Multilevel Modeling and Longitudinal / Repeated Measure data analysis using Generalized Estimating Equations




Repeated Measures

In medical research, repeated measures (Longitudinal) data analyses are very common. Diabetes patients are followed periodically over time (every 6 months for example two years) to study the impact of treatment in HbA1c level. The usual regression analyses assume that the repeated measure data is independent, while they are dependent (correlated) in nature, or the observations within an individual are correlated. Assuming independence while the observations are correlated would provide a narrow CI for the regression coefficients thus paving way for highly significant results. Therefore the right analyses would be to capture the amount of correlation within the individual level and adjust that correlation in the regression analyses, This procedure is called generalised estimating equation (GEE)

Multilevel Models

These are basically regression models of data that have a hierarchical or clustered structure. Such kind of data arises routinely in many fields, for instance in growth studies (where students are nested within class rooms and class rooms are nested within schools. Thus the first level is class room and the higher level is school), family studies (within children nested within failies), and in medical research (with patients nested withinphysicians or hospitals). Clustered data may also arise based on specific research design. For example, in large scale survey research, the data collection is usually organised in some sort of multistage sampling design that results in clustered design. Hierarchial linear models or multilevel random coefficient models, Variance component models or multilevel random coefficient models are other names for the multilevel models. Repeated measures analyses is a single level MLM.

Practical and Software:

Practical will be followed by a lecture and real time data will be used for practicals. Practicals will be taught using SPSS, MLWIN (copy will be provided) and STATA.


Biostatisticians / Statisticians, Epidemiologists, Medical researchers and Social scientists who have basic knowledge in Regression anayses


Multilevel Modelling

Multistage sampling and multilevel theories

Macro, micro level units and cross-level relations

Statistical treatment of clustered data:

Aggregation, disaggregation, Intra-cluster correlation coefficient (ICC), design effect in two stage sampling

The random intercept model:

Definition, within and between group regressions, simple and multiple regression models

The random slope model:

Introduction, analyses and interpretation, explanation of random intercept and slope

Generalized Estimating Equations

Introduction to longitudinal data:

Traditional methods: Repeated measures, analysis of variance, summary measures and relationship with other variables

Generalized estimating equations- The best approach:

Introduction, working correlation structures, GEE for continuous and categorical outcome, interpretation and diagnostics (QIC)

MLM versus GEE:

Comparison between random coefficients and GEE

Course Fee : Rs 6500/- (Inclusive of 18% GST)


Participants will have to bear their own expenses for travel, boarding and lodging. The Organizers will provode Course Kit, Lunch and Snacks. However, the organizers may arrange basic accommodation with A/C facility in the college campus on request. Only limited accommodation is available in the campus and priority will be given to female participants.

Course fee shuld be paid in full by August 23, 2019.

Payment can be made by Demand Draft in favour of “Christian Medical College Vellore Association Account” payable at Vellore.


Please use the “Application Format” for registration to be sent along with the required demand draft.

For brochure and registration form, click here

Address for communication

Mr.M.Mafhan Kumar / Mr. Sasikumar

Clinical Epidemiology Unit

Christian Medical College, Vellore

Pincode 632004


Pnone : (0416) 2282329, 2282759

Add your thoughts

Your email address will not be published. Required fields are marked *