 Okay, well it's bang on 10, so I'm going to get started. So hi everyone, my name is Louise Kaepernick and I'll be leading the Computational Social Science Introductory workshop today. So let's get into the actual content. So throughout this workshop, we're going to be covering three main sections, which include, so the first one is what's up with computational social science? So what does this even mean? Then we're going to discuss a bit about how you actually become a computational social scientist and finally we'll cover the eight steps of CSS, which is where we'll have the most interactive part of the workshop and where you can sort of get stuck in sketching out your own computational social science projects. And of course at the end there'll be time for final thoughts and any questions that you might have. So we have our definition here, so computational social science is the use of computational and empirical methods to address social science questions. So we can break this down a little bit more. So computational social science requires some human thinking and that's to identify important research questions. And because we're going to be dealing with social science questions, we're going to need to understand how people behave and what they want and what they want to achieve. So we need that social science brain in order to formulate those really interesting research questions. And you can see as well that we're going to need a little sprinkle of computer thinking and that's so that we can turn these research questions into computational or empirical methods. And again after your research you're going to need that sort of social science human thinking again in order to effectively communicate those results to other people. So you can see it's this blend of computational methods and social science thinking. Another way to understand computational social science is to think about what its main components are and we can do that through thinking about what computational social science is not. So it's not just using computers within a social science research project. I'm sure all of us have sent an email at our jobs or doing some research, but that doesn't necessarily mean that we've been doing computational social science. And it's also not just using digital versions of purely traditional social science methods. So you know it's not just asking for survey participants to complete an online form instead of handing one out. So yeah it's not just a digital version of something that you would normally do in your social science research. And I'm going to give you a few examples now that are going to sort of untangle these ideas a bit more because I know that they're not always entirely obvious. So on the screen we can see some examples that might help pop this into context a bit. So our first example of a CSS project would be collecting, processing and analysing millions of online news articles to show changing political attitudes. So you can see how we have that human thinking here in order to formulate that social science research question, which looks at political attitudes, but we also have that computer thinking inherent in using a computational method. So in this case the computational method would be web scraping and that's what we use to crawl the web and gather a massive amount of news articles and sort of put them in a data set. We could also, for a second example, we've got using real time weather and traffic data to show how travellers react to events. For example, maybe you hear that there's a storm that's caused damage to a local town and you want to look at how people react and how that event is dealt with in real time. So you can see how this project isn't just a case of using a computer with a social science project or it's not just using a digital version of traditional social science methods, instead we have these uniquely computational methods. And we have some other examples which include combining data from novel wearables or apps that you can think of like Apple Watches, Fitbits to establish a correlation between social media activity and heart rate. So that could be aiming to answer maybe a social science question about how people feel about certain images. You could look at how they react, are they mostly positive or negative feelings, are they feeling irritated and so on. For our final example, we could also be interested maybe in mapping family names over time and we could do this by importing, processing, formatting centuries of parish records that could allow you to explore the movement of certain families in different areas. Or you could maybe look at the changes in family size or whether families have moved away or not. So these are just a few kinds of broad examples of computational social science projects. Moving on, there are key factors that make a CSS project computational. And the first one of these is data volume, complexity, speed, difficulty or novelty. And this is more important than the exact data source or type. So in our previous example, where we had parish records, the source of data is not entirely important. It's about the volume and the complexity. And you know that difficulty or the novelty of that data. Additionally, another key factor, like I mentioned before, is that with a computational social science project, the data must pertain to people, actions, behaviors, choices and statements. That's that key social science ingredients that we need. And just to hone in on that point, the research question should be a social science research question, which uses that atypical data to talk about how people make decisions or what influences their behavior and choices. So the exact research question is not important, but it must be a social science question. And that's where we have this really obvious intersection between the computation and the social science. And that's what we're going to be focusing on today. And here we have a nice little quote, which says, in essence, CSS is an opportunity to do socially valuable research that would not be possible without computational methods and tools. So by this, I mean that we couldn't, for example, manually scan, let's say, years of police recorded statistics if we wanted to try and understand how crime rates have changed over the last 10 years. This physically wouldn't be possible to do manually. But with the use of computers, we can apply advanced statistics and models in order to understand that change in crime rates. You know, we could look at how we could count many types of crimes that have taken place in certain areas and then aggregate these crimes to map them spatially. We could look at exploring the long-term trend seasonality or the noise components of these different crime types. And this type of research just wouldn't be possible without the use of those computational methods. Another example could be, as I mentioned before, web scraping millions of online articles. And you could be doing that to try and understand maybe how political opinions have changed over the last 20 years. In order to get a sense of political opinions, we want to examine the words and the articles and how many words perhaps belong to different categories. And then we can look at what sort of themes are starting to appear. We can look at the proportion of these words as well and how that changes over time. So again, that's something that we wouldn't be able to do without computational methods that I've mentioned like web scraping and natural language processing, which is where you look at, you know, the words and you can track their proportion over time or their sentiment. So in order to see if any of that has made any sense at all, what we're going to do now is have a little bit more interaction. And I'm going to give you guys the chance to vote on whether you think a given project is a CSS project or not. So is it computational social science? So if you want to head back over to Mentimeter, we can get this set up and then you can start voting. So you can see that the code's gone back on the screen if anyone needs it. And I'm sure if you do need it, Emma also is going to post it in the chat. So our first example involves a project where we want to scan historic recipes and use AI algorithms to recognize text with the aim of identifying ingredients and measures used over time. So what do we think? Is this computational social science or not? I'll just give that a minute. So OK, nice. It looks like the majority are saying that it's not enough. There's not enough social science in this question. And I would I would tend to agree. There's, you know, it could be. We'd have to know more. But from this once sentence that we've got here, it doesn't seem to be a social science project, as we're not really asking any questions about people's behaviors and choices. You know, perhaps if we were talking about how imported foreign foods integrate into domestic scenes and how people then take that up, how it relates to wealthy people, then adopting a fancy foreign things is their new favorite dish. It could be exploring how that trickles down, and then those foreign foods become more normalized and accepted. You know, that could be you could see it could be a social science question. But if we're just talking about maybe the proportion of natural foods that people are eating over time, that actually could be more of a biology question. So, yeah, I would actually tend to agree with the majority here that it's there's not enough social science in here for us to say that it's definitely CSS. We do have a few people saying that there's not enough computation. I wouldn't expect that to to be the case word, because if you if in the question, we can see that it does sound quite computational when we're talking about the use of, you know, artificial intelligence, algorithms and the use of, you know, scanning recipes, it seems like we've got the computational component pretty down here. So, yeah, people saying they need more information to decide that I'd say that's completely fair enough, because like I said, there could be a social science question implicit here. You know, there could be one, but we don't have enough info to truly say. OK, so let's go on to the next one. So is this CSS or not using gamified smart home displays in order to understand how people interact with energy saving technologies? OK, we've got some people saying that they need more information to decide, which is fair enough. But it looks like the majority is saying that this is definitely CSS. And personally, I'd have to agree with that. But again, you know, as with the other question, it does depend. It is hard to capture the nuances of a research project in just one sentence. But I would think that, you know, we've got the computational component down with the gamified smart home display that's probably going to capture a lot of digital data about how people and how quickly people react in terms of looking at what time of day it is or who it is that's interacting. It could also possibly act as some sort of tracking data. And again, that would make it very computational. And we've got that social science question implied as well, because we're looking at people's behaviors, right? So we're looking at how people interact with these energy saving technologies. And if the purpose is to try and change or modify people's behavior to make them engage more with those energy saving technologies, then, yeah, this is definitely going to pose a lot of social science and research questions. But yeah, it's very interesting to see that there's not as much division as compared to the previous question. Seems like we're more in agreement that this is CSS. There is some people saying that there's a few people saying it's not enough computation, I think, given that we have the gamified smart home displays and we're going to be sort of analyzing the data that that's captured. I'd say that's that's probably definitely enough of a computational component there. Let's go on to our next one. So here we've got a project which involves advertising for survey participation on social media with the responses being stored in a database. What do you think of this one? Got a few answers in. It seems so far we're saying that there's not enough computation, not CSS, not enough social science. Some people saying they need more information, fair enough. Let's have a look. It seems that the consensus is that we don't have enough computation here. And yeah, I'd also have to agree with that. So if you remember on a previous slide, we talked about how CSS is not just using digitized versions of traditional survey methods. But this line can be hard to draw because there is kind of no clear line between how much computation is enough computation. But a question to ask yourself is if you can do your research without digital tools, then it's probably not a computational social science project. And if we draw the conversation back to surveys in itself, typically the larger the survey samples, the more necessary that there's going to be for computational tools because the analysis will become much harder. But the use of online surveys doesn't automatically make it a CSS project, as you could probably collect roughly the same amount of responses through paper surveys as you would online. But it's interesting to see that a lot of people have said as well that they need more information to decide because that's fair, in that we don't know where this research is going from here. And again, we've got a substantial amount of people saying that there's not enough social science here, which is also valid as we don't know what research topic this is focusing on. So it could be that we're just collecting responses about participants' weight and height, for instance. So that is also equally valid. Let's go on to our second to last project here. So yeah, if you're feeling like you're totally questioned out, don't worry, we're going to be done soon. So this example looks at reading in real-time weather and air pollution data to create complex models of hyper-local air quality. So it seems that we're getting the majority of people saying that there's not enough social science here, which is interesting. And I think I would have to agree. So what we'd have to think is, is there a link between hyper-local air quality and human behavior? It's tough. I don't think I could say that that is just based off of this one sentence that we've got here. We'd have to know more, are we looking for a political impact or is there going to be an overlap between different research fields and schools, maybe with geoscience, perhaps, or with physics? It's not clear by the statement that there's an intent to explore human behavior. So I think I'd probably have to agree that there's not enough social science in this statement. We have a few people saying that, or a few saying that it's definitely CSS. I would just remember to look out for it. Is there, is there an implicit social science question here that focuses on, you know, people's actions, people's behaviors? It is one sentence, so it's fair enough to say that you need more information to decide. But in terms of there not being enough computation, I think, I think we've, I think in terms of like reading in real time, whether an air pollution data to create complex models, I think we've probably nailed it in terms of the computational component here, such as something to think about. OK, final one. So here we're looking to train a neural network on social media data. The aim in doing this is to create a believable chatbot that counters online radicalization. So what do we think? We've got one person so far saying it's definitely CSS. Looking like the majority of you agree that this is definitely CSS. And again, I think I'd have to agree with you. I think this is probably one of those quintessential CSS projects. There seems to be this implicit social science research question, as we're focusing on people's actions, attitudes and behaviors, and we have a computational method to help us explore this. So we've got a neural network that has been trained on social media data. For those that are maybe unfamiliar with this term, neural networks are a kind of artificial intelligence and they teach computers to process data in a way that's inspired by the human brain. This type of machine learning process is also called deep learning that uses interconnected nodes or neurons in a layered structure that resembles the brain. So this has been implemented in the hopes that we can use it to predict the user's behavior based on their social media posts and their messages. So in order to train this model, there's probably been a fair bit of web scraping involved as well. So very computational stuff. But yeah, in terms of we've got people saying maybe there's not enough social science. I think, as I've said, because we're focusing on sort of counteracting people's behaviors and online radicalization, I'd say that there's probably definitely a social science research question implicit in this. And talking more about the sort of computational components, in order to count online radicalization as well, we're going to need to have a good idea about what online radicalization is and how people are radicalized and what kind of mental processes are going on. So that's, sorry, it's probably more social science. But in order to sort of break that down, you can put methods in place from an actual language processing to get more of an idea about the breakdown of language. So if you're going to be countering online radicalization, you're going to be maybe looking for keywords and you're going to want to have maybe some analysis there. So yeah, we're using these kinds of complex computational methods to mimic and interact with individuals. Then that's a clear example of, you know, when natural language processing is going to be involved. And because of that, for me, that this is clearly a computational social science project. So that was our last example. And a fact to everyone for taking part in that, I think it is useful just in terms of exploring the source of components that make up a CSS project. Now we just have a little word cloud. So if you maybe go ahead and put, you know, up to three words about what you've learned about CSS so far, maybe it's something that surprised you or something that you found interesting or anything that you're still curious about, everyone will then be able to see your responses as they build up and flow through. Then what we're going to do is we'll have a little five minute break before the next half of the workshop. Don't feel pressured. You don't want to put anything. That's no worries. OK, nice. So overlaps with AI. Nice. Yeah, we definitely have some of that with these projects. Challenging, yeah, definitely. I can say it's someone that went from primarily, you know, been someone who studies social science to move into more computational methods. It's definitely challenging, but interesting. Yeah, we've got AI that's appearing quite a lot, big data. Yep. So like I've said, it's often if it's not something you could manually do yourself. Like I said, if we're looking at, you know, analyzing loads of police records or, you know, scraping millions of online articles, it's not something that you can manually do yourself. So you're often working with quite big data interdisciplinary. Yeah, for sure. You're often going to be straddling, you know, those different research fields. If you're going to be looking at something more computational, many ways to apply this. Definitely data scraping. Yep, endless possibilities for sure. Crucial ingredients. Interesting complexity is a popular one, too. What skills would I need? Still need more info, all fair enough. And hopefully in the next chunk of this workshop, I'll be able to give you that network analysis, widespread, programming skills. Yeah, for sure. OK, so what I'm going to do now, so 25 past 10, we're going to take a little five minute break. OK, hopefully that was enough time for you to just have a little little break, maybe get a brew. Now let's get back to the slides. OK, so we're going to talk a bit about how to become a computational social scientist. And first, we're going to start by covering what a social scientist is. So social scientists think like people. And what I mean by this is that they use a lot of human type thinking skills like abstraction, inference, and they understand fuzzy, fuzzy concepts and background knowledge. And generally as a social scientist, you know, you don't shy away from gray areas or overlapping categories. And these are just, you know, part and parcel of being a social scientist. And it's no surprise that they need these human type thinking skills because they study people, interactions and behaviors. And that requires a certain skill set. But social scientists also do build up a fair bit of data skills in the course of their research. So if you think about things like response, categorization, encoding, quality evaluation, pattern detection and statistics, they actually do have quite a bit of experience with data skills. But whilst, you know, they often do use computers, this often doesn't involve writing computer code. Instead, it might involve computer programs such as SPSS or Stata for statistical analysis. But like I said, not much computer code. And then we have computer scientists. And again, you know, obviously we're making some big generalizations here. But in contrast to the sort of human type thinking skills that social scientists have, we can say that computer scientists tend to have to think like computers. So the thinking skills that they have are more along the lines of working with concrete definitions and absolutes, thinking in terms of strict hierarchies and categories and clearly defined and scoped variables and rules. And in terms of data skills, computer scientists collect, analyze and manipulate data often through programming scripts and computational methods and technological tools. But unlike social scientists, they're not usually taught to identify or motivate research projects on the basis of societal impact or value. So those of you that are social scientists might about to justify your research, you know, on the grounds that it will make the world a better place, right, or it'll contribute to, you know, knowledge for good. And it could be that you're, as I've said before, in a previous example, maybe you're researching radicalization in online forums and you're doing that to produce insights that could lead to countermeasures. Whereas a computer scientist might be more focused on a logical justification for a particular project. So maybe they say, well, I want to make this algorithm more efficient so that it uses less memory. So in order to do computational social science, you're going to need a blend of social science and computer science skills. And you're going to need these four things in particular. So you're going to need those human type thinking skills that I've mentioned before. You're going to need a sprinkle of computer thinking and a great deal of open mindedness. And you're going to need mixed problems to work on. And you're going to need those skills that we briefly touched on before. So those are skills like being able to identify important problems or knowledge gaps, considering possible solutions, connecting problems to relevant theories or perspectives and then being able to collect relevant information and research to frame your approach. And these are all things that social scientists excel at. The ability to understand context and nuanced perspectives, how to communicate abstract ideas and how to really attack a research question. Whereas this might be an area where computer scientists may struggle as they're more used to those concrete definitions and absolutes rather than, you know, these gray areas or more murky social science concepts. So that's going to be the first thing you need. But you're also going to need that computer thinking. So you're going to need the following skills. The ability to access, organize, process and handle vast amounts of data and complex data. You're going to want to know how to write collaborative code and also how to properly document your workflow, which is often a step that people neglect. These are all the skills that computer scientists find quite easy, as well as data scientists who have been trained in computational methods. But for social scientists, it can be much harder to transition towards these computer thinking skills. But as I've said before, you know, social scientists do have those data skills that they can build upon. And those are those things that I've mentioned before, like coding responses, pattern detection and statistics, formatting surveys. So, you know, if you are a social scientist, you shouldn't argue it as starting from a completely sort of like blank slate. You have those skills there that you can build upon. And this is a really important one. And I can tell you, as someone that has gone from, you know, a social science field to doing more CSS type stuff, that it can obviously be really intimidating at first. And that's why it's good to remember that, you know, no one's going to start out with all of the skills that they need. Nor are you going to start out with all the skills that they that they. Sorry, nor do they know all the skills that they might need to acquire. This happens quite often. So you might start off saying something like, well, I want to scrape tweets for information on the 2016 US election. And you're going to expect, well, I'll probably need some coding ability. But you might not know that that's going to entail learning about APIs or different file formats or ways to then visualize this data. But if you approach CSS with an open mind and a willingness to learn, then you're going to gain more skills along the way. And you'll also start to understand that some skills will have a steeper learning curve than others. And that's going to lead you, hopefully, to collaboration with those from other fields. You know, if you're looking at implementing some complex machine learning, you know, you can still do that in a CSR project, but you're not always expected to do it completely by yourself. A really important research skill is collaborating with other people from different fields. And that's where you start to get this bridge between the social science and the computing world. As social scientists learn more about computing and vice versa, we see more conversation and collaboration between these two fields. And what you'll find is that, you know, you don't need to know everything about computer science, but you'll want to know enough that you're able to have those productive, collaborative conversations with those in the field. So, you know, like I said, you're not maybe going to be expected to build a really complex machine learning algorithm. But you might want to familiarize yourself with the algorithm and, you know, maybe do a little bit of test code on it, on like maybe a dummy data set so that you're familiar enough with it and what you want to achieve with it. And finally, what you're going to need is a problem that requires these CSS skills, and that's a mixed problem. So that's a problem that's going to require that blend of human thinking and computer thinking. So some of you may be here because you've already encountered one of these problems. And I wouldn't be surprised because, you know, as resources are becoming more digitized, these problems are going to become more relevant and more important. You know, the fact that interactions, objects and processes have become in smart or networked. The fact that large volumes of data are now being made available and updated much quicker than ever before. It means there's much more opportunity for interesting research. If you think about a classic social science problem, you know, maybe we're interested in how men and women move through cities differently or maybe how people with disabilities navigate cities. Traditionally, for this kind of social science problem, you might station some interviewers in different places to stop people as they go past or to count how many people go by that are using mobility aids or maybe you'd send our surveys to people's houses. But now there's much more opportunity for us to gather a large amount of data with these novel computational methods. So you could collect data from public transport networks about how many people bought tickets or swiped their card at the tram stop. You could even get sensors, which can track how many cars go past a given point and you could use AI algorithms and CCTV cameras to identify how many people are moving through a space and at what speed. As you can see, these are new ways of approaching traditional social science questions, but it's worth noting that, you know, there's no reason why you have to abandon traditional methods completely. Part of really good research is evaluating different methods and comparing outcomes, and it might be interesting to see whether by using different methods, you get different answers. And if so, you can then go on to ask, well, why is that? Which may then prompt further sub questions for your research. And this is a difficult task. You know, CSS isn't always easy, but in terms of it has a lot of benefits in terms of building upon your computational skills and strengthening those social science skills that you may already have. So there is going to be another short break, but maybe just put in the chat if you if you don't want that break or you want me to continue going, just let me know and I can carry on. But if you're really desperate for another five minute break, let me know. So before we do or don't take that break, I'd also like to quickly introduce an eight step process for how to undertake CSS projects. After this break, we're going to go into more detail about each of these steps. And this will be about identifying problems, exploring the problems, formalizing concepts, collecting data, implementing software and verifying the concepts, then using those concepts to experiment or analyze data, discussing your findings and presenting a conclusion, then communicating, publishing and presenting your findings and sharing your findings as well as documenting and validating your findings. So I'll just have a little look at the chart and see if we want to go through these eight steps now or have a break. Yeah. Pretty much the majority of you have said to continue. So let's carry on. OK, so to make this process useful to you as we go through these eight steps, I want you to start thinking about either a project that you'd like to tackle or a research idea that you've been thinking about, or it could, you know, if you're just, you know, stuck for ideas, it could be a project that you've done in the past. You can then jot these ideas down or maybe even put them in the chat and have it in mind as we go through these steps. So step one, identify the problem. So once you've identified the problem, you're going to want to be as clear and specific as possible about the pattern, problem or lack of insight. So you'll want to identify who is involved, where it is, etc. And what this will do is help you to define your research question. So maybe we have a goal in mind. Let's say we want to get more people traveling actively through city centres. We want less cars on the road and more people riding their bikes or scooters, or just, you know, being able to get from A to B in their wheelchair. The research question might be, what are the barriers to active travel in city centres? So what you need to do then is you need to identify who is involved. So you can start to list who might be involved in this, whether that be potential companies, people or different demographics that may be of interest to you. So you might want to look at city councils or bus companies, have in mind some different businesses to contact or, you know, things of that nature. And it's better to just go all out with these lists as well, as it's going to give you a lot of different avenues to explore. And, you know, if you find some of these leads are a bit of a dead end, you can always cross them out. You've done a bit more investigation into them, and then you decide whether or not they're actually relevant. So, yeah, this is just making those big lists and thinking about, well, who's going to be involved in this, what are we going to be looking at? And the next step will involve exploring the problem. And this is where you'll gather information and perspectives in multiple different ways. So, you know, you can think about surveys, observations, the secretary data analysis, app creation, web scraping and so on. So this might involve conducting a few interviews with people of interest. So, again, if we focus on that travel in the city example, we can think about interviewing the manager for our city's transport network or local council workers, but we probably also need a survey or observations or secondary data analysis to capture how many people are actually moving through the city centre. So it's about using different methods and tools to further enhance your understanding of the problem or the research topic. In this step as well, you're also going to want to spell out sub-problems, processes, relationships, sort of any simplifications, assumptions or related issues. So after settling on your main research question, you're going to need to get more specific in order to make that question relevant and measurable. So if, as we've said, we've got this question, you know, what are the barriers to active travel in the city centre? Let's say that's our main question, right? We might specifically be focusing on, well, what are the barriers to active travel through this specific city centre at this specific time of day, given the way that these specific roads are laid out? So this is where you really nail down the particulars of your research question. So if we head back to MNC, we'll have a little, you know, you can basically tell us a bit about your step one and two and what they involve. So if you want to share what maybe you've got so far, any ideas you're thinking of, no worries if you don't want to share. Like I said, you can always keep a note of this and just jot it down for yourself. So yeah, you can do this if you find it useful, but if you don't, that's fine. And it doesn't have to be well defined. I know it's hard to outline these steps if you've only got a vague idea. But for step one, it could be as simple as, well, people on Twitter from different political camps really seem to hate each other. So what I want to do is look at political polarization on Twitter. And then my step two might be, well, to do that, I'm going to gain access to Twitter or X's API and see what I can find by scraping some tweets. And maybe by doing this, I can build up a big data set. So it could be as simple as that. I'll give that maybe a minute. OK, so we've got wider pupils drop out of schools, national data, which schools have high, low, early school level rates, national data. Yeah, that sounds super, super interesting. What does the eye or the lack of it look like in my industry based on qualitative data? No previous research done. Begin to scrape the surface based on the answers, especially with minority participants. Nice. Employees sense making of organizational culture from looking at the employees, employees reviews on the company. Yeah, nice. So you could look at it in some web scraping for that, maybe some natural language processing to look at the kind of sentiment. There's often people will do like a sentiment analysis where you can sort of tag each sentence, each word as well and see whether it's, you know, positive or negative. And then you can sort of like then have your group responses, right, sort of positive and negative and then look at what sort of words commonly occur. Why are some people more likely to be unemployed than others? Yes, so can you think of maybe a method that you would use to implement that research or, you know, how do you measure it? Why do crime levels differ across different postcode areas of the city? Nice. Step one, the role of social media influences in localization of news information. What's the geography of social media influences on Facebook and Twitter and then all the networks? Nice. Yeah. So you could look at some geotagged social media posts for that, maybe. That could be one way of getting computational with it. And, yeah, you know, trying to scrape those social media networks as well. I will say from experience, it's now become a little bit difficult trying to get access to X's sort of like API. But I think you are still allowed to scrape a certain amount of tweets every so often. But yeah, that's something to look into. OK, nice. So we've got some really interesting responses here. Oh, I'll just read the one that's just complete. So step one, impact on nutritional behavior with and without social media influence. Step two, scrape social media following and monitoring health behavior. Yeah, nice. What does what dose of nature do people use for improving their health and well-being and how does this link with their views of local green space? Nice. OK, I'm going to move on now to step three and we'll have another one of these where I think it's for maybe step three and four or four and five. So keep jotting down these ideas and then. Think of, you know, ways that you can implement that computational method as well. So moving on for step three, you need to formalize your concepts. And what I mean by this is you'll want to make all the concepts and processes explicit, formal and both computer and human understandable. So this is often known in the computer programming world as pseudo code. You don't need to know how to write code. You just need to start understanding how to formalize things if you're going to use computational methods. For example, maybe your research question focuses on trust, which is a very social science sort of concept. If your goal is to get a computer to be able to measure it or model it or represent in a simulation, you're going to have to define it in a way that a computer would understand. So maybe you'll define trust as a variable between zero and 100. You'll need to make rules about how that variable will change in certain situations. So maybe if two parties interact positively, then trust increases for given a negative interaction where one of the parties is judged to be deceitful. A level of trust will decline or reset to zero. So this is a step where you start thinking about how to formalize concepts in your research question so that a computer will be able to understand it and interpret. OK, and step four, this involves collecting data, implementing software and verifying your process. So you're going to need to select and implement one or more methods. Many of you might have thought a bit about these methods in step two. So maybe you wrote down something to do with web scraping. I definitely saw a few that had sort of implicit web scraping methods or, you know, it could be sentiment analysis or agent based modeling. This is the step where you implement those methods and make sure that they work in the way that you anticipated. So maybe the data might come in a different format than you expected, or maybe you're encountering a lot of error messages in your code. So, you know, can you make it work in the way that you expect? Maybe if you're having trouble a lot of time if I'm having trouble coding, what I do is I use a toy data set or I use a very small portion of my data set that I can test out the code on because it's just sometimes it's just a bit easier to see when you break it down. You don't have something iterating over, you know, however many rows of data. And this helps while you find your method. And of course, the choice of method, again, it's going to be highly dependent on the research topic. Lastly, what you're going to need to do is thoroughly check that the selected method has been implemented correctly. And that's what we mean when we talk about verifying your process or method. It's about answering the question, and, you know, did we do the right thing? I mean, sorry, did we do the thing right? Yeah. So let's have a little bit more interaction now. If you like, you can feel free to talk about your steps free or form what they would look like. Again, I know this might be hard as you've not actually undertaken any research or implemented any methods, but you can share some big ideas about how you might formalize those concepts and, you know, how you might implement the methods. So we'll give that a little minute. So to really restate what step three was, we had formalized our concepts. So this is where we make all the concepts and processes explicit, formal and both computer and human understandable. And we had that example about how we don't make trust an understandable concept for a computer. And then for step four, we had collect data, implement software and verify. So this is where you select and implement one or more methods. So I said as well that the choice of that method is going to be dependent on the research topic, and it's about checking that you've implemented it correctly and asking that question, did we do the thing right? Don't worry if you don't want to put anything here, we can just head on on the next slide, but I'll give it maybe a minute. OK, measure the role of social media influences through social network analysis represented by network centrality measures, then investigate role of geography by calculating Euclidean distance between users. Yeah, excellent. Yeah, that sounds great. Use interview data to code construction, deconstruction of engagement at school, e.g. coding engagement once a hundred with negative engagement indicating the code at the point of disengagement. Yeah, nice. So you've got that formalising those concepts so that a computer can understand it. Yeah, we've got the same thing with the first one, investigating role of geography by calculating Euclidean distance between users. Yeah, excellent. Maybe I'll give it a little minute to see if there's any more. And if not, we can head on to our step five. Maybe based on the theoretical model to come up with abstract concepts and based on this, come up with some more specific terms based on abstract concepts, rules, workarounds, promotion. Yeah, that's why this step three is really important because it just helps you define those terms and it helps you sort of get more specific than with how you know, how you to formalise them and then what method you'll use as well. For example, if I'm using a survey monkey or even a Google Doc survey before I send it out, I should be making sure all the questions are being asked correctly. No confusing questions, ethics approval. Yeah, you know, for sure, that's a really important part of step three. Well, what I'm going to do now is I'm going to head over to step five. This is where you're going to want to experiment and analyse the data. So of course, you're going to be running the experiments, building any models that you might need, analysing the data or otherwise use the methods that you've selected in the previous step. And you'll need to identify and explain the results within the context of the experiments and the model or the method that you're using. Then once you've run your experiments and you've analysed the data, you can then start to interrogate the results and form some conclusions. So this means going beyond the experiment and the model or the method that you've used to draw some conclusions about what the results mean. So, you know, you sort of ask yourself, well, what sort of picture are my results forming? Do my findings support any policy recommendations? So it's the research indicating that there might be, you know, some potential changes that we could make. And then we have, who or what do these results affect? And why does it matter and what should change? And if it should change, who benefits from those proposed changes? So if we use that example of looking at how people move through cities, perhaps you've found that people with disabilities related to mobility have more difficulty navigating through particular areas. Maybe there's uneven surfaces or particular roads that are just too narrow, in which case you can then recommend some changes which include wider footpaths in XYZ area. So we're nearing the end of the research process now. And this is where we're going to focus on communicating and sharing that research. And in terms of communicating the research, it's important to understand that all of the previous steps must be communicated to multiple audiences in multiple ways. And what's really important, too, is that you think about short term and long term engagement. So for a lot of us, you know, before researchers, our minds will instantly go to, well, I want to get this published in this academic journal, which, you know, for sure, is definitely important. But it's also really good to think about other forms of communication. So are you going to present your work at a particular conference or are you going to submit a piece about it to a popular blog? And you can then think about whether there's any workshops or classes that you could present on it to then get your research shared more widely. And with that, you know, you're going to want to think about how you adapt your style and your tone of communication to suit those different audiences. So presenting at a conference, for example, is going to be a much different style and tone than submitting it in a piece for a blog. So you're going to really want to think about how you're going to communicate that research to get it shared as widely as possible. And finally, you're going to want to share, document and validate your findings. And what we mean when we talk about validating your research is making sure that the right thing was done by allowing your work to be studied, reproduced and or modified as needed. And to do that, you want to allow as many people as possible to be able to access your methodology and any code or data. You're going to want your research to be as transparent, well documented and open as possible. And of course, you know, there's going to be caveats to this. So you might be working with something like administrative data. You know, someone mentioned before they were looking at data to do with, you know, public schools. Maybe this is restricted. And so you can't release that full data set, even though you could analyze it yourself with permission. So what you could do then, if you're based with something like that, is you could create a dummy data set so that people can still run your code and work for your methodology, but not with the actual data with sort of this like dummy or toy version of the data. And then just to note on the fact that, you know, reproducibility is a really important area of research and it's often neglected, which is why it's good to think about how you'll document your work before your research actually gets underway, which brings me to just a few things to note as we head towards the end of the presentation. And that's that these steps that I've outlined, these eight steps, they're not linear. So there's going to be many points from each steps that you'll need to return to or apply throughout the research process. For instance, you know, documenting your research is something that you're going to want to apply from the beginning. And when it comes to computational social science projects, most or all of them are going to require many iterations. And that means you're going to need to revisit certain steps. So maybe you'll come up with a research question in step one. After you've actually explored that problem further in step two, you might want to jump back to step one to then reformulate that research question in light of something that you've read. Or maybe you're on step four and you've implemented your method, but you need to then go back to step three because you've not outlined the concepts and the processes. So this process, you know, it is iterative and that's why documentation is super important so that you can capture all of these nuances and changes as your research evolves. And, you know, I've often had it where I'll make some really important changes to my code, but I don't end up documenting it. And then when I come to sort of writing up my methodology, I'm struggling to actually explain, well, why did I opt for this particular type of algorithm or this Python package? And it is a shame when, you know, you don't document it, right? Because those are the really interesting insights that you then lose. Having those insights then helps someone else. And it helps you figure out how someone took a certain path and why it is that they took that particular path. It might be, well, that was a method is really obvious to me. Why didn't you use that one? That might be a very good reason, but you don't document then that insights lost. So it is really important to just document each of these steps. And, you know, it's got the other bonus of the fact that you're going to make the writing process so much easier because you've got everything there that you already need. I've definitely learned from my mistakes on that one. So before we head to the end, where we have a little Q&A, I'm just going to go back to Mentimeter because we've got, I think it's one last word cloud kind of thing. And, yeah, let's do that now. But if you want, you can enter your last minute takeaways on what CSS is, maybe why you think it's important. And you can pop these in Mentimeter now. Someone in the chat said, you know, great presentation. I'm going to have to go now. No worries. In fact, we'll come in. Don't worry if you have to leave. Need more like this? Oh, great. Social science, computational, data, collaboration. Yeah, super important. Complex, definitely. But hopefully maybe a little bit less intimidating after this workshop. No standard methods. Yeah, pragmatism, yeah, definitely. It's, you've got to be a pragmatic if you're working with computational methods. Yeah, I've learned that. Need to collab. Yeah, definitely. I hope this workshop has stressed that, you know, I no one expects anyone to be building monstrous, you know, crazy machine learning algorithms. You know, it's enough to learn about that method and then reach out to someone who has more experience with it and can help guide you. But, you know, it's sort of more simple programming. It is worth maybe having a look into that. There's sort of analyzing data. So we've got programming, social science, informative. Think of your audience. Yeah, definitely. It's always good to think about, you know, how you adapt your turn. A big conference, you know, you can often be a bit more theatrical and exciting than you might be in a journal article. So, you know, keep that in mind. Useful, well explained, multi-stage process. Yeah, I'm really glad that it seems like you guys have found this quite useful. We've got ATA, of course, computational and programming. And pragmatism seems to be the most popular words, which would fully agree with. OK, it's not moving as much now. So what we'll do now is we'll go to our Q&A. I should also spotlight as well. Let's just see. There's some references. So, you know, for anyone that's interested, just to say as well that we're going to be making these presentation slides available so you can always go back and look at them in more detail. And it's all going to be uploaded online as well. So, you know, you're going to have access to that recording if you check the events page. So anything you need to go back and check, you can do. As I mentioned, we've got those evaluation surveys. So if you click on continue after you've exited, please fill them in. That would be really helpful. And I'll also leave my contact details as well. So I've got a Twitter or an ex account and also my email there, too, if you want to ask me anything about the workshop or you've got anything sort of CSS related that you want to ask. And like I said, as well, we've got those CSS drop-ins as well. And they're really useful if you just want to connect with other people in different fields that are maybe starting out on CSS stuff or maybe you're the only one in your particular, you know, group that's doing some CSS, come along, ask us anything you want. They're always useful and the discussions are always really interesting. But big thanks to everyone for coming. I hope you found it useful. And yeah, thank you.