56

Derek Chauvin had 17 complaints on record when he killed George Floyd. I have seen many articles, including this NYT article, which use this statistic as evidence that the whole department is corrupt. I don't discount that possibility, but it seems likely to me that every police officer, good or bad, would have many complaints. Every criminal they bust has a strong motive to file a complaint; there is a chance that the complaint will stick and will lead to them being acquitted.

So is the number of complaints a good measure of how abusive a police officer or department is?

I'm not saying Chauvin is innocent, just taking issue with NYT's extrapolation about the complaint numbers.

Ryan_L
  • 885
  • 5
  • 11
  • Comments are not for extended discussion; this conversation has been [moved to chat](https://chat.stackexchange.com/rooms/108909/discussion-on-question-by-ryan-l-is-the-number-of-complaints-filed-a-good-measur). – Oddthinking Jun 05 '20 at 02:55

1 Answers1

71

The association between the complaints against an officer and the bad behaviour of an officer is complicated, but, yes, there is a relationship.

This study looked at the productivity of officers - e.g. how many traffic citations, arrests for felonies, arrests for misdemeanors, made per day, and compared it to the number of complaints. They found a statistically significant correlation, but it was a weak one.* So, being "hungry" to seek out crime and criminals (as one high ranking officer they interviewed put it) may lead to slightly more complaints.

They also found young and inexperienced officers were more likely to get complaints.

*) According to Table II in the article, the total number of complaints was found to be positively correlated at a p < 0.01 two-tailed significance level with the mean number of felony arrests (Pearson correlation coefficient r = 0.327), misdemeanor arrests (r = 0.239), traffic citations (r = 0.178) and offense reports (r = 0.179), and at a p < 0.05 level with the mean number of field interview reports (r = 0.128), juvenile status arrests (r = 0.152) and no-report incidents (r = 0.142).

This looked to see whether the number of allegations against officers was predictive of the number of lawsuits in which the officers were named and the amount the lawsuits paid out in damages

We find a strong relationship between allegations and future civil rights litigation, especially for the very worst officers. The worst 1 percent of officers, as measured by civilian allegations, generate almost 5 times the number of payouts and over 4 times the total damage payouts in civil rights litigation.

So, it seems the number of allegations against an officer is a reasonable measure, but not a perfect measure, that the department will one day be successfully sued because of the officer's behaviour.

I note, with frustration, that the original NYT article gave very little to compare the number "17" against. I could see nothing in the article that would tell the reader whether 17 was a very high number, that should have alerted the appropriate authorities that there was a problem, or actually a very low number compared to officers with similar experience, or something in between.

Ilmari Karonen
  • 1,243
  • 1
  • 9
  • 13
Oddthinking
  • 140,378
  • 46
  • 548
  • 638
  • CNN reports it as 18 prior complaints, so there's that too. –  Jun 02 '20 at 21:32
  • 2
    It would seem around 100 complaints over a year is considered worst case scenario. https://www.chicagotribune.com/investigations/ct-chicago-police-complaints-met-20161013-story.html . In that story it states most misconduct is done by a small number of officers. –  Jun 02 '20 at 21:36
  • A few sources cite "over 20 complaints" against Van Dyke of the Laquan McDonald case. WaPo says 17. About the same. He was charge with 1st degree murder. –  Jun 02 '20 at 21:45
  • 8
    Found what looks like a good resource that could hopefully answer whether 17 complaints is a normal number of complaints for an officer to have. [Citizens Police Data Project](https://cpdp.co/). I'm having trouble using the resource, but it lists quite a few officer's name with many more than 17 complaints. –  Jun 02 '20 at 21:53
  • 2
    So, this answer focuses on civil remedies? Since many of those are settled and otherwise solved without assigning blame, "not perfect" indeed. Did you find any information on criminal action, both in isolation and as it relates to complaints? Only prodding because of the implication that complaints of all types can be used to predict *criminal* behavior, as has been charged in this recent case. –  Jun 03 '20 at 06:59
  • 1
    @fredbend: Yes, the second paper focuses on civil remedies. Settlements are included. I didn't find information about criminal convictions versus numbers of complaints. I imagine the sample size is much smaller. It doesn't seem to be that the claim is specifically about criminal behaviour, but this might be an interpretation issue. – Oddthinking Jun 03 '20 at 12:44
  • 4
    @fredsbend I haven't looked through the data, but it'd also be interesting to see number of complaints normalized by time, if possible. 17 complaints in 1 year would seem a lot worse than 17 complaints over 20 years, for example. – Nuclear Hoagie Jun 03 '20 at 13:48
  • 1
    @fredsbend You have to make do with the data you have. It's not like you can perform controlled experiments on this. – Barmar Jun 03 '20 at 20:00
  • "They also found young and inexperienced officers were more likely to get complaints." I wonder how it changes things if you take all the same data from the study, and throw out complaints from prior to an officer's 5th year. I bet suddenly the correlations get even stronger. – Joel Coehoorn Jun 04 '20 at 16:45
  • 1
    @fredsbend I think you'd also find that there are too many confounding variables in *convictions*, in particular that DAs are often not inclined to prosecute police officers for anything less than the most heinous of actions, and juries are often not inclined to convict police officers. This undoubtedly varies, both across jurisdiction and across time, so any data from convictions would be hard to assess. Civil lawsuits will also have some of this noise, but much less (as they have no prosecutor to convince to bring suit), and with much larger n size. – Joe Jun 04 '20 at 18:08
  • @Joe *"DAs are often not inclined to prosecute police officers for anything less than the most heinous of actions, and juries are often not inclined to convict police officers."* ***citation needed**. Granted, there's a certain level of immunity officers are given, but that's still a big claim, especially since at the moment only correlation is necessary. –  Jun 04 '20 at 23:51