Statistical Research Across Columbia
(mini talks, grand topics)
Teaching,
Application &
Research
at Columbia
Conference 2004
Dr. Tor Wager
Department of Psychology
Topic: Statistics in functional brain imaging
Abstract: Recently, there has been an explosion of interest in using brain imaging as a research tool for psychology and neuroscience. This research investigates the neural correlates of human performance in diverse fields, including cognitive function, decision making and preferences, social behavior, and emotional experience. In fMRI, the most popular imaging technique, data consist of a timeseries of 3-D images (each containing ~100,000 observations) collected during two or more task conditions. These data are replicated across individual participants, with the goal of making inferences about a population of human subjects. Statistics plays a critical role in the analysis in several domains, all of which are currently undergoing rapid development. One is timeseries analysis, which involves modeling the hemodynamic response to psychological tasks and estimating both within- and between-subjects parameters and error. A second area is statistical thresholding of images to determine which brain areas are believed to show true differences in activity among tasks. A third area is multivariate analysis of patterns of activity across space and time. A fourth area is design of experiments for fMRI. And, finally, a fifth area is meta-analysis of imaging data. I will briefly discuss some current approaches and issues in each of these areas.
Discussant: Dr. Martin Lindquist, Statistics
Dr. Andrey Rzhetsky
Deptt of Biomedical Informatics,
Columbia Genome Center
Topic: Chains of collective reasoning in biology
Abstract: n/a
Discussant: Dr. Tian Zheng, Statisitcs
School of Social Work
Topic: Long term effects of an HIV sexual risk reduction interv ention for heterosexual couples
Abstract: This presentation will describe Project Connect, a relationship-based HIV prevention intervention for women and their regular sexual partners. Project Connect was tested in a randomized clinical trial designed to examine the efficacy of the intervention and test whether it was more effective when both members of the couple received the intervention together than when the female partner received it alone. A total of 217 couples were randomized to either (1) a six-session intervention provided to the women and her sexual partner together (n=81); (2) the same intervention provided to the woman alone (n=73); or (3) a one-session health information control condition provided to the woman alone (n=63). Findings suggest that the intervention was efficacious in reducing unprotected sex at six and twelve months post-intervention. No significant differences were observed when comparing whether couples received the intervention together or when the woman received it alone.
Discussant: Dr. Jennifer Hill, SIPA