JavaScript is disabled in your web browser or browser is too old to support JavaScript. Today almost all web pages contain JavaScript, a scripting programming language that runs on visitor's web browser. It makes web pages functional for specific purposes and if disabled for some reason, the content or the functionality of the web page can be limited or unavailable.

Friday, July 18, 2025

Polling the polls

by

20150913

"The poll­sters got it all wrong," said MP Colm Im­bert in a let­ter to the ed­i­tor. So, even be­fore get­ting a min­is­te­r­i­al port­fo­lio, Im­bert was prac­tis­ing mis­lead­ing state­ments.

In fact, H.H.B. & As­so­ciates, the firm con­tract­ed by Guardian Me­dia Lim­it­ed to con­duct opin­ion polls for the elec­tion sea­son, fore­cast a Peo­ple's Na­tion­al Move­ment vic­to­ry of 22 seats to 19. The ac­tu­al re­sult was 23-18.

The poll­sters who got it wrong were Hamid Ghany of UWI's Con­sti­tu­tion­al Af­fairs and Par­lia­men­tary Stud­ies Unit (Cap­su), Vish­nu Bis­ram of the North Amer­i­can Caribbean Teach­ers As­so­ci­a­tion (Nac­ta) and Nigel Hen­ry of So­lu­tions by Sim­u­la­tion (SBS). Tech­ni­cal­ly speak­ing, Hen­ry didn't get it ei­ther wrong or right since, on the eve of the elec­tion, SBS said the race was too close to call. But H.H.B. man­ag­ing di­rec­tor Louis Bertrand didn't mince his words, say­ing: If a poll can­not pre­dict mar­gin­al seats, it is not a poll. This is inar­guable. If as a poll­ster you can't make a fore­cast, then you are just as ig­no­rant as the or­di­nary per­son in the street, ex­cept for hav­ing been paid a few hun­dred thou­sand dol­lars to ad­mit your ig­no­rance.

Un­til this elec­tion, SBS had an im­pres­sive record of ac­cu­rate pre­dic­tions. Their 2015 elec­tion poll, how­ev­er, found that 51 per cent of their re­spon­dents sup­port­ed the Peo­ple's Part­ner­ship and 48 per cent the PNM. How­ev­er, not­ing Amer­i­can re­search which showed that peo­ple's ex­pec­ta­tions of who would win an elec­tion was a more ac­cu­rate pre­dic­tor than state­ments of who they would vote for, Hen­ry found that 37 per cent ex­pect­ed the PNM to win and 36 per cent the PP. Even so, he could not come up with a tech­nique to pen­e­trate that sta­tis­ti­cal­ly tied cur­tain. In­stead, SBS's tele­phone poll of 1,997 per­sons fore­cast that the win­ning par­ty would have 21 or 22 seats: and that was def­i­nite­ly wrong.

Cap­su's fore­cast was far more egre­gious. Re­leased one week be­fore the elec­tion, Cap­su in­ter­viewed 2,328 re­spon­dents, who com­prised 39.2 per cent Afro-Trinida­di­ans, 38.3 per cent In­do-Trinida­di­ans and 21.7 per cent mixed de­scent. This is a rel­a­tive­ly large sam­ple, yet Cap­su found that the PP had a near­ly sev­en per cent lead over the PNM in the mar­gin­al con­stituen­cies of La Hor­quet­ta/Tal­paro, Point Fortin, St Joseph, To­co/San­gre Grande and Tu­na­puna. All these seats were won by the PNM. Cap­su al­so found that Kam­la Per­sad-Bisses­sar was favoured as prime min­is­ter by over 42 per cent of re­spon­dents, as com­pared to 35 per cent who pre­ferred Kei­th Row­ley.

Nac­ta, which did not re­veal its sam­ple size but claimed to have con­duct­ed face to face in­ter­views, said that the elec­tion could go ei­ther way and that 47 per cent of re­spon­dents pre­ferred Kam­la as com­pared to 42 per cent who pre­ferred Row­ley and that the PP had more na­tion­al sup­port.

H.H.B., how­ev­er, was able to make its ac­cu­rate fore­cast by polling just 120 per­sons in each of the five mar­gin­al seats.

So why did every­one else get it so wrong? There may be many fac­tors, but bias could have been key. By this I do not mean that the bias was de­lib­er­ate or con­scious, and it may not even have been po­lit­i­cal. But any­thing from the phras­ing of the ques­tions to the choice of in­ter­view­ers may have skewed the re­sults, ei­ther be­cause the re­spon­dents cho­sen were in some way a bi­ased sam­ple or be­cause their re­spons­es did not re­flect their ac­tu­al opin­ions.

Bertrand al­so got the fore­cast right be­cause he con­cen­trat­ed on the mar­gin­als. The oth­er mis­lead­ing state­ment Im­bert made in his let­ter was that "seats deemed by the poll­sters to be too close to call were, in fact, won by large ma­jori­ties, mak­ing com­plete non­sense of the poll­sters' pre­dic­tions." While this was true of the three poll­sters who got it wrong, H.H.B. was right on tar­get. For Tu­na­puna and To­co/San­gre Grande, Bertrand fore­cast a PNM vic­to­ry of 55 per cent and, for San Fer­nan­do West, his fore­cast was 60 per cent. The ac­tu­al per­cent­ages in all three con­stituen­cies was 59 per cent for the vic­to­ri­ous can­di­dates.

None of this means that polls have been dis­cred­it­ed in T&T pol­i­tics. In fact, the lessons learned from this elec­tion will help poll­sters to be more ac­cu­rate for fu­ture elec­tions. The key in­fer­ence to be drawn, in my view, is that the swing vot­ers in the mar­gin­al con­stituen­cies are just as mono­lith­ic in their think­ing as the diehard vot­ers. It must be so, or else the five or six mar­gin­al seats would not have all swung the same way.

With a vot­er turnout of near­ly 67 per cent, ac­cord­ing to the Elec­tions and Bound­aries Com­mis­sion, the PNM got 378,447 votes while the com­bined to­tal for the Peo­ple's Part­ner­ship was 341,597. So the PNM won this elec­tion by a slim ma­jor­i­ty of 36,850 votes, which is a mere five per cent of all those who vot­ed with a 51.5 per cent over­all ma­jor­i­ty.

This means that cor­rup­tion is not a key is­sue for most vot­ers and that the Gov­ern­ment does not have a strong foun­da­tion to win na­tion­al sup­port for its poli­cies. At a time when hard de­ci­sions will have to be tak­en, this does not bode well for or­di­nary cit­i­zens.

Kevin Baldeosingh is a pro­fes­sion­al writer, au­thor of three nov­els, and co-au­thor of a his­to­ry text­book.


Related articles

Sponsored

Weather

PORT OF SPAIN WEATHER

Sponsored