Assistant Professor of Human Development and Family Studies
231 Health and Human Development Building
University Park PA 16802
I am interested in the way that humans interact with their environment, and more specifically with each other. I'm also interested in the statistics that we use to analyze these kinds of interactions, and the way we treat the data we collect to study it.
It takes a lot of data to study this kind of thing, so I rely on modern computer-vision and machine learning technology to manipulate live interactions to watch how the dynamical system of conversation adapts under different conditions. This involves a lot of messing around with avatars, and making videos like the ones on my web page.
I also want to learn what makes conversations good. Some conversations just seem to flow naturally, while others seem forced or awkward. I’m interested in finding out what kinds of measures let us tell the difference, about how much rapport there is between the conversants, and hopefully about the relationship between the conversants. And I think dynamics are the way to do that.
I like to share, so I’m working to build up the videoconference-analysis tools into a platform for further development. I’m also looking to integrate the analysis parts of them into training tools for other things, like emotional understanding, classroom dynamics, and even physical therapy.
I like to make things that are useful. To that end, I’m part of the core team of the OpenMx software, which is a platform designed to allow scientists to push the boundaries of what’s possible with extended Structural Equation Modeling (xSEM). We’re in the process of adding an easy way to specify multilevel models within the xSEM framework, with all of the goodness of specification and multivariateness that comes with xSEM.
We’re also working on completely subverting the whole paradigm of data collection. The MID/DLE project is an attempt to make it possible for scientists to measure people and analyze their data without the collection step—we want people to be able to own and control (even physically) their own data.