This is fundamentally a quantitative, data-driven, exercise. As such, one might expect a lot of quantitative analysis before the coming election. However, less than four weeks before the election, most of the analysis one sees in the public domain is qualitative and subjective.
There are many pieces of factual reportage — list of candidates by the main parties or the past records of many candidates for example. Newspaper election pages also contain pieces on how a particular candidate is better placed now, or how rebel candidates might affect the chances of an official nominee. While these pieces may contain results of the past elections, in most cases there is little quantitative, empirical evidence supporting the main thrust of the story.
And this lack of quantitative analysis can be extended to the national level. Both major parties claim that they have the popular support going into the poll. But in the absence of any credible opinion poll, how do they know? Our past experience suggests that the losing side claims that they were denied a certain victory by foul play. But where is the hard evidence of the foul play? One hears a lot about the importance of the first-time voters — accounting for slightly less than a third of all registered voters — in this election. In the absence of any credible exit polls, how would we know how the new voters voted?
In a highly politicised country such as ours, parties, pundits and the general public alike would welcome more quantitative analysis. That this doesn’t happen is partly due to a lack of credible data. We are in no position to answer most of the questions raised above because we don’t have the data needed. We have, however, used a dataset put together by the National Democratic Institute for International Affairs, to produce some quantitative analysis.
Firstly, we have identified areas of significant irregularities in the past. One simple example of a significant irregularity is if total votes cast in a centre exceed the number of registered voters in that centre. We find that this is exactly what happened in 111 polling centres in 2001 and in 141 centres in 1996. In 1996, the worst performing centre was Sharanpur GPS (in the then Jessore-5 constituency). Here 950 votes were cast by 111 registered voters, and Jamaat-e-Islami received the most, over three times the number of total voters, in the centre!
Yet another possible sign of irregularity might if the winning margin is incredibly high — say over 98.5%, which was the case in 100 polling centres in 2001. This may seem believable in some polling centres of Bogra-6 or Gopalganj-3 won by two former prime ministers, Begum Khaleda Zia and Sheikh Hasina, respectively. In Bogra-06, Khaleda Zia won about 94% of the total vote in Erulia Union’s Shikarpur Madrasa center. Similarly, in Gopalganj-3, Sheikh Hasina won 98% of vote in Tungi Para Khan Shaheb High School. However, it is not immediately obvious why this should be so in the then Bhola-3 (where Tofael Ahmed of Awami League lost to Md Hafiz Ibrahim of the 4-party alliance).
These irregularities may seem insignificant to the extent that they wouldn’t have changed the national election result, but this can very well affect the result in the particular seat. There are instances where a candidate lost all but s few unions in one constituency and yet won the seat by wining a couple of unions by incredible margins. And that is precisely why analysts seldom want to explore these data one year — it often takes that long for the data to be compiled — after the election. However, in the lead up to the next election, with everyone paying attention, there is value in pointing out these anomalies in the interest of a better election. If nothing else, we can at least be more alert in places where there were significant irregularities in 1996 or 2001.
In addition to the above analysis, we also use the union level data to analyse possible impacts of the electorate redistribution. All else remaining equal, for example, we can calculate hypothetical results of the past three elections under new boundaries. This would be of interest to political analysts and parties alike. Our analysis will also focus on the voting irregularities in Chittagong Hill Tracks. In the absence of a Gallup Poll in Bangladesh, we will also analyse the results of past two elections to predict the voting pattern of incremental and first time voters. Finally, we will also analyse the parties’ manifestos once they are released.
Over the next few weeks, we will present our findings in these pages. While no substitute for credible opinion and exit polls, we hope our analysis helps with a smooth transition to democracy.
The dataset is publicly available here: http://www.ndibd.org/.