“Political parties cannot undermine social media as a channel to reach people and to pass their messages across”

Interview: Dirish Noonaram & Venna Pavaday – Founding Partners VERDE

‘Ethnicity remains an important consideration by voters even today. This is expected to change, although not in the near future’

‘The younger generations feel the need to be a part of something larger – a spiritual belief, perhaps, or a movement to improve the environment, or social justice. They want to be a part of answering life’s big questions’

 


In what direction is the political wind is blowing right now and which party stands a good chance to win the next general election? What are the factors that will determine voter intentions at the forthcoming general elections as well as the impact of social media on the electorate? Mauritius Times spoke to Dirish K. Noonaram and Venna Pavaday, the founding partners of VERDE, which despite being a relatively new research and data consulting company correctly predicted as part of its political research survey the outcome of the Belle Rose-Quatre Bornes by-election in December 2017, for an insight into how things may work out.

Dirish K. Noonaram’ specialty is custom, creative, and complex research. Prior to co-founding VERDE, he worked in the financial services sector and consulted for boutique frontier market asset management firms. He holds an MSc (Distinction) in Operational Research from the University of Kent, UK, and a BSc (Hons) in Mathematics from the University of Mauritius. Venna Pavaday is an expert in custom market research methodologies and information design. She has extensive experience in quantitative methods having worked in the financial services sector in Mauritius and in the UK for major institutions which include the Union Bank of Switzerland and Schroders. She holds a BSc (Hons) in Mathematics, Operational Research, Statistics and Economics from the University of Warwick, UK.


Mauritius Times: The burden of responsibility which rests on the shoulders of a pollster trying to figure out electoral trends and voter intentions on the eve of elections or earlier must be quite a heavy load to carry even after elections results are out. How does VERDE cope with that?

It is important to first put into context why VERDE decided to set foot into a territory where its peers in the market have previously gone terribly wrong. It was a choice that was in itself intricate and risky, but a risk we were willing to take. As a relatively new research and data consulting company at the time (now 4 years since we have been in the market), we did not have a large portfolio of clients and a number of organisations requested track record and credentials. Whilst we do understand clients’ point of view, we decided to invest resources on pre-election polling to nationally demonstrate the value of our methods.

What made us tread on this path was the confidence that we could apply the right techniques to get it right… and the rest is history. The battered and bruised in our market were definitely not anticipating us to get the results ‘bang on’.

* But the risks of getting it all wrong can be very high in this profession. Brexit was wrongly predicted, as were the Delhi state elections in early 2015 and the Bihar state elections in late 2015. The Scottish referendum for separation from United Kingdom and the 2015 British polls were predicted wrongly too. All those polls were conducted by professional and credible pollsters with the appropriate scientific/statistical tools at their disposal, yet they were grossly mistaken. Is there a simple explanation for those mistakes?

Polling is more than just asking a few questions. More importantly, political surveys are complex and respondents will not necessarily divulge their true voting intentions. Mistakes which are often cited include:

  • Not adequately representing the voters
  • Not having the required distribution of the voting population
  • Asking questions about voting intentions in ways that induces specific answers from respondents
  • Adjusting samples to make them representative in terms of age, gender and social class

We researched a number of political polls and why they went wrong before developing our own methodology, which regrouped a number of key elements to better understand people’s views. Our appreciation was that when people answer our political poll, they express an attitude. It is when they mark a ballot paper that they make a choice.

The way questions are asked largely influences results as well. The practice of polling itself should be such that it builds better trust with respondents.

* VERDE got it right as regards the December 2017 by-elections which saw the election of Arvin Boolell, ahead of the MMM candidate Nita Juddoo, and the Reform Party’s Roshi Bhadain, etc. How and why did you get it right?

We applied an attitudinal survey rather than a traditional election poll, stemming from our in-house methodology and testing. We spend a lot of time developing principles and approaches that are in tune with the local market, rather than use methodologies prescribed by international research firms, which very often are not aligned to the specifics of our market.

Our data collection process is also very strict and we have a rigid framework for verification of our raw data. Moreover, our background in quantitative methods gives us an edge in undertaking analytics and modelling on our data sets, as compared to traditional yes/no answers, which remain very limited as a predictor of voting intentions.

* Again, as regards the election of Arvin Boolell in December 2017, you saw it coming, it would appear, as from October with 33% voter intentions and climbing to 35% in November. That was a by-election, and the stakes thereof could not be higher as they would be for a general election. Do you have the feeling that the same trend could nevertheless be expected next time round in the context of a general election?

The dynamics of a general election are completely different to a by-election. The stakes are different and the voter intentions are impacted by a larger number of variables.

Multiple demographic models would have to be used to have a good indicator of trends at the national level. Aspects which result in erroneous data may include, for example, the assumption that turnout would be the same as the previous election, discarding bundling effects when it comes to candidates, or wrongly assuming that the sample should be representative of the population rather than the voting population.

At this stage, we are of the view that the demographics and voter attitudes today are different to those of 2015 and political parties seem to underestimate those differences.

* How open (and truthful) are Mauritians about their electoral preferences? Do you see a marked difference between what obtains in different geographical locations and amongst different sections of the population?

The majority of Mauritians would not openly disclose their electoral preferences. As a pollster, we work around those intricacies and yet manage to extract information in a way which guides us to the voter’s intentions. But definitely such intentions are quite diverse in different locations and amongst different segments of the population.

The variables which are taken into account are diverse: region, age, gender, level of education, family life… and the list goes on. Each segment and even sub-segment of the population behaves in a different way, has different views and has a different decision-making process.

An electoral survey is one of the most complex surveys and so many pollsters get it wrong because they do not account for all these variables and their net effect on voting intentions. For example, if we take profile A – a person who always votes for the same party, irrespective of the candidate and profile B – a person who bases his final decision on the latest information at hand, any number of surveys will not provide an accurate prediction of the profile B segment.

What I mean to say is that it is not as simple as breaking down the population into segments and sub-segments. Each voter has his own characteristics and this has to be taken into account when the field work is undertaken. The ultimate objective is to regroup the population into personality segments and then undertake analysis on this.

* But what if you were told that most of the people we speak to – all of whom unhesitatingly confide to being long-time supporters of X or Y Party – are at a loss as to the direction in which the political wind is blowing right now and whether their own Party stands a good chance to win the next general election? What does this mean to a professional pollster? Is there more to it than the simple explanation that it’s early days to come up with a credible prediction?

We have been gauging the political wind through snap surveys and this is something which often came up in the results. As you rightly pointed out, there is today this general feeling among voters that there is uncertainty in the direction and philosophy of some political parties. As a pollster this remains a key data point and can be applied in numerous models to assess voter intentions.

Whilst it is early days, we feel that the population would prefer that the main political parties go alone for the next elections, but based on previous campaigns, there is a feeling that, at the end, pre-election coalitions are inevitable. This uncertainty clearly plays in the minds of voters and a credible prediction is therefore difficult but not impossible.

To undertake such a prediction will require time and resources, and stringent data collection and validation mechanisms. The methodology for such pre-electoral polls will also depend on the timing as models to be used shall be developed to fit the context. For instance, undertaking a poll now when there is no certainty on possible alliances will be different to a poll undertaken, say two weeks before the elections.

* Conventional wisdom has it however that elections, especially when it comes to general elections, are decided in the preceding week or even on the very eve of polling day. It might be different in urban constituencies compared to rural constituencies. Does the political geography with the dichotomy that still exists between the urban and the rural contexts make it more risky to undertake a polling exercise in the Mauritian context?

Voters are today more informed and have access to information through a number of channels. For example, the radio today has an estimated reach of close to 830,000 people and social media some 715,000 people. TV preferences are seen to be on a decreasing trend as we move towards younger generations, from 80.6% of Baby boomers (Born 1946 – 1964) and older generations preferring TV, to only 24.6% from Generation Z (Born: 1995-2012).

Accessibility to those media channels have made the rural and urban voter the more data and information savvy, and better positioned to make a choice on political preferences. This, we believe, played an important role in Alliance Lepep’s win back in 2015, as people gathered more and more information through all the channels available and finally made a decision.

In essence, the choice of voters is delayed as the information flow is today widely available and they make up their minds once all the data is available to them, at the end.

Nonetheless, there has always been this perception in Mauritius that voters make up their minds at the last minute. This is questionable and leaves us to wonder whether they changed their minds at the last minute, or the population was made to believe this through wrong dissemination of information.

 * What else makes political pollstering a risky business in Mauritius?

When we predicted that a Labour Party candidate would win the by-elections, a number of people affiliated us with politics and asserted that we had been paid by the party to publish results in their favour. Slurring is sadly still very common in Mauritius, but we remain unfaltered and a professional services provider.

We do not however dispute the fact that publishing political survey results has an impact on the final votes being cast, as is common knowledge. This may take different forms:

  • Impact 1: Influence voters to support the front-runner
  • Impact 2: The underdog effect which rallies voters to support the candidate who is trailing in the survey.
  • Impact 3: There is the motivating effect, which encourages voters who did not intend to vote in doing so.
  • Impact 4: There is the de-motivating effect where voters, especially supporters of front-runners, decide to no longer vote because their candidate enjoys a huge lead.
  • Impact 5: Fifth, there is strategic voting, where “voting is influenced by the chances of winning”.

The influencing of results on final voter intentions is also catered for to some extent in our surveys. We are able to discount this effect due to the numerous waves which we undertake to provide more accuracy. A one-off national survey provides a snapshot of voting intentions at one specific point in time. This is useless when we look at the local context and how volatile the population is in terms of his preferences. We undertake what we call longitudinal surveys which instead gather data on an ongoing basis and discard any peaks or troughs which may occur due to specific events.

* When we were speaking about electioneering in rural Mauritius and the dichotomy that still exists with the urban areas or could be gradually fading, we had in mind the influence of the so-called “reunions privées”, which bring together a select audience to listen to the message and “mot d’ordre”, of election candidates. The same could be happening in urban areas. Are there indications from what you saw taking place in No. 18 that the people are moving away from those traditional platforms and gradually making up their minds independently of political propaganda?

Traditional platforms remain inevitable in the local context. However, their propensity to influence people is significantly reduced as compared to some 10 years back. In a society where information is available, people tend to make their own choices rather than abide by a ‘mot d’ordre’.

If you consider the table below (data from VERDE Media Guide 2019), you will see that media channels play an important role across all age segments.

Possibly this time around, political parties which are more strategic in their communications approach and target the right segments across the proper channels are likely to have a more direct impact on the voter. This may have a more widespread impact than ‘reunion privées’ which platform is nonetheless ingrained and serves as a reason for voters to see their future leaders in person. Once elected, it is then quite difficult to see them in person, as we all know!!

* What does your experience as pollsters inform you about the abstention rate at elections which has been on the rise over the years?

All around the world, we are witnessing a fall in voter turn-out, not only in Mauritius. A number of factors have caused this, namely (1) an increase in the number of youngsters who do not vote due to disparate views from politicians and who disengage themselves from any political happenings, (2) an overall decrease in confidence in the democratic process, (3) the logistical hurdles of everyday life which limit the amount of time that people have to do other things today, and (4) the belief that their votes would not make a difference, especially when it comes to casting a vote for a smaller party.

Extensive research has been undertaken across several countries to find solutions to increase voter turn-out. Methods which have proven successful are social pressure, canvassing and texting.

Social pressure may relate to a mailing to inform recipients that their participation in the election is a matter of public record, and that it will become known to everyone whether they voted or not. This increased turnout in Michigan by 8% in the 2006 elections.

Campaigners are sent off to work on how to increase voter turnout. Canvassing through in-person contact and through text can be effective ways to drive people to the polls on election day.

* It was being said at one time, not so long ago, that Facebookers do not make a majority. But given the increasing penetration of social media in Mauritius, would you say that it would be unwise for political parties to discount their influence on the choices that electors will make in the polling booth?

Social media users are quite prominent today and what is worth considering is the time which people spend on those platforms. The diagram below provides an indication of time spent across social media platforms, which has been rebased on a scale of 0 to 100.

We observed from this study that the majority of people use social media platforms at night (68.2%). This is followed by ‘During the weekend’, with a high score of 60.6%.

When broken down across time of day, 18.4% of respondents state that they use Social Media platforms all the time. Political parties cannot undermine social media as a channel to reach people and to pass their messages across.

We have seen during the US elections how the only pollsters who got the results right were the ones who undertook social media listening in addition to normal surveying. This is because people are more and more engaged on social media channels today, where they feel free to express their opinions and engage with other people. As such, these platforms have become important in understanding people’s behaviour and attitude. We today also couple our on-the-ground research with social media analytics.

* But with or without Facebook, elections in Mauritius have since decades been largely determined by considerations based on community, caste, provincial group, etc. Do you see that changing any time soon? What do electoral trends indicate?

The snap surveys which we do mostly to understand behaviours and attitudes infer that ethnicity remains an important consideration by voters even today. This is expected to change, although not in the near future. Younger generations are less inclined to adhere to a specific community or caste. We tread on a fine line here, but nonetheless believe that when baby boomers and the generation x (Born: 1966-1976) have been replaced by the younger generations, we will start to see a change.

Based on regions being analysed, we sometimes see clear ethnic preferences and we sometimes find no correlation between ethnicity/caste of the voter and his chosen candidates. If we look at the younger generations, even today, we see a decrease in their sense of ethnic and religious belonging. Rather, they feel the need to be a part of something larger – a spiritual belief, perhaps, or a movement to improve the environment, or social justice. They want to be a part of answering life’s big questions.

* Besides political polling, we understand that VERDE also conducts surveys on brands, on consumption patterns as well as the media consumption of the population. If you were asked to draw up the profile of the Mauritian millennial of today and that of tomorrow, say 15-20 years from now, what do you think you will you come up with?

We are now moving towards an age of micro-generations. For example, we today speak of Xennials, a micro-generation with people born towards the end of the Generation X spectrum and at the start of the millennial spectrum.

When we look at the typical millennial (born between say 1986 and 1996), we see connected people, always ready to voice out their opinions and give feedback, people who prioritise access to things rather than ownership, people who put off marriage and having kids and live with their parents for longer.

In 15 – 20 years, this millennial segment, according to us, will:

  • Have more time to himself. He will work less as a lot of his current tasks will be delegated to technologies.
  • Telecommute and collaborate remotely, which will result in more productivity and work/life balance.
  • Move closer to their workplace, as they value speed and efficiency. We will see a growth in urbanisation.
  • Use shared services like carpooling to make their lives easier and less stressful.

* 2014 had been mostly a protest vote election, like many of the earlier elections since Independence. If we go by the result in December 2017, that trend should persist. What’s your take on that?

The situations which may arise in our view are the following: (1) protest vote, (2) a continuity vote, (3) general vote for change. Correlating the 2017 by-election to the general elections will not be prudent as the context is totally different.

We tend to believe that voters remain undecided at this stage as they do not have enough information on the dynamics of possible alliances and what measures/reforms will be proposed by political parties in view of guiding the country into the year 2025.

More importantly, voters now emphasise significantly on their bottom line, that is what they are materially getting from each measure proposed and, at the end of the day, whether they will be better off. These will no doubt play a critical role in setting a trend.

Over time, we also tend to see a growing preference to vote for individuals based on their personality, their aims and vision, rather than their ethnic group or political party belonging. It will become less and less probable for general elections to be based on a ‘wave’ or on word of mouth, as has been the case previously. We therefore expect to see a downward trend in straight-ticket voting over time.


*All media data provided are based on proprietary VERDE data, which are included in our Media Guide 2019.


* Published in print edition on 16 August 2019

One Comment

Add a Comment

Your email address will not be published.