Tweet
Making sense of development statistics
Published: Sunday | November 14, 2010
Martin Henry, Contributor
Students of the University of the West Indies (UWI) of my generation, which is quite some time ago, grew up on a little book, How To Lie With Statistics. But let me pause right here, before pursuing "lying" with statistics any further, to pay tribute to that quintessential UWI man, Professor Barry Chevannes, professor of cultural anthropology, who was not unduly worshipful of statistical analysis which has done so much to derail the social sciences as ways of understanding human society and human action. Barry understood the power of narrative and of art in both scholarship and life, which scholarship seeks to understand.
From his breadth of output, I am most interested in his work, both scholarly and activist, on violence, the greatest scourge on Jamaican society, and his work on men. Chevannes, while frankly acknowledging the problems men faced and men caused, defended, with empirical evidence, the Jamaican male against the pervasive theories of male marginalisation and the 'wutliss' man.
Actually, I knew Pauletta, Barry's wife, earlier and better, as we taught together at Charlie Smith High School in Arnett Gardens in and around the dark days of the 1980 general election. Pauletta worked in education with missionary zeal under difficult circumstances, as a matter of national service. At the time, I was more interested in earning an income while completing 'free education' university. I studied mornings and one full day, and taught four afternoons in that shift school located in what the residents themselves called 'Jungle'.
My condolences to the family - and that 'family' is vast if it includes all whose life Barry's touched with the kind of joyfulness present in his music.
As we go back to 'lying' - or explaining - with statistics, one of the basic lessons of Statistics 101 is having a common base number for comparing things. Base 100 is by far the most popular, hence the widespread use of percentages [from the Latin for per 100]. Another core lesson is having sufficient data points for properly figuring out trends and patterns. I raise these points because too many media practitioners break these basic rules in reporting, and end up 'lying' with statistics.
Misleading report
Take this lead from a news report on Jamaica's 2010 ranking on the United Nations Development Programme's (UNDP) Human Development Index (HDI) which was published in the business section (of all places) of a newspaper: "Jamaica jumped 20 positions to rank 80 worldwide [in] the Human Development Report 2010 published by the United Nations (UN) allowing the island to break from medium to high human development status. Its [sic] the best result in 30 years for the country struggling with crime the report ... indicated".
The story does say later on that "the improvement this year was not perfectly comparable with the 2009 report ... " but then did nothing to provide a common base for year-to-year comparison. Instead, Jamaica's raw rankings were given for a number of years on a shifting base of the number of countries reported on. When those rank numbers are adjusted to a common base [base 100 is fine] a very different and more accurate picture emerges:
In 2010, 80/169 is equivalent to 47/100.
For 2009, 100/182 is equivalent to 55/100, which is, in fact, a better ranking than this year's.
But what is the trend? Just using the data in the story itself:
For the period 2007-2008, 101/177 is equivalent to 57/100.
The 1990 ranking of 87/130 is equivalent to 66/100.
What we are seeing here, contrary to the triumphant announcement of Jamaica jumping in development, is a progressive comparative decline when the base for comparison, the number of countries surveyed is fixed and not allowed to float. Indeed, the hard scores on the HDI, as opposed to the relative rankings, show exactly this. Late in the news story we are finally told that the country has, in fact, experienced a drop in its HDI score, 2010 over 2009, from 0.766 in 2009 to 0.688 in 2010, with 1.0 being the maximum and perfect score. This is a significant 10 per cent drop. This is what matters. This is the real news.
Full Hundred
Published: Sunday | November 14, 2010
Martin Henry, Contributor
Students of the University of the West Indies (UWI) of my generation, which is quite some time ago, grew up on a little book, How To Lie With Statistics. But let me pause right here, before pursuing "lying" with statistics any further, to pay tribute to that quintessential UWI man, Professor Barry Chevannes, professor of cultural anthropology, who was not unduly worshipful of statistical analysis which has done so much to derail the social sciences as ways of understanding human society and human action. Barry understood the power of narrative and of art in both scholarship and life, which scholarship seeks to understand.
From his breadth of output, I am most interested in his work, both scholarly and activist, on violence, the greatest scourge on Jamaican society, and his work on men. Chevannes, while frankly acknowledging the problems men faced and men caused, defended, with empirical evidence, the Jamaican male against the pervasive theories of male marginalisation and the 'wutliss' man.
Actually, I knew Pauletta, Barry's wife, earlier and better, as we taught together at Charlie Smith High School in Arnett Gardens in and around the dark days of the 1980 general election. Pauletta worked in education with missionary zeal under difficult circumstances, as a matter of national service. At the time, I was more interested in earning an income while completing 'free education' university. I studied mornings and one full day, and taught four afternoons in that shift school located in what the residents themselves called 'Jungle'.
My condolences to the family - and that 'family' is vast if it includes all whose life Barry's touched with the kind of joyfulness present in his music.
As we go back to 'lying' - or explaining - with statistics, one of the basic lessons of Statistics 101 is having a common base number for comparing things. Base 100 is by far the most popular, hence the widespread use of percentages [from the Latin for per 100]. Another core lesson is having sufficient data points for properly figuring out trends and patterns. I raise these points because too many media practitioners break these basic rules in reporting, and end up 'lying' with statistics.
Misleading report
Take this lead from a news report on Jamaica's 2010 ranking on the United Nations Development Programme's (UNDP) Human Development Index (HDI) which was published in the business section (of all places) of a newspaper: "Jamaica jumped 20 positions to rank 80 worldwide [in] the Human Development Report 2010 published by the United Nations (UN) allowing the island to break from medium to high human development status. Its [sic] the best result in 30 years for the country struggling with crime the report ... indicated".
The story does say later on that "the improvement this year was not perfectly comparable with the 2009 report ... " but then did nothing to provide a common base for year-to-year comparison. Instead, Jamaica's raw rankings were given for a number of years on a shifting base of the number of countries reported on. When those rank numbers are adjusted to a common base [base 100 is fine] a very different and more accurate picture emerges:
In 2010, 80/169 is equivalent to 47/100.
For 2009, 100/182 is equivalent to 55/100, which is, in fact, a better ranking than this year's.
But what is the trend? Just using the data in the story itself:
For the period 2007-2008, 101/177 is equivalent to 57/100.
The 1990 ranking of 87/130 is equivalent to 66/100.
What we are seeing here, contrary to the triumphant announcement of Jamaica jumping in development, is a progressive comparative decline when the base for comparison, the number of countries surveyed is fixed and not allowed to float. Indeed, the hard scores on the HDI, as opposed to the relative rankings, show exactly this. Late in the news story we are finally told that the country has, in fact, experienced a drop in its HDI score, 2010 over 2009, from 0.766 in 2009 to 0.688 in 2010, with 1.0 being the maximum and perfect score. This is a significant 10 per cent drop. This is what matters. This is the real news.
Full Hundred
Comment