advert
home instructors journals contactus
Home

 

Instructors

 

Journals

 

Contact Us

 

JOURNAL OF RESEARCH IN NATIONAL DEVELOPMENT VOLUME 7 NO 2, DECEMBER, 2009

CORRELATION COEFFICIENT IN THE ANALYSIS OF ELECTORAL OUTCOME IN NIGERIA’S 2003 PRESIDENTIAL BALLOT

Emmanuel I. Okonmah

Department of Political Science and Public Administration, University of Benin, Benin
E-mail: eokonmah@yahoo.com
Abstract
General elections conducted in Nigeria have always attracted a mixed reaction from both internal and external observers over the nature of its conducts. Many Nigerians observed that, the 1999 and 2003 presidential elections were characterized by irregularities, fraud and violence, yet the results were respectively declared free and fair. To test the hypotheses formulated using Pearson’s Product-Moment Correlation Coefficient (r) measurement technique, an attempt is made to ascertain the degree of association or otherwise between the election inputs and  any other associational influence on the vote pattern of the electorates, between choice party and candidate’s popularity.  The essence is that, Pearson’s Product-Moment Correlation Coefficient (r) is a parametric technique used for determining the strength and the direction of the relationship between two variables X and Y.  
Keywords: Presidential,  ballot, outcome, correlation


Introduction:
Pearson’s Product-Moment Correlation Coefficient (r), according to Ogbeide (1997), is a parametric technique  used for determining the strength and the direction of the relationship between two variables X and Y. Although, Levin and Fox (1988) did observe that it is not too easy to directly demonstrate causality from co- relational data as posited by Simon (1959), it is possible for a social scientist to make causal inferences concerning the adequacy of specific causal relationship. Tompkins (1975) is also of the view that, in order to draw causal inference, a statistical method  (here using Pearson’s Product-Moment Correlation Coefficient) must involve a finite set of explicitly  defined variables, basic propositions/hypotheses about how these variables are interrelated  causally, and assumptions about  the effect of outside variables (controlled variables) on other variables.

The Pearson’s Product-Moment Correlation Coefficient formula and the Davis’s Conventional description of correlation coefficient  are provided below:-

  

Davis’s convention for describing correlation coefficient values


Value of co. coefficient

Appropriate phrase

+1.00

A perfect positive relationship

+0.66 – +0.99

A very large positive relationship

+0.36 – +0.65

A large positive relationship

+0.16 – +0.35

A medium positive relationship

+0.06 – +0.15

A small positive relationship

+0.01 – +0.05

A negligible positive relationship

0.00

No Relationship

-0.01 to -0.05

A negligible negative relationship

-0.06 to -0.15

A small negative relationship

-0.16 to -0.35

A medium negative relationship

-0.36 to -0.65

A large negative relationship

-0.66 to -0.99

A very large negative relationship

-1.00

A perfect negative relationship

Source: Davis (1974)


Nigeria and elections in Nigeria                         

The Nigerian State dates back to 1914 when the protectorates of Northern and Southern Nigeria were amalgamated to form one unified body. Okonjo (1974) did provide one of the most compelling accounts of the historical processes of state formation in Nigeria. This account locates the dynamic of the character of politics in the flag of independence. The first was the deep division and tension between the colonial administrators on the relationship that should  exist between  the North and the South after the unification. The second was the desire of the British to secure and preserve the Nigerian state for their economic and political interest. The third dynamic which originated from the interaction between the first two, was the failure of the colonialists to produce an indigenous ruling class with the economic credentials needed to support their agenda.

On the first point, the tension between the colonial administrators in the North and the Governor General in Lagos centred on the degree  to which the North would be allowed to maintain its institutions as a single, monolithic  and almost self-autonomous centre of  power within British Colonial Nigeria. Nevertheless, with the love the colonialists  had for the North, it prompted Sir Bernard Bourdillon who became the Governor  General of Nigeria  in 1935, to propose that by 1942, the Northern province  should be granted an equal voice or  have seats in any (Nigerian) central legislation  involving both the Western and Eastern provinces (Okonjo, 1974). This guarantee was subsequently written into the Richards constitution of 1946 and formed the basis upon which the subsequent pre-independence elections, including the crucial one of 1959, were held.  Okonjo (1974) suggested that, all the political difficulties which Nigeria had experienced before and after independence have something  to do with the fact that one region of the federation could by virtue of the number of parliamentary seats allocated to it, dominate, dictate to and hold ransom all the others.  Certainly, this factor enabled the Northern People Congress (NPC) to assume power at the centre in 1960. It also enabled the NPC to use that power in deepening the crisis that rocked the western regional government in 1962. It also enabled the NPC as the government at the centre not only to intervene in the Western

Region elections crisis in 1964, but also seek to continue in power following the federal elections of December 1965. According to Ikelegbe (1995), these elections were massively rigged by the government in power and led not only to massive boycotts and popular revolt, but also to the first military coup in Nigeria in January 1966.
Conclusively, the malnourished indigenous ruling class and exploit the various incapacities of the masses using the coercive instruments of the state to subjugate  and totalize the entire environment to themselves – the fallout being the introduction of money  politics, use of brutality to intimidate perceived recalcitrant opponents to submit to their whims, massive rigging and manipulation of electoral results. Besides, they have sensitized the people to ethnic politics which have dangerously divided the polity to diametrically opposed lines. The people no longer consider the way and manner with which a candidate made his election rather, the end justifies the means. The culture of mistrust and violence has become part of the people. Those who are not disposed to this unacceptable habit are branded ‘unprogressive” and thereafter either detained or dislodged in whatever way necessary. In other words, the nature and formation of the Nigerian state have significantly contributed to the skewed and lopsided pattern of the polity.
The April 19, 2003 presidential election marked the transition programme from civilian to civilian. This period could be regarded as the first time the country recorded a successful transition from civilian to civilian government, without military takeover as in the case of 1983 elections.  According to a preliminary report of the Transition  Monitoring Group (TMG) issued after the conduct of the 2003 president election,  numerous cases of electoral fraud and  malpractices, including underage voting, stuffing of ballot papers  into ballot boxes, snatching of ballots boxes, falsification of results and so on, were observed in many states in different parts of the country (www.nigerdeltacongress.com/articles/2003_elections_audits_aftermath.htm).               

Observed equally was the manner in which voters registration cards (also a determinant factor in election victories) were distributed by Independent National Electoral Commission

 (INEC) to the States eight months before the election, based on a projected population (and not registered voters). That showed a glaring margin of premeditated victory by the incumbent (www.dawodu.com/aluko39.htm-5/7/2005. In spite of these irregularities, there was a deep sense of ethnic consolidation and extension of the continued protection of a clique of northern oligarchs within the ruling class (Okwe, 2002).

This study tries to look at cause-and-effect relationship in the propositions/hypotheses made. Hence, we treat causation as what determines or what causes an event to happen. Simply, according to Abutudu  (1999) causation is the interrelation between two successive events; where the occurrence of the earlier event is regarded as condition for the later event. Therefore, at the heart of all scientific explanations is the idea of causality, that is; an independent  variable is expected to produce a change in the dependent variable  in the direction and of the magnitude specified by the theory. Although, it does not necessarily follow that a cause-and-effect relationship must always exist in that sequence (Nachmias and Nachmias, 1996).

Determination of causation    
In causal relationships between one or more variables, the variables must vary, that is; whenever A causes B, one will expect that at any given circumstance, A and B must be related but, could vary when C appears. So, A and B must be standard and valid. Therefore, in order to prove that A is the cause of B or B, C, D, etc the following conditions must be satisfied: 

  1. The independent variable (IDV) must be a necessary and sufficient condition to bring about dependent variable (DV): For a researcher to establish that A can cause B, C or D to appear, it must be shown that A is not only necessary but also sufficient to cause B, C or D. For example, the cause (s) of war or revolution must be associated with exclusion (Abutudu, 1999). Class-consciousness can be one of the conditions of exclusion or marginal inclusion.
  2. The relationship between the IDV and the DV must be contiguous in space: There must be a physical nearness between A and B to cause C or D. For example, the President’s speech on fuel price increase in Abuja causing riots or demonstrations in Lagos, Edo, Calabar etc. This is known as contiguous affinity.
  3. There must be temporal sequence: The relationship in time-space is that, to establish causation, A- must always happen first before B or C or D. If on the contrary B comes first, there is no sequence.
    

In practice, the demonstration of causality involves three distinct operations:

    1. Demonstrating co-variation: This simply means that two or more phenomena vary together. For example, if a change in the level of education is accompanied by a change in the level of income, one can say that education co-varies with income. That is, individuals with higher education have higher incomes than individuals with lower levels of education.
    2. Eliminating spurious relations: Here, the researcher is expected to demonstrate that the effect of all relevant variables is controlled and the relationships between the original two variables are maintained. When such is established, here the researcher will have strong evidence that there is an inherent causal link between variables, and the observed co variation is not based on accidental connection with some associated phenomena.
    3. Establishing the time-order of the occurrences: Here, it is expected that the assumed cause occur first or changes prior to the assumed effect. That is, phenomena in the future cannot determine phenomena in the present or the past. For example, an interest in politics precedes political participation and not the other way round.

Therefore, in social scientific research, the notion of co-variation is expressed through measures of relations commonly referred to as correlations or associations. Thus, in order to infer that one phenomenon causes another, the researcher must find evidence of a correlation between phenomena. According Lazarsfield (1959), the size of the partial relative associations and the assumed time order between the variables, are the kind of evidence required for inferring causation.

Hence,  a causal model as shown below, displays a direct and indirect relationship which show a direct effect of X1 on X2, direct effect of X2 on X3 and an indirect effect of X1 on X3 through X2. Thus, X1 is the independent variable, X2 the intervening variable (Nachmias, 1979) and X3 (or sometimes denoted as Y) is the dependent variable. It may be correct to say from above that, X2 cannot cause X1 and X3 or Y cannot cause X2 or X1


jkjlkjlk 

X3

 

So, taking an illustration or applying the Pearson’s correlation coefficient in the study of the vote pattern of Nigerians in the 2003 presidential election, using a stratified random sampling of seven states out of 30 and the FCT, we formulated the following hypotheses considering a few selected determinant factors that;

The vote pattern in the 2003 presidential election in Nigeria was determined by party identification and the individual assessment of each presidential candidate.

  1. There is a relationship between ethnic identity and vote pattern of the electorates in the 2003 presidential election in Nigeria.
  2. There is a relationship between the use of money and its influence in the vote pattern of the electorates during the 2003 presidential election in Nigeria. 

Here, the perception of campaign issues (strategies) are important but, are considered a controlled variable (intervening) in that, the nature, mobilization and acceptance of a party and it’s presidential flag bearer by the electorates against other candidates from another party or parties, is a relative choice and decision of individual electorate.


Table 1: The effect/relationship of party identification and individual candidate’s assessment on the vote pattern of the electorates in the 2003 presidential election in Nigeria


Respondents

X

Y

 X2

Y2

XY

Summary

Edo

26

25

676

625

650

N = 7

Delta

18

16

324

256

2858

∑x = 127

Rivers

20

12

400

144

240

∑y = 134

Lagos

19

15

361

225

285

∑X2 = 2487

Bayelsa

22

28

484

784

616

∑Y2 = 2758

Abuja

11

20

121

400

220

∑XY = 2497

Anambra

11

18

121

324

198

 

N = 7

127

134

2487

2754

2497

 

Calculated Result: r = +0.35.

Result


 From the Davis correlation for describing correlation, coefficient value of +0.35 calculated, show that there is a medium positive relationship in the hypothesis  tested at the degree of  freedom – df 5. (n=7-2 =5). And because the calculated level of associational probability (r) is less by +0.35 than the calculated t @ 0.89 error free, we hereby accept the research hypothesis as statistically significant, that there is a correlation between Y and X on the vote pattern of the electorates in the 2003 presidential  election in Nigeria. According to Ogbeide (1996), the ability to predict the exact effect of an independent variable X on a dependent variable Y makes Pearson’s r more powerful than chi-square, χ2   Although, Aluko (2003) through the National Democratic Institute preliminary report opined that, the total number of distributed voters registration cards was implausibly higher in many constituencies across the nation than the actual registered voters, and there was no check on the veracity of the register itself, it should be accepted that during the voters registration exercise of 2002, at many polling stations, the number of registered voters was unknown.


Table 2: Effect of ethnicity and candidate’s assessment influencing electorate’s vote pattern


Respondents

X

Y

X2

Y2

XY

Summary

Edo

9

38

81

1444

342

n = 7

Delta

12

24

144

576

288

ΣX = 134

Rivers

12

20

144

400

240

ΣY = 127

Lagos

25

10

625

100

250

ΣX2 = 2940

Bayelsa

29

14

841

196

406

ΣY2 = 2937

Abuja

23

11

529

121

253

ΣXY = 2019

Anambra

24

10

576

100

240

 

N = 7

134

127

2940

2937

2019

 

Calculated Result: hjkhj


Ethnicity        
As pointed out earlier, the factors of ethnicity  and religion came into prominence during the 2003 like every other elections in Nigeria, except the annulled June 12, 1993 presidential election, although, this is contrary to the finding from the tested hypothesis with a calculated result of r = -0.85 representing  a very large negative relationship. But as it were, President Obasanjo and Buhari represented their major ethnic groups and the two major religious faiths – Christianity and Islam. In the North, Obasanjo lost the North-West zone to Buhari, while  Buhari shared the North-Central States where Islam cannot claim predominance, and where the people have expressed their desire to be seen as independent  of the monolithic North.  On the other hand, Obasanjo clearly won all  the Southern States (Anugwom, 2003).


 

Table 3: Effect of money distribution by presidential candidates during elections on the vote pattern of the electorates in the 2003 presidential election in Nigeria


Respondents

X

Y

X2

Y2

XY

Summary

Edo

40

9

1600

81

360

n = 7

Delta

22

11

48

121

242

ΣX = 218

Rivers

25

7

625

49

175

ΣY = 43

Lagos

29

6

841

36

174

ΣX2 = 7032

Bayelsa

37

5

1369

25

185

ΣY2 = 325

Abuja

33

2

1089

4

66

ΣXY = 1298

Anambra

32

3

1024

9

96

 

N = 7

218

43

7032

325

1298

 


 Calculated Result: jkkljlk

 

Money politics in Nigeria    
The role of money politics during the 2003 presidential election cannot be over-emphasized although, the tested hypothesis show that kjlkjk which represent a medium negative relationship from the Davis’s convention for describing correlation coefficient values. But less than a month to the 2003 general election, the country was agog with harvest of fund raising ceremonies. According to Adesina (2003), this was contrary to section 221 and 225 of the 1999 constitution. Section 221 of the 1999 constitution provides that no associations other than a political party shall canvass for any candidates at any election or to contribute to the funds of any political party or to the election expenses of any candidate at any election. Section 225, sub-section 3 state that, no political party shall be entitle to retain any fund remitted or sent to a political party from outside Nigeria. Section 225 (4) also states that, any fund or other assets remitted or sent to a political party from outside Nigeria, shall be paid over or transferred to INEC within twenty one days  of its receipts with such information as the commission may require. Section 228 empowers the National Assembly to decide the degree of punishment of any person involved in the management or control of any political party found after due inquiry, to have contravenes any of the provisions of the sections above (constitution 1999).

Contrary to  the constitutional  provisions which on oath, Obasanjo and Atiku promised to protect, the fund raising was flagged off by the Obasanjo/Atiku campaign organization, co-coordinating and mobilizing Nigerian businessmen, companies, groups and rich people concerned (contractors) for the focus of the 2003 re-election bid. According to Okoror (2003), close to N5billion earmarked for the re-election bid was realized  under a single mobilization. Although, Chief  M.C.K. Ajuluchukwu defended the mobilization of  PDP for Obasanjo/Atiku as quoted by Okoror (1999) by stating that, even in advanced democracies  of the world, money plays vital role in electioneering campaigns, usually for advertising, printing  of handbills, posters, parties and contestants  winning in election. The summation  is simple, financial mobilization for an election purposes is wrongly used in Nigeria.

The relevance of correlation coefficient in social science research
Correlation coefficient helps a researcher to:      

  1. make causal inferences adequately of specific causal models;
  2. explicitly define set of variables on the drawn causal inferences about variables that are causally interrelated.
  3. make assumptions about the effect of outside variables  on the variables included in the model (Blalock, 1964).
  4. provide an initial test of the credibility of the underlying theory about the topic of research,
  5. shape to size the variables required in each hypotheses formulated.

Conclusion 
Pearson’s Product-Moment Correlation is a parametric technique that is used mostly for determining the strength and the direction of the relationship between two or more variables. It is most useful in social science research because, it helps the social scientist to make causal inferences concerning the adequacy of a specific causal relationship. It is more powerful than chi-square based on its ability to predict the exact effect of an independent variable on a dependent variable. Therefore, Pearson’s correlation coefficient provides an initial test of the credibility of the underlying theory about the topic of research; and it also shapes the size of the variables required in each hypothesis formulated for test.    

Although, Nigeria made its first peaceful transition in 2003 from civilian to another civilian government, there were widespread lapses in the organization of the general elections by INEC which must be considered and corrected in order to achieve credible integrity by

reducing the incidents of electoral irregularities or sabotage.
             
References
Adesina, A. (2003)   Obasanjo- Atiku Donors.
Lagos. The Guardian, March 17.   
Aluko,  M. (2003)  Election 2003: NDI Preliminary Report.  Huston, USA.
Anugwom, E. (2003) Ethnicity, politics and elections in Nigeria: overview of current trends. Nsukka, Department of Sociology  and  Anthropology, University of Nigeria.
Blalock, H.M.   (1964) Causal inference in non-experimental research; Chapel Hill: University  of North Carolina Press.
Davies, J.A.   (1974)  Elementary survey
analysis. New Jersey: Prentice-Hall Inc.

Ikelegbe, A.  (1995)  Political parties, interest
group and elections in Nigeria” in Ola, R.F. (ed)
Nigerian political system: inputs, outputs and
environment, BeninCity, Nigeria: Ambik Press.

Iyayi, F. (2004) The conduct of elections and
electoral practices in Nigeria,a paper delivered
at theNigerian Bar Association Conference,
Abuja.

Lazarsfield, P.  (1959)  The algebra of dichotomous system: items analysis and
Prediction,  California: Stanford University Press.
Nachmias, D. (1972)  Public policy evaluation, New York: St. Martins Press.  
Nachmias, C. and Nachmias, D. (1996) Research methods in social sciences, New York: St. Martins Press.

Ogbeide, U. (1996)  Graphic presentation of research data” in Osemwota, O. et al (eds.) Research and statistics methods in social sciences, humanities and education.  Benin-City, Nigeria, Uto Publishers.
Ogbeide U. (1997) Statistical techniques for social and management sciences, Lagos, Nigeria: Amfitop Books.
Okonjo, I.M.(1974) British administration in
Nigeria, 1900-1950:a  Nigerian
View, New York: NOK Publishers.
Okoror, F. (2003)  April polls and the money
affair, Lagos, The Vanguard Monday March 17.

Okwe, J. (2002)   Shagari endorses Obasanjo for
2003 presidential election. Lagos. The
Post Express, Thursday March 21.
Simon, H.  (1957)  Models of man: social and rational, New York: Wiley Press. 
The Guardian  (2003)     April Wednesday 23, Lagos
Tompkins, G.L.  (1975)  A causal model of state welfare expenditure;  Journal of Politics, No. 37 
Websiteswww.nigerdeltacongress.com/articles/2003_elec
ions_audits_aftrmath.htm­ 05/7/2005
www.ethnonet-africa.org/pubs/papeze.htm. 05/7/2005
www.dawodu.com/aluko39.htm 05/7/2005.