The persuasive power of numbers and the mystery “8.5%” figure

Having decided to write in to Bloomberg.com for including what looked like a clear underestimation of the proportion of Tamils in its news items and to highlight the sensitive nature of a country’s ethnic make up (particularly of course Sri Lanka’s), I wanted to check for myself the veracity of what I was saying. I also wanted to find out where the figure of “8.5%” that Bloomberg routinely refers to derives.

A search on Google threw up some interesting material. The mysterious figure of “8.5%” kept cropping up again and again, in very different contexts and without explanation. I was struck by how the figure was being used, seeking to link the viability of a community’s grievances with its proportion in the country’s population and to question the validity of its collective claim for greater autonomy.

Starting with the numbers, the estimate of “8.5%” came up most recently in this Bloomberg.com article:

“Any peace settlement must be based on a homeland for Tamils, who make up about 8.5 percent of the South Asian island nation’s 20 million people, the LTTE said last month”. (Sri Lanka Says It Wants to End Conflict Through Talks, Not War, 3 October, 2007, http://www.bloomberg.com)

In the following extract the figure of “8.5%” has been used to question the viability of the Tamil community’s demands for a separate state, on the basis of its perceived reduction in population size. In other words arguing that since Tamils only make up a small fraction of the country’s population anyway, should concessions really be made?

“What is more, when Britain came to that conclusion in 1948, it was estimated that the Ceylon Tamils, taken together with the Tamils of Indian origin then resident in the country, amounted to a little under 20% of the total population. Today, due to factors such as the repatriation of large numbers of Tamils of Indian origin back to India and the equally significant migration of Sri Lankan Tamils to South India, the UK, other parts of Europe, the USA, Canada, Australia, and South Africa, the proportion of this ethnic group within the total population of Sri Lanka is estimated to have declined to 8.5%. What is more, of this 8.5%, the Sri Lankan Tamils in whose name a separate state is demanded account for only 3.9% of the total population of Sri Lanka, the Tamils of Indian origin, chiefly employed in the tea plantations of Sri Lanka, representing the other 4.6%. As you can see, considerations of viability were never more pertinent.” (Extract from a signed petition to Gordon Brown, dated 15 June 2007, for the reversal of the British government’s decision to withhold aid to Sri Lanka: http://www.petitiononline.com/TDA0115/petition.html).

In the above example, note that the “8.5%” figure is referred to as the combined total of both “Sri Lankan Tamils” and “Tamils of Indian origin” and the actual figure given for Sri Lankan Tamils is “3.9%”.

The “8.5%” figure has also entered other spheres, this one relating to tourism: “Out of the 8.5 percent Tamils residing in Sri Lanka, a majority follows the Hindu religion with the rest following Christianity”. (http://www.reddottours.com/Activity/FestivalsFairs/index.php).

A Google search for “3.9% Tamils” brought up what might actually be the original source for the 8.5% figure, since articles were also referring to “U.S. government data” as their source (http://www.lankanewspapers.com/news/2007/9/19585.html). Digging a little deeper, it seems that the culprit might be the CIA Factsheet:

“Sinhalese 73.8%, Sri Lankan Moors 7.2%, Indian Tamil 4.6%, Sri Lankan Tamil 3.9%, other 0.5%, unspecified 10% (2001 census provisional data)” (Source: https://www.cia.gov).

The CIA Factsheet cites the “2001 census provisional data” as its source, but there is a curious reference to an “unspecified 10%”. So what did the 2001 census come up with?

Getting an exact figure on the number of ethnic groups in Sri Lanka (particularly the proportion of Sri Lankan Tamils) is clearly difficult because of the lack of access to certain areas of the North East. Although far from ideal, the data collected by the Department of Census and Statistics (DCS) when it conducted its Census of Population and Housing in July 2001 and its subsequent estimates would seem the most reliable data available in terms of population statistics. The census is incomplete because enumeration was only conducted in 18 out of the 25 districts (as a result of not being able to access districts in the conflict affected North East), but despite these limitations, estimates of the populations of the remaining districts of the North East were made on the basis of valid data, such as the register of births and deaths. The DCS official estimate is 11.9% for ‘Sri Lankan Tamils’ and 4.6% for ‘Indian Tamils’, out of an estimated total population of 18.8 million (Source: http://www.statistics.gov.lk).

It seems that the CIA Factsheet, despite taking its figures from the DCS census data, has taken the wrongs ones and come up with the 3.9% estimate. It took the actual numbers for Sri Lankan Tamils from the completely enumerated districts (based on a population of 16.9 million), and then calculated the percentage on the basis of the DCS estimate for the total population (18.8 million). The ‘10%’ that are deemed “unspecified” actually includes 1.5 million Sri Lankan Tamils, around 223,000 Muslims and 130,000 Sinhalese.

Unless an alternative credible source can be given, if one is going to rely upon an estimate for the number of Sri Lankan Tamils in this country, then surely the estimate of the DCS must be the starting point? After all, it is the Government’s official estimate and citing “US government data” or using data from the CIA Factsheet seems remarkable, particularly for the Sri Lankan Defence Ministry: see for example this rather confusing reference to ‘8.5%’ on the Ministry of Defence website:

“A larger percentage of Sri Lankan Tamils who in total make up around 8.5 percent of the island’s population now live outside this region sought for their separate state, as they wish to distance themselves from the self declared sole representatives of the Tamil community, the LTTE, which is responsible for killing large numbers of their own community to eliminate any and all opposing views.” (Source: British Tamils intimidated into funding final war, Ministry of Defence, http://www.defence.lk/new.asp?fname=20070209_02)

The reference to “8.5%” could be understood as either the total percentage of Sri Lankan Tamils who live outside the “region sought for their separate state”, or the total population of Sri Lankan Tamils. Interesting the ‘8.5%’ figure is used to indicate only “Sri Lankan Tamils”, whereas even taking the flawed CIA Factsheet, it is set out as a combination of the percentages of both Indian Tamils and Sri Lankan Tamils.

Some reflections

So what difference does a few percentage points make anyway? The difference between the DCS estimate of 16.5% (11.9% Sri Lankan Tamils + 4.6% Indian Tamils) and the 8.5% estimate is 8%. Taking a total estimated population in 2001 of 18.8 million, this would mean that approximately 1.5 million Tamils are excluded. Since the areas inhabited by the ‘Indian Tamils’ is covered by the census, the 1.5 million Tamils would all be ‘Sri Lankan Tamils’ or taking the language of the above petition, those “in whose name a separate state is demanded”.

We have seen how the incorrect “8.5%” estimate has been used to support political arguments on the internet and possibly elsewhere. Granted that the petition mentioned above only collected 268 signatures, but the danger is that the estimate of 8.5% and the more alarming figure of 3.9% becomes fact through its repetition and republication. This kind of data is becoming increasingly politicised. While we cannot stop those wanting to push a particular agenda, we can try to persuade those that put themselves out to be responsible news carriers to understand that this kind of information is highly sensitive and that by giving a figure without its source or including caveats as to its incompleteness, you risk allowing it to be accepted as the truth. If you are going to cite estimates, then the DCS figure is the official estimate. Until another census is carried out, which may or may not happen after another 10 years, we risk having the reference to 8.5% or 3.9% proliferate, set in stone and used to support positions like a particular community’s grievances are somehow less important because of its reduced percentage in the population.

But the numbers aside, it is dangerous to suggest that the grievances and aspirations of a collectivity are dependent on their share of the population. The link between the proportion of the population and the significance of Tamil grievance must be broken. The demand for a separate state was not based on a group’s proportion or percentage in the population. It stemmed from a complex set of issues connected with the origins of the conflict. It is interesting that percentages and proportions are now being used as a basis to challenge that demand. That is the persuasive power of numbers.

This article was published in the Peace and Conflict Timeline (PACT), the interactive timeline of the Sri Lankan conflict context. Please visit http://www.pact.lk to give your views.

  • Nishan

    Excellent, excellent, excellent. The research and the elucidation of the numbers is much appreciated.

  • veritas

    A Great article addressing an important issue. No one checks facts these days. This highlights the dangers of Googling and “copy and paste” fact finding that most people do.one would expect good journalistic values and deligent truth seeking is the norm for journalists and govt. Unfortunately someones imginary statistics is another ones truth if it is later quoted enough then it becomes a fact
    It can certainly make the problem appear smaller than that what it is in sri lanka.some have gone to war over imaginary statistics It is time we question statistics union black is right on the money-Good job. Let’s hope that this article is picked up when you google!

  • Krish

    Well researched evidence based document which is very well articulated. This also highlights the hidden agendas of some forces and dangers for a hard working highly intelligent loving community whose image is tarnished for political advantages.
    Congratulations for your intellectual contribution.

  • http://www.pact.lk unionblack

    Pleased to report that Bloomberg.com now refers to the correct 2001 Census data:

    “The government has rejected a settlement that divides the country and is offering to devolve power to some provinces. Tamils made up 11.9 percent of the population in 2001, the government’s Census and Statistics Department said. Sinhalese made up almost 74 percent of Sri Lanka’s 20 million people.” (Sri Lankan Military Kills 20 Rebels Preparing Attacks, 14 November 2007)

    http://www.bloomberg.com/apps/news?pid=20601091&sid=aGxOaDRbnFxQ&refer=india

  • feedback

    I emailed paul tighe at bloomberg to correct the 8.5%. He pointed out that they are using U.S. government figures.

    What they are not doing is leaving out what the U.S. government says about the incompleteness of the 2001 census.

    From CIA factbook:
    Sinhalese 73.8%, Sri Lankan Moors 7.2%, Indian Tamil 4.6%, Sri Lankan Tamil 3.9%, other 0.5%, unspecified 10% (2001 census provisional data)

    Please reply back to bloomberg and have them report the fact that the 2001 census is incomplete.

    Thanks.

  • Pingback: groundviews » The persuasive power of numbers (Part 2): “54 per cent of Tamils live outside the North and East”()

  • Pingback: The persuasive power of numbers (Part 2): “54 per cent of Tamils live outside the North and East” « Kannan Arunasalam’s blog()

  • Pingback: The persuasive power of numbers (Part 2): “54 per cent of Tamils live outside the North and East” « Kannan Arunasalam’s blog()