Multilevel Modeling (MLM) is an increasingly popular statistical tool which extends the ordinary regression analysis to the situation of clustered or hierarchically structured data. Application of standard methods to data having hierarchical or nested structure leads to the violation of the assumption of independence of errors. The current study intends to explain the importance of MLM over standard regression techniques by identifying the factors associated with symptoms of depression among urban adolescents in India. The data from cross-sectional school-based study was used to identify risk factors for depression among adolescents. Conventional and different multilevel general linear models were fitted and compared using Likelihood Ratio tests, AIC, BIC along with regression estimates and standard errors. The results showed that ignoring the structure of the data lead to biased inferences and hence analysis should always account for the sampling technique adopted during data collection as well as structure of the data.
Presented in Session 3. Population, Development, & the Environment; Data & Methods; Applied Demography