What’s the literacy rate of Sri Lanka?

What’s the literacy rate of Sri Lanka? I read a post by Kasun Herath over my morning coffee with the neighbourhood meta-aggregator. In the process of laboriously re-debunking the much-debunked myth that humans only use 10% of their brains, Herath makes a passionate plea to Sri Lankans to get their shit together, etc., and notes that our literacy rate of 92% should help in this endeavour. Now, this number is interesting, because it may well turn out to be as much of a myth as the idea that people only use 10% of their brains.

First, the obvious preliminaries. According to the CIA World Factbook, based on the 2001 census, is 90.7% of the total population (male 92.3%, female 89.1%) is literate. These numbers are matched by Department of Census and Statistics, UNESCO and USAID survey figures in the last decade, varying only slightly. Definitions of literacy used in these surveys are generally “age 15 and over, can read and write”, where at least one survey admits that “literacy data are generally based on self-declaration (i.e. one person, usually the head of the household, indicates whether each member of the household is literate or not)”. This isn’t just a Sri Lankan problem (see below paragraph for a very Sri Lankan problem), but a methodological flaw in how “literacy” is measured in general.

According to the Department of Census and Statistics in their special Millennium Development Goals survey, the literacy rate is 95.8% (male 94.8%, female 96.6%) in the 15-24 range. Their definition of literacy is the percentage of the population in the 15-24 range who can “both read and write fluently or read fluently and write with difficulty in everyday life”. My emphasis, obviously. The Department’s own census in 2001 measured that the 15-24 age group to be 19% of the total population, so this indicator is being restricted to the fifth of the population likeliest to be in late secondary or tertiary education (this is presumably because the relevant Millennium Development Goal is about achieving universal primary education, though I don’t see how measuring literacy at secondary/tertiary levels is supposed to help with that) and being given a lax definition of “literacy” -it doesn’t say if they got to simply declare themselves fluent or not fluent, or if there was some sort of test. I doubt, however, that they were more rigourous than all the surveys (including their own) mentioned in the previous paragraph. If you take the pass rates for the GCE Ordinary Level language exams, perhaps that would provide a better picture of how many 16-year-olds in the school system could be considered fluent in their native languages? Discounting English (since it would drive the numbers down badly), the averaged pass rate for Sinhala and Tamil in 2006 was 81.3%. This doesn’t include 16-year-olds who didn’t take the exam or aren’t in school, of course, which would drive the number down again.

Meanwhile, a presentation by Chandra Gunawardena of the Open University, 1995 (PDF) reporting on a survey of literacy in disadvantaged communities in Sri Lanka, says pretty much what I was going to say: the official statistics are optimistic, to put it charitably, and most of the data above relies on self-declarations of literacy, or overly simplified tests. This survey tested 50 households each from 12 disadvantaged communities (including urban and rural working class, slum, fishing and plantation communities), age 10 years and above, meaning that the survey results are biased to begin with, not to mention being more than a decade out of date. That said, the survey is interesting because respondents were tested on both claimed literacy rate (self-declaration) and an “actual” literacy rate, based on several tests of reading, comprehension, oral interviews, writing in context, and so forth. This is the disparity between claimed and actual literacy, averaged, in these communities: claimed 83%, actual 60.1%. Even given the various caveats about the limitations of this survey, this says bad things about the tendency to exaggerate or overestimate in self-declaration.

So, what’s the magic number? Given all of the above, it is probably not below, say, 60%, but it is almost certainly not as high as 90%. We’re not going to know until someone replicates something like the above study for a national sample. And that’s not gonna happen as long as everyone clings frantically to the 90% myth in a desperate attempt to find something to feel good about.