Multilevel Analysis Quotes

“A relevant model here is the random effects ANOVA model.1 Denoting by Yij the outcome value observed for micro-unit i within macro-unit j, this model can be expressed as where μ is the population grand mean, Uj is the specific effect of macro-unit j, and Rij is the residual effect for micro-unit i within this macro-unit. In other words, macro–unit j has the ‘true mean’ μ + Uj, and each measurement of a micro-unit within this macro-unit deviates from this true mean by some value Rij. Units differ randomly from one another, which is reflected in the fact that Uj is a random variable and the name ‘random effects model’. Some units have a high true mean, corresponding to a high value of Uj, others”
― Tom A.B. Snijders, Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling

“If measures are taken only at the micro level, analyzing the data at the micro level is a correct way to proceed, as long as one takes into account that observations within a macro-unit may be correlated. In”
― Tom A.B. Snijders, Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling