Expanding Sarah's Law: Implications
Written by Matt B · first published 26/01/2010
Stop the Press! I almost agreed with Cllr Simon Moores just then.
I find myself somewhat in agreement with Cllr Simon Moores, my some times dance partner, on the issue of available statisitcal information. However Dr Moores does not go far enough in his demands and as they currently stand would produce nothing but sensasionalist headlines in the like of the Daily Mail and the Sun.
In his article Expanding Sarah's Law. He's talking, of course, of the pilot of the so called 'Sarah's Law' a sex offender alert scheme. Cllr Moores suggests that we should be able to enter a postcode and get the number of registered sex offenders nearby.
This is all well and good but what does "nearby" actually mean? Canturbury is nearby to most people's minds but the criminal activity the people at the other end of the 8A bus route is not exactly going to impact greatly on the suitability of letting little jonny go to the shops by himself.
Keep reading as I take a look at what else is needed beyond the popularist demands
Let us asume that I do type in a post code, say, CT9 5TJ (to make one up at random) and it comes back, say, 27. What does that mean? Is that good? Is that just in the same post code area, town, district or county? How far away counts as "near".
The fact is that it is a meaningless figure without far greater definition.
More usefull but still mostly meaningless might be the numbers within, say, 5, 10 and 25 miles of a given location. However this is still meaningless. In a town with just 900 people 20 sex offenders is a lot but in an area with 130,000 people 20 is next to none at all.
Context is everything.
What we need is a descriptive data output such as "percentage of the population" within some well defined meaningful area. Going back to our two examples 20 would be 2.2% in the first instance and 0.015% in the second illustration.
However this figure tells you very little in and of itself. Sure it tells you what the chances of finding at least one sex offender in a random sample of 100 people from the area in question but is that good or bad? Is this, to be blunt, a normal ammount? Can you form a reasonable assesment of safety and potential risk from that figure?
You might be asking things like what is the lowest percentage of registered sex offenders as an expression of the total population where it might be realistic to expect any incidents within a given time scale.
If the twon next to yours has a percentage score of 0.9% and yours is 1.2% many people are going to think seriousely about moving. But in day to day terms is 0.9% significantly different from 1.2%?
What is needed is a figure to compare this too such as a national average. A result of such and such a percent which is such and such above or below the national average is a more useful figure. For example "1.2% which is 0.4 above the national average" now tells us that the town next door's 0.9% is also above average.
However the numeric distance from the "national average" is, on the whole, meaningless still. If you are of a sharp mindset you will be asking is 0.4 a long way or a short way from the national average. Is, indeed 0.4 above the national average so close as to be called "about average" or so far above as to be a crisis in need of address?
What we need is a measure of our distance relative tot he whole.
In stastics we use a measure called the standard deviation which is a measure of how far you are from the average compaired to how spread out the numbers are to start with. If your figure is nine points clear of the average but a single deviation is six hundred then your figure for common use is as good as average ("about average" in common terms) but if a standard deviation is just one or two then you have an extreme figure and in our case a need to ask one's MP what he or she thinks he or she is playing at letting that happen.
If we wanted to go the extra mile we could also insist on the figure as an expressed centile. The average falls on or arround the 50th centile (about 50% of the way though the range of possibilities as you would expect). Your town's centile figure would give you an indication of what percentage of the towns in the UK have less than yours and would, instantly, tell you if your town was unusal. A town in the 99.6 centile or above has the most registered sex offenders per hundred people in all of the UK.
You will have seen a working example of percentiles if you have ever looked inside a new born baby's red book. On the charts for size and wieght you will see a line indication the average (50th centile) and additional lines above and below that are roughly the same shape. Those gap are distribution averages for babies by age and a "normal" or average baby would remain within the white area and out of the shaded areas which reprisent the top 2% and bottom 2%.
So what Simon Moores should be asking for is that with a postcode we could find how many standard deviations our area's count of registered sex offenders is from the national average; the number of registered sex offenders as a percentage of the population as a whole; and the areas placement as a centile perhaps witha small chart to make the data even easier to read. The ideal, I imagine, being any negative number of standard deviations (less than average) to less than one standard deviation more than the mean average and anything below the 75th centile.
This would tell us in one sentance or less what the numbers are like and how "safe" this might be relative to the nation as a whole. The raw data of so many hundred and odd registered sex offenders per town serves no actual purpose especially as this does little to deistiguish lower and higher risk groups.
After all there is a world of difference between knowing that a handfuil of people in your town were found with indecent images on their computers and knowing that a similar number have commited violent acts against others. While both are crimes one crime indicates a far greater risk should they be repeated to the local population than the other.
Such figures will impact on house prices and if Simon (a free market loving Tory by subscription) is going to insist on this data then he has a duty of care to insist that the data be meanginful and understood within the wider scheme of things rather than sensationalist and prone to hurting damaged town economies.
Indeed such a lot of effort to asses the risk of a single type of crime to a child might mean that these data are not worth the time and money to process. When one considers that a full risk assesment can be carried out from looking not at the number of potential risks but the actual rate of ralised risks aka the local crime statisitcs one has to ask is all that effort going to be worth it?
Simon is on the right lines for pleasing a worried elelctorate with this idea but he needs to follow this thought up some more to arrive at a worthwhile and usefull implimentation suggestion. There is a need for good statisitcal information that empowers parents to asses their children's safty however this is already availbale and would need only a regular PC and a few days work with a spreadsheet or a calculator.
Maybe Simon Moores could put his IT influence to work in having the data he himself has pointed out processed into information?
I find myself somewhat in agreement with Cllr Simon Moores, my some times dance partner, on the issue of available statisitcal information. However Dr Moores does not go far enough in his demands and as they currently stand would produce nothing but sensasionalist headlines in the like of the Daily Mail and the Sun.
In his article Expanding Sarah's Law. He's talking, of course, of the pilot of the so called 'Sarah's Law' a sex offender alert scheme. Cllr Moores suggests that we should be able to enter a postcode and get the number of registered sex offenders nearby.
This is all well and good but what does "nearby" actually mean? Canturbury is nearby to most people's minds but the criminal activity the people at the other end of the 8A bus route is not exactly going to impact greatly on the suitability of letting little jonny go to the shops by himself.
Keep reading as I take a look at what else is needed beyond the popularist demands
Let us asume that I do type in a post code, say, CT9 5TJ (to make one up at random) and it comes back, say, 27. What does that mean? Is that good? Is that just in the same post code area, town, district or county? How far away counts as "near".
The fact is that it is a meaningless figure without far greater definition.
More usefull but still mostly meaningless might be the numbers within, say, 5, 10 and 25 miles of a given location. However this is still meaningless. In a town with just 900 people 20 sex offenders is a lot but in an area with 130,000 people 20 is next to none at all.
Context is everything.
What we need is a descriptive data output such as "percentage of the population" within some well defined meaningful area. Going back to our two examples 20 would be 2.2% in the first instance and 0.015% in the second illustration.
However this figure tells you very little in and of itself. Sure it tells you what the chances of finding at least one sex offender in a random sample of 100 people from the area in question but is that good or bad? Is this, to be blunt, a normal ammount? Can you form a reasonable assesment of safety and potential risk from that figure?
You might be asking things like what is the lowest percentage of registered sex offenders as an expression of the total population where it might be realistic to expect any incidents within a given time scale.
If the twon next to yours has a percentage score of 0.9% and yours is 1.2% many people are going to think seriousely about moving. But in day to day terms is 0.9% significantly different from 1.2%?
What is needed is a figure to compare this too such as a national average. A result of such and such a percent which is such and such above or below the national average is a more useful figure. For example "1.2% which is 0.4 above the national average" now tells us that the town next door's 0.9% is also above average.
However the numeric distance from the "national average" is, on the whole, meaningless still. If you are of a sharp mindset you will be asking is 0.4 a long way or a short way from the national average. Is, indeed 0.4 above the national average so close as to be called "about average" or so far above as to be a crisis in need of address?
What we need is a measure of our distance relative tot he whole.
In stastics we use a measure called the standard deviation which is a measure of how far you are from the average compaired to how spread out the numbers are to start with. If your figure is nine points clear of the average but a single deviation is six hundred then your figure for common use is as good as average ("about average" in common terms) but if a standard deviation is just one or two then you have an extreme figure and in our case a need to ask one's MP what he or she thinks he or she is playing at letting that happen.
If we wanted to go the extra mile we could also insist on the figure as an expressed centile. The average falls on or arround the 50th centile (about 50% of the way though the range of possibilities as you would expect). Your town's centile figure would give you an indication of what percentage of the towns in the UK have less than yours and would, instantly, tell you if your town was unusal. A town in the 99.6 centile or above has the most registered sex offenders per hundred people in all of the UK.
You will have seen a working example of percentiles if you have ever looked inside a new born baby's red book. On the charts for size and wieght you will see a line indication the average (50th centile) and additional lines above and below that are roughly the same shape. Those gap are distribution averages for babies by age and a "normal" or average baby would remain within the white area and out of the shaded areas which reprisent the top 2% and bottom 2%.
So what Simon Moores should be asking for is that with a postcode we could find how many standard deviations our area's count of registered sex offenders is from the national average; the number of registered sex offenders as a percentage of the population as a whole; and the areas placement as a centile perhaps witha small chart to make the data even easier to read. The ideal, I imagine, being any negative number of standard deviations (less than average) to less than one standard deviation more than the mean average and anything below the 75th centile.
This would tell us in one sentance or less what the numbers are like and how "safe" this might be relative to the nation as a whole. The raw data of so many hundred and odd registered sex offenders per town serves no actual purpose especially as this does little to deistiguish lower and higher risk groups.
After all there is a world of difference between knowing that a handfuil of people in your town were found with indecent images on their computers and knowing that a similar number have commited violent acts against others. While both are crimes one crime indicates a far greater risk should they be repeated to the local population than the other.
Such figures will impact on house prices and if Simon (a free market loving Tory by subscription) is going to insist on this data then he has a duty of care to insist that the data be meanginful and understood within the wider scheme of things rather than sensationalist and prone to hurting damaged town economies.
Indeed such a lot of effort to asses the risk of a single type of crime to a child might mean that these data are not worth the time and money to process. When one considers that a full risk assesment can be carried out from looking not at the number of potential risks but the actual rate of ralised risks aka the local crime statisitcs one has to ask is all that effort going to be worth it?
Simon is on the right lines for pleasing a worried elelctorate with this idea but he needs to follow this thought up some more to arrive at a worthwhile and usefull implimentation suggestion. There is a need for good statisitcal information that empowers parents to asses their children's safty however this is already availbale and would need only a regular PC and a few days work with a spreadsheet or a calculator.
Maybe Simon Moores could put his IT influence to work in having the data he himself has pointed out processed into information?
Think you can do better? · read more items categorised Politics · Don't miss a thing: Subscribe Today · Help: what's a feed? · This site is operated on set principles of policy and disclosure read them here
Waste and recycling collections may be suspended say council 
