Wednesday, April 28, 2021

How bad are the police at the OIA?

Te Kawa Mataaho Public Service Commission published their updated OIA statistics last month, showing that most public agencies were meeting the basic timeliness requirements of the Act. One of the weird features of these statistics is that the police regularly report amazing timeliness - 95.3% on time, from 36,440 requests! - which causes bitter laughter from anyone who ever actually requests information from them. Because the lived experience of people who request information from police national headquarters (PNHQ) is that requests are invariably late, usually by months, and you might as well just cut out the middle-plod and ask the Ombudsman directly. The suspicion is that this is because the police handle a huge number of routine law-enforcement related requests at a district level - insurance requests, for example - which they do well (because they are routine, accepted as part of the job, and part of the normal workflow), which hides their appalling performance when they get a non-routine request (say, a journalist, blogger, or researcher wanting something from PNHQ).

How could we prove this? My first thought was to get a sample of requests to PNHQ and compare their outcomes to the stats they report to the Public Service Commission. FYI, the public OIA request site, provided a source of data. The site has been used to make over a thousand requests to police, all of them through PNHQ. There's no reason to believe they are not a typical sample of such requests. So I spent two weeks worth of evenings hand-coding every request made to police using FYI in the 2019 and 2020 calendar years for timeliness and outcomes, with reasons for refusal. It painted an appalling picture: 34% and 36% on-time in 2019 and 2020 respectively, with dozens of requests apparently ignored and still outstanding, and median response times of 26 and 28 working days respectively. But was it representative? One of the requests via FYI was for the police's OIA training manuals, which included the user guide for their Information Request Tool, the software they now use to track all OIA and privacy requests. Which showed me I was doing it the hard way. Once I knew the right questions, I could just ask!

The released information shows that the initial suspicion is correct: the police's OIA stats are dominated by huge numbers of routine requests handled at district level. Some of these types - traffic crash reports (for insurance purposes) and speed camera requests (for deciding who is paying that fine) - have 100% on-time rates, which I guess tells you what the police care about. As for "real" OIA requests, sent to PNHQ, the stats are absolutely dismal:


[These numbers differ from those calculated from FYI data, slightly for 2019 and more seriously for 2020. From information on requests received, FYI was responsible for 10.6% and 8.3% of total requests received by police in 2019 and 2020 respectively, so the spreadsheet might not be entirely representative. Also, the police are really bad about received dates, and often try and scam extra time by basing their deadline on the date they bother to read their email rather than the date it was actually received]

These are appalling statistics. And they're completely hidden by Te Kawa Mataaho's rudimentary reporting, which turns the police's utter failure into success (the police seem to have boosted that "success" even further in 2019 via a change in what they chose to report. But that's another post).

So what can be done? The police are outside the usual oversight mechanisms for general OIA practice: they're not part of the public service, so TKM can't touch them, and the Ombudsman can only inquire into complaints about specific requests, and cannot review their general administrative practice as they do for other agencies. In theory the IPCA can inquiry into "any practice, policy, or procedure of the Police", but I suspect they'll be uninterested (I may try anyway). Which is how they get away with it: because there are no effective watchdogs on their performance.

But there is one immediate thing which can be done: the Public Service Commission can stop repeating the police's lies. Since they're reported seperately anyway, they can demand the police break out their stats by district and type, as they do above. This would allow the statistics to fulfil their purpose: to allow areas of weakness to be identified and targeted for improvement.