Making the difference – How many voters do you need?

Presiding officer Isoa Rasousou (left) supervising the queue at the Sawanikula polling station in Naitasiri in 2006. Picture: FILE/ELIKI NUKUTABU

Thinking people must be wondering why on earth the Supervisor of Elections (Mohammed Saneem) and the Elections Office are spending millions of dollars of taxpayers’ money in multiple rounds of voter registration in Fiji and abroad in Australia, New Zealand, Canada and the United States.

I had to drive across Melbourne to get myself and my wife new voter cards from some very pleasant staff members from the Elections Office. The FEO staff may yet go to Ethiopia where I believe there are a few potential Fiji voters and even one VIP.

The Supervisor of Elections has also gone through several costly exercises changing voters cards, voters’ names, and even the names of women on birth certificates to match their married names on voters’ cards.

Recently they even announced that voters will not need their voter cards to vote, since their photos will be on the voter rolls. Wow. Some official will decide that the face of the voter in front of them is that of the person whose photo is on the voter rolls?

Any sensible personmust be thinking and asking: will it make any difference to the election result in 2022 if 99 per cent of eligible voters vote instead of 70 per cent of voters? Sensible people can even ask themselves: take every election that has taken place in Fiji’s history.

Would it have made any difference whatsoever to the result if there had been another 5 per cent of eligible voters voting in addition to the percentage (70 per cent or 80 per cent) who actually voted? The answer would have to be no. The percentage of votes for the various major or even minor parties would remain roughly the same.

Any small changes would not have made any difference to the result. So why on earth are the Supervisor of Elections and Elections Office going through all this expensive rigmarole when the Government cannot be bothered about democracy for the Fiji National Provident Fund? I will have a guess at the end of this article.

But first, some brutal statistics, which the opinion poll companies and the Fiji Bureau of Statistics statisticians know very well, that ought to horrify the cost-conscious voters of Fiji. What opinion pollers know Let us just focus on one question: which political party do you support?

Let us assume that the person interviewed is not afraid to give the interviewer an answer to this question (although that cannot be taken for granted). There are usually only about 10 possible answers depending on how many political parties are standing (we can forget about silly independents now).

The major challenge for the polling company is: how many persons should be interviewed to give a reliable result for the election outcome?

To try and predict the Fiji national election result, the polling company knows that it must ensure that its sample comprises all the important characteristics of the voting population, which might have a bearing on the political party they represent.

Once upon a time, before the 1986 elections, virtually the only important characteristic was ethnicity. Indigenous Fijians voted one way (Alliance), Indo-Fijians voted another way (NFP).

Then class interests (workers’ party) and indigenous Fijian groupings in Fiji (West, Vanua Levu, etc.) also became factors. Age (young, middle ages, elderly) and gender (males, females) could be important.

What sample size for opinion polls?

So assume that the sample of people interviewed covers all the different regions of Fiji, the ethnic groups, the age groups, the class, the gender variables.

How many people should the polling company interview? I believe a good opinion poll company like Tebbutt used to interview just over 1020 persons or potential voters.

Given that there are about 600,000 potential voters, this is a mere 0.2 (that’s right, zero point 2) per cent.

A 1 per cent sample would require a pollster to interview six times as many persons, or 6000 persons.

Yet that tiny 0.2 per cent sample was good enough to give the Fiji public a fairly good idea of the support that the major political parties could expect from the voters of Fiji.

So here is the question for policymakers who worry about the costs of registering voters and running the actual elections: How many voters should a good opinion poll company interview, in a random sample obtained throughout Fiji where all the votes are going to be cast, to find out which party should be in Parliament and in what numbers (given that we have a roughly proportional electoral system?

The good polling company would stake its reputations on less than 3000 persons – not the 600,000 that the Elections Office is registering (nor the 400,000 who will be voting).


In case the Fiji public mistrust opinion polls, especially where they are associated with media companies who may be biased towards one political party or another, I suggest that they look at the sample size that the Fiji Bureau of Statistics uses to get accurate statistics on the details of income and expenditure habits of all 200,000 households in Fiji.

The FBS goes through an incredibly elaborate exercise where they use detailed census data on the location of all households in Fiji to obtain a random 3 per cent sample for the Household Income and Expenditure Surveys (HIES) or their Employment and Unemployment Surveys (EUS).

I kid you not – that even I was astonished that a mere 5000 households, if randomly selected out of 200,000 households, could give such incredibly accurate statistics on all of Fiji’s households: detailed breakdowns of incomes and expenditure, by area, ethnicity, age, gender etc.

For instance – what was being consumed of dalo, cassava, kumala, yams, flour products, different meats, etc. For instance, what kinds of incomes they received and how much.

One can even estimate (as I have done to within an error of 1 per cent) how much FNPF should be receiving annually in contributions, and even exactly how much was it missing out on.

The accurate statistics that a mere 2 per cent sample of households could provide on all Fiji households is truly astonishing.

One does not need 400,000 voters to tell us who should form government in this country. Even 10,000 voters, randomly selected, would be enough.

Registering another 20,000 voters at great expense, is a waste of time and money.

So why does the SOE do it?

The point I am making in the two examples above is that going from 400,000 votes to 500,000 votes or 600,000 votes is not going to make any difference whatsoever to indicating the extent of political support for the major or even the minor parties.

Nor will it change how many MPs any party should have in Parliament given a proportional electoral system.

All the current exercises in registering voters in Fiji and overseas are a waste of time, and will make no difference to the outcome in the 2022 elections.

Remember that 10 per cent of all potential voters would amount to 60,000 voters. So the opposition parties could ask Mr Saneem, how many extra voters have you actually registered with all these exercises?

I suspect that the answer may be another 10,000 voters at most. So why is the Supervisor of Elections doing it, if registering another 5 per cent of potential voters to go from 70 per cent voter turnout to 75 per cent voter turnout will make no difference virtually to the percentage of votes received by each of the major or even minor parties?

Of course, the answer would be: we are strengthening democracy in Fiji so that all voters can have a say in who runs Fiji over the next four years.

How can you deny that? Of course you cannot, until you extend that question to the most important financial institution in Fiji, the Fiji National Provident Fund.

Democracy in FNPF?

The Fiji National Provident Fund has more financial assets than all the commercial banks put together.

t is the most important institution controlled by the Government in Fiji (although it is supposed to be independent). It has loans to numerous public and commercial enterprises.

It is responsible for the retirement funds of all the working people of Fiji.

For decades now, union leaders and representatives of workers (like Felix Anthony) have been calling for democratic representation of the FNPF contributors on the board of FNPF.

The Bainimarama Government has steadfastly refused. It has been many years since there was a single worker representative on the FNPF Board.

No democracy for Fiji’s workers retirement funds. Instead, the Minister for Economy (and lots more) has appointed all the Board members.

They do not answer to the workers of Fiji, whose money it is. Given that FNPF knows the names and addresses of all the FNPF contributors, why has FNPF not organised an election to elect half of the FNPF Board? Because the Bainimarama Government does not wish to do so.

Yet the Bainimarama Government is quite happy for the Supervisor of Elections to use millions of taxpayers’ money to register potential voters all over Fiji and abroad, whose votes will make no significant difference to the 2022 Elections outcome. What next?

• Prof WADAN NARSEY is an Adjunct Professor at James Cook University and a former Professor of Economics at the University of the South Pacific where he worked for more than 40 years. The views expressed are the author’s and not necessarily of this newspaper.

More Stories