Wednesday, September 04, 2019

Still juking the stats

The State Services Commission and Ombudsman have released another batch of OIA statistics, covering the last six months. Request volumes are up, and the core public service is generally handling them within the legal timeframe, though this may be because they've learned to extend rather than just ignore things. And the police - who are counted seperately - have learned how to report timeliness again (they also seem to have reported a whole year's worth of requests, twice as many as the rest of the public service put together. And its amazing how despite reporting 96.3% timeliness, they're almost always late with anything I ask them. But I guess that, like Mark Hanna, I'm just a statistical outlier, right? But I suspect what's happening here is that all of their court / legal process OIAs are handled in a timely fashion, while their political / policy ones are absolute shit. The only way to fix this is to report them separately so they can see where the problem lies).

The real problems seem to be in Crown Entities and DHBs. Some of these are small, and a single late request can make a big difference. But South Canterbury DHB stands out for its sheer shitness. They're not a low-volume agency; 60 requests a year is bigger than some ministries. But they manage only a 56.7% on-time rate. If these statistics are to mean anything, then SSC must respond to this appalling result and sort them out. Otherwise what good are they?

Meanwhile, the Ombudsman's statistics tell an interesting story. Complaints for full refusals are up again, reinforcing the idea that agencies are meeting their timeliness requirements by just issuing a blanket "no" (another example of why SSC needs to report outcomes rather than just timeliness: so we can identify such behaviour. But that's probably exactly why they don't want to). And the Ombudsman has moved even further towards informal resolution, with a massive 57% of complaints this half-year resulting in no investigation. The number was elevated by last half-year's massive number of complaints against school boards, and in a break with past-practice, they're claiming credit for remedies even when no investigation is undertaken. Which is both fair - getting a pre-investigation call from the Ombudsman is often all that is required - but also irritating, in that it destroys my time-series on the proportion of resolved complaints that resulted in a remedy (i.e. the proportion of noticed poor decision-making). While I could go through the data set and produce a back-compatible one, to be honest I can't be arsed dealing with it; I'll just wait to see if they produce compatible statistics next time, and start again.