Skip to content

“Collecting or collating? Not the real issue” (7 April 2018)


Collecting or Collating: not the real issue (7/4/2018)

These last three weeks, there has been heated discussion in Parliament about what the Minister for Economy (and the Fiji Bureau of Statistics) actually said in parliament about the availability of census data on ethnicity.

Did he say that the FBS does not “collect” data by ethnicity (as was reported in the media) or did he say, as he asserts, that FBS does not “collate” data by ethnicity?

Note that “collect” ethnicity data means that the FBS has recorded the “ethnicity” of each person counted in the 2017 census.

On the other hand “collate” the ethnicity data means that the FBS puts together and presents tables of statistics disaggregated by ethnicity, and perhaps also by age, province, male/female, rural/urban, mortality, etc.

No doubt Hansard will eventually come up with some “record” of what was said by Khaiyum and which politicians will no doubt debate till the cows come home.

What matters is what the Government Statistician (Epeli Waqavonovono) said to the public.

What really matters for the Fiji public is that the 2017 Census Reports must also include tables disaggregated by ethnicity, as they have always done for more than a hundred years.

What matters is that Government Ministers must not interfere with the work and output of the FBS.

What the Government Statistician said

When a political leader criticized the non-reporting of the 2017 Census data by ethnicity, the GS was reported as saying: “On ethnicity information, we are guided by Chapter 1, Section 5 of the Fiji Constitution that states that all citizens of Fiji shall be known as Fijians.”

The Government Statistician also said “We did ask about people’s ethnicity, but the decision to release it has to come from higher authority. Once we are instructed then we will release it.” Clearly indicating that such statistics were available but merely not reported to the public.

Minister Khaiyum on the other hand, said initially that there was no need for information by ethnicity.

Then when pressed further in Parliament, Khaiyum said that the Fiji Bureau of Statistics would issue a statement.

The FBS website then duly reported: “During our review of the data and methods of collection, we uncovered a number of anomalies in the way data on ethnicity were collected that severely compromised the objectivity and completeness of the data set on ethnicity… It was found that many enumerators failed to verbally ask for the respondent’s ethnic background…  Instead, the data was collected based on the enumerator’s observed assumption of the respondent’s ethnicity…  Many enumerators also assumed the entire household’s ethnicity based on one household member, failing to account for households with multiple ethnicities…..  many respondents refused to disclose their ethnic background.”

Unfortunately, this statement muddied the water further, while casting grave doubts on the professionalism of the FBS technical staff (including the GS himself), the soundness of their preparatory processes, and doubt on the quality of their statistics in general.

Field testing and normal errors

All the supposed weaknesses in the ethnicity data that the Government Statistician alludes to have been present at every census.

They have never prevented the FBS from still producing reasonably accurate estimates of the Fiji population by ethnicity, as they have done for more than a hundred years.

The ordinary public would not know, but before any census and even sample surveys of 5% of all households in Fiji are conducted, the FBS does “field tests” of the questionnaires (getting an even smaller sample of households to respond to the questionnaires) and the results analyzed.

Any weaknesses found are then used to iron out the questionnaire and the interviewers method.

Given the public releases that were given out before and during the census, this last census was supposed to have been the most technologically advanced.

Interviewers were supposed to be working with laptops and tablets in the field, sending information back to FBS head quarters which would have even then at that late stage picked up any weaknesses and corrected for them.

I personally doubt that the weaknesses were sufficiently significant to warrant not issuing data by ethnicity, especially when one notes that the census is a hundred percent (100%) sample of the entire country, and small weaknesses are not enough to render the results “inaccurate”.

FBS data usually good

Over the years, the FBS has produced wonderful and remarkably accurate national estimates of all kinds of variables, using a mere 5% sample of households.

This ability to perceive the “truth” from only a 5% a sample is something which only raw data technical analyzers can appreciate, as I have done over the last fifteen years.

I have personally analyzed four household sample surveys’ data on employment and unemployment and incomes and expenditure and produced voluminous reports used by government, and civil society stakeholders.

Given that the 2017 census (and every census) is a 100 percent coverage of all the households in the entire population (not just 5%), even a small proportion of errors will not make the census data on ethnicity unusable by social scientists, health professionals, demographers and economists.

For the “explanation” by the Government Statistician that the data was too poor to be at all accurate there would have to be an enormous amount of errors in the ethnicity data, which I personally feel is quite unlikely.

If it were true, it would be an unfair and incorrect indictment of the hundreds of trained, dedicated and efficient FBS staff who were involved in the 2017 Census and many also in the 2007 Census, which never reported such weaknesses.

It would also be an unfair self-indictment of the conscientious, dedicated and efficient Government Statistician (and Census Commissioner) who I have personally worked with for more than a decade.

Collecting and collating

The ordinary member of the public may not understand, but obtaining census tables by “ethnicity” requires no extra work at all by the Fiji Bureau of Statistics.

In all their software programs, it would required merely substituting the variable “ethnicity” for “male/female” or “age” or “rural/urban” or any other variable that all individuals are labelled with and the computer would generate these ethnicity tables, in the blink of an eye.

So there is no reason why the FBS cannot generate reasonably accurate census data aggregated by ethnicity, as they have done for a hundred years.

Is there a need for such data?

Need for data by ethnicity

Of course, a few politicians might want to obtain ethnicity data on voters, however little use it is going to be to them, apart from knowing how many young voters there will be  in Fiji by ethnicity.

But just as gender, age, rural/urban location, and even specific geographical locations are useful for development policy in the fields of demography, labor force, health, education,  housing, aging, insurance, so also is it useful to know by ethnicity:

* how many children do households have

* the ages structure of the population and population projections

* their infant and adult mortalities, and life expectancies

* their housing needs

* their retirement needs

* their religions

* their water supply

* their school attendance

* their educational attainment

* their extended family structure

* their province of residence and origin

* their rates of rural urban drift.

All this is obvious if you just read the previous 2007 Census Reports.

Media, please ask the experts

The public deserve honest unbiased opinions on the matter from our experts and it is the duty of the media journalists in television and radio to seek these experts out for their professional opinion.

Sadly, with the exception of the Fiji Times, the rest of Fiji’s media are not making any great effort to educate the public through informed debates and interviews, as they used to a decade ago.

While some politicians may be seeking census breakdowns by ethnicity merely because of its perceived importance in the coming elections (ho hum), there are many experts in government, universities and NGOs who fully comprehend the development policy need for census data by ethnicity.

Dr. Neelesh Gounder (Senior Lecturer at USP) has very responsibly given his views on the matter.

Please, media, interview other experts such as: Dr. Kesaia Seniloli, Dr. Akanisi Kedrayate, Dr. Martin Bakker and Dr. Isimeli Tukana, and nudge them out of their understandable reticence.

The autonomy of the FBS

It is unfortunate that this Ministerial interference with the normal work of the Fiji Bureau of Statistics also raises another legitimate question: what other statistics is the Bainimarama Government influencing because of their political objectives?

I can certainly say that the FBS is no longer producing data on ethnicity data on poverty using their sample surveys on household income and expenditure.

What does the Bainimarama Government fear from the facts? That perhaps the facts will show that indigenous Fijians are not doing as well economically as other ethnic groups? (I suspect this to be the case).

But also, is the national GDP data being “massaged”? I suspect now that this may also be true, given that the FBS statistics are themselves showing that despite the seven years of alleged positive GDP growth, formal sector employment is barely changing as is clearly from the latest Employment and Unemployment Survey.

What other statistics is the FBS sitting on not collecting or not collating, because the Bainimarama Government does not wish for those facts to be known?

It is scandalous that there is a virtual silence on these vital issues from the intelligent members of the public, immersed in their mindless focus on rugby sevens and other trivia, all happily encouraged by the media, both government owned and controlled and independent.

Fiji is well and truly a banana republic.






Comments are closed.

%d bloggers like this: