Statistician on election result: 8,000 of each party’s 2013 voters switched sides

He said that in reality, the gap between the two remained the same over the course of the entire legislature

Vincent Marmara estimates that there was a shift of roughly 8,000 voters towards both parties (Photo: Facebook)
Vincent Marmara estimates that there was a shift of roughly 8,000 voters towards both parties (Photo: Facebook)

The Labour Party lost 8,000 of their 2013 voters to the Nationalist Party, however this was cancelled out by the fact that an approximately equal number of voters made the opposite switch, according to statistician Vincent Marmara.

He made the comments in an interview with it-Torca where he discussed the same paper’s voting intention surveys, for which he was responsible.

“When one considers that the Nationalist Party lost by a big margin in 2013 but was still losing voters, it shows that the Labour Party was increasing its share of the vote and still convincing people,” he said.

In addition to this, he said that Labour won a larger proportion of new voters, while PN fared better with those who hadn’t voted in the past.

“At the end of the day all the shifts between the parties and various voters cancelled each other out and the gap remained the same over the course of the past four years,” he said.

While all surveys carried out by newspapers in the run-up to the election showed the Labour Party winning, albeit by a smaller majority, it was only it-Torca that accurately predicted the size of the win.

The last survey published by the paper two days before the election had the Nationalist Party’s vote at 44%, with Labour 9.5 points ahead with a 53.5% share of the vote.

The final official result had Labour at 55% and the PN at 43.7% meaning the survey misjudged Labour’s vote by 1.5 points, while the PN’s predicted result was just 0.3% greater than the its actual result.

Marmara said that the surveys carried out by it-Torca were the only ones to have analysed, with some detail, the crop of respondents who claimed to be undecided.

“The multiple imputation technique was used,” said Marmara. “It’s a method where one finds a technique to predict, as accurately as possible, how undecided respondents will vote.”

He said that in order to get an accurate result, the selected sample of respondents was always representative of the age, sex and district distribution of the Malta’s eligible voters. Moreover, he said, this ensured the survey’s results matched the 2013 election results.  

On respondents who would not reveal their intentions, Marmara said that while a lot was said about how their numbers were larger than in previous elections, in reality there were roughly 10% less when compared to 2013.

“The result, and the fact that the gap between the parties remained constant, indicates that - in reality - undecided respondents and those that didn’t reply simply did not wish to reveal which party they supported,” he said.

Marmara added that the level of undecideds and no replies was a normal one and that through the method employed in his surveys, an accurate prediction of how they would be voting was made possible.