 OK, so hi. My name is Rika. I'm a lecturer at the University of Manchester. The stuff I'm going to be talking about, I started doing while I was doing my PhD in London. And my approach is not so much looking at the use of Fix My Street, but something that came up in one of the discussions earlier today. Actually, it was like, what are people doing when they're participating in these platforms? And what kind of inferences can we make about their experiences from the data that they generate while they're participating in these sorts of things? And in particular, I was interested in people's perception of disorder in their neighborhoods and also their ability to act as guardians. So to give a little bit of context or background, the realm that I work within is something called crime science. It sounds weird. It's basically an engineering approach to the study of crime. So really looking at what it is about the situation that creates an opportunity for crime to occur, rather than things like personal history or reasons for offending those sorts of things. So we're really looking at the time and the place where a particular crime happens and what can be done, introduced into that environment, to block that opportunity for crime to occur. And my particular interest within that field is to look at, can we apply that kind of situation specific approach to people's perceptions of crime and perceptions of disorder and what people are experiencing as they go about their everyday lives? Another thing that I get quite excited about is sort of these new and emerging forms of data. And I don't know how new this is anymore, but things like data from online interactions, crowdsourced data, tracking data, satellite imagery, those sorts of stuff. So something that gives us a slightly different insight than the traditional social science approach of surveys. So that's kind of my interest. And in particular, I want to know, okay, what can we do with these new and exciting forms of data that can tell us more about people's experiences and that to do with perception of safety, crime, and so on? So traditionally, these exciting forms of data tend to be reduced to an approach to measuring the ambient population in an area. So this might sound slightly trivial, but it's actually a relatively exciting problem in criminology to try and estimate an accurate denominator for calculating crime risk. Because if you think about it, the best way of calculating crime risk would be to use a dumb nominator that best approximates the number of opportunities present. So if we're talking about pickpocketing, we want to know the number of pockets to be picked in an area at a certain time. And traditionally, you would use things like residential population or workday population, these sorts of things, which are not always the best representations of the opportunities. So these new and exciting forms of data have been used to approximate how many people are there that can be robbed or crime offended against. But my question was that I really wanted to look into is that is it important what people are doing while they're there? So where and when people are, is something that's important? But depending on what they're doing, it might mean different things. So for example, if you consider this guy, he's tweeting on his way to work about his coffee. This sounds like a pretty good pocket to be picked, right? So in London, mobile snatching of, moped snatching of mobile phones is a pretty big issue. It'd be relatively easy to take this guy's phone, right? So this is a good sort of vulnerable target or suitable target for a crime to occur, right? On the other hand, if you consider somebody who's walking around in their environment paying attention to it, and it's something in that environment that causes them to then take out their phone and participate in something like Fix My Street, that is a qualitatively different indicator of what that person can be in that situation. And it's possible that their interactions with their environment can actually tell us something different than just they are there and their phone can be stolen. So I looked at data from Fix My Street, which I am assuming everyone is familiar with, just in case, here's a screenshot. So it's a problem reporting website. People can go and log their problem and give it a little title and a description and you get location, time, and so on. So I remember I asked for this data back in the day and I was very excited to see that data slide this morning if it will be available. At the time it wasn't, but luckily your reports are sequential. So I could just iterate through each one. So it's report one, two, three, four, five, one, two, three, four, six, one, two, three, four, seven, and so on. So thank you for that. I'm sure it was there intentionally for me to scrape. So I went through and got data for all of the UK but I decided to focus in on London. So I looked at over 55,000 entries in London. There were more. My scraping potentially introduced a lot of noise so I had to exclude some reports that way. But I thought that's a pretty decent end to go for. Most of them were like this. So most of them are reports about potholes and broken street pavements and so on. So when I'm talking about people's perception of disorder in their neighborhood, I'm not sure that's the best kind of indicator of that. But more exciting for me, around 30%-ish were about incivilities. So incivilities in sort of criminological literature, there's signs of disorder in the environment. And the Home Office defines them as environmental antisocial behavior. So basically, antisocial behavior against the environment, which includes graffiti, littering, I think dogfouling is in there, all these sorts of things. And one of the really interesting things about trying to measure this from this crime science approach is that if we send out somebody to go survey the environment, they will record every single instance of graffiti or every single instance of litter. But the nice thing about Fix My Street is it's reported by people who've encountered this and qualitatively assess this to be a problem. So they are not noting the nice mural on the wall that they are happy with or the litter that is always there because it's just there. They're recording things that are out of the ordinary and that could be interpreted as signals of, okay, there's something here that I'm unhappy with. So there's this concept of signal disorder, that signals that something's not quite right in the neighborhood. And I was trying to look at whether we can use Fix My Street to look at people's experiences with these events. So the first thing I wanted to look at with these, so now I'm focusing in on the incivilities, is the time of reporting. So there's an assumption made that time of reporting is when they actually experience the thing. I did some shady calculations where I looked at if broken street lights were reported when it's dark and they tend to mostly be reported when it's dark. So I use that as a good verification that I can use this time as time of experience. But more excitingly, when I looked at when people were reporting, just generally reporting in the day, it reflected people's routine activities. This sounds like a very no-duh kind of moment. But remember that this is data of people reporting stuff on a Fix This website, right? And this, but them doing this as they are out and about doing everything else and this reflecting their actual routine activities is pretty exciting. And this happens to the extent that you can see changes in routine activities in the reporting data. So I compared Weekend and Weekday reporting and you can see that on the Weekday, you get up early, you go to work and on the Weekend, you have a bit of a lie-in and you stay out a little bit later. So this again, this isn't data of when people are out, this is reporting data. So I thought that was really neat. So based on this, okay, well when are people reporting incivilities, right? Let's have a look at the percentage of all these reports that are these incivilities during time of day. And as fear of crime and all those sorts of literatures which suggest, I was like, it'll be evening, it'll be early morning, it'll be at times when it's dark and scary to walk down that back alleyway. So my peak time was 7 AM and I wasn't quite sure why. That was not what I was expecting at all. So I went into the sort of qualitative detail associated with these data and it turns out that the people reporting at 7 AM are people who are on their way to work in the morning and they're seeing signs of the night before, right? So they're walking through the alleyway and that, oh, there was definitely street drinking here last night. That kind of scares me. That's not the activity that I want associated with my favorite back alleyway kind of thing. So that was really interesting because it's not necessarily the street drinking that's the problem if we're thinking of people's perception of safety. It's actually seeing the signs afterwards when these people walk through the space who are perceiving this. So that was interesting about time. I also wanted to look at sort of spatial distribution of where reports are being made and looking at where there's high density, low density of reports. This tends to just resemble a population map, but there we go. But the interesting thing was here is that I started to separate out these by the type of incivility. And it tends to generally, again, correspond with routine activities. So you can see dogfowling in the more residential areas and litter, if anyone's familiar with London, that's sort of the King's Cross and Pancras area of, there's a lot of litter. It's a busy train station, so that's why. But if you look at the corner there where it says abandoned vehicle, we get abandoned vehicle hotspot in Covent Garden. So if anyone doesn't know London, this is Covent Garden. It's a shopping area. It's pedestrianized. It's not really a place to abandon your vehicles, because this is what generally abandoned vehicle should look like, right? So again, it's like, okay, this is exciting. Let's go into the reports. What's happening there? And this is what people are reporting as abandoned vehicle. And I mean, technically, they're right. It's a vehicle. It gets you from point A to point B. But I think technically this is fly parking, and which falls under fly tipping category. But to people, this was a vehicle, right? And so it's very important how people interpret the categories that they're being provided. And it's not necessarily going to be the same way that we would have intended. So my sort of lessons from this are that there's a motivated perceiver who is important in terms of their perception of the environment and them coming across the signal. So it's when they see them and also what they decide to define as this issue for them. So one of the sort of downfalls of these new forms of data and so on is that you don't always have all the data that you want, right? So I was very interested in who participates and how they participate. But when you report your problem on Fix My Street, you don't get asked about your age, your gender, your ethnicity, whatever other demographics researchers might want. But some people do leave their name. And there's ways of inferring this information from a name. And that's again a bit messy and probabilistic and so on, but it's better than nothing. So from a name, from a first name, you can infer gender relatively well. So that's quite strong. Age, because there's trends in baby names. I don't know if anyone's seen the famous baby names graph. You can start to infer age from the first name from the surname you can infer ethnicity. So based on that, I had to look at the names left with the reports. So two thirds of the reports were left anonymously. A quarter were male names and 8.6% were for female names. So there's two conclusions we can draw from that. Either men report more, which I think has been found in these kinds of data. Or women report more anonymously, right? So maybe they're less likely to leave their name. So I had to look at what different people are reporting. So men were more likely to report things like potholes, pavement and road issues. Whereas women were more likely to report litter, dogfouling, parks, public toilets, things that I think were more associated with walking. But I know nothing about gender and use of space. So if anyone does and is interested, then do come speak to me. The interesting thing to me was that graffiti in abandoned vehicles, so two of these environmental, anti-social behavior and civility, they were reported more anonymously than anything else. So when people are reporting these things, they don't necessarily wanna leave their name because maybe they're worried that their local teenager tagging their window is gonna come back and tag some more. So that was one thing of trying to infer people's perception of the neighborhood. The other thing is, okay, so when these people are walking around and they're looking for these signs and taking stock of these signs and so on, can they also act as a sort of guardian of their area? Are they monitoring their environment? Are they out there as a sort of neighborhood watch kind of thing? And the traditionally research into guardianship tends to look at guardianship within a neighborhood. So you can think of a guardian as someone who by their presence can block a crime opportunity from happening. So we often talk about stay at home parents and burglary. So if somebody stays at home during the day, then their house is less likely to be burgled just by virtue of them being at home. So the burglary will choose a empty house somewhere else. And these are always measured with surveys and they tend to be associated to one neighborhood. And in the UK, one measure of neighborhood is a lower super output area, which is a small sort of spatial area that we use as a proxy for neighborhood in research. So I was wondering to what extent do people stay in their neighborhoods, right? Is that true that I'm going to be a guardian in my little neighborhood, my area, and then probably not so much elsewhere? So to look at this, I had a look at the, again, I took the people who left their names with their reports, and I took the top 1% of reporters, which is actually like a wide range of reports. So people who left 20 reports or more were in the top 1%, but the top 1% also included people who left 800 reports. So there's a bit of a range in that top 1%. But I took their reports and I wanted to see, okay, where are these people reporting? How far do they go? And I actually found different spatial patterns. So that was actually really exciting. So we have people like this. So this person, they have a main neighborhood, presumably where they live. And then they also have a lot of other neighborhoods where they report, right? And there's some hubs. So maybe where they work, where their friend lives, where their partner lives and so on and so on, but also as they're traveling amongst these. And so these people who have this spatial profile of reporting, they have a high connectivity score. So the LSOAs that they report in, which is each one of these, tend to be connected. So they almost have this one neighborhood, but it's a pretty big neighborhood where they're reporting in. So that was one spatial profile. The other one that I found was actually people who were reporting in much more sort of scattered neighborhoods. So they had this sort of main area, again, probably the LSOA that has their home in it. And then the neighboring ones, where they go to do their shopping and so on. But then also there's reports in other places where they probably had to travel to. So these people potentially have a very different approach to their perception of themselves as guardian of this environment, right? So to them, wherever they see a pothole, they will report this pothole, right? Whereas maybe for this person, they take ownership of these areas, these connected areas. And this is where they will pay attention, but outside of that, they don't really care so much. So as I was looking at this data, I also thought that there are quite a few points of caution to raise. So I'm saying I was speaking very much about the strength over traditional social science data, collectivist surveys and so on. But there are other limitations that need to be kept in mind. And one of these comes to the mode of production at this data. So the sample, if we're treating these people as a sample, they're a sample of involved and interested citizens, right? It's a self-selecting sample. And so it's gonna be people who are particularly interested, who select because of a variety of reasons. And it's their voice that is going to be heard through these data. So that's something to keep in mind. We can't be used to sort of make generalizations about everybody who lives in that neighborhood and so on. And the other interesting thing is the how, then this history for all crowdsourced data is that how the Pareto principle is hugely amplified. So the Pareto principle is referred to as the 80-20 rule. So it's in all sorts of social science in criminology. We often say that 20% of street segments account for 80% of the cost of police and so on. So it's just a law of concentration. But in crowdsourcing projects, they're hugely amplified. So for example, who's ever been on Wikipedia? Okay, who's contributed to Wikipedia? Okay, so we're hugely overrepresented in this room, which is kind of to be expected considering the location we're in. But actually, considering Wikipedia, so 99.8% of people go to Wikipedia to read the article. 0.2% of people participate very occasionally or like contribute. And it's 0.0003% of users who contribute most of the content on Wikipedia. So I quite like that, that's pretty shocking to me at least. So it's the same for Fix My Street as well for named participants. This is just a Lawrence curve, sometimes used to plot inequality and income. But it shows you that the majority of people, they contribute once, so that's where you see the straight line. So most people contribute, they send one contribution, then two, then slightly more. And it's the small top percent that account for a large volume of the contribution. And also, who contributes is not random either. So there's all sorts of factors associated with increased contribution, one of those being wealth, I guess. So in the UK, there's this index of multiple deprivation which ranks all the neighborhoods from the highest deprivation score to the lowest deprivation score. And so I tried to plot this cloud of number of reports by the rank on this index of multiple deprivation. And we see a lot more reports from better off areas. And again, this is the same across variety of crowdsourcing projects. There's also a lot of ethical considerations. So all the discussion around privacy and surveillance is very important here. So I was able to take these people's names, link it with other data to get their age, to get their gender, to get all sorts of things. I could have potentially gotten some other data that I might have had access to and link it to that and so on. And that's something that maybe they didn't consider when they started participating. Also data ownership, who owns the data, how can it be used and so on. The linkage ties back to the first point there. And also this implication of funding. So as I said before, it'll be sort of the better off people who will be more active in these platforms. And if we use this, for example, to task services for street cleaning or something like that, this would be very much tasking services towards people who are shouting the loudest and not necessarily towards people in the areas where these things are actually happening the most. So that's just something to keep in mind. So I think I'm over time, but I'll take questions. I'll take the one time. All the questions at the end. So.