An interview
We recently talked with Sandrine Müller and Heinrich Peters, about their article titled, “Investigating the Relationships Between Mobility Behaviors and Indicators of Subjective Well-Being Using Smartphone-based Experience Sampling and GPS Tracking”, which is scheduled to appear in an upcoming issue of EJP. Sandrine is a postdoctoral fellow and Heinrich is a doctoral candidate at Columbia University, New York.
Read on to learn more about their work on mobility and well-being!
Q: Hello Sandrine and Heinrich! Can you tell us a bit about yourselves and how you became involved in personality psychology?
Heinrich Peters: I'm super interested in how people interact with their environments through technology and how we can use technology to study human behavior as it happens in the real world. For that to happen, I believe, we have to expand our methodological toolkit and integrate insights from other disciplines like data science, for example. That's the motivation behind the research I'm doing and the path that I've taken so far.
I did my undergrad in psychology at the University of Hamburg, Germany, where I had a brilliant Professor who taught personality psychology. That’s what first got me interested in the field. Around the same time I became very passionate about some of the research that came out of the Psychometrics Centre at the University of Cambridge and I was fortunate to be able to work there as a research associate. Eventually, I decided to do my PhD at Columbia with Prof. Sandra Matz, who has done some amazing research in the field of personality psychology.
Sandrine Müller: I got fascinated by technology– Smartphones in particular give us an unprecedented ability to just track people 24/7. We were never able to do anything like that as psychologists in the past, so I find it really exciting as a methodology.
I did my undergrad in Dresden, Germany, and my master's and PhD in Cambridge, where my dissertation work focused on using smartphone data to study people and their experiences in different environments. I spent some time at Austin and Stanford as a visiting researcher, where I worked on smartphone sensing studies, and I've been at Columbia’s Data Science Institute as a post-doc for the past year-and-a-half. I also work with Sandra Matz, which is how Heinrich and I started collaborating.
Q: What aspect(s) of your job do you enjoy most?
SM: For me it's the freedom to be able to look into whatever interests you most; To really be able to define your own work, both with regard to the questions you ask as well as how you are going to address them. But, even in the way you do the work, you are free to work where you want – for example, in your office or the library, in a coffee shop or from home -, and with whom you want. You can define every part of the job and I think that's pretty amazing.
HP: Yeah, I feel very similarly. I am in the second year of my PhD, which takes at least 5 years in the US, so I basically have 5 years of time to explore exciting topics, expand my skill set, and to do (almost) everything I want. I think that's really an amazing opportunity! I also like that you can pick your own projects, so you can really take ownership of what you're doing.
Q: What kind of things do you like to do outside of your job?
HP: A lot of things actually! Aside from my academic pursuits, I really enjoy making electronic music. I also like to work out. Back in Germany and also in England I used to box a lot, and I found a boxing gym here in New York - which is unfortunately closed right now, due to COVID19. I moved to Brooklyn about a year ago - there is so much to explore, so I obviously do that as well.
SM: My interests are actually similar. Also music and sports, though slightly differently expressed. I play the clarinet, in a chamber music ensemble and by myself. And I row, although that's also on hold for the time being, so now I'm just taking e-gym classes, going for runs, and doing yoga.
Q: Can you tell me about the study?
SM: I got the idea for this study almost a bit exploratively. I started working with smartphone sensing in my PhD and I initially conducted my very first study in Cambridge with about 30 students - it was really small and a bit messy. I tried to collect everything I possibly could. I worked with a computer scientist who wrote an app for me, and then I just started poking around in the data.
I looked at the different apps people had on their phones and what they were doing, and as I started looking at the location data and mapping it out, I just felt so strongly that I knew so much about these people. I would see very different patterns; there was one student for example who would go rowing every morning, which was the same thing I'd be doing too, and I felt like I knew what his life was like. And then there was someone else who would go to a different pub every evening and I felt that that person must be super social and hangs out with their friends all the time. I just got such a strong feeling of knowing so much about these people just by looking at where they went, how regular their patterns were, and so on. That's how I really became interested in mobility behavior. Next, I started thinking about how I could quantify it and how our mobility patterns might shape our well-being. So that became the focus of my dissertation, and this paper expanded on that.
For this paper, we looked into the links between mobility and well-being in young adults. We tried to characterize mobility in general, so we measured mobility behavior first through smartphone-sensed GPS records and also through self-reports with regard to the places they visited. We looked at the links between mobility behaviors and well-being on different kinds of dimensions of mobility (taking a wide variety of different metrics that have been established in computer science) and well-being (depression, loneliness, and emotions that are experienced in the moment). We also examined a temporal component in the study, looking at mobility behaviors on a daily level as well as over an aggregated period of time.
HP: One of the interesting things about working with GPS data is that we have objective data and we have a lot of data points per person, so a lot of longitudinal information encoded in that data. GPS data basically consists of a time series of just two variables, latitude and longitude, which capture the coordinates that people visit over time. One of our questions was: "How can we extract as much information as possible from those two variables?" So we looked at a range of different metrics that describe how people move from location to location over time. For example, one of the more obvious things to look at is how far do people travel, but also how their time is distributed across location clusters that we identified in the GPS trajectories, or how much regularity we can find in their movement patterns, etc. In total, we had about 20 of these measures that describe the latitude-longitude time series in a pretty comprehensive way.
The measures were based on a thorough analysis of the previous literature. Most of the previous literature looked at a subset of these metrics and one of our motivations for the paper was to bring them all together, and to analyze the latent structure of the mobility behaviors that are captured in these metrics.
SM: Computer scientists in particular have proposed many ways for how to measure movement patterns. But another question is of course how to use those metrics in psychological work, especially in relation to well-being, and which of these metrics are even relevant, and how you can organize the breadth of what's out there... That's what we tried to do.
HP: We performed factor analyses to uncover that and found that there are three major underlying factors, which are related to the distance that people travel, the entropy with regard to how their time is distributed across the different locations, and the routines that people follow. With routine we mean that the temporal structure of people’s movement patterns is similar from day to day.
SM: There are also more specific findings in terms of how the different mobility and place patterns relate to well-being on both temporal levels (the daily and the aggregate).
One main finding was for depression and loneliness, where we see a relationship with more routine behaviors. And there were a couple of patterns that also held across the different temporal levels. For example, we saw that people were more stressed at work – that's something that we see both on a day-to-day level as well as in the aggregate trend over time – and that people feel less energetic at home.
HP: Additionally, and very broadly speaking, distance traveled, entropy (distributing one’s time evenly across different location clusters) and spending time in social places seem to be related to more positive outcomes in terms of well-being. We have to mention, though, that the effects that we found are actually very small. Of course, you also have to take into account that well-being is a multi-determined construct and that the mobility data is very noisy.
Q: What do you think are the implications of your findings?
HP: I think there are different kinds of implications. There's the question, "What are the implications for our field of research?" and, "What are the implications for real-world applications?". With regard to the previous question, I think we did a good job at pulling together a lot of different predictors as well as dependent variables, and we have painted a pretty holistic picture of the relationships between mobility behavior and well-being. And I think showing the size of the effects for all these relationships is also a contribution, because it tells us what to expect in future studies.
SM: While it was not something our study was aiming to do, I think it definitely gives ideas for future studies on interventions and real-world applications. Interventions related to behavior or movement or spending time in certain places are very exciting. There's potential for, for example, learning individual patterns and showing that on the days where people go to certain places, they feel better, and giving them suggestions to try certain things, to try to make them feel better. With regards to distance travelled and spending time in social places, there could be opportunities for trying interventions and prompting people to go out, go for a walk, or go out with friends.
HP: I would actually caution a bit when it comes to applications. Some people, for example, claim that it's easy to predict mental health outcomes and make diagnoses at the individual level, based on similar data, but that's really not what we think our findings suggest. Our results show some small relationships in the given sample, but I don't think it's anywhere close to being able to gain high-confidence insights into individuals’ psychological states - say predicting whether an individual is depressed or not.
I do, however, think that tailored interventions are an interesting idea - even if the effect sizes are small. Since nearly everyone owns a smartphone, there is the potential to analyze people’s mobility behavior and deploy interventions at a large scale. Even if these interventions make only a tiny contribution to an individual’s well-being, they may still have a pretty significant impact on people's lives in the aggregate.
Q: Is intervention research also the next step for you in this research?
SM: Definitely! We are also working on a project around depression that is using a larger dataset from all around the US. At the moment we are in a similar phase, where we are working on identifying predictors of depression, but the next step of the process will definitely be to work on interventions around mobility as well.
Q: Do you have any tips or advice for young researchers?
HP: Maybe I should start, since I'm still at the very beginning of my career – so you can have the last word, Sandrine. What I would advise anyone to do is to really see what you are interested in and to try to pursue that. And don't be shy to try to meet new people who do exactly what you are interested in. When I was in undergrad, I had some amazing professors at the University of Hamburg and I was really interested in what they were doing. One of my professors got me interested in differential psychology and psychometrics. But then I realized that there were other people all over the world who were doing things that were perhaps even more interesting to me. So for instance, when I started reading the papers by Michal Kosinski and David Stillwell about predicting personality from social media data, I was fascinated by their work. I already knew how to code and felt like I had a good grasp of research methods, so I decided to reach out to David Stillwell. That's how I started to work with him at Cambridge. It really helped me to figure out what I want and it opened up a lot of opportunities for graduate school. So my most valuable piece of advice would be to try to create opportunities for yourself and reach out to people who do exactly what you want to do and what you are passionate about.
SM: I think my advice would be along a very similar line. My advice is to collaborate and work with other people. If you bring people to the table who have different kinds of expertise or background, it allows for work that you wouldn't have been able to do by yourself. I started working with an interdisciplinary team of psychologists and computer scientists early on in my PhD and it made all the difference. For example, this paper stemmed from my previous collaborations with Gabriella Harari and Weichen Wang, a computer science PhD student at Dartmouth College! So I think collaborating is really game changing.
HP: Maybe also learn as much as you can as long as you have the time for it. I am right now still in this phase where I have time to explore and learn a lot of new things. Considering how quickly psychology is currently moving forward in terms of the methods that are being used, it really makes sense to start doing that early on and not to be intimidated by coding, statistics, machine learning, etc. I think it's a really good time to learn those things, especially as long as you don't have too many other obligations.
Q: What would you tell yourself if you could go back to the start of your PhD?
HP: I don't really have advice for my past self yet!
SM: I would tell myself not to worry too much. I think that especially when you are doing your PhD, just the thought of having to hand in that dissertation weighs really heavy on you. At least that put a lot of pressure on me when other things should really be the focus. Like Heinrich said, just learn and enjoy doing research. Just immerse yourself in that and don't worry too much about the administrative constraints of it.