Title | Generalized Multilevel Models for Binary and Count Outcomes I |
---|---|
Course | Statistical Methods |
Institution | Valdosta State University |
Pages | 5 |
File Size | 101.9 KB |
File Type | |
Total Downloads | 21 |
Total Views | 124 |
Download Generalized Multilevel Models for Binary and Count Outcomes I PDF
Generalized Multilevel Models for Binary and Count Outcomes Lecture Notes
Generalized Multilevel Models Relatively new methods for modeling non-normal outcomes in longitudinal or cluster sample studies (mid- to late-1990s)
Combines ideas of generalized linear models with random-effects modeling
More highly complex alternative to GEE Using random-effects to account for within-subject variation, rather than specifying covariance matrix Generalized Multilevel Models Similar to MLMs Subset of fixed-effects assumed to vary randomly across clusters Random-effects represent heterogeneity due to unmeasured factors Mean of outcome is conditional on random-effects and is linearly related to fixed-effects (covariates) via a link function Continuously distributed outcome related via ‘identity’ link Assumes highest-order clusters are independent Generalized Multilevel Models Differ from MLMs
1
Generalized Multilevel Models for Binary and Count Outcomes Lecture Notes Distribution of random-effects unlikely normal Assumption of normally distributed residuals untenable Binomial: Can only assume 1 of 2 values (0,1) Predicted values of outcome are restricted within certain bounds, depending on distribution of outcome E.g., Probabilities of a binary outcome are between 0 and 1 and predicted values of a Poisson model are non-negative integers Variance of random-effects depends on predicted values of outcome and are not homogeneous Generalized Multilevel Models Three components… Conditional distribution of Yij (on random-effects) belongs to a member of the exponential family of distributions (e.g., Binomial, Poisson) Conditional mean of Yij is assumed to depend on fixed- and randomeffects and is linearly related to fixed-effects via link function
Structural linear model consisting of covariates Random-effects are assumed independent of covariates Generalized Multilevel Models Now distribution of random-effects represents unobserved latent variable(s)
2
Generalized Multilevel Models for Binary and Count Outcomes Lecture Notes Simple assumptions about distribution are made (i.e., normality)
However, for models with non-normal outcomes, interpretation of regression coefficients can be highly sensitive to difficult-to-verify assumptions about the distribution of random-effects, particularly dependence of latent variable distribution on covariates Generalized Multilevel Models For computational convenience and simplicity, random-effects are typically expressed in standardized form Random-effects variance term becomes part of generalized MLM equation (not case in standard MLMs) Regression coefficients and variance are on same scale…logit Thus, coefficients from generalized MLMs are amplified by amount of variance in random-effects, those in standard MLMs and GEEs are not Generalized Multilevel Models β represent influence of covariates on a specific subject’s mean response, holding other covariates, as well as that individual’s random-effects, fixed or constant
3
Generalized Multilevel Models for Binary and Count Outcomes Lecture Notes Thus, prediction of outcome must include random-effect related to cluster to which participant belongs, in addition to linear combination of predictors Interpretation of parameter estimates is now subject-or unit-specific (I prefer cluster-specific), as opposed to population-averaged, as in marginal (GEE) models
Fixed-effect coefficients represent effects for the ‘average subject’, rather than averaged across subjects Generalized Multilevel Models Interpretation of time-invariant or between-subjects effects is difficult as there is no subject-specific change occurring (models are subjectspecific) Change might be inferred from individuals who posses same randomeffects, but differ by 1-unit in terms of fixed-effects
Generalized Multilevel Models Why was this not the case with standard MLMs? When an identity link function is used (outcome is continuously and normally distributed), regression coefficients can be interpreted as either subject-specific or population-averaged
4
Generalized Multilevel Models for Binary and Count Outcomes Lecture Notes When averaged over distribution of random-effects, population means flow from a linear model with regression coefficients β Mathematically, this relationship does not hold for non-linear links used in presence of discrete outcomes Generalized Multilevel Models Why was this not the case with GEEs? GEEs models are only marginal/population-averaged No possibility for subject-specific interpretations Estimates in GEEs represent true averages of individual effects (regression lines)
5...