Ever since it was put into use in Indian elections, the electronic voting machine has been universally acknowledged to have simplified the election process, quickened the pace of counting and perhaps, even enhanced voter turnout.
The clearest advantage of the EVM has been its ability to eliminate the category of “invalid votes”. These votes, cast in varying numbers across various constituencies in Assembly and Lok Sabha polls across India, were defined so because of manual flaws in the process of voting. If the voter did not put in the seal on the ballot paper in the allotted slot or double/multiple-voted or damaged the ballot paper, the vote was deemed “invalid”. The EVMs eliminated the possibility of casting invalid votes, as only one button can be pressed to register a vote, ruling out manual errors of the kind prevalent during the paper ballot days.
The question is whether this was a major or a minor change? Did invalid votes really matter in the Assembly elections? Were such votes capable of affecting the mandate itself?
Thanks to the availability of data from the Election Commission website, eci.nic.in and the collation of invalid votes (compiled by political scientists Francesca Jensenius and Gilles Verniers at the Trivedi Centre for Political Data), we can attempt to answer these questions. The invalid votes dataset has tabulated invalid and valid votes for each constituency in Assembly elections between 1961 and 2003, when paper ballots were finally phased out in Assembly elections.
The question whether invalid votes affected the election or not is a tough one to answer. A rather simplistic but interesting way of showing this is to calculate the difference between tallies of the winner and the runner-up and see if it is less than the total invalid votes in each constituency.
First and second place If invalid votes are greater than the difference between the tallies of the first and second-place candidates, it could be assumed that they had an effect on the final mandate. A look at the data shows that of 35,937 constituencies that went to the Assembly polls between 1961 and 2003, nearly 4,993 saw the polling of more invalid votes than the margin between the winner and the runner-up. This is a rather huge number suggesting that invalid votes could have affected even the final formation of the government in some States.
The attached table shows how many constituencies were affected in each State that went to the polls between 1961 and 2003 and in which invalid votes were greater than the margin between the winner and the runner-up.
But there is a flawed assumption in the above method. We wrongly assume that the invalid votes must have been cast against the winner if done properly.
‘Effective parties’ method A better statistical way of showing whether invalid votes mattered is to see if the invalid votes constituted a “substantive” chunk of the votes relative to other candidates (or parties). I try to do this using what is called the “effective number of parties” method.
The Laakso-Taagepara method calculates the “effective number of parties” in every election. This is a measure that shows how many candidates had an “effective” vote share in any election as opposed to non-serious candidates/or candidates with minor vote shares.
What I try to do is to calculate the effective number of parties considering only the valid votes and the same considering the invalid votes as well.
If the difference between these two measures is significant (more than 20 per cent), then invalid votes were “effective” in themselves in some elections. This measure is an indication that invalid votes cast would have gone for an “effective candidate” who mattered in the polls and would not be a spoiler if such votes were not cast at all.
This calculation — a bit more involved one — suggests that 313 of the 35,937 constituencies were strongly affected by the presence of invalid votes. Of these, 109 were in Uttar Pradesh alone, 36 in Assam and 18 in West Bengal. Invalid votes did not matter much in the big picture, but it did in select States and in not so minor numbers.
The EVM has indeed turned out to be a “valid vote” machine.
