Praveen Chakravarty, the chairman of head of the Congress party’s newly constituted data analytics department, categorically says that the Congress is a clear favourite in three of the four poll bound states – Rajasthan, Madhya Pradesh and Chhattisgarh. In an interview with Ashlin Mathew, he says that according to his department’s internal survey the reasons for disenchantment with the governments vary – it is farmer distress in Rajasthan, youth anger over unemployment in Madhya Pradesh and Dalit/OBC atrocities in Chhattisgarh. With an eye on the elections, he talks about how the analytics department has helped the party.
Edited excerpts from the interview:
What kind of analytics have you done for the poll-bound states?
At the Data Analytics department, our endeavour is to supplement the party’s campaign efforts with information and analytics at the booth level.
We have, through our big data work, categorised every booth in these election-bound states into strong and focus categories. Then for every vidhan sabha, we formulate a strategy of which booths to focus on. We arrive at a specific vote target for every one of these focus booths which will then help us win the vidhan sabha overall. Once we have this target at every booth, we have also identified SHAKTI booth coordinators in every one of these booths. The primary goal of these booth coordinators is to ensure we undertake door to door campaigning and achieve the vote target in their respective booths. We are also piloting a Ghar Ghar Congress mobile app that is given ONLY to these SHAKTI booth coordinators to enable them go door to door in their booths.
Every candidate receives a personalised letter from us along with their nomination form which outlines all these details and the strategy of how we can help them win their seat.
So our hope is that this sort of ground game will complement and supplement the party’s air game campaign efforts of rallies, public meetings, social media campaigns and so on.
Have you conducted surveys in the four poll-bound states? What do the surveys show? Could you break it down for us?
Yes we have ongoing surveys in all these states. We follow a certain pattern for surveys during elections. In empirical science, these are called panel surveys where we try to observe and measure a trend. We don’t go out and ask every time who will a voter vote for. That is too banal and “noisy” as we say in data science. Since this is a fairly scientific approach, it takes time for us to glean meaningful and actionable results from our surveys.
We have done this in Rajasthan, MP & Chhattisgarh. Telangana was announced late and hence we are still in the process. We should be ready with Telangana in a week.
We undertake a village level survey that is well stratified and randomised and is a large sample. Some of the findings are:
1. Congress party is a clear favourite in all three states.
2. Disenchantment with the government in Rajasthan is very high. But surprisingly, in Madhya Pradesh the disenchantment is double that of Rajasthan.
3. The top ranking voter issue is different in different states – farmer distress in Rajasthan, Youth anger over unemployment in Madhya Pradesh and Dalit/OBC atrocities in Chhattisgarh.
4. In general, economic issues of jobs, agriculture distress and price rise trump issues of identity all across.
According to the surveys, especially in Madhya Pradesh and Rajasthan, could you share what could be the vote split especially among the upper castes and the other backward castes?
Our surveys and its detailed findings are for internal use only. We have shared with every candidate detailed village level feedback in their specific vidhan sabha. Unlike an external commercial agency, our goal with these surveys is to use this as inputs for actions for the party, not mere information collection.
I can say that there continues to be a split in voting behaviour across Upper Castes and OBCs. OBC votes are getting increasingly more consolidated than the Upper Caste votes. Also, now the upper castes are split 60-40 towards BJP-Cong versus 75-25 in the previous elections. At least, 3/4ths of OBCs are likely to vote for one party, as per our findings.
Will the consolidation of votes among other backward castes help the Congress. How would it help in terms of projected vote percentage?
Yes, our view is it should help the Congress since it is seen as the party best positioned to offer alternative to the BJP and form a government. But this is my inference and judgement, not empirical.
Do you get any indication from the available data about the possible impact of the BSP-Ajit Jogi alliance? Will it eat into the Congress votes?
Since Dalit anger is palpably high, it is tempting to think that the BSP-Jogi factor could impact the landscape and wean away votes. It is also clear from our findings that Dalits want to replace the incumbent government at all costs. It is also clear that the only viable alternative to replacing the incumbent BJP government in these states in the Congress party.
Which then leads me to infer that perhaps the OBC voters will vote for the party that has the best chances of overthrowing the incumbent government.
Do you think the current emergence of sadhus and sants in Madhya Pradesh politics will change the vote dynamics? If yes, how?
No we don’t have specific data or evidence of this impact. That could mean two things – either its not an issue or we don’t know about it
How will the data analytics wing help in the functioning and ticket allocation in Congress?
As you know we have the SHAKTI initiative in the party that is aimed at connecting grassroots workers with the leadership at the state and AICC level. We have 40 lakh workers across 4.5 lakh booths registered in SHAKTI across the country. This is an ongoing initiative and we add one every second.
The Congress President wanted to get inputs from grassroots party workers in ticket allocation this time. In SHAKTI, we have 5 levels of members depending on their activity and affiliation levels in the party. Level 1 is the base, Level 2 and 3 are active party workers at the block and booth level that we measure using SHAKTI points system. Level 4 and 5 are office bearers and leaders.
We solicited the inputs of Level 2 and 3 SHAKTI workers across these four states and asked them one pointed question – “Name just one person who you think should be the candidate in your vidhan sabha”. We received more than 5 lakh responses across these four states. Using artificial intelligence technology, we analysed and summarised who the top choice was in each vidhan sabha. This was provided as just ONE input into the candidate selection process. These results were made available ONLY to the Congress President, the Central Election Committee and the AICC General Secretary in charge. Nobody else even in my own department is ware of the results. I reiterate that this was just ONE input and not by any means the sole determinant.
This has had an impact at various levels.
· Lakhs of party workers felt thrilled that their inputs were being solicited for the first time in candidate selection.
· It provides us a sense for who the popular choice is. That does not always mean they are the best choice.
· Also the total number of responsed we received in various vidhan sabhas gave us an indication of the enthusiasm of workers there which in itself is a useful indicator.
· In some places there was an overwhelming favorite, in some others there were many equally strong choices and in some a lack lustre response. All these are useful inputs
Will the grassroot Congress workers benefit from these analytics. If yes, how?
As I said, our primary goal is to work with booth workers and coordinators. We have made an app exclusively for them, We are constantly trying to engage with them and helping them go door to door and canvass. We are collecting inputs from them regularly. And most importantly, these workers now know that if they perform the tasks assigned to them, they are measured objectively and rewarded commensurately. That has been the biggest change. But all of this is still work in progress. I do not want to sound like we have perfected this art into a science completely. We are after all just eight months old in the party. But we have certainly made a beginning.