<B>A true representation of the present political climate</B>

The opinion poll conducted by Synthèses-l?express on political parties needs to be taken seriously by all concerned. Both this opinion poll and the recent by-election in constituency nº 7 (Piton/Rivière-du-Rempart) call for comments in order to establish (1) the values of our political parties in the public eye; (2) the truth behind the success of the Labour party at this by-election.

Notwithstanding that lots of developments in infrastructure had been done around the whole island, the electors in constituency nº 7 took a different view when casting their votes simply because they felt that their MPs (not the government) had let them down by neglecting the bread-and-butter tasks and priorities to solve the problems of rising unemployment, increasing homelessness and exclusion that affect their living standards. In return the electors blamed the government, which lost the by-election despite its enthusiasm to go ahead with progress as set out in the electoral manifesto.

Most people thought that the government is not at fault as they are fully conscious of the massive developments taking place in the island and outer islands. So when one examines meticulously the political situation, it is clear that the facts and figures in the Syntheses-l?express opinion poll are not misleading at all. This by-election was, according to some political analysts, an isolated case and should not be regarded as a signal or forecast of defeat for the government in the next general elections. There should be no room for complacency on the part of the government either, as one day is long enough for anything to happen in politics.

According to the Labour party, the government has failed in its duties vis-à-vis the population and is constantly losing ground. Whether this is hypothesis or fact remains to be seen. The MSM-MMM alliance lives to fight another day. On the other hand, if ever one wants to make a constructive criticism on the values of our political parties, it should be said that the Labour party is gradually getting back its popularity. This increasing popularity is endorsed by the Synthèses-l?express opinion poll. The questions to be asked are: In what way and by what means is the Labour party gradually getting back its popularity? Was this Labour victory at nº 7 won by fair means or devious emotional means to the extent where communalism was used? If the election was won by fair means, it is an achievement. On the other hand, if it was won by devious means, then the Labour party would appear strategically a threat to its main political counterparts in future.

The MSM party is perceived by many - and it has been rightly confirmed by the opinion poll ? to have lost its popularity. As it stands today, it has no political structure. Most of its activists from central and regional committees are not bona fide MSM-oriented simply because there is a general lack of communication. There is a decline in solidarity and trust among most of its members and activists. Moreover, the quality of Pravind Jugnauth?s leadership is being watched, not to say monitored closely, by the public at large and on his performance will depend his popularity. Given the chance, young Jugnauth could in time become a good leader as he has already proved to be as minister of Agriculture.

The MMM has its die-hard supporters because this party cultivates and brainwashes them in an effort to give them an identity and a recognizance as valuable members of the MMM. The same applies to the Labour party due to its history and tradition because its supporters will stay die-hard members and will never abandon the ship. These two parties, MMM and Labour have an impact on political tradition and culture in this country. So they will always have their electorate due to their well-organized political structure. As for the MSM, Pravind Jugnauth needs to work hard to rehabilitate his party by being more involved and taking into consideration vital information from the grass-roots level of his party. Therefore, there is still time for young Jugnauth to make up for the weaknesses and losses.

Only the three main political parties, namely the MSM, MMM and Labour are often given the choice to govern the country. This choice by the electors is going to stay for quite some time before another fourth political force arises. The people are not ready for it yet.

<B>A problem of credibility</B>

While it is probably futile to hope that the political class will ever be satisfied with their popularity level, Syntheses and its client would be misguided to shrug off the credibility deficit of the recent ?Baromètre Politique? as mere political noise. Beyond the political fray, a genuine inferential critique of this poll can indeed be formulated.

The first issue stems from the ?middleman? problem. Unlike most international polls of this nature (Gallup, Ipsos, Mori), this one was not released by the polling company itself but intermediated by its client (in this case a newspaper) which means that the presentation, argumentation and interpretation of results was left mostly in the hands of non-specialists. Unfortunately this meant that newsworthiness crowded out relevant information on methodology. Even the bare minimum vital bio-data of such polling exercises, i.e. the sample size, the level of confidence and the standard deviation (margin of error) has not been made public to this day. This leads to the second issue, that of ?opacity?.

Again if we refer to best practices observed by the likes of Ipsos, Gallup etc, political polls are usually accompanied by at least a website address where interested parties can obtain detailed information on methodology, actual questionnaires used etc. In the case of the Syntheses survey, this is not the case. At a time when we?re clamouring for increased transparency and a Freedom of Information Act, it is indeed regrettable that this political poll is shrouded in unanswered interrogations which can be grouped as the third main issue, i.e. ?inferential validity?.

For having spent too many hours of my actuarial training exploring probability theory, statistical testing, design of statistical experiments etc, I have a lot of sympathy for the difficulties faced by the pollsters. However, as anyone who has taken Statistics 101 will concur, numbers emerging from random sampling exercises can only be a proxy to reality if and only if certain basic principles of statistical sampling rigorously hold. The overarching, sacrosanct principle of random sampling - simple sampling, stratified (quota) sampling etc ? is the EQUAL PROBABILITY OF SELECTION of the target population. This means is that every single person in the target population (in this case some 800,000 voting adults in Mauritius) must have an equal probability of being selected as part of the sample. With only about 200,000 genuine household fixed lines to sample from (after excluding corporate and small business lines), it is debatable that this represents a satisfactory number for the equal probability of selection principle to hold. Moreover, there are no indications as to how a number of situations are treated. For example, how are no-answers treated? Are these numbers called back? This is important because if the sample only considers calls that are answered the first time, this may introduce a no-response bias.

In addition, it is worth remembering that the elimination of all possible sources of bias lies at the very core of random sampling. Hence other pertinent questions crop up, such as: Question wording? At Gallup, Ipsos and Mori, countless hours are poured over survey questionnaires to dissect whether these are worded in a strictly neutral manner. Are we assured at a 100% level of confidence that this has been the case for this particular ?barometre politique?? Is there a comfortable level of transparency and therefore trust, that Syntheses? methodology and processes for questionnaire design are aligned with international best practices?

Moral Hazard? In these polls, it is quintessential that the interviewee is convinced beyond any iota of doubt that the caller has no vital information about him; and furthermore that he fully trusts that the interviewer is in no way connected to any decision making that may impact his life. In Mauritius, if you are a Public Sector officer or an unemployed person looking for a job and someone questions you on your political affiliation, will you readily communicate your true preferences particularly if they happen to be against the ruling party? Doubtful - especially in a country where a) people have an entrenched risk aversion for voicing opinions and b) you?re lucky if the next person you talk to has at least three degrees of separation from some circle of influence. With observations suggesting that the category which most appreciate the ruling alliance are, er?. the unemployed, this should at least introduce some level of discomfort with the results.

In fact, the very title of the report (?La Surprise des Chiffres?) indicates that the discrepancy between actual results and expected results had been duly noted. Therefore it is rather amazing that the numbers have been taken at face value without the least curiosity as to why this could possibly be the case. In any scientific experiment (and statistical sampling is one of them) if results do not concord with what you would originally expect, the automatic next step would be to investigate the reasons for which this may be the case. When the gap between expected and actual results of a scientific exercise cannot be plausibly explained away, one cannot exclude the chance that the experiment may be flawed.

Lastly, trailing just two months after the government?s admission that it faced a significant perception problem and vowed to be more attentive to the management of public opinion, the ad-hoc timing of this survey has inevitably given rise to the question: Mere coincidence?

To take out the sting of this credibility deficit (a formidable enemy of trust), it would be in Syntheses? and its client?s interest ? and indeed in the interest of the nation ? if they could start by setting regular, pre-determined regular, and publicly communicated timings for these ?Baromètres?. At a time when ministers are frenziedly trying to patent new methodologies for calculating the unemployment rate or inventing new strands of feminism, it is all the more desirable that politically independent stakeholders adhere to the most rigorous professional standards and best practices so that they can engender the kind of unchallenged trust needed to oil the wheels of a vibrant democracy.

Nita Deerpalsing

London Business School Sloan Fellow (feedback: [email protected])