Rearguard action by regional parties and local factors halted the BJP's juggernaut in Haryana and Maharashtra. The party fell short of a majority in Haryana, but its alliance with the Shiv Sena was victorious with a reduced seat tally in Maharashtra.
A tempered victory The BJP-Shiv Sena alliance managed to romp home with a diminished majority compared to the 2014 elections. The combine faced setbacks in the Marathwada and the Western Maharashtra regions where the NCP garnered a significant number of seats. Otherwise, the combine fared well across geographical regions.
Carrying the other Both the NDA and the UPA turned out to be somewhat lopsided coalitions, shows Table 1 . The BJP won 63% of the seats it contested with a vote share of 44.6% in those seats, much higher than its partner, Shiv Sena (45.2% and 38.4% respectively).
The NCP had a much better hit rate (43.8% vs 30.6%) and contested vote share (38.3% vs 32.8%) compared to the Congress.
Outside the region of... Except for Konkan, the NDA suffered a drop in vote share in all other regions of the State. The INC-NCP combine lost shares too, but smaller UPA parties picked up shares. (Table 2)
Less lopsided The BJP and the Shiv Sena, even if they contested separately in 2014, won more than 50% of the vote shares in nearly half the 288 seats in the state, when their shares were combined (Table 3) .
This number reduced to 78 seats (less than a third) in this election. The vote share was between 40% and 50% in the bulk of the seats, indicating much closer competition.
Pan-geographic Despite a significant decrease in vote shares across both rural and urban areas, the NDA still managed to upstage the INC-NCP combine. These votes were lost to smaller partners in the UPA (Table 4) .
Cultivator angst In areas where farming was predominant (more than 45% of the population were dependent on agrarian income), the Congress-NCP combine won substantially. In other areas, the NDA fared better (Table 5) .
Less dominant The NDA had more lopsided victories (margins more than 10%) than the UPA, but these were fewer compared to the 2014 elections (Table 6) .
Dalit shift The NDA's vote share among seats reserved for Scheduled Tribes was marginally lower than in 2014, but was more substantially less in general and Scheduled Caste seats. Again, the smaller UPA allies mopped up the bulk of these lost votes (Table 7) .
In all the above calculations, the vote and seat shares for the BJP and the Shiv Sena in 2014 were tallied together for comparison purposes, even though the parties had competed separately and not in alliance.
This enables a good comparison with the 2019 elections when these parties were in a seat sharing alliance. The INC-NCP combine held in both in 2014 and 2019; however smaller parties such as Swabhiman Paksha among others were also part of the UPA.
Close call Belying expectations following Lok Sabha polls when the BJP won 58% of the votes, the party was reduced to 40 seats. It is now the single largest party (five seats short of the majority and 36% vote share).
The Congress gave the party a tough fight, but the JJP's strong showing in western Haryana was a blow to the BJP, which could still return to power with support from independents.
The JJP's western frontier Despite a better performance across regions, the BJP could not convert the leads in vote shares into seats, as the JJP's strong showing in West Haryana dented the BJP's tally, shows Table 1 . The INC also grew marginally across regions even as the INLD collapsed compared to 2014.
Spoiler effect among farmers While the BJP and the Congress both gained at the expense of the INLD in places with a significant number of farmers and those without, the JJP's strong performance in the rural areas restricted the other two parties (Table 2) .
Advantage, cut into BJP was the better performer in terms of vote share ranges, but the JJP's dominance in four seats (>50% vote share), helped it become the spoiler (Table 3) .
Urban surprise The Congress was a major gainer in urban areas where the BJP suffered a 6 percentage point decline compared to 2014. Rural areas were more keenly contested by all three parties (Table 4) .
Much closer than 2014 The BJP's dominant performance in 2014 was reflected in its winning 28 seats with a margin more than 10%. That number fell by seven seats in 2019. In six seats, the vote margin was less than 1%, indicating a close poll (Table 5) .
How agriculture income share was calculated Rural households directly report whether their primary income is from agriculture, small enterprise, wage work, or any other source. Researchers behind the SHRUG data lab aggregated this and reported the share of households in a village that draw their income from agriculture. Higher the figure, more the income from agriculture.
How rural-urban data was calculated Night lights are used as a proxy for electrification or economic activity, which in turn differentiates a rural area from an urban one. Gridded night lights data were matched to village and town polygons, and aggregated into totals, from which mean night light was calculated by the SHRUG team.
Sources: Maps from Datameet Github; past election results from Trivedi Political Data Centre; regional classifications from CSDS; rural-urban divisions using night light data and income from agriculture data from THE SHRUG Development Data Lab; 2019 data from Election Commission of India; maps and analysis rendered using Tableau.
Compiled with inputs from Varun B. Krishnan and Naresh Singaravelu
Illustrations by Satheesh Vellinezhi
