Tuesday, 3 June 2008

Service improvement program

There was a BBC report at the weekend about the decision of four police services to drop the government targets on crime and to use their own 'common sense' approach to prioritising. Government targets and the problems associated with them are a regular feature of the older posts on coppersblog, along with almost any blog dealing with people working in public services, as well as numerous documentaries, reports and so on.

Essentially the complaint is that by recording and judging services (and departments, and people, and everything) on particular formulaic and numerical criteria, and using the results to produce a ranking from 'best' to 'worst', the targets encourage cheating. Not because the people cheating are trying to disrupt the statistics, or because they are incompetent, or because they are evil, but simply because it makes sense.

Hospitals who are judged on the proportion of patients who have beds and who sleep in wards need to find ways to increase the numbers of beds and wards. The intention of the targets is to prompt hospitals to spend their money doing the best for the patients by paying for these new facilities, but what actually happens is that the permanently-strapped-for-cash hospital reclassifies gurneys as beds and corridors as wards. Hey Presto! More beds, more wards, and everyone's numbers go up. The patients don't "lose out" as such, because they were sleeping on gurneys in corridors anyway, and the money can be spent on other (equally vital) areas.

Similarly, police forces that are judged on the number of 'detections' they make, are being encouraged by the targets to find detections everywhere. Why commit 4 officers to a murder investigation that may take weeks and never get solved, when you could have those officers "solving" domestic disputes, that take far less time and so can provide a far better crime-prevention rate. A kid kicks a ball through a window? Criminal damage. The parents get into a fight about it? Two counts of assault. One incident, one morning's work, three detections, and everyone's numbers look better.

The police officers and hospital staff are not being intentionally divisive in this regard, but they are simply doing what they're told. Their bosses know that if their service comes near the bottom of the 'rankings' they are going to be in trouble, so they need to get their numbers up however they can. This pressure feeds down through the organisation, and everyone is encouraged to concentrate on the numbers rather than on the 'common sense' approach to the tasks. The price, of course, of not following this pattern, is that your numbers don't look so good, you are reprimanded by a watchdog, and heads roll at the top. Everyone pats themselves on the back - once again, the numbers spotted a low-performing service, and action was taken to improve it.

Clearly this is a hugely sub-par way to run public services, and it is a problem caused entirely by the implementation of the government targets. So what are the options to improve this? Well, firstly, there is the option most often spouted by exasperated commentators: to remove the targets system entirely, reduce the amount of paperwork dramatically, "get the police back out on the streets" and go back, presumably to some sort of inspection system. I disagree with this approach because I think the 'targets' system has some advantages that it would be a pity to lose. I agree that the police, for example, need to have far less paperwork (as I mention previously here), but I think that that is a separate problem from that of the targets. I think that the collection of process data is hugely useful, and if it can be done with an efficiency that allows officers to effectively carry out their jobs at the same time, then it is something that should be continued.

The problem seems instead to be with the way that the data is collected. Rather than asking for detection rates, and ranking organisations on their relative performance, something that breeds a subservience to the formula that calculates rank, I think it'd be much more effective to have the formula itself be hidden. Data collected from the services would be analysed in conjunction with the organisation themselves to look for areas of improvement using a number of different criteria that might change over time. The most important aspect of it, though, would be that the organisations were not aiming for a simple measurable factor. Rather than being able to attempt to maximise 'detections', they would be reporting all their raw data, and the analysis would be ongoing. If it was reported that people were attempting to skew the results in a particular way, the analysis could be altered to take it into account.

In addition, the results of the analysis could be presented as a series of recommendations, rather than a score or ranking. Recommendations that would be tailored to an organisation's particular situation. The main argument against this, would, I suppose, be that the organisations would have less motivation to provide good services without the public shaming of a low ranking. However this could be combated through the additional recognition of particular improvement or decline in service quality. Again, it would be related to the local situation, so a police service in rural Wales would be judged on different criteria than the Met.

Finally, the involvement of the organisations in the process would be an indication of the point of the analysis, not to name and shame "failing" organisations or people, but to provide an indication of performance and pointers for improvement. In particular this is important for public services, since the point is not to increase profit margins, but to provide a good service to the public. In this regard, then, the government and the organisation have the same aim, and should be able to work together. With the involvement of the organisations, local targets could be ensured to be appropriate and achievable.

After the system had run for a few years, it might be possible to make national comparisons, or to provide more general analysis of multiple organisations, but only if it can be done without losing the focus on improving services.

Not only am I now rambling, but in talking about "improving services" I'm starting to sound like a government press release. In any case, I think that things can be done to improve the system of government targets without scrapping the whole thing, and it's something that niggles me whenever people talk about the target system as though it was an absolute evil. I'm not an expert (or even an amateur) in the area, and, for all I know, what I've written is complete bollocks. What do you think?

No comments: