When party policies do not matter: Examination, the ambivalence of voting behaviors in the Zambian presidential elections

Debates on whether people in Zambia cast their votes for a presidential candidate based on the good policies of the party or the qualifications of their candidate are peppered with tales of ethnicity, tribalism, corruption, and the education levels of the voters. These problems have undermined the credibility of the winning candidates as being put into office based not on their qualifications, but on the desire for individual voters to have someone of their tribe as president. While some scholars have argued that people are not naïve to vote for a candidate irrationally, others hanker on the fact that party policies are barely known to the Zambian voter who takes different forms of communal identities. The two approaches underscore the nascent debates of voting behaviors in Zambia today. Therefore, the aim of the study is to examine the voting behaviors of Zambians in the 2011 Zambian presidential election. Quantitative evidence suggests that party policies and manifestos in the Zambian elections do not matter because people base their votes on ethnic alignments.


INTRODUCTION
The 2011 presidential elections marked the second time in Zambian history when the opposition party won against the ruling party.The Movement for Multiparty Democracy (MMD) was the first in 1991 with Frederick Chiluba winning against the 27-year reign of President Kenneth Kaunda.While it might be common that the opposition wins elections in developed nations, Zambia"s cases are exceptional in Africa.That is why the International Institute for Justice and Peace (IIJP) declared the 2011 Zambian elections as "setting an example for African politics" (October, 2011).In the 2011 elections, the ruling party (MMD) conceded defeat with a vote share of 36 (15%), while the opposition emerged victorious with a vote share of 42 (85%) (EISA election observer mission Zambia, 2011).In spite of these developments, critiques have questioned whether the 2011 presidential elections were indeed free and fair; and whether the people voted based on the principles of a "free and fair" election as defined in the constitution.
Arnold Moyo (2014) asserts that "Free and Fair" elections are a key defining feature of a vibrant multi-party democracy and peace.They provide a public mechanism for regular peaceful institutional competition for power and the opportunity for people to change or review the government through their freely expressed will (Anyangwe, 2013).Moyo dissects the phrase "Free and Fair elections" by arguing that "Free" elections encompass an electoral process that is characterized by the enjoyment of civil and political rights, while "Fair" refers to electoral processes where the electoral conditions such as the law and access to information are the same for all participants (p.3).Article 21 of the Universal Declaration of Human Rights, on the other hand, states that one of the characterizing features of free and fair elections is the knowledge that the voters have about what they are voting for.The enticement for voting should not be based on promises that the candidates know they have neither the possibility nor intention of fulfilling, but rather on grounded well-articulated and achievable policy scheme and development objectives.These will become the important part of the criteria by which the people judge the suitability of candidates and ultimately their choice of a president (Maitra, 2009).However, the question as to whether party policies matter in Zambian politics is still relevant to many scholars.Unlike developed nations, many political parties in Zambia do not differ on policies (Von-Soest: 2007).During presidential campaigns, there is barely/less emphasis on the differences in party policies.The presidential candidates base their manifestos on promises of fulfilling what the ruling party has not fulfilled.This is the same manifesto that the ruling party gave when they were in the opposition (Moyo,p. 11).Such circumstances raise questions as to whether people vote based on the personality of the candidate or the kind of policies the party is propagating.For Daniel (2007), Zambian politics uses ethnic cleavages for electoral mobilization.Other scholars such as Naomi (1982) have argued against this thinking, contending that sometimes what we see as ethnic cleavages are just ways of expressing cultural and linguistic solidarity and have no bearing on voter preference.The content of party policies and the type of candidate are the main factors that explain voter preferences (p.467).Many others have advanced the literature by separately discussing economic endowments, language composition, and education levels as shaping voter preferences (Zuber: 2013;Lupia and McCubbins: 1998) in Zambia and Africa at large.
Therefore, the main aim of this paper is to examine the reasons why people in Zambia choose to vote for one candidate over the other.It inquires whether ethnic identities or political party policies can better explain the voting behavior of the 2011 Zambian presidential elections.It seeks to answer questions of whether Zambian voters have firm opinions of the political parties they voted for, or cast their votes based on ethnic affiliations.Thus, understanding the degree for which Gondwe 11 Michael Sata"s Patriotic Front (PF) and Hakainde Hichilema"s UPND parties attracted votes from their counterpart"s tribe becomes relevant.Theoretical approaches have been designed to respond to such overarching debates about elections in Zambia.This paper is structured in six parts.It starts by giving the profile of Zambia; highlighting the 73 major languages and how they get clustered into seven officially recognized languages.It then reviews existing literature on the subject and provides theoretical frameworks for studying voting behavior in Africa.The next section describes the data and method of operationalization.This is followed by the presentation of the findings/results and a discussion.Finally, the study concludes on the findings and suggests that Zambian elections be driven by informed policies made readily available in political manifestos and election campaigns.

Profiling Zambia's ethnic divisions
Zambia as a country is almost the size of Texas, with a total area of 752,618 square kilometers.It currently has an estimated population of 14.9 million, with an increased population of 11 million people since independence in 1964.The country is ethnically complex with 73 different tribes and languages currently identified.The complexity lies in the fact that Zambians are affiliated in various ways: through language, tribes, ethnicity, and religion which overlap closely with one another.While all the languages are considered relevant, the 73 languages are subsumed into 7 official languages.The national news is aired 8 times every day in the 7 languages as well as in English; another official language.The division of these languages is geographical in the sense that linguistic dialects are concentrated within ethnic localities.Table 1 and Figure 1 show a map and a chart summarizing the 73 languages and how each is linked to the 7 official languages.
Table 2 demonstrates the alignments of the seven official languages of Zambia towards the presidential candidates of the two major parties in Zambia.While President Michael Sata seems to only have two languages under him, he still stands to have more people under him because of the composition of the two languages.On the other hand, Nyanjas and Bembas are concentrated along the line of rail where development first started in Zambia.Although these places seem to be metropolitan for the most part, many people have converted themselves to either Bemba or Nyanja.In the early 50s and 60s, many people moved to these places to work for the rail line or the mines, and the children born in these areas identify themselves with either Bemba or Nyanja (Gertzel et al., 1984).

LITERATURE REVIEW
Studies on voting behaviors characterized by ethnicity, tribalism, and language composition, are among the   classical fields of social science inquiry, but remain an under-research area in Zambia (Posner, 2005) and Africa at large.Gertzel et al., (1984) and Adjei (2012) asserts that it is because the post-independence African countries gravitate towards authoritarian regimes adopted during their eras of independence (p.3)

Ethnic grouping and voting in Zambia
Zambia"s pattern of ethnic affiliations is a complicated phenomenon.It carries with it different patterns that are deeply rooted into one another.Essentially, the people see themselves as ethnically affiliated in two different ways: First, as members of one of the 73 tribes of the country and second, as speakers of one of the 7 official languages (Posner, 2005).Therefore, this complexity raises questions as to which affiliations they are most inclined to.Spanning a wide range of scholarly traditions, the conventional wisdom of Zambian politics is primarily characterized by patrimonial pattern exchanges that are not only deeply rooted in the Zambian cultures, but also perpetuated by the British colonial era (Bradley, 2011;Finnis, 2014).Since the country"s independence in 1964, the United National Independence Party (UNIP) has had a dominant position in national politics.Initially, Zambia embraced multiparty politics, allowing various political parties to exist.Multiparty politics did not go well with the patrimonial cultures (Racker, 2003;Lindemann, 2010) because people were less exposed to other ethnic tribes, and thus were not receptive to candidates from other tribes.This led to factions in which people felt obliged to vote for people of their own tribe.At that time, each tribe was confined to its own province.Under such circumstances, President Kenneth Kaunda declared Zambia a "One-party" participatory democracy state in December, 1972.This establishment was a move aimed at managing and structuring political conflicts (Chinkulo, p. 29).Similarly, the electoral system was revised to structure and control competition within the one-party framework.For example, the Election Act of 1973, and the electoral rules and regulations which followed, strengthened the role of the ruling party, UNIP, in the electoral process by assigning it the tasks of: (i) candidate selection, (ii) formulation of the election programs, and (iii) handling election campaigns for both the Parliamentary and Presidential elections.The Presidential elections operated on a two-stage, single candidate system.The first stage involved only the party.The person elected as Party President at the UNIP General Conference automatically became, by virtue of his post, the sole presidential candidate.The second stage provided for a direct affirmation (or rejection) of the sole candidate by the electorate.The voters had a choice of marking "Yes" or "No" against the name of the presidential candidate.To retain or attain the Presidency, the candidate had to receive at least 51% "Yes" votes out of the total votes.
Frederick Chiluba"s 1991 landslide victory of 76% vote share ousted President Kenneth Kaunda, who had ruled Zambia for 27 years, and marked the restoration of multiparty politics and democracy.Zambia has now held more than 5 legislative presidential elections and it is Gondwe 13 heralded as the most peaceful and champion of democracy in Africa (Baylies and Szeftel, 1997).

Language composition and voter influence in Zambia
The declaration of Zambia as a One-Party state in 1972 affected the voting pattern and voter turnout.According to Chikulo (1988), only 39.8% of the 1.7 million registered voters turned out to vote in the 1973 elections.This low turnout was the result of two main factors: First, voters did not see a candidate from their tribe that they could vote for, second, they assumed that a single tribe was taking over the leadership of Zambia.
1 The 1978 elections, on the other hand, presented a completely different picture.Of the 1.9 million registered voters, 66.9% cast their vote.As in previous elections, some provinces" turnout was higher than the national average: the highest polling provinces were Lusaka with 72.7%, Central with 72.5%, Eastern with 71.7% and Copperbelt with 71.7%.The lowest polls were recorded in Western Province with 44.6% and Southern Province with 59.8% (Chikulo, p. 39).
The polls with the highest number of votes for President Kaunda came from provinces that either shared his ethnicity or tribe.The Western and Southern provinces of Zambia barely did not speak the language of President Kaunda, which explains why turnout was much lower here.The Northwest was a terra incognita to the politics of Zambia.Because of its geographical location and had less to contribute to the national economy, development and other national activities were almost cut off.It is little wonder that it recorded low votes.Although the economy was badly hit, it was surprising that the resentment for voting decreased among the Zambian citizens, from 19% in the 1978 polls to 4.5% in 1983.Despite this fact, the general voter turnout was low.For instance, 60.2% in 1973For instance, 60.2% in , 33.1% in 1978For instance, 60.2% in , and 34.5% in 1983. .There is barely any study that exists to explain this resentment.Chikulo points out that if we accept the plausible thesis that low voter turnout reflects opposition to the regime, testing the thesis would require access to survey data showing the reasons people give for why they did not vote.
Nonetheless, the low turnout in 1973 was held to be clear evidence of the widespread voting preference based on ethnicity and thus opposition to the introduction of the one-party state.It was widely assumed that the low poll was a protest vote, a "silent No vote" which gave the one-party state uncertainty (Croke, 2016).However, although the 39.8% turnout was a sharp drop from the impressive numbers registered during the sixties, it cannot be attributed solely to the establishment of the one-party state.The decline in voter turnout, as we have earlier argued can only be understood against the background of developments which have been taking place since independence.
which might explain the declining poll.However, ethnicity and language composition, in this case, become a major factor.
A similar trend of low voter turnout continued in subsequent election polls until 1991 when President Frederick Chiluba won a landslide victory.Many scholars such as Posner (2005), Bratton (1999) and Rakner (2002) have argued that Chiluba"s landslide victory was not influenced by any ethnic affiliation, but by the shift to multiparty politics and the quest for the Zambians to escape the IMF sanctions that were brought about in the Kaunda era (Chipenzi et al., 2011).For many scholars, this was the only time when Zambians based their voting on party manifestos and policies (Chipenzi et al., 2011, p. 61).Notably, while Kaunda"s policies insisted on government monopolies, Chiluba believed that privatization was the solution.Chiluba"s policies attracted Western attention and support because they reflected the ideas of democracy.However, after the 1991 victory and the dawning of multi-party democracy in Zambia, the society reverted to the old trend of voting based on tribal affiliations and not on policies.To date this debate of whether people in Zambia vote based on policies or tribal affiliations has not been reconciled.This is because when one talks about the issue of tribalism in Zambian politics, others challenge them as inciting violence by playing an ethnic card.Thus, it is only through evidence that it can be concluded that voters cast their votes based on tribal affiliations and not on party policies.

Corruption and tribal cleavages in the Zambian politics
The tendency of pioneering ethnicity as the variable to explain why people would vote for one candidate over the other has taken root in the Zambian politics.However, scholars such as Kasoma (2009) refute the proposition that people vote on ethnic lines, asserting that corruption is instead a driving factor (This means that one can vote for any candidate regardless of their ethnicity if they are bribed).Kasoma perceives corruption in form of offering or accepting gifts for a favor.While this is a generally accepted definition, Kotecha and Adams (1981) and Burke (1991) assert that it is impossible in Africa to talk about corruption without linking it to tribalism.The two authors define corruption as a means of consolidating incumbents" personal clientless and coalitions (p.86).The term here is used as a synonym for certain specific types of behavior as proposed in McMullan (1961)"s straightforward definition: A public official is corrupt if he accepts money for doing something that he is under duty to do anyway, or that he is under duty not to do, or to exercise legitimate discretion for improper reasons (p.319).Unlike Kasoma"s perspective, this definition includes the acceptance of bribes and kickbacks, as well as extortion by public officials.
The last phrase of the definition accommodates a wide variety of ways in which an official may use discretionary powers for his/her own political or economic benefit.However, McMullan fails to include one important form of corruption that is common in Zambia, namely nepotism or the practice of putting one"s kin on the public payroll.
According to Wilson (1966), bribery and nepotism were common forms of corruption in Great Britain during the 18 th Century.This was a time when Britain was also attaining colonies including Zambia.Wilson thus asserts that there is a higher chance that Britain manifested tendencies of nepotism to their colonies.Similarly, Posner (2005) observes that nepotism is an important factor in Zambian politics.For Posner, Zambian voters seek to maximize the resources they can get from the state, such as jobs, development funds, agricultural subsidies, feeder roads, health clinics, relief food, and schools.The game of politics is thus understood to revolve around the transfer of these scarce and highly desired benefits by politicians in return for voters" political support (p. 3).
Posner is drawing his argument from Young and Turner (1985) who believe that politics in Africa was like "cutting the national cake."Young and Turner perceive the output of the state to be divisible into slices of possibly unequal size, sweet to the taste, and intended to be eaten (p.147).However, Young and Turner ignore the role ethnic identity plays in determining the size of the slices.As Posner (2005), Barkan (1979), and Kanyinga (1994) assert, Zambian voters assume that access to resources will depend on electing politicians who belong to their ethnic group.That is why the competition for jobs, and other social benefits, becomes a struggle among ethnic communities to put one of their own into a position of political power (p.4).
In the same way, voters in Zambia also seek access to state resources by allocating their electoral support to members of their own ethnic groups, who they assume will be more likely than non-co-ethnics to redistribute those resources to them (Posner, p. 1304).This in some ways compels candidates not to ignore the ethnic card, but play it through bribery and nepotism.In this way, ethnicity, through illicit kickbacks, assumes a position of respectability in election campaigns not because of people lack information, but because voters see that as the only way they could have a share of the national cake.Posner concludes by saying that corruption provides a cue that helps voters distinguish promises that are credible from promises that are not.In other words, the bribes from politicians provide an assurance for sharing the national cake because either the candidate or the voter believes someone owes them something.

Education and tribal cleavages
Plethora studies have pointed to the importance of education in the way people choose to vote for one candidate over the other.For instance, Teixeira (1987), Wolfinger andRosenstone (1980), andVerba et al. (1995), found out that education was the most and important predictor of voting behavior in the United States.In Zambia, education is believed to give people an appreciation for opinions different from their own (Moyo, 2011).A working paper from the Afrobarometer supports this theory by arguing that increased levels of education within the same country decrease ethnic alignments except in situations where a country is a quasi-democracy or an electoral autocracy (Kwenzi and Lambright, 2005).The two authors used data from Zambia and found out that better-educated citizens had a better ability to engage in civic life and were more likely to support free and fair elections that are based on good party policies.In contrast, less educated people did not care about the policies of the party they chose to vote for, but on how well the candidate articulated themselves.Therefore, the study questions the degree to which education levels affected the voting behavior of the Zambian people.

Policy-based electoral debates
Apart from the need for free and fair elections, there is also a question of the quality of the people seeking to be elected.One feature of Zambian politics that has been noted by many critiques is that it lacks substance.A Zambian based think-tank, Jesuit Center for Theological Reflections on the essentials of the democratic process, observes that political debates in Zambia are characterized more by character persecution than policy discourse.They continue to argue that, politicians in Zambia spend more time and energy slighting their opponents than in articulating their views on development and other issues of similar importance (p.11).Hence, people vote for candidates even though they have little or no knowledge of the policies of the party they are voting for.Their choice of candidate tends to base on factors and criteria that are of no immediate relevance, such as, the tribe of the candidate, his/her perceived wealth, or illusionary and dishonest promises of immediate political and economic gratification (Posner, p.1304;Von-Soest: 2007).Based on the aforementioned overview, the study advances the argument that "Free and "Fair" elections can only be attained when people vote based on the comprehensive understanding of party policies.Therefore, the question of why people get to vote for one candidate over the other becomes important.As argued earlier, various reasons might be at play.The standing hypothesis at this point is that the people of Zambia do not vote for a presidential candidate based on the policies the party adopts, but on ethnic identification and alignment.

Theoretical framework
Various frameworks about voting behavior have come to light since the 1960s when Lipset and Rokkan (1967) developed the structural theory of voting behavior.The three authors were responding to the trend of voting Gondwe 15 behaviors that emerged in Western Europe during the 1960s.The authors observed that at that time, Western European politics were characterized by social identities which were centered on class inequalities, sectarian and regional cleavages.Horowitz (1985) advanced this theory by directly linking social cleavages to ethnicity.He thus argued that ethnicity had a very strong impact on the psychological behavior of voters.Unlike Green and Shapiro (1994)"s pathologies for rational choice, voting for Lipset, Seymour, & Rokkan, entails, an expression of group loyalty and identity.
Around the same period, Tajfel (1981) developed a more specific framework entitled, "Social Identity Theory" (SIT).For Tajfel, people behave according to the standards and dispositions of the groups they align themselves to.This implies that identity emerges from individual"s sense of belonging that is exhibited through self-esteem.This tie is so strong that the behavior of an individual is translated into the behavior of the group (Mitchel, 1981).That is why Adjei (2012) asserts that individuals in such situations tend to protect their tribes by building solidarity and normative protective culture that socialize tribes into believing that theirs in the best (p.9).It is no wonder that they vote for a candidate from their ethnic classification who is, in their belief, the best.While it is clear that the voting behavior can be influenced by tribal and ethnic alignments, it is inconclusive as to whether this might be the only determining factor.We thus pretested other explanatory variables such as apathy and employment.The argument is that the level of education determines whether someone would be voting based on ethnicity identities or not.
Therefore, we can draw some hypotheses from the discussed theories by asserting that the people of Zambia massively voted in the 2011 elections based on a candidate"s affiliation with the voters" tribe or ethnic language.In our hypotheses, we contend that: 1.People in provinces with a higher composition of Bemba voted for Michael Sata and those with higher Tonga composition voted for Hakainde Hichilema.2. The districts with higher levels of corruption had voted more on ethnic lines than those that showed fewer levels of corruption.Existing literature indicates that tribalism and corruption are deeply intertwined.The assumption is that corrupt constituencies had a higher chance of encouraging election behaviors that were based on ethnicity.3. The provinces with high levels of education were knowledgeable enough not to vote based on ethnic affiliations, but for a candidate whom they considered qualified to foster national development.

Data selection
The availability of secondary data from the Zambia Electoral 16 Afr. J. Pol.Sci.Int.Relat.
Commission enabled us to examine voting behaviors in the 2011 presidential elections.The 2011 presidential election results were given precedence because of the way they treated independent variables.Unlike other presidential elections in Zambia, the 2011 elections did not exhibit a lot of nascent intervening variable.Other elections, such as the 1991 elections have other variables other than language composition, corruption, and education that can explain why people voted for one candidate other the other.However, percentage results of the 2006 and 2008 presidential elections were included, but only as a reference and supporting evidence for the 2011 study results.

Dependent variables: Sata and HH's shares of the 2011 total votes
The focus of this study is voting behavior in the Zambian 2011 presidential elections.Behavior is measured by a single question about whether people voted for Michael Sata/Hakainde Hichilema based on their good policies, or on how ethnically aligned they were to them.Therefore, President Michael Sata"s share vote and that of his rival Hakainde Hichilema were statistically regressed in Stata against the three independent variables: Language, Education, and Corruption.Further, percentage votes from the clustered 7 ethnic languages of the 73 Zambian tribes were equally quantified.This was done so as to determine where each candidate got the most votes.The results of the 2011 presidential elections were later compared to two other overarching ethnically aligned presidential results in which Hakainde Hichilema had contested.

Independent variables: Education, language, and corruption
In the analysis of this study, factors identified in the literature as influencing voting behavior in the 2011 Zambian elections were examined.While literature identifies a plethora of variables as affecting voter behavior, only education, language, and corruption were found to have a positive correlation with the dependent variables.Other variables such as apathy, employment, and social economic conditions were equally pre-tested but found not to have a significant correlation.

Operationalization of the independent variables
Hypothesis testing within a quantitative environment requires both sets of variables to be readily inputted as factors of an equation.In the case of social science, many variables are not easily reduced to numerical scales due to their multi-dimensionality.In this case, language composition and corruption percentages were complex to compute even though we had existing data at hand.Such qualities are therefore difficult to incorporate into a single dimension equation.
In order to retain a simple single dimensional analysis, it is common for only one element to be crudely quantified and then tested in the equation.Such an approach is, by necessity, a crude approximation of "degrees" and "percentages" and yet enables simple statistical analyses.Consequently, the variables require indicators that not only represent some quality of the original form but are also able to undergo statistical analysis.The official document from the Electoral Commission of Zambia (ECZ) provided data on the 2011 presidential results, language composition and employment levels for each constituency and district.The Zambia Bribe Payers Index (ZBPI) document provided data on the levels of corruption in each district and constituency.In calculating the Bribery Index, they quantitatively used "Excel pivot" tables and cross tabulation to disaggregate the data so as to provide an empirical evidence base that could enhance knowledge of bribery, anti-corruption success, and indications of social groupings most affected.Demographic and corruption knowledge data from the questionnaire was interrogated against questions on the ZBPI survey key performance indicators (KPI).Calculation of the aggregate bribery index of the ZBPI was based on the weighted average of the 3 key performance indicators, which are incidence prevalence and frequency of bribery, for all the target public institutions covered in the survey.First, the weighted average score for each individual target institution within a constituency was computed and then the aggregate bribery index was derived from it.The respective formulae are: Denotations: I = Incidence; P = Prevalence; and, F = Frequency.Σ denotes SUM (addition of a group of numbers) Note that, the KPIs weighted average is multiplied by 100 to reduce it to a percent.And for an individual institution, this provides the aggregate index.
Data on the education level of each constituency was collected from the Zambia Education Statistical Bulletin which processes data through a Microsoft Access based software Ed*ASSIST.This software has inbuilt formulas to compute all the necessary indicators.Indicators not captured under standard reports can be generated through queries to the database.Before data are disseminated to stakeholders, they undergo cleaning where sample questionnaires are taken from headquarters and compared with those in the schools to validate them and ensure that they have the same information.Population projections used to calculate data within this report have been harmonized with the Central Statistics Office"s (CSO) official population projections.

Contextual variables
The study also reviewed and analyzed the contents of the manifestos for Sata"s PF party and Hakainde"s UPND party.The objective was to assess whether the two manifestos and policies are different.In this way, the study will be able to argue as to whether people in Zambia vote based on policies or ethnic affiliation.Samples for both Sata and Hichilema were drawn from 150 constituencies of the 10 provinces of Zambia.Through scientific clustering and sampling of the 73 tribes into seven (7) languages, the results project to the population of voters across the country with an equal representation.Therefore, to the best of our knowledge, this paper represents the most robust study of Zambian voting behavior ever attempted.

RESULTS
The results of the study are purely based on the analysis of a model built upon variables influencing election results.Language composition, education levels, and corruption levels of each constituency were identified as the main independent variables.Data were collected from the Zambia Electoral Commission, The Zambia Bribery Index, and the Zambia Education Bulletin of the election year.Since a change in any independent variable might affect a change in the results, everything was limited to the year 2011.Therefore, the model is as follows: Gondwe 17 Table 3A.Correlation of Sata"s share against Independent variables Table 3B.Correlation of HH"s share against Independent variables.

HH"s share
To test the hypotheses, the share of each of the two candidates was registered separately against the three independent variables.The sample was split into Sata and Hakainde"s shares and a simple regression model used in Stata to determine the correlations.Initially, each independent variable was separately correlated with the dependent variable.This was done so as to determine which independent variable had a strong causal effect.Tables 3A and 3B indicate the above-mentioned correlations.
Table 3A shows the results of test.A simple regression was run with robust standard errors for President Michael Sata"s share of the 2011 elections total vote.Various independent variables, such as employment, apathy, and economic conditions were pre-tested against the dependent variable.Their lack of correlation resulted to the decision of ruling them out of the equation.It is also believed that their inclusion would water down the p-values of other significant variables.
Tables 3A and 3B exposes several interesting patterns worth talking about.While the results of coefficients present a similar pattern, their R-Squares are rather very different.Table 3A yielded an R-Square of 0.75% to indicate that the independent variables have the capacity to explain the dependent variable by 75%.However, Table 3B"s explanatory power is limited to 38%.Nonetheless, the fact that language demonstrates a correlation with a  shares were collated as percentages from the seven tribal affiliations.In Table 4, columns A and B present data on the percentage of the vote by a tribe and party support within the same tribes.Column A examines the contribution made by each tribe of the seven groups to the party.Using the standard formula, the effective number of these tribes was calculated by dividing one with the sum of the squared share of each tribe.Based on the calculations of the effective number, HH"s share indicates that tribalism accounts for less than a quarter of Sata"s.Significantly, column A demonstrates that both parties receive huge amounts of support from the tribes they are affiliated to.
On the other hand, column B examines party support within each of the seven tribes, against Sata"s and HH"s share.Interesting patterns emerge in this column as well.While the Nyanja tribe is believed to be overwhelmingly correlated to the Bemba tribe, about 20% of the votes seem to rest on neither Sata nor HH.In fact, HH's share from the Nyanja-speaking localities is reasonably high.It is interesting to also note that a Nyanja with a small party also contested for the presidency, but did not get higher votes as Sata or HH.In other words, while the voting pattern in "some" provinces is arguably impartial, one cannot help but notice patterns that describe ethnic voting in provinces like Northwestern, Luapula, Northern and Southern.

The 2011 results verses 2006 and 2008 presidential results: A comparison
A number of elections have taken places in Zambia since 1964.While there are several that would exhibit an ethnically aligned voting behavior, the split between Michael Sata and Hakainde Hichilema's votes exhibit a pattern of tribal voting behaviors.Table 5A and B presents the summary result of the 2006 results as presented in the "African Election Tripod" database.
Tables 5A and 5B present data that is designed to support the argument that people in Zambia cast their votes based on ethnic ties and not policies of the parties as others would argue.As observed in our statistical regressions, Table 5 presents a pattern voting that correlates with ethnic affiliations.While the patterns in some constituencies are arguably impartial, it is impossible not to notice the turnout in provinces like Copperbelt, Northern, Muchinga and Luapula that align with Sata's tribe, and The Southern province, whose alignment is with Hichilema.This demonstrates a split in shares that are hard to interpret when we assume that party policies were at play.For example, the 82.09% of Hichilema over the 1% of Michael Sata does not reflect a voting behavior driven by policy or party manifesto, but by how much the voters felt affiliated to the candidate.
In the same way, the 2008 presidential election results indicate that Sata and Hakainde"s votes were skewed towards their alignments.For example, in the Northern province, President Sata scored 65% while his opponent, Hakainde Hichilema had 1.3% in total.However, HH also scored 72% in the Southern province, while Sata had only 4.8%.Such discrepancies in the share of votes cannot dispute the argument that the 2011 presidential elections followed suit.Similar patterns are observed in other candidates such as Rupiah Banda, who scored a number of votes from his regional places.It is however disputed that Banda"s votes might have been more influenced by other variables and developmental projects of the MMD.

UPND and PF party manifestos: A comparative analysis
By textually analyzing the two policy documents of the Patriotic Front (PF) and United Party for National Development (UPND), we discovered that both documents only differ in name and not content.While UPND"s manifesto was entitled "Realize the dream of a better Zambia through real change", the PF had "For lower taxes, more jobs, more money in your pockets" as their theme.An introspection in the documents is that both manifestos pledged to achieve the following objectives: Education development, the creation of jobs, agriculture development, poverty reduction, good governance and

45.40%
Source: Africanelectiontripod.comwomen empowerment.In their campaign messages, the creation of jobs for the youth was the most pivotal issue.However, none of the two candidates explained how they would do this.One among the most remembers campaign message for Michael Sata was to empower local investors and only international investors that would pledge to employ a lot of the youth from within Zambia.According to the Zambian Daily mail, Sata appealed to many Zambians in his campaign messages by playing the "job creation" card; a challenge that everyone faces.The paper noted that this was the most salient problem that the paper brought about.Second, agriculture was the second important thing in their manifestos and campaign messages.While the focus was on increasing food supplies, there was less emphasis on the how.
Most of their descriptions were legal, and broad in nature.The UPND focused on subsidies and loans for farmers; a thing that PF equally talked about in their campaign messages.However, PF"s manifesto was vague in the sense that they did not specify on how they intended to diversify agriculture as stated in their manifesto.While it is argued that other policies were equally covered, the divergence in the policy characterized both the UPND and PF campaign messages.This led to many questioning whether those promises were indeed achievable as simply stated.Therefore, the two issues mentioned above dominating the campaign messages leave behind a lot of questions.Most prominently, one would question about the degree to which "job creation and agriculture strategies" influenced the voting behavior of the Zambian people.Many would argue that these campaign messages had no bearing on the voting behavior of the Zambian people during the 2011 elections.

DISCUSSION
In a sharp departure from previous elections, the findings in this study present voting behaviors characterized by irrational voting as defined by Downs (1957).The results presented give credence to two main things about the voting behavior in the 2011 Zambian presidential elections: First, is that people voted for a candidate in relation to ethnic/tribe alignments and second, is that party policies did not affect the way people chose to vote for one candidate over the other.Essentially, evidence shows that the total election votes were split between the Bembas (Sata's group) and the Tongas HH's group.Most of the losing votes went to other less popular political parties, which the study asserts came from people that might have voted based on either apathy or a good understanding of party policies.In a strict sense, such trends are detrimental to multi-party democracy that Zambia claims to embrace.The reality is that multi-party democracy seeks to project fairness and equity in the distribution of all forms of country resources.This includes the right to information about the manifestos and viable policy implementations of political parties.If tribal alignments characterize and influence Zambia's voting behavior, credibility, competence, and achievement will be sacrificed.National identity is a pillar for multi-party democracy (Anderson, 1991).It is therefore logical to conclude that a country divided by ethnic alignments will lose its national identity.As buttressed by Nueberger (2000), you cannot have both ethnic identifications and democracy.Nueberger gives four reasons why he thinks this is impossible: (1) That it is hard to define a state"s interests if the people within a country were divided; (2) That in ethnic voting, people base their arguments on identities and not policies; (3) That voting behavior characterized by ethnicity is a recipe for election fraud in a zero-sum situation; and (4) That ethnic alignments are recipes for civil wars.
Therefore, the trend shown in the Zambian 2011 elections is detrimental to Zambia"s quest for multi-party democracy.The first step Zambia needs to take is to accept the prevalence of a voting behavior characterized by ethnic alignments.It is only through such steps that remedies will be sought.Otherwise, if not addressed, it will be perpetuated and bring about many other vices detrimental to the development of the country.This has been witnessed in many neighboring countries such as Angola, Democratic Republic of Congo, and even Burundi and South Sudan.

Limitation
Two major limitations were observed in this study: The exclusive use of secondary data from the Zambian government documents and the employment of a single year case.These two problems can be improved by complementing one-on-one interviews with the voters (e.g.Asking them why they voted for one candidate over the other, and how much they know about party policies) and comparing the 2011 results with other years of elections.Specifically, 1973Specifically, , 1996Specifically, , 2006Specifically, and 2016 would be interesting years to compare with.An ideal study would be to study all the elections years of Zambia and analyze the tribal alignments.Above all, it is important to note that studies on voting behavior in Zambia and Africa at large will always be contested.

CONCLUSION AND IMPLICATIONS
Broadly, this study has argued that an ethnic/tribal aligned voting behavior characterized the 2011 Zambian presidential elections.We started with the purpose of finding out whether people in Zambia vote on tribal affiliations or on the good policies manifested in the party manifestos.Given the complexity of tribal affiliations, the study tackled ethnicity, tribalism, and language as synonymous to one another.These variables are highly correlated to a degree that using one for the other makes no difference.The discussions and data presented demonstrate a potential phenomenon to the fledging "free and fair' policies of multi-party democracy.Therefore, the theory of ethnicity is supported by the evidence presented in the statistical results.
On the other hand, the study establishes the fact that the policies found in the Zambian political parties" manifestos do not influence the voting behavior of the Zambian people.Policies should matter, but the textual analysis of the two manifestos of the PF party and UPND indicated no difference in content and execution.

Figure 1 .
Figure 1.Map of Zambia with its provinces.The map shows the divisions of Zambia by province, and the key table next to it presents the dominant languages in those provinces.While people have relocated in the past 50 years, the languages presented are still dominant in these areas.

Table 2 .
Mapping languages affiliated to the languages of the two presidential candidates

Languages spoken in each province PROVINCE LANGUAGE Luapula Bemba Northern Bemba North-Western Kaonde/Lunda/Luvale Western Lozi Muchinga Bemba Eastern Nyanja Central Bemba/Tonga Copperbelt Bemba Lusaka Nyanja Southern Tonga
significant p-value, is important in this case.This demonstrates that tribal alignments highly characterize the voting behavior of the Zambian people.This argument is equally supported in Table4that is based on data compiled from the Zambia Electoral Commission document.For easy interpretation, party 18 Afr.J. Pol.Sci.Int.Relat. statistically

Table 4 .
Percentage of the vote and party support by tribal group in the clustered 7 languages.

Table 5A .
The 2006 election results showing an ethnic correlation in voting behavior.

Table 5B .
The 2008 election results indicating an ethnic correlation in voting behavior.