Some Thoughts on the Use of Statistics in Ethical Matters

Some months ago, I was discussing Donald Trump’s election chances with a couple of friends, and I happened to mention a voter I had read about somewhere, a young man in California, who described himself as a feminist. He was concerned about the power of false rape allegations to destroy the lives of men, and was on that account leaning towards a vote for Trump. I said that while I didn’t sympathise with the vote, I certainly understood and sympathised with the concern behind it. One friend responded that such false accusations were surely statistically far less significant than real cases of the crime itself that go unpunished. The other friend agreed, and that seemed to settle the matter, at least for the two of them.

For me, well, it got me thinking. I have since found myself noticing how arguments that proceed by means of statistics are often used as though they provide a sort of final knock-down in ethical matters, unanswerable and definitive – and yet this is something they can almost never provide.

To see why, let’s indulge for a moment in a bit of philosophy. Most undergraduates who have done a philosophy course should have some idea about what utilitarianism is. We can define it as the principle that the greatest good should be done for the greatest number of people possible. Our undergraduates should also have some idea of a potential problem with utilitarianism: it is compatible with an acceptance of radical evil, provided, of course, that the evil is done to a minority. You could literally exterminate people while achieving the greatest good for the greatest number. (The responses of major utilitarian philosophers to this problem is something we can leave aside here.)

Now here’s why I bring this up: it seems to me that the sort of claims you can make with statistics are precisely the sort made by utilitarianism. Statistics give us a coarse overview of a subject; they tell us that something is the case on the whole, and for the most part. And to the extent that we confine ourselves to this view, we open up the possibility of radical evil, so long as it is done to a relatively small number of people. So if we’re going to commit ourselves to statistical state of affairs as a desirable aim of policy or activity, we should think quite clearly about what we’re actually advocating.

In the case I began with, we are indeed dealing with a readiness to do real evil to innocent people in the name of achieving what is taken to be a statistically desirable result. Feminist activists often speak as if the purpose of the justice system is to bring about a certain proportion of convictions relative to the number of complaints made. That is, they claim that only a very small fraction of rape accusations result in convictions, and that this small proportion indicates a failure on the part of the justice system. The implicit aim is a system that produces convictions in some much larger proportion, say, 50% or 98% (though I have never seen an actual number given). There is no room in such calculations for the possibility that the lives of innocents might be utterly destroyed by baseless accusations. Indeed, that a callous indifference to the possibility of crushing innocent people is in fact at work here is confirmed by such well-known cases as the Duke lacrosse team rape trial, or the travesty of reporting about another such supposed crime that Rolling Stone committed. If we were to follow the statistical point of view to its logical endpoint, the proper question to ask is not about the guilt or innocence of the accused at all, but simply whether or not the appropriate number of guilty verdicts have been delivered so far relative to the overall number of accusations.

The purpose of a justice system worthy of the name, however, is not to produce a statistical result, but rather to answer a specific question: what is the truth of the matter in this particular case? Such an approach is not likely to produce the most pleasing proportion of guilty and innocent verdicts, and is likely to fall short of its goal regularly, but it allows the rights of accuser and accused to be balanced against one another. Perhaps most importantly, it keeps the justice system, as much as possible, from the active commission of injustice. The words primum nil nocere – “first, do no harm” – are supposed to apply to doctors, but they should be a principle of the justice system no less.

Certainly statistics might inform and guide us in subsidiary ways, but only to the extent that we observe the more important principle of a system that does not itself do active harm to the innocent. When discussing ethical matters, then, we should always be cautious when confronted by arguments that proceed only on the basis of statistics.

Statistics are dangerous in other ways, too. They can give a seductive, but misleading impression of being simply factual; they can seem to clothe a particular view in the robes of science, giving it an apparently insuperable status in comparison with non-statistical arguments. (I’m reading a book right now that seems to suffer from this very failing; a blog post reviewing it will probably follow in the weeks to come.)

Of course, statistics can be massaged, and where crime is concerned it is clear that they often are massaged. Consider Rainer Wendt, federal head of the German Police Union: “if I, as a senior officer, want narcotics-related crime to go down in my city, then I send the officers responsible for it out to police [vehicle] traffic infringements. Then I promise you, that narcotics crime will go down – at least statistically.”

In fact, things may be far worse than Wendt’s words suggest. There is reason to believe that the arrival of statistics-based crime fighting has corrupted our justice system in a most fundamental manner. Here I will allow Theodore Dalrymple to speak, as he relates his wife’s attempt to report a crime:

She noticed some youths setting fire to the contents of a dumpster just outside our house, a fire that could easily have spread to cars parked nearby. She called the police.

“What do you expect us to do about it?” they asked.

“I expect you to come and arrest them,” she said.

The police regarded this as a bizarre and unreasonable expectation. They refused point-blank to send anyone. Of course, if they had promised to make every effort to come quickly but had arrived too late, or even not at all, my wife would have understood and been satisfied. But she was not satisfied with the idea that youths could set dangerous fires without arousing even the minimal interest of the police. Surely, some or all of the youths would conclude that they could do anything they liked, and move on to more serious crimes.

My wife then insisted that the police should at least place the crime on their records. Again, they refused. She remonstrated with them at length, and at considerable cost to her equanimity. At last, and with the greatest reluctance, they recorded the crime and gave her a reference number for it.

This was not the end of the matter. About 15 minutes later, a more senior policeman telephoned to upbraid her and tell her she had been wasting police time with her insistence on satisfaction in so trivial a matter. The police, apparently, had more important things to do than suppress arson. …

It is not difficult to guess the reason for the senior policeman’s anger. My wife had forced his men to record a crime that they had no intention whatever of even trying to solve (though, with due expedition, it was eminently soluble), and this record in turn meant the introduction of an unwanted breath of reality into the bogus statistics, the manufacture of which is now every British senior policeman’s principal task.

Once the job of the police is to fight crime statistics rather than crime, the possibility of almost infinite corruption has been opened up – indeed, a motivation for corruption has been provided.

As one reads other Dalrymple essays, one encounters similar episodes: police refusing to respond to crime, or even to record it – or, alternately, responding to minor infractions, committed by individuals likely to respond to the authority of the police, rather than to violent crimes committed by people likely to respond in a violent manner (e.g., here or here or here).

Statistics, then, will not only fail in most cases to provide a decisive argument when considering ethical matters, they can also be the cause of great harm. Any insight they provide has to be balanced against direct contact with the problem in question. Localism – taking power away form massive bureaucracies and putting it in the hands of smaller communities – would no doubt be helpful. The more I read Dalrymple, and the more I read of the views of police officers in England, America and in Germany, the more I think that policy on criminal matters should never be allowed to be made on the basis of statistics alone, and that there should be a requirement that people who deal directly with crime are given some substantive power in the production of that policy.

And while I’m dreaming, I’d also like a private jet.


Author: Babbington

Citizen of the English-speaking world, resident of the German. Refugee from academia, writing a blog because, well, "in my heart there was a kind of fighting/That would not let me sleep."

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s