Research and Scholarly Activity
Dr. Torabi's research areas are: Spatial and Temporal Models, Cluster Detection, Small Area Estimation, Longitudinal Data Analysis, Survival Analysis, Generalized (General) Linear Mixed Models, Measurement Errors, and Robust Statistics.
In Spatial Statistics, he is interested in developing the smooth methods to describe the geographic distribution of mortality/incidence rates. To this end, the new methods are developed to compute reliable rate estimates by borrowing information over geographic regions, while the spatial aspect of the data is also taken into account. This work also focuses capturing time trends which manifest in sequences of maps of mortality rates produced over time. The mapping of mortality rates is then used to suggest factors which may be linked to various causes of mortality. An overall description of mortality may be used by policy makers to allocate health funding.
In Cluster Detection, the interest is to detect geographic regions with high/low rates of incidence, and statistical tests are used to identify geographic regions with higher/lower incidence rates than expected by chance alone. He is interested in developing new methods to detect the regions with high number of incidence or incidence-related events.
In Small Area Estimation, his interest is to find model-based estimates and their precision including prediction intervals. Model-based estimation, via suitable linking models, is used to borrow strength across related small areas and thus improve on the traditional area-specific direct estimators. Such estimates have many applications, e.g. in public health, agriculture, economy, policy making and allocation of funds. Examples of small area estimation include, but not limited to, poverty counts of school-age children at the county level, income for small places, health-related estimates for local regions, and so on. The main tool to do inference in small area estimation is generalized (general) linear mixed model approach.
In Longitudinal Data Analysis, the repeated observations for each individual are collected over time which has many applications in public health and medicine. He is interested in analyzing the longitudinal data particularly where there are missing observations and/or covariates measured by error.