hi I'm Chris Martin and I'm Andrew McIntosh and we are your hosts on the primate cast today we're going to talk to Dr Roger mundre who's the resident statistician at the MOX plunk Institute for evolutionary anthropology yeah so one of our major objectives here at scasp which I'll remind everybody is PR's International Center is to bring in kind of international speakers that can help to our educational purposes uh at The Institute here you know we often have intensive lectures done by
researchers from within Japan but now we're starting to have the opportunity as well to bring people in from outside and we hope to kind of broaden the perspectives of our students as well as faculty here at the Institute that's right and so Andrew you and I both work for scasp the International Center and the first speaker that was brought in by scasp was actually in the podcast Dr Lawrence Anthony yeah and you can look that up on the podcast number the primate cast number three um and he came
to give a morning workshop on kind of introducing the use of Corpus Linguistics and Corpus Tools in helping people develop their technical writing skills and that was really great to have here and so Roger mundre then was somebody that we invent invited from Germany to spend actually more time here it was a it was a workshop that spanned over a week that's right and he was here to talk about statistics and specifically some of the pitfalls involved in statistics like multiple
testing data data dredging and also giving a broad overview of the use of General linear models within the framework of our statistical software mhm and it it was very um great educational opportunity for a few uh reasons I think for me the main uh thing that stood out was that he is very familiar with the kind of statistical problems that primatologists face so he's not just a normal statistician he's a statistician that deals every day with primatologists and
our statistical problems and so it was really nice to kind of get his perspective here at the PRI yeah and that's something that'll come out a bit later during the interview but that's one of the reasons why we targeted him specifically as well you know it's very rare to see within primatology or anthropology departments or research institutes a dedicated statistician um and so many of the problems that he is forced to deal with on a day-to-day basis are very relevant to us to our
Institute and hopefully to the listeners out there so we hope to gain some insights absolutely so without further Ado here is Roger mundre so Roger how long have you been at MPI a little more than 6 years now yeah and so what is your official capacity uh officially I'm responsible for helping people from the Departments of Developmental Psychology and primatology with Statistics and that's pretty much what I'm doing there all the time occasionally I'm helping also
people from the other departments and I'm also quite busy with teaching right it sounds like you're very busy yes I am but one of the things I think a lot of people here appreciated um in hearing your talks was that it they were presented by what everybody could see very clearly was a biologist rather than a pure statistician so yes I I am a biologist by training and I never got any formal uh training and statistics so I taught all that stuff to myself in a way and of course I
benefited a lot from meeting people who some only just gave me books to read or articles that I were asked was asked to check whether the could be helpful but also there were some people I learned a lot from so but basically I'm a biologist and I got interested in statistics because I wanted to learned something about the animals I studied and that um yeah meant that I had to worry about statistics and then I discovered statistics also to be a very interesting field and so now I'm working as a
statistician currently in just taking the world of primatology for an example I think MPI now is is a bit of an example because I don't know of too many anthropology or primatology departments or institutes that have their very own dedicated statisticians yeah uh to my knowledge most of them don't have a dedicated stati statistician and that's uh something I suffering from occasionally because people from all over the world tend to approach me um the the primatology world is pretty small so I
mean you you don't you just get to know be known by many people pretty soon from collaborations which you have and then you got these emails from people from remote places youve never heard about those people and they send you emails asking questions about statistics I try to do what I can but you can imagine that this it can be quite overwhelming at occasions and um it's not easy for me to always catch up with all the issues coming in but there would be definitely
more need for statisticians in other places so um it's good that they have one light and and um actually we're even searching for a second now because that's so much work to be done um but there will be would be definitely more places not only in patology but in biology in general and maybe also psychology where more regular statistical advice would probably be a good idea absolutely beneficial I think having here at PR has maybe kind of laid that kind of seed into other people too so
that we can get either bring you back on a regular basis would be fantastic or even take some some lead from MPI but can you give us an idea then about some of the projects that you're that you have been involved with at NP in yes um say I mean first of all I'm involved in pretty much every kind of project dealing with genetics cognition ecology Behavior culture pretty much everything you can think of also Linguistics and development of children and pretty much
really everything I personally at the moment I'm particularly engaged in more large scale ecological studies some of which really cover the entire great brange in Africa and um they are say pretty computation intensive in the first place so it just needs a huge amount of programming and computers may be running for more than a month to get the analysis completed stuff like that and also from the perspective of data processing they might be fairly complex because it's a whole lot of Fairly
different data so maps for instance showing certain ecological gradients over entire Africa some of them having a resolution of 200 by 200 M so it's also large amounts of data to be processed but also data about abundance or presence of Apes in certain areas protection measures um economical variables so we also incorporate things like gross domestic product and Corruption indices so they complex analysis and I like them a lot so um complex models to understand pretty
complex questions and they they need more input from my side because uh the students at least initially they are quite overwhelmed by the by the amount and also the complication of the tasks to be done so merging all these very different kinds of data geographical economical data ecological data so that takes some time and effort but I like these studies a lot and they challenge me me to some extent they made me thinking about the Ecology of great apes and what can be done about their
protection so I think it's also very relevant so these studies are not just um say you do something and eventually you get some significant effects or nice models but they really might matter also for future decisions which um might be made about how to protect wayes which is a very big issue so what person so it seems like there's a few aims from these studies there's a conservation aim and then there's also kind of scientific aim figuring out distributions of Apes
across Africa so what's what's kind of the main priority for that for that study um first of all it's a whole bunch of studies so it's not a single study but um I went when when I came here I considered for a moment to give out talk what I'm doing personally and that would have been I know didn't give it because I didn't find the time to repair but it would probably have been my first talk as a primatologist so coming out as a primatologist originally I was an
oologist and I still very interested in birds so my heart is still beating for the birds and when I see something outside the window then I need to have a look at it and I can get very excited about birds but I I realized when I when I thought about what could I talk about that actually I'm much more involved in primates primatology by Apes than I in birs now MH and um so then I thought about what could be the different projects I was involved in or am currently involved in um I could
say a few words about that and they covered a whole bunch of problems so some of them deal with uh habitat changes in Africa over the enti great range and um how that might impact the suitability of the habitat for great apes and the change of that seem to be quite dramatic MH in the past decad decades um so that's more like a general question about what's going on in Africa so I mean we don't really know too much about um habitat suitability for great apes over the entire range there's a
whole bunch of studies for certain areas but it has never been combined on a larger scale that's interesting but other studies also deal with say Countrywide models where we look at for instance um human impact and it's not always such that areas where there are more human activities are those which are verse for great apes so it seems that there are places where great apes and humans can get along quite well with one another so that's another perspective but we also look at for
instance infectious diseases and their impact on AE abundance protection measures we had one recent study which is published already which shows that if you have protected areas if you don't do any measures like putting guards in there or having activities of NGS you have a high risk of great apes to get extinct within the protected areas so these are studies I'm involved in and I for me it would be hard to classify them into say more applied or more fundamental research most of them they
have some faet related to both aspects I would say okay that's interesting yeah our own field site um in Guinea Basu West Africa um we have there's this problem of isolated Forest so there's a group of chimps 14 chimps and they live in this very small pocket of forest that's surrounded by Savannah and they're just kind of stuck there they can't get to the nearest Big Forest because they're worried about Crossing to Savannah The Chimps don't do that so I think that might be a problem around a
lot of West Africa is these isolated groups and then they run into genetic problems in the long run yeah I guess it is and actually just right now I mean before I came here I was involved in a project and I will be involved again when I come back where we try to um just not not just measure say habitat suitability but also say connectivity of the habitat so that we don't look at an isolated patch is it good or bad but we try to measure the quality of the patch also in terms how
Val it is connected with other places and how easy it would be for a great a to pass from one patch to the other it appears to be more complic at than you originally thought but that's part of the fun of the job that you are challenged with new questions and problems again and again that's interesting absolutely so I think from a researcher perspective we typically have you know maybe a couple of main topics that we that we focus in depth on and maybe we get to know those topics really
well and even the data that we end up using for the main purpose of those topics but for somebody like yourself you're confronted with so many different types of challenges a very diverse set of research questions and then trying to figure out ways to analyze those independent or independently or in some or whatever so what was that like for you when you first came to NP or had you already started along that path long before that I mean when I came to MBI I was initially absolutely
overwhelmed uh I I didn't have much knowledge about primatology and psychology of course I was reading papers and journals like an behavior and and knew to some extent what the people were doing in The Institute but of course not much details and before I came there was a two month Gap where there had no statistici so from the first day on I had just a huge crowd in my office pretty constantly uh asking me questions and I mean you know that um people have their
specialized terminology so they spoke with me about B in Africa I had no idea what a b was later learned that it's a huge clearing which is there from natural reasons and then many elements go there occasionally because there are some things they like to eat I don't really know what it is but they are pretty attractive for elephants and gorillas and stuff like that but people very people were speaking with me about these buys as if it is completely known to everyone on the want to buy this I
never heard about that and there were many other things like false belief tasks which the psychologists do and things like that and I didn't know that so there were lots plenty of new things to Le so but it was at the same time it was also very exciting to get to know about all these Concepts and things people investigate now it's easier for me because I mean after 6 years I kind of have an idea what these people are doing I'm kind of acquainted with their with
the experimental designs and psychology and the questions the primatologists are addressing from a statistical perspective it is it is less diverse so there's lots of say cross breing uh between different challenges tasks and quite frequently happens to me that I spend a day or two or a week with particularly focusing on one project and I do something pretty weird unknown to me new with regard to the statistics and just a few days later someone having a completely different question um coming
to my office it appears that actually what I just learned about recently could be very helpful in that context as well so the statistics that I'm using are not too diverse and some of the tools like linear models they just provide a very general framework with which you can address a whole bunch of questions which are not really related at all so I'm using points ofal models a lot in various context yeah and it was interesting you know last night at dinner Dr fichi made the comment that in
you know for example the field of physics things tend to move towards these unifying theories and now at least in some fields of Statistics he noticed a potential parallel in general and generalized linear models which within the same or similar statistical framework are are able to tackle you know so many different types of problems it is a very unified uh very general framework this generalized linear model and it's a little sad actually that we didn't find the time to speak about the
generalized linear model in this course because it really provides a very very flexible framework for addressing questions and it's also um say the the mathematical Machinery behind that is pretty much used in the exact same way in a frequentist approach as well as in an information theoretic approach to statistics but also the basian use pretty much the same underlying mathematical Machinery so it's a very general flexible framework still um I'm also using a whole bunch of other things
like for instance non-parametric statistics I still think they have a sense they have a meaning they are useful they are applicable only to very simple designs research questions but then they also give very simple answers which is good so these complex models they they also can give complex answers and sometimes people say design studies which are too complicated for the questions they want to address and then they get kind of overwhelmed by the complexity of the analysis and also
the answers that they get and if they would have designed a more simple study then maybe just a simple lookx and test or spearm correlation would have pretty much expressed quite well what they wanted to know and in such cases I I think the simpler me methods have still some right to exist and I still use them so just briefly going back into this uh Workshop that you gave here at the PRI it was very clear that for you um maybe it's the way that you do statistics but
also the way that you teach it uh was by combining Theory and then practical application using R okay so for maybe for you those two things kind of go very very well together so do you think that's something that you would highly recommend to people yes definitely I mean first of all R is just the future of Statistics I don't see any space for commercial packages in the near future I mean there will still be some institutions and people buying them but R provides just so much more
flexibility and it it allows to handle so much more kinds of data than any of these commercial packages that I don't really see much need in using them and it makes on the longer at late makes life easy because you don't have to learn like three four five 10 different programs and how to use them and all that related to dealing with them but you just have to learn one program and it is pretty much the same logic that you use for handling geographical data quistic data any kind of data you could
even think of so R is just a great thing first of all and then secondly I think at least uh on the immediate inter intermediate term are also really makes it easy to explore data understand data understand the output of statistical model so everything is pretty much at your fingertips so if you ran a model and you want to understand what does it reveal is it a good model does it fit well are there any issues with that model it's pretty easy to extract all that using R it has absolutely fantastic
graphic capabilities another reason to use it and with regard to say teaching Theory and practical application pretty much um how you say that um just closely tied together interwoven is that the right yeah I think it is very useful because I mean first of all teaching theory for 68 hours a day everyone would be asleep after 90 or 120 minutes so just having practical things in between it wakes people up it makes them doing something active but also I think the
practical application that helps a lot in understanding the models right so these theoretical Concepts they are theoretical concept Concepts and dealing with them practically and making experiences about say what is an interaction and how is it modeled and exploring what it means practically using some data set it makes it much more accessible for the people to understand what these models are doing so that's why I think teaching two things at the same time statistics and R
is a challenge for many students but I think also that eventually it pays maybe one more word about teaching just r so I was occasionally invited to give a course just about using R um how to run linear models in in R and uh whenever I taught such a course it appeared that there were also quite considerable gaps in the knowledge that people had with regard to the underlying Concepts and that's why I don't really like to do it so I mean of course you can teach people how to run linear
models but then they might maybe uh use stepwise uh in conjunction which with linear models which I would not advise at all for a whole bunch of reasons so um or they might misunderstand concepts related to interactions and building models with interactions included and and and so that's why I feel not very comfortable in just showing people how to do things without also explaining some of the concepts most people are not as familiar with the concepts as they
need to be so what is in store for the future of Roger mundre nothing particular I'm extremely happy at The maxbank Institute I really like it a lot to work there it's a very interesting place with regard to the scientific part of the work but it's also very pleasant to be there it's a very very casual atmosphere I mean the the Institute is very challenging to me and they definitely expect all of us to do a great work but they also give us all the support we need so it is a new
experience for me to work in a place like that before I've been working at universities in Germany and there just to get connected to a printer it can take like 3 weeks or something like that and then still it doesn't work and light things like that are impossible if you have a problem that's solved right away so people are very friendly and it's a really nice place to work so I hope to stay there as long as I can and um that's the plan so I don't really I never really made plans for my life and
I ended where I ended and I'm happy about it so we're going to see what the future brings I know we're also happy that part of it ended landed with you being here in Japan with us so thanks for coming thank you very much thanks for inviting us pleasure to be here great and thanks for joining the podcast so we hope to see you in the future okay for