Skip to content

“Are there really 53,400 more voter registrations than voter age persons according to the 2017 Census?” (9 Nov. 2018)


“Are there really 53,400 more voter registrations than voter age persons according to the 2017 Census?” (9 Nov. 2018)

Last week, ABC Radio ran a news item in which Dr. Neelesh Gounder pointed out that the number of voters registered by the Fiji Elections Office appeared to be some 30 thousand more than the number of people of voting age recorded by the 2017 Census conducted by the Fiji Bureau of Statistics, a strange discrepancy I had pointed out in a January FT article (the actual discrepancy in 2017 was about 39,400).

The ABC suggested that these extra 30 thousand voters could influence the outcome of the 2018 Elections, although it was not explained how.

In the ABC interview, Dr. Gounder offered two possible explanations, the first being that censuses are never 100% accurate  and that there may have been an under-reporting of population numbers (I say possible, but not likely to that extent).

His second possible reason was that some people registered to vote might have registered again with a different address in order to collect cyclone assistance grants from government (also possible but I doubt this explanation).

In any case, Gounder suggested that the FEO database “was not able to delete the duplicate names” (also possible).

[One possible factor not considered is rural:urban drift over the last ten years with movers being registered in both their original place of residence and the new place of residence- read below].

But Mohammed Saneem (Supervisor of Elections) informed ABC that the FEO recorded each voter’s “biometrics” as well as “iris” (eye?) measurements; and that they cleaned up their database regularly, including eliminating records of deceased people (the latter is probably correct to some extent).

But then Saneem also cast doubt on the reliability of the census data stating that people needed to understand “what census data was collected for”.

So far so good: an academic responsibly pointing out possible voter registration discrepancies and a Supervisor of Elections giving his “explanation” for the apparent discrepancy.

Of course, if asked, the FBS could truthfully state that one reason “why census data was collected” was to know as accurately as possible how many people are in Fiji on Census night and their personal details including age (plus many other variables that demographers and social policy experts are interested in).

Hence the FBS census data ought to estimate fairly accurately how many persons there were of voting age (18 and over) on census night, within errors of less than 1%.

I suspect that they have estimates of these 2018 numbers already.

But then the SOE went further.

SOE demands an apology

Instead of welcoming a comment from a respectable academic, the Supervisor of Elections  sternly notified  Dr. Gounder:  “We have noted … your allegation that there are close to 30,000 duplicates on the Voter Roll… when people changed addresses… The FEO would like you to furnish us with evidence of your allegations above before the end of the week” [and failing that] “a proper statement is issued by you withdrawing your statements and admitting them as falsified”.

But Gounder had merely said that voter registration duplicates might be one of the explanations, and even admitted that there might have been some undercounting by the Census, the very point that Saneem was strangely implying.

But Saneem also alleged on Fiji Broadcasting Corporation that “freedom of speech is being used as a curtain to make wild unsubstantiated claims.”

Did Dr. Gounder really make “Wild unsubstantiated claims”?  I don’t think so.

In any case, Dr. Gounder promptly sought legal advice from a reputable firm of lawyers.

 But the responsible public should be asking: surely Dr. Gounder was doing a service to the public by raising questions about the accuracy of voter registrations?

The responsible public should be asking, is it in the public interest that a professional academic, purely in the pursuit of his totally unpaid service to the community, be made so insecure by a tax-payer funded Supervisor of Elections that he feels the need to obtain legal advice?

Of course, Gounder would know like many others that the SOE can have and does have discretionary recourse to the powerful prosecuting authority of FICAC (whose ire was recently felt also felt by academic and former Vice Chancellor of Fiji National University, Dr. Ganesh Chand ).

But more importantly, I suggest that Dr. Gounder was not making “wild unsubstantiated claims”.

I present data here to suggest that the discrepancy between registered voters and people of voting age was probably far more than 30,000 (a number I myself had suggested in a FT article in January) probably more than 50,000 in aggregate.

The strange voter registration and population numbers

The Fiji Elections Office should be commended for putting on their website, detailed voter registrations as at 1 October 2018, not only in aggregate, but also by age groups and by divisions.

This is an extremely rich database which researchers should find useful for all kinds of reasons.

The 2017 Census of the Fiji Bureau of Statistics also gives the detailed numbers of persons in Fiji of voting age on 2017 Census night, also by division (which I do not give here), which I have aggregated into the same age groups as the FEO website.

Table 1 here compares the two sources in aggregate (take my word that the division disaggregates paint exactly the same picture).

RegistrationsFirst of all, note that excess registrations in total amount to some 57,195 far more than the 30 thousand I roughly estimated in January.

The voter registrations are higher at all the age groups, except for the strange result that at the 18 to 20 age group, the FEO appears to have 13,456 fewer voters than were of that age group in the 2017 Census.

Of course, the 2017 Census numbers need to be projected to 2018 and I have done so using my demographic projections from the outdated 2007 Census data.

My rough estimates suggest an overall increase of only 3,480 people of voting age between 2017 and 2018 (no doubt the FBS may have more accurate estimates).

You might ask, why is there such a small increase of voting age population from 2017 to 2018?

Well, my demographic projections indicate that the increase is small because the annual increase in indigenous Fijian voters of about 5,300 is countered by a decrease in Indo-Fijian voters of about 2,300 (due to emigration and low population increase), and with an increase of about 480 “Other” voters, giving a net increase of potential voters of a mere 3480.

So the overall discrepancy between registered voters and persons of voting age, would decline slightly, but still be a very large 53,715 or 8% of total voter registrations.

Who is more accurate?

Of course both the FBS and FEO can have inaccuracies in their databases- their staff are humans after all.

But the FBS aims for 100% coverage, and usually achieves more than 99% coverage of all households and persons in Fiji.

Their own estimates of the likely limit of under-counting is less than 1%.

All demographers in Fiji trust these FBS numbers, whatever Saneem may think- and he is no demographer, despite his apparent knowledge of “biometrics”.

Summary: Table 1 indicates that FEO voter registrations in 2018 are almost 9% more than the estimated number of persons of voting age, far more than can be expected by statistical randomness of errors.

The responsible and voting public (and the Multinational Observer Group) must therefore ask, is there any independent authority which checks the accuracy of the voter registrations vis a vis the census figures of the FBS?

Surely, the Fiji Electoral Commission has the responsibility to request that the Supervisor of Elections (Mr. Mohammed Saneem) enters into dialogue with the Fiji Bureau of Statistics (and GS Epeli Waqavonovono) to clarify why there appears to be such a large discrepancy, and whether my estimates may be in error (after all, I am also human and capable of statistical errors).

Another possible reason for over-registration?

One additional factor which can be considered by the Fiji Elections Office and the FBS is that in the last twenty years there has been a rapid rate of urbanization with large number of rural people moving to urban areas, no doubt also to different places of voter registration.

I suggest that the FEO examine whether there are significant numbers of duplicates, not just to voters dying, but due more to internal migration in Fiji, from rural to urban areas and from one Division to another, between the 2014 Elections and October 2018, with voters registered in both the previous and the current residence, and not picked up by the “cleanup” program of the FEO.

The extent of the problem could even be ascertained by a small sample survey conducted by the FBS (the FBS can explain what that entails to the SOE).

What is in the public interest?

In every society, when complex problems or statistics emerge, the media request “experts” to comment with their professional views.

Such expert opinions should be encouraged so that the public can be enlightened, even if at the end, the experts “agree to disagree”.

A reasonable Supervisor of Elections,  whose salary is paid by taxpayers, should not be arrogantly blasting such academic experts, like Dr. Neelesh Gounder, for freely giving their professional opinion.

In Fiji today, it is already difficult to get any academic or professional in the private sector (such as economists, accountants, auditors or engineers), to advise the public on matters of concern to their disciplines and professions (we all know why they might be reluctant).

A reasonable Supervisor of Elections would call on his own database experts and the FBS demographic experts to get together and come up with some consensus understanding of where the data discrepancies could be, including looking at what I have suggested above, including rural : urban drift and people registering in both places.

A responsible Supervisor of Elections and a responsible Government Statistician  would then issue clarifying statements which the public can assess on their own merit, even if they still do not agree with each other.

Is that too much to ask for?

But how can rigging occur with excessive voter registrations?

My understanding is that when a person votes, that voter’s finger is marked with indelible ink.

If such a person tried to vote twice, that would be picked up by polling station officials and observers.

But while there may be human error in the FEO not picking up duplicate voters because of rural:urban drift. Fiji’s Political Parties have every reason to worry (and the MOG should ask): can “extra votes” or “bogus voters” be “electronically inserted” into the counting process in favor of some candidate and not be picked up by innocent observers, especially  if voter registrations are indeed 53,000 more than the number of actual voters alive?

The Political Parties must demand to know, as is their lawful right, does the Fiji Electoral Commission which is tasked with ensuring that the elections are genuinely “free and fair”, have any procedure in place to ensure that such “electronic rigging” cannot and does not take place, especially when they know that the SOE, who has amply demonstrated his aggression towards Opposition parties and candidates, was ridiculously and forcefully appointed by the General Secretary of one of the political parties?

Post-script 1

The public (and the current MOG) may remember that Mohammed Saneem was unilaterally appointed as the Supervisor of Elections, against the wishes of the former Fiji Electoral Commission, by the Minister for Elections (Mr. Aiyaz Khaiyum) who also happened to be (and still is) the General Secretary of the Fiji First Party.

The current MOG ought to ask themselves, as the previous MOG obtusely did not while having a holiday in Fiji, would such a partisan appointment be accepted by the voting public in Australia or NZ?

Post script 2:   What about the 2014 Elections?

My rough estimates indicate that even for the 2014 Elections, the voter registrations may have been some 20,000 more than would be indicated by the 2017 census data projected backwards (the FBS can be asked for their backward estimates as well as their forward estimates).

Post-script 3: The strange silence of the Fiji demographers

All good demographers know that for the FBS to estimate know how many persons of voting age there were likely to be in Fiji in 2018, the FBS needs accurate population projections, which need accurate estimates of base populations (usually collected at census time), fertility rates, mortality rates, and emigration.

In Fiji, because Indo-Fijians have significantly lower fertility and significantly higher emigration rates than indigenous Fijians, their total population has been declining since 1987, but their voting numbers only began declining after 2008 (which demographers can easily explain to Mr. Saneem if he is puzzled).

Far more easy to comprehend is that both the indigenous Fijian total population and persons of voting age have been rising all along (especially after the voting age was lifted to 18).

But demographers must ask, how can the FBS do such projections if they allegedly do not have accurate ethnic numbers in the 2017 Census, as has been strangely claimed by the Bainimarama Government.

Sadly, I must ask why Fiji’s demographers (like Dr. Kesaia Seniloli and Dr. Martin Bakker) and health specialists have not pointed out to the Bainimarama Government that statistics on ethnicity must be collected in order to monitor a large number of essential MDGs,  including that for the indigenous and First People of Fiji.

To refuse to count the number of indigenous Fijians in Fiji is a first step in denying their existence as a distinct culture and people. How outrageous.

Just as Australian Aboriginals only began to be counted as an identity in Australian censuses in 1967.

Post-script 3: the Fijian language (there is no such thing as the iTaukei language)

How many people wondered, like me, that while Prime Minister Bainimarama (and Khaiyum) have decreed that the indigenous Fijian language must not be used in the Fiji Parliament, Bainimarama speaks to his military troops only in the Fijian language.

How many people wondered, like  me, that for the 2018 Elections, Bainimarama only debated with Opposition Leader Rabuka on the Fijian language government-owned television and radio station (FBC), in the Fijian language, not in English.

Post-script 4    The MOG

It is to be hoped that members of the MOG, following the elections, do not  venture their innocuous and superficial opinions on the elections in academic publications, as did one of the NZ representatives for the 2014 MOG, Leonard Chan.


Comments are closed.

%d bloggers like this: