Benford’s Law Application: Case of Elections in Sri Lanka

Sri Lanka, like many developing countries has been involved in a circle of allegations of election fraud. Usually these claims are pronounced more by losing parties. This study uses Benford's law, a law of probability distribution of digits, to investigate whether the election fraud claims might have merit. A sample in this study is made of 808 election data. This data comes from the 2010 Presidential election for representatives from three major political parties and from 2010 General Election data. All of the data points were obtained through reliable government sources, two of which are, the Department of Elections website of Sri Lanka, and the National News Paper statistics (2010). The study contrasts the distribution of the first digit of election results against the Benford's Law benchmark. After obtaining the results, we organize the data and find median, mean, mode and standard deviation. The preliminary results showe that the data does not align with Benford's law predictions. In other words, it shows that the data does not follow the law where the mean is larger than the median and there is a positive skewness then it likely follows a Benford's distribution. The distribution of the first digit of actual data for three parties disagrees with Benford's law. This misalignment is more pronounced for the winning party than for the second and third place parties, respectively. We, therefore, look forward to run the data through several critical analyses and observing if there shall be any fraud or manipulation in numbers.