Lies, damned lies and dodgy statistics


Disraeli; credit where it is due…

There are some stories that are too good to pass up without further comment; even if they have little relationship to local government.

This from Yahoo news is one such gem. Under the headline ‘25,000 men need an obstetrician or gynaecologist every year’ the story, originating from the Telegraph, reports that:

Official figures from the NHS appear to have proved that old adage true, as they show that Britain has tens of thousands of ‘male mothers’. They discovered that 17,000 men were recorded as having been admitted to hospital for obstetric services -a specialism for pregnant women and their babies – and 8,000 to see a gynaecologist; while another 20,000 apparently needed to see a midwife.

They also identified a steady increase in the numbers of children and teenagers attending geriatric services, to over 3,000 between 2009 and 2010, and more than 1,600 adults over 30 using child psychiatry services.

The problem for the NHS, and for other public bodies who use statistics like this to run their services, is that bad data like this can then lead to bad decisions. Increasingly public services rely on large quantities of statistics.

This is entirely rational. After all, public services don’t have profit to base their decisions on and the other ultimate means of determining success are elections and they are fairly blunt instruments.

If statistics are so important to public services and yet these statistics are prone to be impacted by human error, as the above statistics were, what can be done about it? Here are a few suggestions:

1)    Don’t rely on macro stats to run your service: Large sample sizes are really helpful when trying to create narratives and understand the context of your service. However, if you are trying to make day to day decisions then these massive statistics are not going to be that helpful. Much more helpful is a series of manageable and current statistics

2)    Publish your working: Statistics can go wrong and human error is always a risk. But if the data that underpins the statistic is open to everyone it becomes much easier to notice if something is wrong. It also makes it easier to trace back and identify anomalies

3)    If your measures are incredibly complicated then you’re probably doing something wrong. A simple measure that a member of staff can easily understand is less likely to go wrong than something that requires multiple datasets to be combined together or requires large amounts of calculation.

4)    Don’t use contrived statistics for benchmarking. Trying to create statistics to meet someone else’s requirements is a sure fire to make your numbers unusable for your own service and also incredibly complex.

5)    Treat your statistics as the beginning of the conversation not the conclusion to it. Only an idiot would have introduced male only gynaecology services but in some other areas the statistics can be equally misused without the right answer being so obvious. However, if we use the statistics to start the conversation then at least we are not bound by the errors within them.

Oh yeah, and don’t forget that numbers are only as helpful as the humans handling them.

Welovelocalgovernment is a blog written by UK local government officers. If you have a piece you’d like to submit or any comments you’d like to make please drop us a line at: welovelocalgovernment@gmail.com

About these ads
Explore posts in the same categories: We love the Council

Tags: , , ,

You can comment below, or link to this permanent URL from your own site.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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


Follow

Get every new post delivered to your Inbox.

Join 241 other followers

%d bloggers like this: