[This article is a response to the lead news-cum-article written by Sanjay Kumar and Suhas Palshikar and published in The Indian Express as well as Jansatta on 7th October about the pre-poll survey related to the Bihar Legislative Assembly Election. We had sent this article to the The Indian Express, but they could not give any space to our views. – Authors]
Politically, Bihar is one of the most complex states in India. It is often difficult to provide a substantially cogent electoral prediction because of the multivariate factors that impact the political outcome in the state. The other obvious reason is the political maturity of the electorates of Bihar, who decide the fate of the candidates based on several considerations including caste orientation and the candidates’ performance in the past. Still many analysts have tried to provide a picture regarding the possible outcome of electoral fray for the Bihar Assembly Elections based on quantitative surveys. The opinion poll conducted by Lokniti-CSDS and published in The Indian Express and Jansatta, two of India’s most respected newspapers, on 7th October, 2015 was one such attempt. In the last two decades election studies in India has seen a dramatic evolution with poll surveys gaining immense popularity among both analysts as well as electorates. Lokniti-CSDS has been one of the most reliable institutions for such studies because unlike other market oriented institutions it has always focused on serious academic and intellectual understanding of electoral competition. Many reputed academicians have been part of its election studies and its publications have given new dimensions to the study and understandings about the dynamics and churnings of Indian democracy. However, the pressure of media as well as the rush to publish opinion polls seems to have affected the way CSDS-Lokniti is known to release its analysis.
The news-item in the front page of The Indian Express read ‘Advantage BJP as Bihar gets ready’. It was claimed in that news-cum-article that BJP led NDA had an advantageous edge in the forthcoming Bihar Assembly Elections over Nitish Kumar’s Grand Alliance. We are not sure whether it was the editors who chose the headline to attract attention of its readers or it was consciously decided by the poll conductors based on their analysis. Whatever the case may be, the projection of ‘Advantage BJP’ exposes fissures at several levels, most of which are evident from the data itself. The publication also forces us to pose significant questions about the way in which such opinion polls are conducted both in terms of methodology as well as the analytical categories used to understand electoral politics in a complex society as in the case of Bihar. It further creates doubts about the aim of such published opinion polls.
First, it is really strange that in this new item the complete analysis is based on the survey of 30 constituencies of Bihar and the sample size was only 2079. Now the analysts may claim, as masters of quantitative analysis suggest, that till the time proportionality of factor variables is maintained, the sample size does not matter. Even then there are grave issues attached to the sample selected and presented. One of the important issues is that the current Bihar elections are closely contested one where even a small error could lead to drastic difference in outcomes. Further, the nature of each constituency is different and political arithmetic based on caste, class and political affiliations could reflect varying results even in closely situated constituencies. Keeping both these aspects in mind one can ask how representative are the samples used by analysts of CSDS-Lokniti. To make things worse, even the constituencies which became part of the survey are not mentioned, and hence it is not clear whether it really reflects the mood of the state at large. Interestingly the methodology part is just mentioned in a box at the end of the article. It is true that generally a reader may not bother to understand the methodologies involved in conducting such surveys. But it is equally true that the issue of methodology is primary to establish the authenticity and validity of any such survey in the public sphere. Also because it was published by two important news-dailies of India, that have played the roles of opinion-makers for years, such lacunae needs to be seriously questioned. Further, the name of author and institution working on the survey being so credible, there is no doubt that it would affect voting preferences of a number of voters who have yet not made up their minds. In that context the challenge of being methodologically upright and rigorous becomes more pertinent.
Second, certain propositions and conclusions of the article do not flow from the data presented. One intervention in this regard has already been made by Shivam Vij (in huffingtonpost.in) where he argued that the data actually reflected ‘advantage Nitish’ as according to the data the BJP alliance is shown as leading only in urban constituencies and urban constituencies make only 12% of all Bihar constituencies. Further, the authors consider that one of the reasons why the Grand Alliance is trailing at this time is because BJP’s campaign about ‘Lalu Yadav’s Jungle Raj’ have gained support among common voters, However, they fail to recognize the kind of polarization that issues such as Mohan Bhagwat’s statement on reservation or the question of beef eating, among others, have allowed. Such issues have definitely created ripples in the minds of voters. The survey contains more such anomalies. For example, according to the article, on the question about voter’s preference for Nitish Kumar as Chief Minister, 44 percent Lower OBC, 44 percent Mahadalit, 53 percent Yadav, 54 percent Kurmi, 57 percent Koiri, and 56 percent Muslim supported his candidature. Now, if any leader has such support in the numerically dominant groups of the state, the clear conclusion should be that he is really popular and there is possibility of his return in the power. The data is remarkable because even after battling ten years of anti-incumbency, Nitish Kumar is the favourite candidate for Chief Ministership in the state. This is one of the factors that puncture the ‘Advantage BJP’ claim made in the article.
Third, the points mentioned in the section titled ‘Limits of BJP’s advantage’ (IE, October 7, 2015, p. 2) reflect the limitations of such surveys in closely contested elections. The ‘probable changers’ among the pro-NDA respondents is shown to be 12% and among those favouring Grand Alliance as 8%. If even 20% of these respondents actually change their current preference, the alignments will completely change (Thanks to the first-past-the-post system and as reflected in several election outcomes including the 2014 general elections). More importantly, when the respondents know their candidates, their voting pattern may vary drastically based on regional, caste and other factors. It is not for no reason that BJP has fielded a sizeable number of Yadav candidates, and on the other hand JD (U) has given tickets to several upper castes mainly Bhumihars. Keeping these factors in mind one can conclude that the survey data is inconclusive. But if it is only ‘trends’ as the authors have claimed, what is the relevance of the title ‘Advantage BJP as Bihar gets ready’?
Undoubtedly election studies and surveys have played a crucial role in the better understanding of electoral politics and democratic churning in the India. There is, however, need to be more cautious against the pressure of media and market forces in the publication of such studies. Usually scholars of election study also claim that they follow a kind of ‘value neutrality’ and only present ground realities related to the changes in the mood and priorities of the electorate in a particular state or the whole country. But as we have tried to underline that the use of data, presentation of news, focus on certain aspects and avoidance of some crucial facts in such surveys not only register methodological problems of such studies, but also put a question mark on the very notion of ‘ideology-less’ and ‘value neutral’ election study. At the least such surveys need to be rigorous and multivariate in nature and the methodology as well as sites of study should be well spelt out for the convenience of the reader and to maintain high degree of legitimacy among the electorates.
Both authors are Assistant Professor of Political Science at Dyal Singh College, University of Delhi