Close
Stephanie Dickinson 0:30
My name is Stephanie Dickinson. I am the statistical science green champions. And that is for the Department of statistical science at UCL. So I'm interviewing Professor Jim Griffin, about his work in relation to sustainability. So let's get started. How is what you do environmental, Jim?
Jim Griffin 1:00
Yeah, so I wanted to talk today about my work on the environmental DNA. So in environmental DNA, people collect samples from the environment, so things like water, soil, or even air. And then in the lab, they can, they can do an analysis, using PCR, to try and find which trace kind of trace elements of DNA for particular species. And so you can either do that in terms of a single species, so you can say, well, you kind of have some signature DNA signature of a particular species. Or you can look at multiple species, or you can just get lots and lots of samples and kind of try and test again, there's lots of different possible species. And so that's, that's kind of an emerging technology that people have been looking at, for the last n 15 years. And so the idea is that it's much cheaper than other kinds of ways of monitoring species. So traditionally, people would just go in and count count birds or, or observed particular types of species, or maybe they would kind of go get into the same place again and again, and try and see whether they see the same species there or not. But those things are quite expensive. Because it involves employing someone. So being able to do to use these kind of technologies is increasingly attractive to to researchers who kind of don't have limitless budgets. And so, that kind of is interesting from an ecological point of view. But also statistically, it throws up some kind of statistical challenges, because the data involves errors. So you can have false negatives. So you don't observe particular species in your sample, because maybe they're not actually present there, when you went and collected your sample. Or maybe they're, they're kind of they're in small amounts. And the data actually kind of show up when people who do the analysis in is quite complicated, in some ways. And so that kind of leads to challenge. And also, one thing we're interested in is as well as understanding whether the species is there trying to understand how much of the species is there. And for various reasons to do with the way that the data gets analysed. It's quite difficult to make that link. But but there are some possibilities. So yeah, that's the kind of environmental aspect of my work.
Stephanie Dickinson 3:43
Thank you very much. Very, very complex. But a lot of that makes sense in many different ways, especially the efficiency in cost, I think, shows where perhaps, we're heading in types of research. So that's, that's really interesting.
Jim Griffin 4:04
Yeah, yeah. So there's other types of, there's other types of measurements that people take. So they can take kind of acoustic measurements and listening for different types of Birdsong, or different noises in, say, the jungle that different animals are making different calls. And then and then they try and understand them which species are present. So people are looking at all these other ways to kind of monitor things and they're trying to kind of look at all these different methods and try and combine the results to really get a much clearer picture of what's what's going on. And you know, he doesn't doesn't often people go people go and they kind of disturb in some way the the environment, right, maybe the environment doesn't behave in the same way as normal if you put some money in it. So he's kind of promotes fencing. Yeah, it's going to become increasingly important.
Stephanie Dickinson 4:56
Yeah, no, that's that's really positive, too. To understand how people are doing things differently to make sure that they're not causing a mistake, because they're making the environment react differently to them. So yeah, that's, that's, that's really interesting and positive. And it also, for me, it seems quite inventive to think of all of these different ways of measuring where the animals are and what they are in.
Jim Griffin 5:24
Ecologists are very good at this. I'm not sure that's the statistical thing. But yeah, ecologists have kind of come up with all sorts of different ideas, because they understand very well, how the species behave, right? And what kind of characteristics of that species that maybe you could try and observe in someone?
Stephanie Dickinson 5:44
Really interesting. But thank you. Let's go to question two, what process will your work benefit?
Jim Griffin 5:53
So I think it's just about monitoring, really. So initially, the kind of work came out of people who want to monitor great crested newts. So great crested newts turned out to be an important species for understanding what was thought this the way that government and various agencies working trying to understand the effects and they're they're kind of a protected species, they're particularly important. And when people want to develop land, they have to worry about what to do with if there's great crested newts there. It's that kind of interested. So that's kind of how it started. But also kind of being able to look at many species, it kind of then lends itself to ideas for player diversity and understanding changes in biodiversity, then you can get into much more kind of complicated kind of analysis in terms of understanding the communities, it's kind of interesting whether an animal there is there or not, right. But also, it's interested in how they interact, right. For ecologists, that's very, that's very kind of important point area of research. And also just, you know, if you're, if we're going to, people are going to try and intervene in some way, either by building something, or by trying to kind of reintroduce species or trying to, to try and kind of recreate in areas, then it's important to understand how those will affect the community rather than just a specific species. Because these things obviously, don't don't work in isolation. They work this really complicated processes cutting off. Really, those two, those two sides of I think the main, practical outputs of this.
Stephanie Dickinson 7:36
That's really interesting makes me think a lot about housing development, and how we need to be obviously very careful about where we're building new homes, and communities of humans. Obviously, we need to pay attention to the other communities that exist there already. And so yeah, I can see how that research would help us to think a little bit more about that, and hopefully make the right decisions.
Jim Griffin 8:12
Yeah, yeah, of course. So, I mean, I think government is always sending out regulations around around building, right, and the Environment Agency, and people are kind of looking at how to regulate things. And so having a better understanding of what you should be measuring when you regulate is kind of isn't, isn't important thing. And, and this kind of, I mean, that kind of that kind of area of research is very practical for for doing.
Stephanie Dickinson 8:41
Okay, let's move on to question three. Why did this need to happen?
Jim Griffin 8:49
So began, as I said, it began it began with some people who are interested in great Christian news. So when I worked at University of Kent, there's kind of a lot of interest in great crested Newt, because there's a lot of great crested newts in Kent. And so they were interested in using this kind of new technology of environmental DNA, and trying to understand how reliable it was really. And so they kind of became aware that there's these different types of error. So you can have false negatives that you don't see the species, but also you can have false positives, that sometimes either you get contamination of your sample, and you kind of start to detecting species that can't be there. But also, in reality, they don't really find species, what they find is some kind of DNA signature, which they then call a species. And so then there's some possibility that through doing that, you don't kind of match up these DNA sequences with the species properly. And so there's there's some positive there's some there's some potential for false net negatives and false positives. And so that, so somebody in my department who was working with this group, and then she came along with this kind of problem. And and it seemed to me that it was an interesting statistical problem where there was a kind of natural Bayesian statistical approach to doing things. So in plays in statistics, you want to use prior information. And the prior information we could use there is that you kind of expect to see more true positives than false positives. So if you're kind of doing your, your experiment reasonably well, that's what you would hopefully expect to see, right. And so we could use that information in the analysis. Now, that was enough in a way to help us to be able to, to kind of understand this data and come over overcome some of the statistical challenges of this data. And so then they are able to kind of go forward and use this model to try and understand how reliable using environmental DNA was for great crested newts. And also to understand how you should do the monitoring. So should you kind of go to the same site multiple times? Or should you be going to different sites? And should you kind of do do retrofits within the lab of your study? And how many retrofits you should do? They could kind of answer questions of the study design, as well. So that's kind of what what came out of it really? And then and then that led on to, to working with other people who are interested in environmental DNA.
Stephanie Dickinson 11:37
Great, thank you. What are your reasons for doing this work?
Jim Griffin 11:46
So I think a lot of reasons are statistical, really, it seems to me an interesting statistical challenge. So I'm always interested in interesting statistical problems. And so as I said, the the original work had the kind of natural Bayesian Bayesian solution as a Bayesian statistician, that was, it was nice to be able to find that. And then subsequent work has looked at looking at multiple species, and trying to model to a certain extent, the process that happens in the lab. So when they, when they do the analysis, it goes through various stages. And these kind of they then that then introduces biases, I guess, really, and maybe errors into the, into the results that kind of come out of each of the stages of this analysis in the lab. And so we wanted to kind of adjust for those things in a model. And so that's kind of an interesting modelling challenge. But it's also nice that it has a an important practical application in the environment. That's also nice. And it means that people are very young people you work with, are very interested in what you're doing, because they're very interested in getting the results and getting reliable results, so they can make a scientifically justified conclusion.
Stephanie Dickinson 13:12
That seems like the the ideal interdisciplinary collaboration between you know, people who are doing kind of good worthy sustainability work and statisticians who really, really like the statistical side of what you're doing.
Jim Griffin 13:29
I think he's one of the one of the beauties of being a statistician that you're able to work in, in different areas include the work of very different applications, often the problems are similar. And you end up working across these different areas, and you kind of find unexpected similarities, or that you can bring across different techniques or methods.
Stephanie Dickinson 13:58
Great. Thank you. And what is your future wish for the results of this work?
Jim Griffin 14:05
Well, we've developed various packages that people can use to do the analyses of their data. And so I guess the hope is that these become a standard part of the toolkit for ecologists to analyse their data. So, I'm not sure we got there, but people are interested. So that's good. That's a good start. And yeah, that will be that will be the kind of best possible outcome.
Stephanie Dickinson 14:36
Great, thank you. How could you apply what you have done to other areas?
Jim Griffin 14:44
So one obvious similar area is the study of the human microbiome. So that that there rather than in ecology, you have species and in in the microbes If you have bacteria, so you're interested in how so for example, people are interested in the gut, the microbiome in the gut and how bacteria interact there. So there's kind of differences. Because in the kind of, in the microbiome you're interested in, whether that is connected to people having particular conditions or particular health kind of outcomes, maybe if you kind of kind of just kind of see this, a lot of people are interested at the moment, right, that if you can adjust your microbiome, then you can somehow have better kind of you can become healthier or stronger, or sleep better or something like, and so they're kind of interested in all these kinds of physical outcomes. So in some ways, the analysis in the lab is, in many is similar, very similar. And a lot of the statistical challenges are quite similar as well. And so that's kind of one one on one area that's kind of separate. But it's similar. Yeah, yeah. And the other is, the other thing that I'm kind of involved in is anti doping. And there, you have a similar thing, because you have mostly you have false negatives. Because people are very worried about false positives, for obvious reasons. But there, you have the same kind of problem, you have false negatives, and you have a lot of kind of noise noise in the way that things are happening. So you're kind of very interested in these types of errors. So there's kind of a similarity there as well.
Stephanie Dickinson 16:38
All right, let's go to question seven, I heard someone once saying that they want to solve the world's problems with math. Do you think environmental statistics could be an example of something similar?
Jim Griffin 16:59
Well, I think I mean, as we were talking about earlier, these kinds of ideas of remote sensing are potentially very powerful for monitoring environments. And those are always going to come with challenges, that that has to be dealt with, through math, through mathematical modelling, or statistical modelling, or kind of AI type of approaches, or probably, in reality, some combination of all of those kinds of different ideas and different ways of thinking. And so I think, yeah, that's kind of a, that will become increasingly important, just to try and quantify what's what's going on, right? Because it's just very difficult to understand particularly, you think about insects and things like that, then it's just incredibly difficult to monitor them in any other way, then, through these types of approaches, you know, even now, people are discovering new species of insect rights. Yeah, we just don't very little about about what's going on. And when you kind of get into the soil, what's going on in the soil and different types of animals, they're, they're nice things, we really know very little. So it's going to really throw up a lot of a lot of interesting, interesting work, he seems to. And I think Statistics has an important role to play there. And that's being assigned early also, in terms of, if you want to intervene in some way, either through through kind of trying to recruit in things or trying to build or try and do all sorts of different ways they can intervene in the environment, then you can be hopefully, you can begin to understand what the effects of those things are. I think that's incredibly challenging, but also, potentially incredibly important. And maths in its general mathematical sciences, in its general sense, will all play a key role in that.
Stephanie Dickinson 18:57
Does sound like there could be a lot of future work for people wanting to get into this type of statistics. We try and solve our climate mess.
Jim Griffin 19:13
Yeah, and I think even different approaches, right. So I think, you know, there's kind of now and approach towards kind of a marketized version way of trying to address these problems right through kind of green finance and it is building a market. And somehow, then you need to quantify what's going on in the environment, in order for people to be able to actually kind of run a market right, they have to have some idea. If if you kind of say I'm going to improve in some some area, then in a way and this you quantify what you mean by that, does it mean that you have more of a particular type of species or somehow you have more species in general or then the it becomes And then then quantification becomes very important. And so at the moment, that's kind of a direction, or that's one direction that people are moving in, right, particularly, I think in terms of government and people. So that's kind of also another on top of the kind of science side of things, and the kind of things that do academic is also that side of things as well.
Stephanie Dickinson 20:23
Right. Question Eight. Imagine a future where your work has become a standard method used all around the world for many different things. What would that world look like? And what would be different there to what it is like now?
Jim Griffin 20:43
More hopefully, I mean, as I said, a lot of our work is really about trying to adjust for these errors and biases in data. And so hopefully, then, from that, you can get a much better understanding of what's going on. And a much more realistic understanding of what's of what's going on and the effects of, of various factors in terms of deciding kind of the communities within the environment. And so having a better understanding of that will lead to people making better decisions, it will lead to better science in the future, hopefully. And I guess that's, that's kind of that the outcome? I'm not sure it will be radically different, but hopefully a bit better.
Stephanie Dickinson 21:27
Hopefully, people will know, what's going on a bit more, and what they're doing, what they can do that won't cause harm.
Jim Griffin 21:37
Yeah, yeah. So of course, I mean, I think that that's kind of the next stage, right? Once people are able to get better science, you know, as we improve the science, and we have better techniques for for measuring things and for understanding things, then yeah, of course, that can play into how people make decisions.
Stephanie Dickinson 21:54
Question Nine? Do you think there is a difference between the way you think about environmental statistics and other kinds of statistics?
Jim Griffin 22:08
So no, I think for me, I kind of come at it very much from the point of view of a statistical methodologist. So I think I always I always see these these problems as data that data comes from subsystem and needs to be modelled in a stochastic way. And so my interest is always in building these stochastic models and building methods for making inference in these stochastic models. And so whether that's in the environment, or in anti doping are in econometrics, in many ways. The toolkit that you use is different for each of those different areas. But there's lots of things that are in common, and a lot of common kinds of techniques that you can bring across. So that's kind of my my view on it.
Stephanie Dickinson 23:00
And question ten, what unintended consequences of environmental statistics could potentially occur, generally not relating to your own specific project?
Jim Griffin 23:15
Yeah, so I think as we as we've talked, I've emphasised the importance for adjusting for the kind of biases and errors that come in data. And I think environmental data often has a lot of that, because firstly, there's kind of an issue of, of design. So you kind of have biases that we would usually think about in statistics now that you're going to decide to sample in a particular location, and in a particular way, and maybe that could potentially introduce biases, but also, as people start using your kind of remote sensing or more complicated types of data, maybe that also needs the deeds that need some some kind of modelling. And if you don't do that, you may start getting erroneous, erroneous results. I think, also, in terms of we talked about measurement and regulation earlier. So it seems to me that always in regulation, there's a danger that the measurement replaces the thing you're trying to measure. So So you may be interested in biodiversity, but at some point, you build a measure of it. And somehow the measure becomes the thing rather than the actual bias, which has to necessarily be a simplification, right, of the actual thing you're interested in. And so always we quantification, but that's kind of the danger, that the people kind of, they just become obsessed by the actual measurements rather than rather than the actual thing you're interested in, in the first place. But I think I think those are I mean, I think I think the possibilities, there's huge possibility that I think for environmental statistics really huge advances. It's recently.
Stephanie Dickinson 25:03
That's really interesting. And I do think that we need to definitely think about unintended consequences. Because from what I can see that with other forms of research in the past, there have been lots of unintended consequences when people thought that they were doing the right thing. And I think that environmental statistics is another one of those where we are caught up in the idea that we're doing this because it's got a purpose, and you know, we're doing the right thing. So we need to be careful that we don't get swept up and do the wrong thing. While trying to do the right thing.
Jim Griffin 25:49
Yeah, I completely agree. I think I think Statistics has some idea of objectivity. That kind of isn't always true, right? Because at the end, somebody is making a lot of decisions. And they may be bringing a lot of a lot of opinions that maybe are not, are not quite, you know, they're not made explicit, and kind of somehow implicit. Yeah, I agree. I think statisticians are kind of aware, aware, often a lot of these a lot of these problems. It's often when he gets out, I think, of the statistical community, and then kind of those ideas get picked up, and people maybe have less interest or less awareness of the the kind of nuances of method.
Stephanie Dickinson 26:38
I think this is the last question. I think question 10 was last question. Let me just check. Oh, no, I've got another question. Maybe more than one I can't remember. Right, question 11. And it's become pretty evident that we all need to become more environmental, and more sustainable. Do you think we should all try to improve our knowledge of statistics and get more people involved in similar work to do this?
Jim Griffin 27:08
So I think always, it's good for people to improve their knowledge of statistics, I think we live in a world where people throw a lot of statistics at us. And I think that that can be good and bad. And as we've, as we've already said, statistics are not neutral. And people often want to make a make a particular point, particularly around a critical point, right? And then their job is not really to present a fair picture of what's going on, it's to present a picture that backs up their point, right. So it's useful for people to have a kind of good to be statistically literate, important. But I think I think also in environmental, particularly in kind of thinking about kind of ecological things, then, then statistics have been kind of quite useful, I think. So you often see these kind of results of surveys, right, people monitoring birds, and bees and various types of insects and things and, you know, they're there, people are interested, right, that comes up in the national media. That's, that goes beyond the kind of academic or, you know, especially as kind of the audience. And so there that seems to be it's been quite useful for our people to understand what's going on and what's changing. And, you know, those those kinds of those kinds of statistical kind of statistical work has been very, I think, important for, for, for raising, raising awareness.
Stephanie Dickinson 28:38
So what you've just said, as sparked off a couple more questions, I hope you don't mind me asking these. Firstly, to what level do you think it would be useful? People try to aim to improve their knowledge of statistics in order to understand the statistics that are constantly thrown at us from various channels? And then as a follow on? Actually, no, let's just focus on that first. Yes.
Jim Griffin 29:07
I think it's good these days in GCSE maths, and that kind of kind of level people increasingly have parts on statistical literacy. And so I think that's, that's very good for, for young people to, to be exposed to different types of data and the issues that are around that. So I'm not sure people need to have an amazingly technical knowledge. It's just to be aware of some of the issues and the biases and some of the ways that people try and slant results, right. So often, you hear often you hear, right, that something could this could be true or if this could happen. Often that means that it may still be incredibly unlikely to happen, right? It's just, that's the most extreme thing that could could happen. That could be true, but probably not true. And often often people present things in terms of this most extreme thing, and kind of end up concentrating on that being reality that's probably not going to happen. And I think it's kind of issues a bit like that. There's, there's, you see common things come up in newspapers. Right. But because I guess the people, journalists also understand a little bit how to make a more exciting story. So I can that's kind of under the label.
Stephanie Dickinson 30:25
Okay. So you're saying, so kind of GCSE level is, would be a good level for people to aim up to?
Jim Griffin 30:34
Yeah, the kind of character uses D and things like that. But I think often, there's kind of courses around statistical literacy out there, I think. I think that's, that's that will be, that'll be a really good thing. For a lot of people, I think it's becoming increasingly important data is increasingly part of our lives. Right. And that means that it's increasingly something that people useful to have a kind of basic understanding around.
Stephanie Dickinson 31:04
And as a follow on from that, if the general population did start to improve their statistical knowledge, and to think more about statistics, how would that change? What happens in the universities? I'm sorry, if that's a difficult question.
Jim Griffin 31:25
I mean, I think it's quite, it's quite separate, right? Unless, you know, some universities get involved in this idea of trying to teach people about statistical literacy. So before when I was a kid, we had kind of courses that people within outside not academics, but professional services, and people would come to courses to build up their, this kind of basic statistical literacy. And so I think in a way, the people we work with in the university are often an outside university are often highly statistically literate, particularly in areas they're working. So thinking about this kind of environmental work I had, working with an outside company and their people have a really good understanding of what the statistical issues that they're facing are. So I think in terms of university, we often are working with people who are very honest, I am working with people who are very aware of what's those kinds of issues, but it will be good societies as a society in general, it seems to me people to be more aware.
Stephanie Dickinson 32:26
Okay, great. Thank you very much. I agree with that as well. I think that it would be fantastic if more people, including myself could improve our knowledge of statistics, because you know, I work in the department, but I I'm not a statistician. I'm in professional services. And I do the green champion role for the department. And I see what I'm doing here is trying to draw out information from you and other academics, to try and place that information for everyone to learn from on our website. So thank you very much. I really appreciate your answers to the questions. I'm pretty sure that is the last question. Yes, it is. Thank you, Professor Jim Gryphon, do you have any last comments that you want to make?
Jim Griffin 33:16
No, I think I've said everything that I wanted to say. So thank you for your time, Stephanie. Great.
Stephanie Dickinson 33:20
All right. Hope you have a nice day. Bye.
Unknown Speaker 33:26
UCL minds brings together the knowledge, insights and ideas of our community through a wide range of events and activities that are open to everyone.