In September 2022, the Mayor’s office of Amsterdam decided to re-introduce a politically heated and previously abandoned police practice – the random weapon searches – when she ordered a pilot project to start up in November.
In the past, such weapon searches were criticized for being racially biased towards minorities and/or areas with a high minority population. To ensure due process and prevent ethnic profiling, it was therefore decided that “citizen observers” would observe the searches. The police union, however, disapproved and in turn sent out their own observers to observe the observers.
Consequentially, it was not without political tension when this practice was to start up again in the fall of 2022. Following the renewed debate, independent researchers, more precisely the NSCR, were asked by the municipality of Amsterdam to monitor the police’s behavior: to tag along and observe the controls in order to eventually, and based on objective measures, conclude whether members of ethnic minorities are searched more frequently than white people.
Initially, our research group, led by prof. Marie Rosenkrantz Lindegaard, was asked to accompany police on their shifts, join their briefings and observe the weapon search practice as it unfolds. However, we argued that this would severely compromise the validity of the study: how do we know that the police (and citizens!) do not change their behaviors when they know they are being observed? And assuming this to be the case, how can we then objectively measure the effect of ethnicity on the probability of being searched by the police? Following discussions within our research group, we concluded that we could not conduct truly independent research if we were to join the police briefings, drive with the officers to the scene and set up our observational study in their presence.
Norman Denzin wrote in 1968 that the goal of any science was the “advancement of knowledge and explanation. Any method that moves us towards that goal is justifiable”. Inspired by this idea, our team concluded that the only feasible way to “advance the knowledge” about whether there is ethnic profiling during random weapon controls was to covertly observe the searches. The American Psychological Association states that covert observation, when made in public, is ethically justified when it is not harmful to the observed. One of the ways to ensure this, is to anonymize the observed people. We ensured anonymity through our strictly quantitative coding instrument in which we recorded only static characteristics based on our visual assessment (in this case: assessed age, assessed gender, and assessed ethnicity). In a next step, we asked ourselves – is this research even possible? Can we really “spy” on the police and remain unnoticed? So, we went out to find out – and agreed in advance that the observations should stop when the officers spotted us.
I observed such weapon searches on two different days. The first day, a fellow NSCR researcher and I went out to find out: is it feasible to observe undercover, and how many observations (of people getting searched, and people present in the area) can we record before the police “bust” us? We received the location of the searches from a police coordinator via WhatsApp, and we headed out by metro. As we arrived, we quickly located the search area in underpass by the train/metro station nearby. We observed the police setting up their gear: around 15 officers, some with handheld metal detectors, others without, and a metal detector gate. To get an overview, we moved through the underpass once. Once on the other side, we decided it was best to split up to cover to most ground, and to get the most observations possible. My colleague sat down in the open right outside the entrance of the underpass. She had a clear view of everything and was able to code, with high accuracy, the age, gender, ethnicity, whether or not a person was with a child, and whether or not the person was controlled by the police. However, her outlook came with a cost, as she was compromised in a manner of minutes, when several officers were looking at her, talking to each other and clearly noticing her being “out of place”. This was later confirmed by the coordinator of this specific shift. I, on the other hand, opted to sit down by a portable bakery, a considerable distance away from the underpass. While I sacrificed some efficiency – I spent significantly longer to code a single person than my colleague – to stay “hidden”, it was later confirmed by the coordinator, that I remained unnoticed the entire time.
Based on this, and in accordance with the police literature, we concluded that it is possible to observe covertly if we are able to blend into the environment. “Doing nothing” would mean you are out of place, and the police would notice this. After all, they are trained and conditioned to observe, and indeed suspect, those who do not fit into the situational landscape.
The second day, another colleague and I went to an area to observe. The first thing we witnessed, as we arrived, was that the police had entered a bus to control everyone on it. They put up their signs – “the police are doing weapon searches in this area”. While observing the scene, we tried to “do something” to not do nothing. Soon after, the police had arrested at least one young man. Then they packed up and left. We were continuously updated by our program leader about the locations the police were heading to. In the first area they focused on, we were able to observe two search events. In both cases, the practice seemed to be limited to a very small, geographical space. In the first search event, they cornered off the outside eating area of a fast-food restaurant and the side street on which it is located. Struggling with how to blend in, we were relieved that several other on-lookers paid notice, and took up their phones, and thus offered us an activity to follow too while staying under cover. We thus stuck around to observe these searches. Shortly after, the police packed up and moved on.
It should be noted that a couple of areas where the searches would take place had clearly been communicated. This is another advantage of our method: we could observe everyone who could have been controlled for weapons, hence we could establish the ratio between potential and realized searches per individual category. Unfortunately, due to searches outside authorized areas, after two weeks any further weapon searches were canceled. Therefore, our analysis could not be conducted with a well-powered sample. However, we experienced and learnt a thing or two about doing observational research. More precisely, the practice of observing people in public space without their knowledge. The most important insight was, perhaps unsurprisingly, that “fitting in” was key. So: if you aim to conduct hidden observations, don’t bring your clipboard and just stand around! Feel out the environment, drink a beer if you’re observing drinking culture, wear a jersey or scarf if you’re observing football supporters, or do what the general crowds do if you’re observing in public space.
We also note that none of the observed individuals – the citizens that were stopped and searched, those who passed by and the officers we were researching – run any particular risks due to our research. Their identities were kept confidential by nature, and it would be practically impossible for anyone (including us) to approach any of them after we left the scene. This shows that when observational research, either overt or covert real-life observations, as well as video research, uses strictly quantitative coding instruments it can almost by default reduce the risk of harming the subjects’ privacy to near-zero.
Hans Myhre Sunde, PhD Candidate NSCR
Denzin, N. 1968. “On the Ethics of Disguised Observation”. Social Problems 115:502-504.
Notes observed selection procedures weapon control Amsterdam 2022
Full analysis: Citizen profiling during weapon controls in Amsterdam: An observational analysis of practices and potential biases
Interview with prof. Marie Rosenkrantz Lindegaard about undercover field work