Over the past four weeks The Malta Independent has been providing its readers with a weekly scientific survey commissioned to Business Leaders, a leading company specialising in polls and scientific data. The iSurvey started in The Malta Independent in 2014 on a quarterly basis and switched to a weekly survey the week before Prime Minister Joseph Muscat called a snap election for 3 June.
The process of collecting and processing data in a scientific manner is very laborious and requires ample resources. The iSurvey generally gathers, on average, a population of 600 respondents, which many times requires triple that amount of telephone calls if not more. For accuracy’s sake, it encompasses both landline and mobile numbers for it to be as representative to society as much as possible.
Does the iSurvey give a clear picture?
Like all scientific surveys the methodology applied is crucial to get help pollsters get as close to the actual result as possible. However, over the past years we have come to witness global polls failing to predict what people actually do when they’re alone in the voting booth. What are the reasons behind such inaccuracies? The basic problem with surveys arises when a significant chunk of your population refuses to participate, declares it is undecided or simply lies on their voting intentions for personal reasons. While pollsters have developed methods on how to deal with those who refuse to participate and those who declare to be undecided, the later is a nightmare as it skews the data to one side or the other. When the amount of people lying is low, pollsters have a way of weighting a survey. This is done by comparing the replies of respondents on how they voted in the past election to the actual result of that election. If it is found that people declare in a survey to have voted a party more or less than the actual electoral result than the pollsters need to adjust the weighting of the survey to reflect that of the election. This helps to give an accurate picture of the situation, however, it reduces the efficiency of the survey and widens the margin of error.
When as many people, or more, than those responding to the survey refuse to participate coupled with those who give a false reply on their intentions, the exercise in weighting does not give a clear picture. When this happens, pollsters advise to ignore the findings and look within the cross-referenced data to extract other useful information.
The iSurvey has given a clear picture of the political status since its inception back in 2014. When compared to other surveys such as those published by MaltaToday, the iSurvey has a high level of credibility. However, during the past four weeks the data has been skewed and weighted and it is clear that many respondents are not divulging their true intentions, neither on how they voted in 2013 nor on how they intend to vote on 3 June.
This doesn’t mean that the exercise is not worth analysing so as to extract other information relevant to what the pollsters want to discover.
What have we found throughout the campaign?
The biggest revelation in the iSurvey throughout the campaign was the mistrust people have in polls. Week after week, the skew in each wave was becoming increasingly noticeable. The amount of people reached by mobile and landline to reach a sample of 500+ respondents was multiplying each week. One clear picture came out: too many people are refusing to say how they intend to vote. When analysing the swing from one party to the other, or what is commonly referred to as the switchers, one finds that the swing from PN to PL was growing each week. This is hardly credible for many reasons outside the survey because as journalists our job is to analyse the public sentiment. We do this in several ways including a daily snap shot of how people are engaging on social media. But for this exercise we prefer to stick to those reasons that are identifiable within the survey itself.
For example, when cross-referencing the trust rate received by the leader of the Opposition with the voting intentions of switchers, the numbers don’t add up. So, for example, people would rate Simon Busuttil over Joseph Muscat but than would say they will vote the PL on 3 June. It is these types of inconsistencies in respondents’ replies that indicate that too many people are skewing the data.
How did they fare?
As regards the two main leaders, Joseph Muscat has retained the trust rating, losing only 1 point throughout the campaign while Simon Busuttil increased his trust rate just below 2 points. He however remained less trusted than Joseph Muscat. One would argue that if the survey results for ‘how will you vote in June 3rd’ are skewed than so are the trust ratings, but a thorough analysis of the data keeps showing consistency between the trust rating and how respondents claimed to have voted in 2013 while the voting intentions for June 3rd swings to the other side.
Throughout the election campaign the iSurvey sought to identify other information traits such as those pertaining to which issues are pressing on people’s minds. Corruption resulted to be the top concerning issue for respondents. Yet again, another indicator that people tend to lie when declaring their intentions on how they will vote on 3 June otherwise how can one explain that they are worried on corruption but prefer voting in the PL which is the only party facing corruption allegations on its top three officials?
Who will win the next election?
If we take the latest surveys at face value without challenging the skew evident in the survey, the PL will be winning again by a landslide. But when factoring the skew and how the leaders fare in their trust rate barometer it is clear that neither party will be claiming a landslide victory. The prediction is that whoever wins will be doing so by a small margin when compared to what happened in 2013.