Resh­ma Ra­goonath

With mere hours to go be­fore the polls open, this coun­try's poll­sters are pro­ject­ing vary­ing re­sults, ce­ment­ing the fact that the 2015 elec­tion is too close to call.

Cit­i­zens are no bet­ter off than they were at the start of the cam­paign, since they are still left won­der­ing just which par­ty has the edge in to­mor­row's gen­er­al elec­tion–the rul­ing Peo­ple's Part­ner­ship (PP) or the Peo­ple's Na­tion­al Move­ment (PNM).

How­ev­er, H.H.B & As­so­ciates poll­ster Louis Bertand be­lieves that even though there are slight vari­a­tions in the pre­dic­tions of the lo­cal polls, cit­i­zens can make an as­sess­ment of who will win to­mor­row since lo­cal polls were proven ac­cu­rate in the past.

"Polls like any oth­er things you are try­ing to fore­cast, by and large, have been proven to be more ac­cu­rate than in­ac­cu­rate. It is like the weath­er­man who tells you it is go­ing to rain and it does not, but it rains the next day," he ex­plained.

Over the last week, poll­sters have an­nounced vary­ing pro­jec­tions for the out­come of the high­ly an­tic­i­pat­ed Sep­tem­ber 7 polls.

H.H.B & As­so­ciates, which con­duct­ed the Guardian Me­dia Ltd (GML) poll, is pro­ject­ing a PNM vic­to­ry, with the op­po­si­tion par­ty tak­ing 22 seats and the PP cap­tur­ing 19. This find­ing con­trast­ed with the North Amer­i­can Caribbean Teach­ers As­so­ci­a­tion (Nac­ta) poll by Dr Vish­nu Bis­ram, which pre­dict­ed yes­ter­day that the PP would win 21 seats, the PNM 18 seats, with two mar­gin­als up in the air. The UWI Con­sti­tu­tion­al Af­fairs and Par­lia­men­tary Stud­ies Unit (Cap­su) poll re­leased last week al­so found that the PP is lead­ing in five mar­gin­al seats in the elec­tion race. So­lu­tion by Sim­u­la­tion track­ing poll by Nigel Hen­ry pro­ject­ed yes­ter­day that the PP was lead­ing by 38 per cent, the PNM had 36 per cent, and a whop­ping 25 per cent were still un­de­cid­ed.

Dr Vish­nu Bis­ram, in his Nac­ta poll, said the per­cent­age of un­de­cid­ed vot­ers had shrunk con­sid­er­ably from a month ago, from "a high of 31 per cent to just six per cent now," and they could very well de­cide the out­come of the elec­tion.

Po­lit­i­cal an­a­lyst Dr Maukesh Bas­deo agreed that the un­de­cid­ed vot­ers could swing the elec­tion ei­ther to the PP or to the PNM.

He agreed that with the vary­ing re­sults, one might be left won­der­ing if polls were ac­cu­rate, es­pe­cial­ly giv­en the sit­u­a­tion with the UK elec­tions in May.

How­ev­er, he said, it was "very dif­fi­cult" to pre­dict hu­man be­hav­iour.

"Polls are a pos­si­ble out­come, it is not to say 100 per cent ac­cu­rate, be­cause you are speak­ing about hu­man be­hav­iour. The poll is a snap­shot in time, a per­son's view."

He ex­plained that some­one may give an opin­ion on a po­lit­i­cal is­sue like whom he is vot­ing for and 24 hours lat­er, they could be ex­posed to new in­for­ma­tion which could change their opin­ion and their be­hav­iour would change.

In the UK, poll­sters came un­der fire af­ter they pre­dict­ed a hung par­lia­ment with the Con­ser­v­a­tives, led by Prime Min­is­ter David Cameron, need­ing a coali­tion part­ner to re­main in gov­ern­ment. How­ev­er, the op­po­site oc­curred, and Cameron's par­ty re­ceived a land­slide vic­to­ry se­cur­ing 331 seats, five more than need­ed for a House of Com­mons ma­jor­i­ty.

Dr Hamid Ghany ex­plained that the UK sit­u­a­tion was dif­fer­ent from T&T's be­cause poll­sters there used method­olo­gies such as on­line polling, tele­phone polling, and lim­it­ed face-to-face in­ter­cept.

"We have had, in T&T, tele­phone polling and face-to-face in­ter­cept," he said.

The key to polling, he said, was the method­ol­o­gy used for gath­er­ing the da­ta and the size of the mar­gin of er­ror.

"If your mar­gin of er­ror is more than +/- five per cent then the poll is un­re­li­able, if is it less than +/- five per cent then the poll is re­li­able. In or­der to get a mar­gin of +/- five per cent, that is de­ter­mined by the num­ber of ques­tion­naires," he said.

Hen­ry, man­ag­ing di­rec­tor of So­lu­tion by Sim­u­la­tion, said polling had a long his­to­ry of suc­cess­es, but re­cent­ly it had been fac­ing chal­lenges for a num­ber of rea­sons.

How­ev­er, he said, it was very im­por­tant be­cause "we do not just need to know the num­bers at the end of the day, but we need to know why peo­ple are do­ing what they are do­ing in or­der to im­prove gov­er­nance."

He said the ques­tion should be: how can we make the polls more ac­cu­rate? �2See Pages A6 & A8