Why polls go wrong

The Asian Age - - Story Behind The News - Dr Deven­dra Ku­mar ( The writer is a po­lit­i­cal an­a­lyst)

Psephol­ogy as a dis­ci­pline was in­vented and evolved in the United King­dom in the late 1940s and was first used in the early 1950s to read and pre­dict elec­tions. As a tool to read elec­tions, it is es­sen­tially a west­ern con­cept. Given the fact that in most of the west­ern democ­ra­cies the na­ture of pol­i­tics and so­ci­ety is very dif­fer­ent from In­dia, it was al­ways a huge chal­lenge for psephol­o­gists to adopt this tool in the In­dian con­text. Even in coun­tries like Bri­tain and the US, where polls of­ten pre­dict elec­tion out­comes very cor­rectly, they have gone off the mark sev­eral times. For in­stance, exit polls pro­jected John Kerry’s win in 2004 but he lost.

In In­dia, opin­ion polls start pour­ing down a cou­ple of months prior to the elec­tions, pro­jec­tions keep tak­ing dif­fer­ent shapes and colours till the exit polls. Mostly, the ac­tual re­sults are very dif­fer­ent from any one of these pre­dic­tions.

No won­der, a TV chan­nel re­cently called an opin­ion- poll based pro­gramme Siyasat Ka Sensex. Dif­fer­ent agen­cies come up with en­tirely op­po­site pro­jec­tions at the same point of time.

Thus, if psephol­ogy is a sci­ence and uses sci­en­tif­i­cally de­signed sta­tis­ti­cal tools then at given point of time all agen­cies should have the same pro­jec­tions and exit polls should never go wrong. As a fa­mous psephol­o­gist from the US once said, “Exit polls should make news when they go wrong, not when they are right.”

The big ques­tion is, why do exit polls go wrong in In­dia. There are many fac­tors which limit the use of psephol­ogy in the In­dian con­text. Pol­i­tics in Bri­tain and the US are dom­i­nated by two par­ties while In­dia has a multi- party sys­tem be­sides dom­i­nance of in­di­vid­u­als at the clus­ter, Assem­bly and Lok Sabha lev­els. This makes cal­cu­la­tion of the swing fac­tor a com­plex ex­er­cise due the shift of votes be­tween more than two par­ties.

Swing anal­y­sis is fur­ther com­pli­cated by the plu­ral­ity of the elec­torate across so­cio- eco­nomic seg­ments. Then, vari­a­tions due to the ed­u­ca­tion level, oc­cu­pa­tion, level of ur­ban­i­sa­tion etc. cre­ate mul­ti­lay­ers of vot­ers and their mo­ti­va­tions and in­flu­encers.

West­ern so­ci­eties have a rel­a­tively ho­moge­nous elec­torate, though there are vari­a­tions due to eth­nic­ity, re­gion and pres­sure groups. Thus, due to rel­a­tive ho­mo­gene­ity, only a 2,000- re­spon­dent sam­ple can be rep­re­sen­ta­tive to make pre­dic­tions with very nom­i­nal “mar­gin of er­ror”.

In In­dia, due to the plu­ral­ity of the elec­torate, a rep­re­sen­ta­tive and sta­tis­ti­cally sig­nif­i­cant sam­ple may run into hun­dreds of thou­sands and may still have a large mar­gin of er­ror. Thus when opin­ion polls in In­dia are done with just a few thou­sand sam­ples; how can we ex­pect them to be cor­rect?

In the West, ev­ery elec­tion is closely fought as there very few in­stances when a huge shift in vote takes place. This is due to the fact that a ma­jor­ity of the elec­torate is strongly aligned with their par­ties. If a poll man­ages to cap­ture the mood of the “swing elec­torate”, which isn’t large, it can make an ac­cu­rate pro­jec­tion. On the other hand, due to the lack of strong po­lit­i­cal af­fil­i­a­tions and dom­i­nance of lead­er­ship, caste and sev­eral contemporary is­sues, we of­ten see a huge shift in ev­ery elec­tion in In­dia.

For ex­am­ple, the BJP in­creased its votes from 18.8 per cent in 2009 to 31.3 per cent in 2014 and the Congress tum­bled from 28.6 per cent to 19.5 per cent. This makes the use of swing anal­y­sis very dif­fi­cult be­cause of the huge shift in the base vote in ev­ery elec­tion.

The BJP base shrank con­sid­er­ably be­tween 2009 and 2014, which made the task of poll­sters very dif­fi­cult in pre­dict­ing the post- 2014 Assem­bly elec­tions. Par­tic­u­larly in states ruled by the Op­po­si­tion par­ties, most of the opin­ion and exit polls un­der­es­ti­mated the BJP’s per­for­mance. The main rea­sons for this seem to be method­olog­i­cal com­pul­sions in es­ti­ma­tion of the swing and its ap­pli­ca­tion in the base vote ( votes polled by a party in the pre­vi­ous elec­tion).

In most of the Op­po­si­tion- ruled states be­fore 2014, the BJP was ei­ther non- ex­is­tent or was a con­sid­er­ably weak force. The BJP had a base of just 15 per cent votes in UP, 9 per cent in Haryana, 16 per cent in As­sam and 20 per cent in Jhark­hand. The BJP needed a more than 20 per cent pos­i­tive swing for vic­tory in all these states.

Prac­ti­cally speak­ing, no pre- poll sur­vey can es­ti­mate such a huge swing af­ter tak­ing into ac­count sam­pling er­rors. Prob­a­bly this was the rea­son that most of the poll­sters couldn’t pre­dict the BJP’s vic­to­ries.

In all these states, the BJP had a sig­nif­i­cantly higher vote share in the Lok Sabha elec­tions. If the Lok Sabha votes were used as a base, the pre­dic­tions could have been more ac­cu­rate. But mak­ing the Lok Sabha elec­tion vote as a base for the fore­cast would have been like com­par­ing ap­ples with or­anges and over­es­ti­mat­ing the Modi fac­tor.

The Kar­nataka elec­tions are the most re­cent ex­am­ple where the polls went hay­wire. Due to the split of votes by the KJP and the BSR Congress, the BJP vote was re­duced to just 20 per cent in the 2013 state polls. With this as the base vote, the BJP needed a 16- 17 per cent swing to get a sim­ple ma­jor­ity. If the 43.4 per cent votes polled in 2014 was made the base, the BJP could have reached close to ma­jor­ity even with a neg­a­tive swing of 7- 8 per cent, but then it would have been the ap­ples vs or­anges story.

These are some of the tech­ni­cal is­sues which limit the abil­ity of psephol­o­gists to read the In­dian elec­tions. Due to the po­ten­tial to the opin­ion polls to in­flu­ence the elec­torate’s mind, the ob­jec­tiv­ity of psephol­o­gists has come un­der the scan­ner. Given the his­tory of elec­tion pre­dic­tions in In­dia, in­her­ent lim­i­ta­tions of psephol­ogy as a dis­ci­pline and the ob­jec­tiv­ity of psephol­o­gists, opin­ion polls and exit polls should be read care­fully, in­clud­ing those for the five states for which re­sults will be de­clared on De­cem­ber 11.

Though psephol­ogy is a sci­ence, no two polls throw up sim­i­lar re­sults. Also, given the com­plex­i­ties of pol­i­tics in In­dia, this es­sen­tially West­ern con­cept has been found to floun­der more than once. Nev­er­the­less, it has emerged as a nec­es­sary evil. Also, the use of swing anal­y­sis is very dif­fi­cult in In­dia be­cause of the huge shift in the base vote in ev­ery elec­tion

IN IN­DIA, DUE TO THE PLU­RAL­ITY OF THE ELEC­TORATE, A STA­TIS­TI­CALLY SIG­NIF­I­CANT SAM­PLE MAY STILL HAVE A LARGE MAR­GIN OF ER­ROR

— S. SUREN­DER REDDY

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.