Beth Tipton: Appreciating the Forest and the Trees
Published in Inside - Volume XVII, No. 2
Interviewed by Siddhartha Mitter
You’ve come to TC with a Ph.D. in statistics and a record of research on statistical methods. What will you be teaching at TC – and what does it have to do with education?
I will teach sections every semester of Applied Regression (HUDM 5122). It’s the second class in the statistics sequence. I think of it as introducing students to how to make sense out of data. You might want to know what the relationship between education and income is, or between socio-economic status and achievement. This class focuses on how you do that: What are the assumptions, what is the theory behind it, and how you do it in practice and interpret what you get out of the computer and make sense of it. Going forward, I would like to teach a meta-analysis class. Meta-analysis is used a lot in education to aggregate information across studies. I’d also like to teach a class on causal inference. We have all this evidence, but how do we know that what we are finding is causal, rather than just correlational?
You describe yourself as a “social statistician.” What does that mean in practice?
The questions I’m interested in come out of real problems that people encounter when doing research. I think of the social part as being really grounded in the questions social science researchers ask. As a statistician, I take these questions and formalize them. I ask questions like: What are the assumptions we need? What are the conditions? Is this problem similar to another problem we’ve already solved? I then focus on developing a new methodology for tackling the problem. You can think of all statistical methods, from regression analysis to factor analysis, as coming out of this process. At the end, as a social statistician I return to the initial question and ask: What does this result reveal about the original question? Does this help social scientists move forward with their research?
What are some examples of these real problems that you’ve been drawn to in your research?
One problem I’ve been working on very recently is how to choose a sample for an experiment. We’ve got all these social experiments, but who is actually in them? That really matters. If we’re really interested in the treatment effect of an intervention, say, in Texas, but we did our experiment in Illinois, how do we extrapolate? How do we know if the information from Illinois is relevant, and what the treatment effect in Texas would be?
How did you find your way from an undergraduate major in math to a vocation as a scholar of education?
At first I thought I wanted to be a quantitative social scientist. Then I had a great opportunity when my future husband and I moved to New Mexico. He did Teach for America on the Navajo reservation. We lived on the reservation for three years. We were immersed in education. He taught elementary school, and I worked at the University of New Mexico-Gallup, leading the adult basic education program and teaching some remedial classes. Sometimes I would have in my class the parents of a student at his school. I came back thinking that I like this work, but I also miss doing math. So I settled on statistics. I get to do math problems, but I think and hope that the methods I’m developing will facilitate better education research that will improve schools and communities.
What did you learn from your teaching experience on the reservation and elsewhere that you carry with you into your work at TC?
I anticipate that people are probably math-phobic when they walk in the door, and they have a lot of anxiety when it comes to statistics. They don’t trust themselves. As a teacher, I try to lay out the logic of things, because I think everyone is capable of understanding the underlying logic of statistics. That also helps me draw the line in terms of what’s important and what’s not important for you to know in my class. Interpretation is really important. Just being able to do the homework isn’t really the point. One day you’ll have your own research questions and data, and you’ll have to perform the mechanics within the context of those questions. You’ll have to decide what the numbers mean in the real world. This appreciation for both the forest and the trees, and an ability to go back and forth between them, is what I want to impart.