by RAJA SWAMY
The Pew Research Center released a new survey that reveals a very favorable perspective of Narendra Modi among Indians. In fact, the header for the report reads: “The Modi Bounce: Indians Give Their Prime Minister and Economy High Marks, Worry about Crime, Jobs, Prices, Corruption.”1 According to the results 87% of Indians have a “favorable view” of Modi, as opposed to 52% for former prime minister Manmohan Singh in 2013. Furthermore, the report’s authors claim that the positive views extend not only to Modi but also to his economic policies and the overall direction of his government’s rule. This support apparently also crosses party lines, with Congress supporters expressing high regards for Modi’s governance on a range of issues. It would seem that the only areas that Modi scored low were on communal relations, with 47% of Congress supporters as opposed to 57% of BJP supporters expressing a positive view. The overall message of this report is that Indians by and large love their prime minister and view his government very positively.
To see how Pew came up with these rather striking claims and glowing conclusions about Modi and his government, let’s take a quick look at its survey’s sample size and methodology.2 First, the survey interviewed a grand total of 2,452 respondents spread across urban and rural areas, though Pew admits that the survey is disproportionately urban. This disproportionate urban bias in the sample is “weighted” to “reflect the actual urban/rural distribution in India.” The question arises then as to what effects this disproportionate distribution of the sample may have on the results. We cannot know for sure because Pew does not provide specific data as to which urban and rural districts were surveyed and also how weighting was utilized to overcome urban bias in the sample. According to Pew’s brief methods statement the interviews were conducted “face to face” with adults over 18 years of age, spread across “15 of the 17 most populous states and the Union Territory of Delhi.” Curiously, the survey did not include the states of Kerala and Assam (no explanation given) and excluded “a district in Chhattisgarh” due to insecurity.
The more deeply problematic side to the methodology used in the survey is its weighting criteria. Weighting is a term used in statistics to refer to adjustments made in order to accommodate one or more factors that might distort results. So for example if fewer women were interviewed than men, we may end up with results that disproportionately reflect the views of men, so the solution would be to increase the “weight” assigned to women respondents in order to make up for the distortion. Pew’s weighting variables for this survey included “[g]ender, age, education, region, urbanity and probability of selection of respondent.” Curiously absent are variables such as religious identity, class, and caste. To highlight how these variables may well have produced very different results, consider the use of the category “urban.” “Urban” is a catch-all term referring to city dwellers that could include everybody living in a city. However, “urban” does not tell us anything about who these city dwellers are. For instance Mumbai counts among its urban population somebody like Mukesh Ambani, the richest man in India, as well as somebody like Shaikh Mobin, a carpenter who resides in Dharavi, the largest informal housing cluster (slum) in Asia. What would “urban” mean then without reference to caste, class, or religious identity? The word “urban” treated thus as a homogeneous category has little explanatory value unless informed by reference to stratifications within the urban population.
Monthly Review Zine for more