Texas Tech University

Making Structural Equation Modeling Fun: Todd Little Goes to Camp

Heidi Toth

July 2, 2015

Little, an educational psychology professor, started Stats Camp because of a dearth of statistical methodology experts at universities throughout the nation.

Stats Camp
Stats Camp was founded by Texas Tech professor Todd Little.

There are no campfires, s'mores, snipe hunting or tents at Todd Little's camp.

There are statistics – lots and lots of statistics.

Stats Camp, which Little founded, is sponsored by Texas Tech University and supported by the Institute for Measurement, Methodology, Analysis & Policy, of which Little is director, and the College of Education. It came about because many colleges nationwide lack experts in the advanced statistical techniques that social and behavioral science researchers should be using.

But don't worry, it's fun too. So fun, in fact, that many people come back year after year.

“I joke that this is the one time when recidivism is a good thing,” Little, a professor of educational psychology, said.

Stats Camp, which just finished its 14th year in early June, is an opportunity for researchers from throughout the world to get a crash course on the methodological programs and techniques needed to ensure the most accurate, most effective results in social science research. It's held annually, this year at the Grapevine Courtyard/TownePlace Suites complex outside of Dallas. It's a much needed and increasingly popular event aimed at educating researchers on the newest statistical methods.

The need

When Little was working on his doctorate in the late 1980s, reviewers asked why he used structural equation modeling instead of more “tried and true” methods like ANOVA or Regression. He responded to those reviewers and journal editors explaining the advantages of the techniques he used and the problems associated with the techniques reviewers requested.

He converted them to his way of thinking, but found himself writing similar letters frequently. Over and over again, he explained the mistaken assumptions on which the old techniques were based and how these new techniques not only corrected those errors but also opened new avenues of discovery.

Todd Little
Todd Little

“There's a lot of fear of uncertainty and risk aversion,” he said. “If researches aren't trained in these best-practice techniques, they're going to use what they were trained to use – what their advisers taught them. Their advisers are using what they were trained to use. There's a fear of moving onto new techniques, particularly because they require advanced training to learn and understand.”

The mistaken assumptions include accepting all data as equally relevant and assuming a piece of data, once collected, is pristine. Especially in social sciences, where Little works, this assumed quality of data frequently is not the case. Anytime questions are being asked, there will be errors.

The danger of outdated, ineffective statistical methodology is researchers and others base their findings on those results, and if the results are wrong, well, so is everything built on top of the results.

“All the relationships you're looking for are going to be kind of masked by that measurement error,” he said.

Not all of this is the result of intentionally holding onto bad methodology, he said. Not every university has a professor who's an expert in advanced statistical techniques like structural equation modeling and those universities that do may not have enough experts to teach all of the graduate students who need it.

That lack makes Stats Camp, which gathers several experts as well as helps students build relationships with those experts, a necessity for many researchers. For Kristy Soloski, an assistant professor of marriage and family therapy at Texas Tech, Stats Camp was important enough to write into her starting budget.

“As a professor and researcher, it's important to me to stay appraised of the current statistical methods in the field,” she said. “With best practice methodologies constantly changing and improving, I wanted to be sure I was up to date.”

The camp

The first Stats Camp was in 2003, when Little was a professor at the University of Kansas. He put out a notice on various websites to see if anyone would be interested in an intensive one-week course that would cover a semester's worth of statistical methodology. At the end of the week he had the entire camp – all 13 of them – over to his house for a barbecue.

About five years in he had to nix the backyard barbecue.

Stats Camp
Stats Camp is a crash course on the methodological programs and techniques needed to ensure the most accurate, most effective results in social science research.

“We got to the point that we had so many people we had to rent a bus and bus them to my house for this massive barbecue,” Little said. “I got really good at flipping burgers.”

This year's camp had 285 campers, including some who stayed for both weeks. Most are in academia, with a fairly even split between graduate students who are learning how to do methodology and professors who have been using less effective methodology throughout their careers. Depending on the subject, they get a few industry people, but it's mostly academics.

Little teaches a course each week and brings in professors from throughout the world to teach others. Courses include structural equation modeling, multilevel modeling, psychometrics, meta-analysis, missing data, causal inference, program evaluation and cost-benefit analysis and other programs and methods.

“If people come to Stats Camp, they're not going to get yesterday's understanding of these modern statistics,” Little said. “They're going to get tomorrow's understanding, because these are the people who are publishing the articles, who are moving the field of statistics forward.”

He keeps the ratio of experts to participants low, as many of the participants bring their data and get help during the camp. Students and faculty members from Texas Tech's research evaluation measurement and statistics (REMS) program are on hand to help people, as are all of the world-renowned professors who are teaching courses.

The evaluations are always good, Little said, with which first-time attendee Martin Goetz, a doctoral student at the University of Switzerland, agreed.

“I highly enjoyed and greatly profited from both courses I took due to their breadth as well as their depth,” he said. “Although the pace and level of the courses were quite demanding, the instructors managed to keep everybody on board and, probably more importantly, keep everybody motivated.

“I especially enjoyed all instructors grappling with each question from the audience and providing valuable answers.”

Plus, it's camp, so it's fun – really! Little works hard to make camp fun. He and the other teachers wear scout camp uniforms, complete with sashes decorated with patches like the Stats Camp logo. Starting next year, participants get a patch for each course they complete.

Plus, Stats Camp is a great way to meet other people in similar fields, put together collaborative projects or find a like-minded spouse. With no small amount of pride, Little said he's had a couple of bachelor and bachelorette parties during Stats Camp, and one couple went on their honeymoon there.

The future

Soloski said she would recommend the camp to any researcher or consumer of research because it helped her not only in her research but also in advising students.

“I mentor many graduate students who are in the process of conducting their own research, and I am now able to counsel them on additional ways to conceptualize and model their research questions and the best practice methods in the field,” she said. “In my research, Stats Camp has helped me consider new means of examining my data and inspired me to continue to pursue complex methods.

“Overall, it has provided me with the confidence to continue with and improve my current line of research.”

Stats Camp Participant
Stats Camp instructors donated several books to a participant from Ethiopia to start a library in his country.

Little hopes other researchers will listen to Stats Camp alumni. For the next few years, at least, it likely will grow incrementally. He's adding three more courses to the second week of camp next year, and he would like to see participation increase by about 10 percent each year. If that happens by 2018, the last year he has the Grapevine Courtyard/Townplace Suites reserved, the hotel and conference space will be about maxed out. If he wants to keep growing, he'll have to find another space.

He's also looking into additional Stats Camp dates. Since many of his attendees come from Europe, he may plan a smaller camp there in 2016. It may be somewhere warm and sunny so he can hold the camp in January, or it may be a castle available for rent near Konstanz, Germany.

Wherever students participate, it's a valuable resource. Goetz said when he met Little he thought Little was just talking up his own program, but he only needed a few conversations to realize this was an experience that would significantly move his research forward. Despite all the intensity, participants have a good time on top of benefiting their research.

“The energy in the classroom when people have traveled far and wide to be there is different from the energy of somebody who's a student and has other concerns,” Little said.