Center for Health Services Research in Primary Care
Data Analysis
The HSR&D Biostatistics Group supports
researchers with standard techniques for statistical analysis. Our
experienced staff also has experience the following, more specialized, areas:
Categorical data
Outcomes that are not continuous and violate
normality assumptions often require complicated analyses, and many of the recent
advances in statistical methodologies have been for the analysis of non-normal
data.
Our staff has experience and expertise in many of
the methods used for analyzing categorical data (i.e. logistic regression,
negative binomial regression, random effect models with binary data)
Longitudinal
Longitudinal data is characterized by repeated
observations over time on the same set of individuals.
Our staff has experience and expertise in
longitudinal data models to help researchers meet study objectives and
characterize the process of change.
Survival analysis
Survival analysis methods are designed for
longitudinal data on the occurrence of events, and statistical methods for
analyzing survival data are explicitly designed to incorporate censoring and
time-dependent covariates.
Our staff has experience in many of the methods used
for analyzing survival data (i.e. Cox regression, Kaplan-Meier).
Principled methods for missing data
In biomedical research studies, datasets where
missing data occur in some or all variables are common.
Multiple imputation is a principled statistical
method for analyzing datasets with missing data that accounts for the missing
values and the uncertainty they introduce in an analysis.
Our staff has expertise in the analysis of missing
data using MCMC (Markov chain Monte Carlo) multiple imputation methods.
