Archives
UB summer institute offers alternative for advanced statistical study
By SUE WUETCHER
Reporter Editor
Their goal is admittedly ambitious: To compete with the Summer Program in Quantitative Methods of Social Research offered by the University of Michigan—considered to be the gold standard in advanced statistics education.
But UB sociologists Michael Farrell and Tai Kang are well on their way to making UB's Summer Institute in Advanced Statistics and Methods, now in its third year, an East-Coast alternative for faculty members and graduate students interested in learning the latest techniques in statistical analysis.
"Research methods are constantly evolving. If we don't work to keep up with recent developments, we will fall behind," says Kang, associate professor of sociology and co-director of the UB institute with James Donnelly, assistant professor in the Department of Counseling, School and Educational Psychology in the Graduate School of Education.
During its first two years, the UB program attracted participants from Roswell Park Cancer Institute, the Research Institute on Addictions and other UB units, as well as some from Rochester and southern Ontario. This year, advertising outside the region has brought inquiries from as far away as Toronto and Virginia, Kang says.
Farrell, professor and chair of the Department of Sociology in the College of Arts & Sciences, notes that one of the motives behind the decision to establish a summer institute at UB was the notion that researchers shouldn't have to travel to Michigan to learn new statistical techniques.
"There are many new techniques in data analysis and many new computer programs that have come out since people left graduate school, but they often don't have time to go to Michigan to learn them," he says. "We started looking and realized we have wonderful faculty at UB who can teach these things, so why should people have to go all the way to Michigan?
"Michigan has a 50-year head start on us, but on the other hand, I think there is a regional market for these kinds of workshops. We have a chance to carve out a niche."
The UB institute does not overly emphasize the mathematical underpinnings of statistics and research methods, Kang points out, but instead focuses on providing "hands-on experience" using these new computer programs.
Farrell explains that statistics often is taught in a way that emphasizes the mathematical underpinnings of the tests social scientists do, rather than their practical applications.
"We take people who have had those basic courses—where they've learned the basic mathematics and the assumptions of the statistics, but they haven't mastered these more advanced techniques," he says.
The institute devotes one day in each workshop to refreshing participants on "the foundations of the statistical methodologies," he says.
The remainder of the time is spent introducing the new computer programs and showing participants how they're applied to different types of statistical methodologies, such as structural equation modeling.
Farrell describes structural equation modeling as "a way of examining how one factor, let's say family stress, influences an intervening factor—parental support—which then influences a concluding or outcome factor—adolescent drug and alcohol use.
"It's a way of testing these complex, causal theories where you have indicators of underlying factors and a causal model linking factors to an outcome factor.
"Many people in the social science disciplines think 'I should know these latest techniques, but I haven't had time to learn them.' In one week's time, we give them a day-long refresher course, then walk them through these programs, then we spend a day helping them apply the programs to their own data. So when they're done, they're independent—they can go off and do it themselves."
Kang and Farrell say that institute participants not only learn new statistical methodologies, but also discover other researchers with similar interests who are using the methodologies as well.
"Many researchers are not aware that others share similar interests and are using similar research methodology," Kang says. "In a way, we're developing a community of scholars and researchers with shared interests and language."
Adds Farrell: "Often there might be only one or two people in a department who use these techniques, but (through the institute) they find other people in other departments who are using a method and find out who knows it well" and who they can call when they need assistance, or when they're looking for research collaborators. "So there's a community that's building as a consequence of the institute," he says.
The ultimate goal of the institute, Farrell and Kang say, is to move beyond offering only quantitative data-analysis methodologies by adding to the curriculum a wide range of methodologies used by social scientists, including methods of analyzing spatial data, census data and qualitative data.
This year's institute, scheduled to run for three weeks, from May 10-May 28, will present four modules: structural equation modeling, growth curve modeling, meta-analysis and hierarchical linear modeling.
For more information about the summer institute, go to http://www.cas.buffalo.edu/surveylab/ or contact Kang at taikang@buffalo.edu or Diane McMaster at dhm1@buffalo.edu.
The institute is sponsored by the Department of Sociology and the CAS Survey Research Lab, with the support of the CAS, the Office of the Provost and the Office of the Vice President for Research.