The Coupling of Parliamentary and Presidential Election Results is A Potential Tool to Assess the Credibility of Elections in Ghana.

'I am not the owner of historic election data in Ghana; I am just the analyst and the story teller according to the numbers'.

I find it particularly frustrating about my seemingly transitory impression of the credibility of elections in Ghana, which is, of course, dependent on which of the two major political party's propagandist that I listen to, and how the election results panned out for his party. I believe this problem is not unique to me.

It's about time Ghana builds an electoral system, with a standard scientific approach to assess the credibility of elections, instead of people relying on the subjective opinions of politicians. Toward this end, I propose a novel model that will test the credibility of elections exclusively predicated on the total presidential and total parliamentary results. The efficacy of the proposed model is demonstrated using the historic election results of the fourth republic.

It is important to state that, with the exception of the 2012 election results, which is not yet published on the website of the Electoral Commission of Ghana, all the historic data were acquired from www.ec.gov.gh. Also, the 1992 election results were excluded from the analysis because parliamentary results for that election were not available. Data for the 2012 election were, however, gathered from the 2012 election petition. Since they were accepted by the court, they were assumed to be reliable, at least, for the purpose of the demonstration here.

I will, at the outset provide a brief overview of the workings of the electoral process in Ghana, which will build the context for the analysis that follows.

The standard protocol for voting on election days in Ghana is that after a voter is verified, he/she goes into the voting booth to cast the vote for a presidential candidate and then proceeds to the next booth to cast the vote for a parliamentary candidate. This standard protocol of voting in Ghana informed the hypothesis of this research that 'the near concurrent nature of the presidential and parliamentary voting activities has established an intimate connection between the outcome of the parliamentary election and those of the presidential election'.

To explore the hypothesis posed in this research, I must state that I always prefer to use numbers because I am a strong proponent of the philosophy that 'NUMBERS DON'T LIE'. Also, the law of uniformitarianism states that 'the present is the KEY to the past'. I also think that the reverse of this law is particularly true. Hence, I will study trends in past parliamentary and presidential election results in an effort to gain quantitative insight that will possibly reveal the assumed inherent connectivity between parliamentary and presidential election results in Ghana.

Hence, analysis of Ghana's historic election results of the fourth republic reveals a historic trend that suggests that anytime a political party won the total parliamentary votes, that party's presidential candidate also won the presidential election, which is somewhat consistent with the expected outcome based on our understanding of the underlying standard protocol of voting in Ghana.

For example, in 1996, the National Democratic Congress (NDC) won the total parliamentary votes by 1,248,019 votes and the party's presidential candidate also won the presidential elections by 1,263,865 votes. In the first round of the 2000 election, the New Patriotic Party (NPP) won the total parliamentary votes by 169,256 votes and the party's presidential candidate also won the presidential election by 236,164 votes. Similarly, in the 2004 elections, the NPP parliamentary candidates garnered 705,360 more votes than those secured by their NDC's counterparts. The NPP presidential candidate went ahead to win that election by 673,706 votes. The first round of the 2008 election saw the NPP winning the total parliamentary votes by 272,993 votes and their presidential candidate also won the presidential election by 102,805 votes.

The 2012 elections, however, deviated from this historic trend. Particularly, the NDC lost the parliamentary election by 121,241 votes, but the party's presidential candidate won the presidential election by 325,863 votes. Based on our understanding of the underlying standard protocol of voting, the disagreement of the 2012 election with the historic trend is quite interesting.

That notwithstanding, while this deviation is a red flag, due to the sensitive nature of the individual party's parliamentary and presidential votes to changes in voting patterns (i.e., the phenomenon of skirt-and-blouse), further analysis is required to ascertain whether this deviation is really an anomaly, or it's actually the true reflection of the will of Ghanaian voters.

It must be emphasized that I consider the overall total parliamentary and the overall total presidential votes, as the most robust and informative parameters that summarize the true story of elections in Ghana, since they are insensitive to shifts in voting patterns. For instance, if a voter casts the ballot for party A in the presidential election and decides to vote for party B, or even for an independent candidate in the parliamentary election, his/her vote will still be captured in the overall total presidential and total parliamentary votes count, without any particular reference to which party he/she voted for in either elections. Hence, the overall total parliamentary and total presidential votes will be used for the analysis in the sequel.

In fact, Ghana's election can be viewed as a scientific experiment; therefore, there should be scientific methodologies for analyzing the credibility of its outcome. I will, therefore, propose an Election Governing Equation (EGE), based upon which election results can be analyzed for errors.

The standard protocol of voting (i.e., the election experiment) intuitively suggests that barring any unforeseen circumstances, at the end of every election in Ghana, the overall total parliamentary votes cast should be equal to the overall total presidential votes recorded. Of course, errors are unavoidable in every human endeavor. Therefore, the coupling between the total presidential votes and the total parliamentary votes will lead to the following EGE:

Total Presidential Votes = Total Parliamentary Votes + Error

An intuitive interpretation of the above EGE is that, the error term is basically the ONLY reason why the total presidential votes recorded will not be equal to the total parliamentary votes observed. It is seemingly envisioned that the error term will comprise three major sources, namely: (i) clerical errors; (ii) invalid votes; (iii) and results massaging.

First, since human beings are involved in manning the elections, clerical errors are unavoidable. However, clerical errors must be randomly distributed. Any evidence of trends in the distribution of clerical errors is a red flag. If trends are evident in the clerical errors, its designation changes to systematic errors, which cannot be ignored.

Second, errors due to invalid votes will arise when a ballot is considered spoilt, and, therefore, does not count. Unless invalid votes observed in the presidential election is severely greater or less than those witnessed in the parliamentary election, the aggregate impact of this category of error on the overall EGE is expected to be minimal.

Third, errors due to results manipulation is the most problematic among all the errors, and probably the motivation for this research. This category of error arises when either the presidential or parliamentary results is manipulated to favor a particular candidate. This event also will increase the disparity between the total presidential and total parliamentary results recorded, and by extension, the error in the election.

Intuitively, the EGE suggests that an increasing disagreement between the total presidential and total parliamentary votes in any election is an indication of increasing INCREDIBILITY of that particular election. The working principle of the proposed EGE will be illustrated utilizing historic election data of the fourth republic.

Thus, the disparity between the total presidential and total parliamentary results for the 1996, 2000, 2004, 2008, and 2012 elections were 132,246; 269,992; 82,681; 81,687; and about 454,300, respectively.

It is important to note that while the total presidential votes exceeded total parliamentary votes in the 1996 and 2012 elections, the reverse were true for the 2000, 2004, and 2008 elections, suggesting that there were more errors in the presidential elections of 1996 and 2012 as compared to those of 2000, 2004, and 2008 elections.

More specifically, while the 1996 winning presidential candidate's total votes exceeded those of its party's total parliamentary votes by 15,846, the 2012 winning presidential candidate's votes, however, exceeded those of its party's total parliamentary votes by 447,104 votes.

I must emphasize that I find the 15,842 votes difference of the 1996 election to be quite insignificant and, therefore, could not be an indication of incredibility of the 1996 presidential election, since that margin could easily be due to some other factors such as NDC voters voting for a pro NDC independent parliamentary candidate.

However, the 447,104 votes difference of the 2012 election is quite significant, suggesting that the 2012 presidential election is probably the most incredible presidential election in the fourth republic, which plausibly explains why the 2012 election severely deviated from the historic trend as established earlier.

It is also necessary to point out that, in 2008, parliamentary results for Akwatia and Asutifi South were not available. A voter turn-out of 70% was, therefore, assumed in order to estimate the total parliamentary votes from the total registered voters representing those constituencies.

In fact, even though the results of the error analysis as presented above are informative, they do not make intuitive sense, to the ordinary eye, in terms of using them as basis for comparing the credibility of elections representing different years. Therefore, a variant of the EGE that could be employed as a standard tool to characterize and compare the credibility of elections is formulated as follows:

For comparison of elections representing different years, the normalization factor should be either the total presidential or total parliamentary, whichever is found to be the minimum. Their maximums could also be used; however, the choice has to be consistent. A negative outcome of the above model indicates that the presidential election was somewhat more credible in contrast to the parliamentary election, and vice versa.

This model predicts that the credibility indices of the 1996, 2000, 2004, 2008 and 2012 elections were 0.019, -0.042, -0.0096, -0.0097, and 0.043, respectively. An intuitive illustration of these results is presented in the graph below. These results indicate that the 2004 election was perhaps the most credible election, while the 2012 election was seemingly the most incredible election in the fourth republic.

I recognize that this model does not reveal anything about the category of error present in an election. However, just like in any scientific experiment, further research is required to characterize an error threshold, beyond which an election results could be deemed unacceptable, especially, when the total presidential votes exceeds the total parliamentary votes.

The model also cannot detect what I call 'even results manipulation'. That is, when the results of the parliamentary and presidential elections are massaged evenly. Nevertheless, while it is somewhat difficult, but not impossible, for the parliamentary votes and presidential votes to be manipulated evenly on a bigger scale, further research is, however, required to devise a strategy that can detect possible even results massaging in a scientific fashion.

This model, in my knowledge, is the first proposal of a scientific approach to assess the credibility of elections in Ghana. Further research is required to model the impacts of the individual error categories, in order to set an overall credibility index threshold, beyond which an election results will be deemed unacceptable.

I hope this proposal will start the conversation about building an electoral system with acceptable scientific standards to test the credibility of the outcome of elections, instead of people relying on the subjective opinions of politicians.

By: Erasmus Kofi Oware.
The author is a Ph.D. Student in Environmental Engineering & Science at Clemson University, SC, USA.
Email: [email protected]