Sunday, June 17, 2007

A world without numbers

It's not available online (the link might work, though, if you have a Columbia affiliation), but at the end of a strikingly interesting and full account of Steven Johnson's book Ghost Map, Helen Epstein (writing for the NYRB) turns to related matters in an imaginative turn that made me envision a most wonderful middle-school math curriculum unit (one of my alternate selves is teaching eighth-grade math):

What would it be like to live in a world without numbers? I got a sense of this on a recent trip to East Africa, where the failure to register births and deaths and monitor diseases is responsible for thousands of deaths each year. I was visiting a program called the Tanzania Essential Health Interventions Project, or TEHIP, which was established in two of Tanzania's 127 districts in 1997. By 2002, the death rate of children in both districts had been cut dramatically but it had remained stable and high elsewhere in the country. In 2005, the Tanzanian government began to implement the program throughout the country, and by 2007, the national child mortality rate, which had barely improved at all between 1990 and 2000, had fallen by 25 percent. Elsewhere in East Africa, no such decline has occurred. TEHIP's budget is tiny by foreign aid standards —the districts spent on average about $1 per resident on the program, an amazing bargain compared to other health initiatives funded by donors such as the US government, the World Bank, and the Gates Foundation.

What was the secret of TEHIP's success? Record-keeping and statistics. The first thing you see when you walk into a TEHIP clinic is a wall chart, drawn in magic marker on butcher paper, showing graphs and tables indicating the case rates of different diseases. From this, health workers can tell which drugs they will need the following month. They also keep detailed ledgers of spending on salaries and procurement of supplies and spare parts for vehicles and radios, staff absences, and so on. Inspectors from the district headquarters visit each clinic from time to time, to ensure that records are being kept properly.

Before TEHIP, it sometimes seemed as though chaos reigned. Medicines were supplied in pre-packaged kits shipped from Europe, but their contents had been decided on in the early 1980s, and since then, the epidemiological situation had changed considerably. The population had grown and rates of malaria had increased in the lowlands, while rates of pneumonia had risen in the highlands. Thus the drugs to treat these conditions often ran out before the next kit arrived. Bizarre mistakes sometimes occurred: "Once we opened the kit and it was full of epilepsy drugs, but there are no cases of epilepsy around here," one doctor told me. Medical staff members complained, but because the consequences of this epidemiological shift were not being measured, nothing changed.

The arrival of salaries was desultory, so some doctors and nurses, unpaid for months, sought other livelihoods. The system for procuring spare parts functioned poorly, so radios and vehicles often broke down, making communication with the health care system all but impossible. Measles and cholera killed hundreds of people because reporting an outbreak often involved a long trek by foot through the bush to a main road followed by a public bus ride to a faraway town where the district health authorities were located. Help might take days to arrive. Now epidemics can be reported instantly because the supply chain for spare parts functions; since the program started, there have been virtually no deaths from measles or cholera in the TEHIP districts.

I don't think the connection between statistics and real-world consequences of the gravest kind is always made clear to students, even at the college level. My friend A.'s most serious advice to college journalists looking to make a professional career of it is to take a statistics class, and that seems to me very sound. (I say this in almost complete ignorance myself of statistics! Ah well...)


  1. Regarding your last paragraph:

    I hate to have to be the one to tell you this, but I don't think you'll get this from a statistics class, at least not from an intro class. I do, however, recommend our book, Teaching Statistics: A Bag of Tricks, actually a fun read for students as well.

    Getting back to the usefulness of numbers, I agree completely. But we don't do much of that at the university level, certainly not in departments of statistics. I was just chatting a couple of days ago with a NYPD detective who was enthusiastic about the police department's data-gathering program (in statistics jargon, "quality control")--whenever they notice a spike in crime, they shift resources to handle it. I told him that we're very very far from doing that at universities. (Can you imagine: whenever we find some students not learning some important material, rushing resources over to fix it?)

    Record keeping takes effort--and there always seem to be better things to do--and also there will be people who don't believe in the measurements and don't want to keep records. It takes a huge effort (as can be seen implicitly in Epstein's article that you excerpt).

  2. Yes, I am sure this is quite right, really I kind of know this! I will look forward to checking out that book, though!