization vs. Political Pluralism in Sri Lanka. Modern
Asian Studies. 24 (2): 341-383.
PAFREL (Peoples Action for Free and Fair Elections). (1999) Voters Rights. Colombo:
PAFREL.
Reilly, Benjamin (2001) Democracy in Divided Societies: Electoral Engineering for Conflict
Management. Cambridge: Cambridge University Press.
Rhodes, Arthur. (2005) Sri Lanka’s Presidential Election: The irony of the LTTE boycott.
Asia Media. Available at: http://www.asiamedia.ucla.edu/article.asp?parentid=33499
Samarakone, P. (1984) The Conduct of the Referendum. In James Manor (ed.). Sri Lanka in
Change and Crisis. New York: St. Martin’s Press.
Reuters News Agency (2006) Truce dead, Sri Lanka war set to escalate-experts.
27
Jan 3, 2008. Available at:
http://www.reuters.com/article/featuredCrisis/idUSCOL273680
Statistical Pocketbook of the Democratic Socialist Republic of Sri Lanka. (1989) Colombo:
Department of Census and Statistics.
The Daily Mirror. (2005) P-TOMS or Mahinda-TOMS, joint mechanism with LTTE
essential. Available at http://www.dailymirror.lk/2005/10/19/front/1.asp
The Global Information Gateway. (2005) Sri Lanka: Tamil Rebels to allow campaigning for
Presidential Elections. Available at:
http://www.adnki.com/index_2Level.php?cat=Politics&loid=8.0.222101744&par=0
The Indo-Sri Lanka Accord. (1989) Indo-Sri Lanka Agreement of July 1987, in Shelton
Kodikara (ed). Sri Lanka: Sridevi Printers.
The Island (2005) SC Rules Presidential Poll in 2005. August 27, 2005. Available at
http://www.island.lk/2005/08/27/
The Sunday Leader. (2005a) CBK-Mahinda in fierce clash. Available at:
http://www.thesundayleader.lk/20051002/news.htm#CBK
(2005b) Ranil takes moral high ground as SLFP and UNP get closer to a deal.
Available at: http://www.thesundayleader.lk/20051016/politics.htm
(2005c) LTTE wants candidates to campaign in Vanni.
Available at: http://www.thesundayleader.lk/20051030/news.htm#LTTE
Warnapala, Yajni & Yehiya, Zufni. (2005) ‘Minority Electoral Politics: A Sri Lankan
Case Study’. Journal of Asian and African Studies. 40 (6): 439 – 462.
Warnapala, W. A. Wiswa (2001) Electoral Politics in Sri Lanka: A study of the
Parliamentary General Election of December 2001. Godage International
28
Publishers (PVT) LTD
Wickremasinghe, Nanda (2005) Factional infighting in Sri Lankan opposition following
electoral defeat. World Socialist Web Site
Available at: http://www.wsws.org/articles/2005/dec2005/sril-d03.shtml
Wilson, A. J. (1980) The Gaullist System in Asia: The Constitution of Sri Lanka, 1978.
London: Macmillan Press.
Wilson, A. J. (1975) Electoral Politics in an Emergent State: The Ceylon General Election of
May 1970. London: Cambridge University Press.

1
Since 1989, the Tamil terrorists have not allowed the government to conduct census in their areas so we used 1989 census data.
2
It is important to note that Sri Lanka does not have an Electoral College system that is similar to that of the US.
Presidents are elected directly by the people; therefore it is technically possible for a candidate to win the
presidency even by losing the majority of the 160 polling divisions.
3
A. J. Wilson (1975) in Electoral Politics in an Emergent State: The Ceylon General Election of May 1970
categorizes the 145 electorates in to urban, rural and quasi-urban areas, based on multiple definitions such as local governmental classifications.
4
Replacing the dead politician with the bereaving widow or a member of the family is a very common South
Asian phenomenon.
5 Kumaratunga alleges that it was her own Prime Minister Mahinda Rajapaksa who encouraged JHU to go to
courts, as he was waiting for an opportunity to run for presidency and believed that he had a better chance of
winning in 2005 and did not want to wait till 2006 to see Kumaratunga, plotting to give party’s nomination to a
candidate of her liking, such as her brother. JHU is an ultra Sinhala-Buddhist party; all its members in the
parliament are Buddhist clergy.