The graph above displays the predicted vote for NDA and UPA by identity group for 2009 and 2014. The data are calculated from the Lok Survey. We see that there is indeed a very strong relationship between identity group and support for NDA vis-a-vis UPA.
In both 2009 and 2014, the difference in support for UPA as compared to NDA is least among upper castes, then OBCs, SCs, STs, and most among Muslims. However, simply focusing on this fact would mask the huge reversal of fortune for the UPA between 2009 and 2014. Change in electoral outcomes is driven by increases in support for NDA within identity groups, not by the relative support for NDA across them.
Indeed, there were double-digit increases in support for NDA among upper castes and OBCs, and an increase of 8 percentage points among SCs (and virtually no change in support among STs and Muslims). The magnitudes of these changes are far too large to be explained by a few well-organized subcastes switching allegiances.
The problem with a pure identity-based logic for electoral outcomes is that it masks the volatility in Indian elections. Identity is relatively fixed over time, electoral outcomes are not. In fact, Leigh Linden has shown that, in the post-1991 period, MPs actually face a strong anti-incumbency bias.
So, if identity doesn’t predict movement in electoral outcomes, what does? Well, that’s the subject of my next TOI piece with Milan Vaishnav and Devesh Kapur, so you’ll have to wait until Sunday to find out.