Question 18 of the consultation on “Transforming Legal Aid” asks:
Which of the following police station case allocation methods should feature in the competition model?
· 1(a) – cases allocated on a case by case basis
· 1(b) – cases allocated based on the client’s day of month of birth
· 1(c) – cases allocated based on the client’s surname initial
· 2 – cases allocated to the provider on duty
Give the reasons for your choice
This has been a question that I was really looking forward to answering, for no other reason that I can be a bit of a geek when it comes to statistics. I have replied as follows:
"It makes one despair to realise that the level of thought that has gone into the above question is almost certainly indicative of the level of thought that has gone into the consultation as a whole. The sheer impracticality of the suggestions beggars belief
Let me show you how foolish these proposals are...
1(a) - allocation on a case by case basis: This will result in numerous instances of one client having several providers representing him. It is not unusual for suspects to be arrested for a number of offences over a number of occasions, as evidence comes to light. Repeat offenders, particularly for minor offences, are commonplace. Even in areas with only four providers, the risks of two providers representing the same client on similar charges are high. The more providers per area, the greater the problems will be
1(b) - cases allocated by day of month of birth. I am simply astonished at this proposal, clearly drawn up by someone with little knowledge of basic mathematics, let alone statistics.
Four months have 30 days in them - thus equal amounts of work for clients born in such months can only be given to areas with 5, 6, 10, or 15 providers
Seven months have 31 days - 31 is a prime number. That means it can only be divided equally by itself or 1. Thus work can only be equally divided amongst 31 providers. There are no areas with 31 providers
February has 28 days. 28 can only be fairly divided between 4, 7 and 14
You will note, I am sure, that there are actually no areas where work can be allocated equally based upon the date of birth of the client and thus this suggestion cannot work.
1(c) - cases allocated on client's surname initial. Again, anyone with a basic grasp of statistics will tell you this is also unworkable in giving an equal share of the work to all providers
There are 26 letters of the alphabet. 26 is divisible only by 2 and 13. There are no areas with 2 or 13 providers
Further, the authors seem not to realise that
a) there is no equal spread of surname initial letters – many are more common than others
b) some surnames are considerably more common than others, and
c) there is a geographical spread of surnames around the country, with some far more prevalent in one area than others.
For example, over 25% of the 25 most common surnames in the UK start with the letter W. Accordingly, a provider assigned the letter W will receive an exponentially higher number of cases than other providers.
10 letters of the alphabet - almost 40% - do not feature in the 25 most common surnames in the UK. For example, you have to go down the list to 46 to find the first A in Adams
The person with the letter S is fortunate because you are 5 times more likely to get a Smith arrested than you are a Lee, yet Lee is the 25th most common surname in the UK!
The most common name across most of England is Smith, or a derivative of it, such as Smyth. However, the most common name in Greater London is Patel. The most common name in Wales is Jones.
An aside, albeit quite an important one – the surname Smith is not the most common name in the North East or Cumbria. The surname Wilson takes those honours.
2 - Allocated to the provider on duty. This brings me back full circle to the point I made at 1(a). It will become very common to have more than one provider representing the same client
Can I suggest a fair way of allocating cases? Of course I can ... it is called client choice"