2019 India Lok Sabha Election Forecasts

Roberto Cerina and Ray Duch

The 2019 India Lok Sabha elections are expected to be a victory for the National Democratic Alliance (NDA, led by Modi’s BJP).  CESS Vote India is predicting the NDA coalition of Prime Minister Modi will retain control of a majority in the Lok Sabha.   The simulated average number of seats predicted for NDA is 304.  CESS Vote India is predicting that the United Progressive Alliance (UPA, led by Gandhi’s Congress) will achieve 119 seats.  All other alliances are predicted to get 120 seats. This is based on a total of 12,500 voting intentions that were surveyed.  Our forecasts are the average of the estimates obtained from surveys (conducted with the CESS India subject pool and MTurk workers in India) and the traditional published polls in India.

Figure 1: Histogram of predicted seats by party for 500 simulated elections. The faded histograms in the background represent the two components of the forecast, namely our online surveys and publicly available traditional polls. The point estimate is identified by the average of the two sources. The dark green line represents the 272 seats threshold - anything above that gives a majority in the Lok Sabha.

Vote India is an election forecasting project that proposes novel strategies for generating vote and seat predictions employing convenience online subject pools – specifically the CESS India online subject pool.  CESS India and 0ptimus Consulting have jointly funded the Vote India project. CESS India is a collaboration between the Nuffield Centre for Experimental Social Sciences and FLAME University, Pune India.  0ptimus Consulting is a Washington, D.C.-based consulting firms.

The novelty of the CESS Vote India forecasting project is three-fold.  First, we incorporate multiple and quite diverse online convenience samples as part of the estimation strategy – we complement information from the CESS India subjects with very regular surveys of MTurk workers in India.  Different convenience samples will add complementary information to forecasts of this nature – the broader issue is how to identify the convenience samples that provide optimal complementarity to forecasts of this nature.  We propose estimation strategies that integrate quite disparate subject pools.  Figure 2 summarizes the differences between our population data and the data obtained from our CESS and MTurk samples.

Figure 2: Differences between our population frame and the re-sampled, non-probability samples, in per- centages. The percentages that these pop − sample difference are calculated from sum to one by category (i.e. for gender, % male and % female sum to 1; similarly for income and education categories, etc.). Above the dark green line we are under-sampling; underneath it we are over-sampling.

Second, we organize the Indian nation into a stratification frame that has 6611 cells – essentially these are defined by the number of demographic categories in our forecasting model that includes income, religion, caste, gender age and education.  We generate estimated vote probabilities for all of these cells (or, if you wish, demographic categories)  feeding survey data collected from our different online subject pools to a random forest.  A novelty here is estimating the vote probabilities separately for the CESS Online and MTurk subjects and then combining the estimates.  Figure 3 presents the expected vote share by alliance for the online sample.

Figure 3: Expected vote share by alliance: the NDA is in red/orange; the UPA in blue/skyblue; the others in black/grey. From left to right: a) 500 simulations of the expected vote share over the 13 weeks monitoring period; b) the breakdown of expected vote share by source (CESS subjects and mechanical turks); c) 500 simulations of the national swing since the 2014 election.

Thirdly, we forecast the national vote share for the major parties by simply applying the cell probabilities of voting for each party to population estimates for those cells obtained from the Indian Human Development Survey.  This allows us to estimate a national swing for each of the major parties.  The novelty here is that we use historical data on the relationship between national and state swings in vote shares to estimate the seat shares for the state parties. The state level multipliers of the national swing are presented in Figure 4.

Figure 4: Graphical representation of the state level multiplier of the national swing with uncertainty bounds (2 standard deviations). The dotted green line represents the National Swing.

The counterpart to our seat-estimation by online surveys is a forecast based on traditional opinion polling. We analyse 42 polls from the 2009, 2014 and 2019 election, and attempt to quantify the polling house error, and remove it from a moving average of the polls. The results of this effort are then averaged with our online surveys. Figure 5 shows the expected number of seats according to our traditional opinion polls model, over the 2019 campaign. The symbols on the plot represent specific polling houses.

Figure 5: Expected number of seats by alliance for the 2019 Lok Sabha election, net of house bias.

Details on the design of the India Vote study and the forecasting methodology are available HERE.