Nigel Hen­ry

If the In­de­pen­dent Lib­er­al Par­ty (ILP) did not "split" the vote–as is the pre­vail­ing in­ter­pre­ta­tion of the lo­cal gov­ern­ment elec­tions re­sults go–in­stead of win­ning eight of the 14 cor­po­ra­tions, the Peo­ple's Na­tion­al Move­ment (PNM) would have won sev­en of the 14 cor­po­ra­tions.That's right.Even if all 102,814 votes cast for the ILP in the lo­cal gov­ern­ment elec­tions were added to the Peo­ple's Part­ner­ship (PP) to­tal, the PNM would still con­trol the Por-of-Spain, San Fer­nan­do, Ari­ma, Point Fortin, Diego Mar­tin, San Juan/Laven­tille and Tu­na­puna/Pi­ar­co Cor­po­ra­tions to­day.

Those who have em­pha­sised that the PNM failed to win a ma­jor­i­ty of votes cast do have a valid point. In many cas­es–17 dis­tricts to be ex­act–a PNM can­di­date was elect­ed coun­cil­lor when a ma­jor­i­ty of peo­ple want­ed some­one else.

The fact that a can­di­date can win with a plu­ral­i­ty, rather than a ma­jor­i­ty, of votes is con­sid­ered by many to be a ma­jor flaw in the first-past-the-post (FPTP) elec­toral sys­tem, which can on­ly be cured by strate­gic vot­ing or con­ver­sion to an­oth­er elec­toral sys­tem such as a run-off sys­tem or pro­por­tion­al rep­re­sen­ta­tion (PR).The con­cept of the split vote is sim­ple.Sup­pose 40 of 100 vot­ers pre­fer can­di­date A, and the oth­er 60, while op­posed to can­di­date A, are split 35-25 in sup­port be­tween can­di­dates B and C. In the FPTP sys­tem, sup­port­ers of can­di­date C must choose be­tween vot­ing for their pre­ferred can­di­date "to make a state­ment," or vot­ing for can­di­date B to avoid can­di­date A from win­ning. Such a vote for the "less­er of two evils" is called "strate­gic vot­ing." Of course, strate­gic vot­ing re­quires ac­cu­rate pre-elec­tion mea­sure­ment (polling) and a so­phis­ti­cat­ed and in­formed elec­torate.A two-round "run-off" vot­ing sys­tem solves this prob­lem by al­low­ing vot­ers to choose the can­di­date or par­ty of their pref­er­ence on the first bal­lot, and then re­turn to the polls to chose be­tween the two front run­ners in cas­es where no can­di­date wins a ma­jor­i­ty on the first bal­lot.

This sys­tem en­sures that the win­ner is elect­ed by a ma­jor­i­ty vote, and still al­lows the pref­er­ences of the elec­torate to ac­cu­rate­ly re­flect in the re­sults of the first bal­lot.So what hap­pened in the lo­cal gov­ern­ment elec­tions? How would the re­sults have dif­fered if vot­ers were able to chose be­tween two can­di­dates, with­out the im­pact of a con­found­ing third choice? It is very dif­fi­cult to de­ci­pher vot­er in­tent with­out ask­ing them di­rect­ly (in oth­er words a poll), but we can look for hints in the da­ta we have.When com­pared against the 2003 lo­cal gov­ern­ment elec­tion re­sults, which were near­ly iden­ti­cal to this year's re­sults in terms of the dis­trict-lev­el out­comes, the PNM lost 11 points from 53.3 per cent in 2003 to 42.2 per cent this year.The Unit­ed Na­tion­al Con­gress (UNC)/Na­tion­al Al­liance for Re­con­struc­tion (NAR) ac­com­mo­da­tion lost 12 points from 46.4 per cent to 34.5 per cent this year for the PP.This means that each par­ty or coali­tion con­tributed rough­ly equal­ly in terms of his­tor­i­cal sup­port­ers to the ILP's 23 per cent sup­port in 2013.

How­ev­er, there is no ques­tion that the ILP) per­formed bet­ter in his­tor­i­cal­ly UNC ar­eas, than his­tor­i­cal­ly PNM ar­eas.There is a very strong pos­i­tive cor­re­la­tion be­tween the ILP vote and the UNC vote, and a very strong neg­a­tive cor­re­la­tion be­tween the ILP vote and the PNM vote. The ILP did best in Ch­agua­nas and Cou­va/Tabaquite/Tal­paro and worst in Port-of-Spain and San Fer­nan­do.

It takes some tricky analy­sis to see how the two can be true at the same time. The an­swer is that in PNM-dom­i­nat­ed ar­eas the ILP took most of their votes from his­tor­i­cal PNM sup­port­ers, in UNC-dom­i­nat­ed ar­eas the ILP took a large chunk of votes from his­tor­i­cal UNC sup­port­ers, and in bat­tle­ground ar­eas such as those in Siparia and Ma­yaro/Rio Claro, the ILP ac­tu­al­ly per­formed quite poor­ly.In 42 com­pet­i­tive three-way races, de­fined as dis­tricts in which no can­di­date won a ma­jor­i­ty, the Peo­ple's Part­ner­ship won 23, the PNM 17 and the ILP 2. Of those 42 seats, 11 were won by the PNM in cor­po­ra­tions that the PNM would have won any­way, and 18 were won by the PP in cor­po­ra­tions that the PP would have con­trolled any­way.

For ex­am­ple, nine of these "split-vote" seats fell in PP-dom­i­nat­ed Cou­va/Tabaquite/Tal­paro Cor­po­ra­tion, and the PP claimed eight of these nine.In oth­er words, the analy­sis shows that there prob­a­bly al­ready is a fair de­gree of strate­gic vot­ing in the pop­u­la­tion. Every­one wants their vote to be count­ed, and no­body wants their vote to be "wast­ed."In terms of turnout, there is a great cor­re­la­tion be­tween the dis­tricts where there is sig­nif­i­cant­ly larg­er turnout, and the dis­tricts where the ILP did well.Of the top one-third of dis­tricts with great­est in­crease in turnout over the last lo­cal gov­ern­ment elec­tions three years ago, the ILP cap­tured 25 per cent of the vote, com­pared to 21 per cent in the dis­tricts where turnout was mar­gin­al high­er or stag­nant.This com­pares with a vote dif­fer­ence of +2 per cent for the PNM and -7 per cent for the PP in dis­tricts with the high­est in­crease in turnout. This hints that the emer­gence of the ILP had some im­pact on cre­at­ing the record-break­ing vot­er turnout this year.While the ev­i­dence sug­gests that vote "split­ting" did not af­fect the over­all re­sult of the lo­cal gov­ern­ment elec­tions, there were some ar­eas where it did.Two of the most fierce­ly com­pet­i­tive three-way races were in San Juan East and Val­sayn/St Joseph.These are two of the four dis­tricts that in­ter­sect with the par­lia­men­tary dis­trict of St Joseph that votes to­day.Val­sayn is par­tic­u­lar­ly friend­ly to third-par­ty can­di­dates where Con­gress of the Peo­ple (COP) can­di­date Gillian Lucky ac­tu­al­ly cap­tured more votes than her two com­peti­tors in a los­ing ef­fort in 2007.

This ar­ti­cle was writ­ten by Nigel Hen­ry of So­lu­tion by Sim­u­la­tion, a pri­vate firm that us­es com­put­er mod­el­ling to pro­vide in­sight in­to po­lit­i­cal con­texts. So­lu­tion by Sim­u­la­tion ac­cu­rate­ly pre­dict­ed the out­come of the To­ba­go House of As­sem­bly elec­tion, pub­lished in our Jan­u­ary 14 edi­tion.