CHAPTER 22 ELEMENTS OF HIERARCHICAL REGRESSION LINEAR MODELS
- Slides: 7
CHAPTER 22 ELEMENTS OF HIERARCHICAL REGRESSION LINEAR MODELS Damodar Gujarati Econometrics by Example, second edition
HIERARCHICAL LINEAR MODELS (HLMs) ØOther names for these models (or ones with similar features) include: ØMultilevel models (MLM) ØMixed-effect models (MEM) ØRandom-effects models (REM) ØRandom coefficient regression models (RCRM) ØGrowth curve models (GCM) ØCovariance components models (CCM) Damodar Gujarati Econometrics by Example, second edition
BASIC IDEA OF HLM ØData often have a hierarchical structure: ØMicro-level, or lower-level, data are often embedded in macro-level, or higher-level, data. ØThe primary goal of HLM is to predict the value of a micro-level dependent variable (i. e. , regressand) as a function of other micro -level predictors (or regressors) as well as some predictors at the macro level. Damodar Gujarati Econometrics by Example, second edition
BASIC IDEA OF HLM (CONT. ) ØAnalysis at the micro level is Level 1 analysis. ØAnalysis at the macro level is Level 2 analysis. Damodar Gujarati Econometrics by Example, second edition
HLM ANALYSIS OF THE NAÏVE MODEL ØIn HLM, we assume: Ø Where Y is the outcome, i is the micro-level observation, and j is the macro-level observation. Ø We further assume that the random intercept is distributed around its mean value with the error term vj: Damodar Gujarati Econometrics by Example, second edition
HLM ANALYSIS OF THE NAÏVE MODEL ØWe obtain: Ø Composite error term wij is the sum of macro-specific error term vj (Level 2 error term) and micro-specific error term uij (Level 1 error term). Ø Assuming these errors are independently distributed, we obtain the following variance: Damodar Gujarati Econometrics by Example, second edition
INTRA-CLASS CORRELATION COEFFICIENT ØThe ratio of the macro-specific variance to the total variance is called the intra-class correlation coefficient (ICC): ØGives the proportion of the total variation in Y attributable to the macro level. ØHigher ICC means macro differences account for a larger proportion of the total variance. ØSo we cannot neglect the influence of macro differences. Damodar Gujarati Econometrics by Example, second edition