In Science News, Julie Rehmeyer writes a short piece on Humanitarian Statistics, with a focus on the “controversial” Iraq war studies carried in the Lancet. I haven’t posted about the Lancet studies before; I recognise that the Lancet studies have an important role to play in tallying the cost of the Iraq war, but anything I could add to the debate would be largely redundant, since it’s been driven by political rather than humanitarian interests.
Although Deltoid characterises the article as being “about the Lancet studies” - and fair enough, that is his particular interest - it is thankfully wider than that, noting the increase in the use of statistics in the human rights (and to a lesser extent, humanitarian) sector while also being aware of the limitations:

But humanitarian crises pose huge challenges. Little information may be available—even from before a crisis—about how many people live where. Even if a previous census was taken, the high birth and death rates in developing countries tend to quickly make censuses outdated. Areas within continuing war zones can be unsafe for survey workers.

Examples from Sierra Leone and East Timor are referenced in the article. The latter case is particularly interesting because it wasn’t just based on a straight survey - which is what we generally think of when we think of statistics - but on pulling together separate and incomplete datasets to build a bigger picture, which is the norm in humanitarian crises, particularly in developing countries.

In the comments section at Deltoid, commenter Jeff Harvey laments

I can only shake my head in disbelief. Who will do the survey? The US and British governments, who are responsible for an illegal invasion that has turned Iraq into a country of wreck and ruin? This is the bitter irony. Aggressing nations do not tally the numbers of their victims. Ian Gould summed it up in the thread below this: because the real death toll of civilians conflicts with the well-cultivated myth of US benevolence, western crimes are not a part of history because they are never allowed to become a part of history. They thus get sent straight down the memory hole.

Jeff misses the point that (I think) Julie was trying to make. Although he gives many examples of past victims of war who have been lost to history, we don’t live there any more. There are more people working on these issues than ever before, and we have a better idea of how to approach these problems. However it’s this attitude - that information gathering and analysis should be a political project - that is likely to prove the biggest obstacle to moving forward.

The only way to do justice to the victims and to persuade belligerent parties to accept the results is to treat these issues as impartially as possible - and to do so with the perspective that our work is at the service of the beneficiaries, rather than of our own political interests.

As we discussed before (see Electoral Geography and Political Violence in Zimbabwe), Zimbabwe goes to the polls this Saturday. The long term field monitoring efforts of Zimbabwe Election Support Network and many others should ensure that the government’s tight restrictions on accreditation of international journalists doesn’t create an information draught. Whilst the usual international outlets (HRW, Economist, ICG) have already released rich contributions about the election, it still remains challenging to get a feel for ongoing events. Perhaps the lack of international media on the ground will widen the space for citizen journalism, and force observers to rely more on non-traditional sources of news.

I had hoped to have some of the maps from the Violations Early Warning System  (ViEWS) of Zimbabwe Peace Project, but these have not hit the Net. The next best thing is Sokwanele.com’s Google Maps mashup of election-related incidents:

You can also read their blog post introducing the project. Like with Ushahidi.com, Sokwanele’s map will drive human rights documentalists (myself included) mad: its primary sources are unverified from the media, its mapping is necessarily imprecise since Google gazeteer for Zimbabwe is far from extensive, categories are overlapping, confusing and sometimes hyperbolic (”political cleansing”).

But that’s really not the point: it’s attractive, accurate and expressive enough, and provides easy links into the source materials. It’s an example of where information design trumps documentation. In Sokwanele’s own words, in the caveat about their data:

The map aims to give an impression of the scale and range of challenges facing Zimbabweans as we head towards the March 29th elections. Even though this is based on a small sample of information we have logged since July 2007, it clearly shows that conditions in the country are not conducive for a free and fair democractic elections.

The only thing I would suggest adding to it is a filter-by-date widget, so we can see what happens on election day.
Via Zuckerman.

Public transport doesn’t often provide pointers for the humanitarian community. The recent cracking of the London OysterCard (following hot on the heels of the earlier crack of the Dutch transit card system) came as no surprise to digital security experts, but it should teach us fundamental lessons about information security and personal privacy issues.

Security researchers say they’ve found a way to crack the encryption used to protect a widely-used smartcard in a matter of minutes, making it possible for them to quickly and cheaply clone the cards that are used to secure office buildings and automate the collection of mass transportation fares.

No electronic identification scheme is secure. It doesn’t matter how good your technology is, any system which is built by humans can be cracked, and the only defense is to make the cost of cracking it as prohibitive as possible. (The kicker is that you never know if you’ve successfully achieved that - until somebody cracks it and it becomes embarrassingly obvious that you haven’t.) On top of that, the more complex and expensive a system is, the more difficult it is to fix it when something like this happens.

In themselves, these obstacles aren’t insurmountable - largely because they’re technical in nature - but you see the real issues when you look at how these schemes are implemented. Governmental (and intergovernmental) organisations are notorious for a) thinking that technology can fix problems which are not technical in nature (for example, running a public transport system) and b) frequently mismanaging technology projects, often with the assistance of the vendor.

In a public transport system, this is not a life-and-death issue. What if this was a tracking system for food aid, though, where RFID has begun to be introduced as the solution to our logistics inefficiencies? Or a refugee registration database in a country where human rights abuses are endemic? Or an employee identity card scheme in a country where terrorists are targeting UN and NGO offices? You start to see where this might be going…

There was also related news that MI5 have requested “full automated access” to the OysterCard database. In a liberal democracy where the rule of law holds, that might not be too worrying - but there are a number of countries in the world that don’t fit that description, and where giving access to this sort of information to the government might not be in the best interests of the beneficiaries.

The fear of cyber-warfare has climbed Whitehall’s agenda since last year’s attack on the Baltic nation of Estonia, in which Russian hackers swamped state servers with millions of electronic messages until they collapsed. The Estonian defence and foreign ministries and major banks were paralysed, while even its emergency services call system was temporarily knocked out: the attack was seen as a warning that battles once fought by invading armies or aerial bombardment could soon be replaced by virtual, but equally deadly, wars in cyberspace.

It’s only a matter of time before humanitarian organisations come under similar attack - and we’re not prepared for it in the least. None of this means that this technology shouldn’t be used - it absolutely should be. What it means is that we need to be a lot more savvy not just about the technology issues but about the entire range of processes - procurement of the system, implementation within the organisation, sensitivity to the situation (including security concerns), and so forth - in order to make sure that we’re prepared to address these situations when they arise.

The government are very keen on amassing statistics. They collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams. But you must never forget that every one of these figures comes in the first instance from the village watchman, who just puts down what he damn pleases.

Josiah Charles Stamp was many things during his lifetime, including President of the Royal Statistical Society between 1930-32, and his view on government statistics is well worth bearing in mind whenever we look at the sort of statistics that tend to crop up in humanitarian and development work. International organisations tend to act in much the same way as governments when it comes to statistics, which we usually refer to as indicators.

Until very recently, all the information we had during a disaster was based on eyes on the ground - from residents in or visitors to affected areas. With the advent of remote sensing, we have a new source of information that doesn’t rely on actual presence - but even so, it’s worth noting that remote sensing without some form of ground truthing is frankly useless. I remember when we were looking at crop patterns in Afghanistan - an expert could pick out opium crops from a satellite image, but it was still necessary to send people to verify (and get shot at, of course - all part of the fun). So we still rely on eyes on the ground, which means that there’s always a human factor involved in data collection.

Where there’s a human factor, there’s always the scope for creativity that Stamp noted, or for deliberate manipulation. On the micro scale, that’s unlikely to make a huge difference, since when you aggregate up to a national level many of the irregularities will be levelled out - unless everybody at the micro level is fiddling the numbers in the same way. For example, anybody affected by a disaster is likely to exaggerate their needs if they think it will mean more assistance; any organisation responding to a disaster is likely to accept those exaggerations if it means they are likely to get more funding. As always, it’s good advice to follow the money.

So what does this mean for the poor information manager, tasked by his bosses to tell them what’s going on? Well, Stamp wasn’t saying that statistics were useless, only that we need to remember where they come from, which is rule number one: scrutinise your sources. He wasn’t saying that the village headman is out to cheat you, only that the village headman is human, so rule number two is: minimise the ways in which errors can be introduced to your data collection. As I noted above, those errors tend to get levelled out when you aggregate up - or at least to get less obvious amongst the mass of data - so rule number three is: where possible, always cross check your data against other sources.

Rule number four, of course, is: don’t expect statistics to solve all your problems. We tend to get a little fixated with a fetish for figures in this field (and I’m as guilty of that as anybody - look at the pretty pictures!) but, if our analysis or presentation aren’t solid, those figures aren’t going to be much use to managers.