Christiansen CL (Center for Health Quality, Outcomes, & Economic Research (CHQOER) and Boston University), Loveland S
(CHQOER and Boston University)
Much of health services research requires study designs and data analysis methods that account for multiple levels of variables and outcomes. This workshop will provide an overview of multi-level models starting with proposal writing and study design and ending with the analysis and interpretation of results from multi-level modeling. The workshop participants will • learn the meaning of terms used in multilevel modeling • understand the effect of clustered data or clustered sampling designs on statistical power • learn to identify the level of analysis required for the study’s objectives • review examples of data and research questions in health services research that require the use of multi-level statistical models. Typical goals of analyzing this kind of data include estimating the effect of patient-level, provider-level and facility-level characteristics or outcomes. • learn how to interpret results from SAS and Stata multilevel software packages. Participants will receive a glossary of multilevel modeling terms, a list of resources and websites for further information on this topic, and a short guidebook on using and interpreting examples analyzed using the multilevel software.
We will use examples from health services research for all of the objectives. Results from different software packages will be presented, interpreted, and compared. The participants are encouraged to ask questions, contribute to the discussion, and provide feedback throughout the workshop.
The workshop is designed for health services researchers who use statistical models in their work, for decision-makers who rely on output from statistical analyses, and for applied statisticians who are unfamiliar with this area of statistics. In particular, it will help researchers who are interested in multilevel modeling to gain better understanding in this area.
Assumed Audience Familiarity with Topic:
Participants should be familiar with the statistical term 'variation' and have an understanding of standard statistical methods such as t-tests, least squares regression, and logistic regression.