Cell­phones ham­per US poll­sters

Pre­dict­ing vote made much harder

Weekend Argus (Saturday Edition) - - FRONT PAGE -

BOS­TON: Many opin­ion polls show Demo­crat Hil­lary Clin­ton lead­ing Repub­li­can Don­ald Trump in a tight race for the Novem­ber 8 US pres­i­den­tial elec­tion, but any one of four fac­tors may make the out­come harder to pre­dict.

Among the chal­lenges for poll­sters: the his­toric un­pop­u­lar­ity of both can­di­dates, the po­ten­tial Elec­tion Day voter re­sponse to the polls them­selves, the grow­ing aban­don­ment of land­lines for cell­phones and the rise of on­line polling.

An av­er­age of polls compiled by RealClearPol­i­tics.com shows Clin­ton beat­ing Trump by 5.4 per­cent­age points. High­light­ing the dif­fi­cul­ties, the range varies from plus-13 for Clin­ton to a straight tie.

Trump has said the elec­tion is rigged against him and this week he ac­cused me­dia or­gan­i­sa­tions of tilt­ing the polls de­lib­er­ately, but he has yet to of­fer ev­i­dence to back up these claims.

Voter turnout in the past few pres­i­den­tial elec­tions has been about 60 per­cent. But given both can­di­dates’ low over­all pop­u­lar­ity, turnout this year may fall to as low as 52 per­cent, said Cliff Zukin, a pro­fes­sor emeritus of po­lit­i­cal sci­ence at New Jer­sey’s Rut­gers Univer­sity. That makes it hard to guess who might stay home.

A sec­ond pit­fall is the ef­fect of the polling it­self on vot­ers. So­ci­ol­o­gists be­lieve polls can weaken pro­jected win­ners by mak­ing their sup­port­ers more con­fi­dent of the out­come and, there­fore, less likely to vote. Poll­sters cau­tion, how­ever, that the ef­fect of polls on the elec­torate can only go so far.

“If it were show­ing Clin­ton up by 2 points, then it’s cer­tainly pos­si­ble that it would be within the mar­gin of er­ror that Trump might win,” said Dou­glas Schwartz, di­rec­tor of the Quin­nip­iac Univer­sity Poll.

One of the big­gest fac­tors in polling to­day is the preva­lence of cell­phones. About half of Amer­i­cans have only a cellphone and no land­line, ac­cord­ing to Fed­eral Com­mu­ni­ca­tions Com­mis­sion data. This makes it harder for poll­sters to gather a truly ran­dom sam­ple be­cause US law pro­hibits com­put­erised auto-di­alling to cell­phones and there is no cen­tral direc­tory for cell­phones.

This has led polling out­fits to gen­er­ally rely on lower sam­ple sizes to come up with re­sults.

Oth­ers con­duct sur­veys on­line. This al­lows them to reach large num­bers of peo­ple at lower cost. But be­cause par­tic­i­pants are vol­un­teers in many cases, rather than se­lected at ran­dom, seg­ments of the elec­torate may be left out.

The lower or skewed re­sponse in both cellphone and on­line polls can pose a chal­lenge with poll­sters hav­ing to ad­just re­sults to match the real world. To ac­com­plish this, they weight more heav­ily the opin­ions of types of vot­ers un­der-rep­re­sented in their sur­veys. – Reuters

Hil­lary Clin­ton

Newspapers in English

Newspapers from South Africa

© PressReader. All rights reserved.