You are using an older browser version. Please use a supported version for the best MSN experience.

The message in the increased voter turnout

LiveMint logoLiveMint 12-05-2014 Karthik Shashidhar

Three days ahead of counting, the one thing we can say for sure about the just-concluded elections is that they showcased democracy at its best.

According to data put together by an open data advocacy group called Datameet ( ), the 2014 general elections saw a massive increase in voter turnout compared to the 2009 one.

For the 470 seats for which we have data (we don’t have data for the seats that went to polls on Monday, a number of seats in Andhra Pradesh and a few others), the average turnout was 67%.

The average turnout for the same seats in the 2009 elections was about 58%, which means that the overall national turnout has increased by as much as 9 percentage points (we must point out that we are using arithmetic average here in which each constituency has equal weight—the more accurate method of calculating national turnout would involve weighting the turnouts in each constituency with their respective size).

So, are there constituencies where turnout has actually fallen?

There are, a total of 47. In 10 of these constituencies, the difference in turnout is less than 1 percentage point, and we will discard these as not being significantly lower.

Among the 37 constituencies in our database where voter turnout has fallen by more than 1 percentage point, eight are in West Bengal, seven in Tamil Nadu and five each in Kerala and Punjab. The biggest drop in turnout has been in Nawada in Bihar, where it fell from 42% in 2009 to an abysmal 29% this year. This is followed by Madurai in Tamil Nadu, where turnout has fallen from 77% to 68%, and Outer Manipur, which has seen a drop from 83% to 74%.

One of the claims I made in the analysis of the Karnataka turnout, the subject of an earlier column, ( ) is that there are constituencies that are “fundamentally high turnout” and those that are “fundamentally low turnout”. In other words, each constituency has a “natural turnout”, and the turnout across elections varies only slightly from this “natural turnout”.

Based on the data we now have for 470 constituencies, we have a chance to test this hypothesis.

In Figure 1, we plot the turnout in 2014 against the turnout in 2009. Each dot represents a constituency, and its “x-value” (along the horizontal axis) shows the turnout in that constituency in 2009 and the “y-value” (along the vertical axis) shows the turnout in 2014. If our hypothesis of “natural turnout level” is correct, constituencies should lie as close to the red line (drawn at 45 degrees) as possible.

What we see from this figure is that our hypothesis seems right—past turnouts can be a good predictor of future turnouts (a regression shows that the turnout in 2009 predicts about 80% of the variance in the 2004 turnout). However, the interesting thing to see here is the distance of the points from the red line.

What we see here is that constituencies that saw a high turnout in 2009 have mostly “held” their turnouts—there has been no dramatic increase in turnouts in these constituencies. The big difference, however, is in the constituencies that saw a sub-50% turnout in 2009—they have seen a massive increase in turnout in 2014.

This is confirmed by Figure 2, where we look at the median change in turnout as a function of turnout in 2009.

What we have seen in these elections is that constituencies that voted badly (in terms of turnout) in 2009 have significantly increased their turnouts. Constituencies that anyway performed well have kept up their performance, and this has led to the massive turnout (relatively speaking) in these elections.

While we see a 9 percentage point increase in turnout at a national level, are there any particular states where the difference in turnout is significant? For this, we will look at the increase in the mean turnout across states compared to the 2009 elections. Ignoring small states (those with less than 10 seats), we see that the state with the biggest increase in turnout is Gujarat, which has seen an increase in turnout by over 15 percentage points (from 48% to 63%). This is closely followed by Rajasthan (again 48% to 63%) and Chhattisgarh (55% to 70%).

At the other end of the scale, we have Tamil Nadu, West Bengal, Punjab and Kerala, all of which increased their vote share by less than a percentage point. It is interesting to note here that apart from Nagaland and Sikkim, which send one member each to parliament, there is not a single state where the turnout has actually fallen compared to 2009. This is further proof that the increase in turnout is secular and broad-based.

The difference in turnout by state can be seen in Figure 3.

More From LiveMint

image beaconimage beaconimage beacon