Profiling is often seen as highly controversial by many. It is the act of looking into the common properties of a particular type of person (criminal, terrorist) and using that knowledge to direct resources into particular areas or people in order to get a maximum return on your investment (cost of stop-checks, border controls, airport security etc.).
This means that, often, people who would more likely appear to be, say, Muslims, or who are black, are stopped more often than others when searching out misdemeanours. I generally (with some later caveats) agree with some of this, and certainly the logic behind it. And it turns out, so probably do you, unless you are willing to adhere to double standards. I have never read up on profiling, not even Sam Harris’s defence of it, so I am very possibly wrong. This has come from my own independent thinking, philosophically. I would be happy to be shown to be wrong. Please point out any problems in the comments section and I will amend accordingly.
You see, we use profiling all the time across many areas of society. I will take education, since I work in this field, and show you how it is used.
Profiling Children
In both primary and secondary school education in the UK we have something called Free School Meals (FSM). FSM can be given out to your children if you get an income-based benefit (for example Income Support), or Child Tax Credit only. You can sometimes qualify if you get Working Tax Credit as well. But it depends on the income your tax credits have been worked out on, and where you live in the UK. Esssentially, you can qualify on account of (but not always) receiving the following:
- Income Support
- income-based Jobseekers Allowance
- income-related Employment and Support Allowance
- support under Part VI of the Immigration and Asylum Act 1999
- the guaranteed element of State Pension Credit
- Child Tax Credit (provided you’re not also entitled to Working Tax Credit and have an annual gross income of no more than £16,190)
- Working Tax Credit run-on – paid for 4 weeks after you stop qualifying for Working Tax Credit
- Universal Credit
The important part is not receiving FSM per se, but what happens with that information, and how it is used to profile children to give them extra benefits or provision. FSM children who have been eligible over the last 6 years (called “Ever 6”) get extra funding for their school (certainly at primary level). This funding is for the school to spend as they will, but they do need to show the inspectorate that it is being used to drive up FSM achievement and close the gap between FSM attainment and non FSM attainment. (Funding is also given to schools for children of Armed Forces families since they are seen to have needs due to parents often moving every two years or being absent).
What this is essentially doing is looking at the sorts of pupils and families who are FSM (income support etc.) and generalising that they, on average, will attain less because of their context.
Here is a table from a 2009 paper:
As the government says as to why it measures FSM attainment:
This indicator is a measure of the attainment gap associated with economic disadvantage. Disadvantage remains strongly associated with poorer performance, on average, at every key stage. It is a leading Government priority to narrow the attainment gaps between disadvantaged pupils and their peers.
In 2013 there was an 18.8% gap in Year 6 levels (aged 11) and in 2014 an 18.1% gap.
In other words, the children are being profiled on account of what their parents earn and are having resources focused on them to make educational attainment improved. The money in education is finite, and it has to be focused intelligently so that there is optimised return on investment. As you can see, one of the ways of doing this is to find out which subgroups of children under-attain and direct resources towards that.
I have some Ever 6 children in my class who get FSM and who are actually high attainers, for a variety of reasons (parents getting divorced, income assumed on single mother etc.). These particular few children are not representative of the norm. This will be important for a later point.
Profiling Places
We profile places, as society, an awful lot. We look at socio-economic demographics and pour money into social and environmental regeneration in particular areas. This happened to the London Docklands, for example (being the largest case in Western Europe). The point of doing this is to address social needs, such as crime, aspiration, social inequality, employment and so on. In other words, with finite resources, geographical areas are profiled and local or national governments decide which ones are deserving of resources to improve the aforementioned things.
I could mention many examples of this, but the point I think is obvious.
Accurate Information and Data
What is of crucial importance here is the need for accurate information. If the data is poor in the profiling process, then the resources you pile in are essentially being mis-allocated. For example, if it turns out that you are not getting correct socio-economic information about parents (a similar case to some of my children in class, though not perfectly analogous) of children who end up being FSM, or you incorrectly connect the whole area of low socio-economic demographics with low attainment, then you get a false picture of what is going on. The profiling is thus wrong. Not the logic of it, but the context of that particular scenario which is built on faulty data and premises.
These sorts of profiling examples concern things which, when joined together, give an ideal of social cohesion: aspiration, attainment, education, employment, criminality etc. They have big ramifications, so it is important to get that data right. Put good stuff in, get good stuff out. Put rubbish in, get rubbish out, and annoy lots of people in the process.
Applying this to Crime and Terrorism
Terrorism
If the logic is sound in carrying out the above simple examples of profiling, then does it work for adults and adolescents involved in crime and terrorism? I cannot see that there is any philosophical difference per se. The only differences appear to come in later, in the consequences, or indeed, in whether the data in the first place is correct, or correctly manipulated (and I gather with racial profiling, this could be a problem). 
To exemplify this, I am going to caricature or overly simplify scenarios and I will admittedly make up numbers; but I am merely showing how it theoretically works.
Let’s take terrorists. This is what would happen.
- Government has finite resources to fight terrorism
- They need to secure, for example, airport safety
- This requires checking passengers for bombs (very simply)
- Most terrorists in this scenario are Muslims (Data Alert! – is that claim accurate?)
- Most Muslims have X profile – appearance, properties etc. (Data Alert! – is that claim accurate?)
- Authorities should focus searches on people who fit that profile to get higher successful searches per £/$
This, as mentioned, requires claims 4 and 5 to be defended with accurate data (for example, how do conversions from other social/ethnic groups feed into 5? Can they be profiled to?).
Let’s pragmatically exemplify this. A security authority has £1000. They need to check passengers for bombs. Each check costs £50. This gives them 20 checks. They have 20,000 people to check. This means only 1 in 1000 people get checked. They, therefore, have to make that checking worthwhile. If they did random checks, assuming 1 terrorist in 20,000, they would have a 1/20,000 x 20 chance of finding that person. They would have a 0.001% chance of finding them. Further, if no terrorist had ever been found to be Japanese and on a Japanese flight, would it be money well spent to check a Japanese flight full of Japanese people? One would think not. That would be money wasted, probabilistically. If there are 400 people who fit profile X, and there is a 75% chance that the terrorist fits profile X, then the chance of finding them in checking only profile X is 400/20,000 x .75 = 0.015% (please correct me if my maths is wrong here!). In other words, there is a 15 times greater chance of catching the terrorist. If these figures are (symbolically) correct, then this means that profiling is, for human safety and financial reasons, the most sensible approach.
If these points hold, then there is nothing wrong with profiling for terrorists in such a way. What there obviously needs to be is robust and accurate data which tells us what most terrorists threats are, and what profile such terrorists have. The data collection and conclusions need to be accurate. [EDIT] I have also purposefully left “profile X” nebulous because that can change. In fact, using this simplistic airport example, terrorists could recognise a popular profile and get around it by using, say, Western-looking/white converts, for example. The beauty of keeping the profile nebulous in the example is that it is thus reactive. If the profile changes and remains a distinguishable profile, then profiling works. If not, then profiling will fail because there will be no profile. As mentioned in a comment below, retaining a random element to checking also acts to mitigate this changing profile somewhat.
Of course, if airport profiling is to work at all, those points 4 and 5 must contains significant enough useful data and recognisable profile. My point is more abstract at this point, rather than one about the technicalities of airport profiling, for example.
I have never really looked into profiling, and I do not necessarily know the stats. This falls down of there is no statistical case for a profile (in points 4 or 5 above, for example). If they stand, however, profiling stands, unless there are larger, more consequential outcomes to following such an activity (which I can foresee in criminal and racial profiling). Trying to think what these could be would be stigmatising Muslims, driving them through feeling more like the out-group into feeling otherised and towards terrorism etc. However, one would have to quantify this with evidence, too!
In the next post I will look at the complex area of criminal profiling with its controversial history with race and ethnicity. I will admit that the case for racial profiling looks to be a lot more complicated, and there will be a lot more guarded claims with caveats. I will be interested to see where that goes… I certainly don’t want to start with a conclusion and build from it; I will attempt to start with ideas and seek evidence, and then build from that to a conclusion.
