Hierarchical generalized linear model

From Wikipedia, the free encyclopedia
Jump to: navigation, search

In statistics, hierarchical generalized linear models (HGLM) extend generalized linear models by relaxing the assumption that error components are independent.[1] This allows models to be built in situations where more than one error term is necessary and also allows for dependencies between error terms.[2] The error components can be correlated and not necessarily follow a normal distribution. When there are different clusters, that is, groups of observations, the observations in the same cluster are correlated. In fact, they are positively correlated because observations in the same cluster share some common features. In this situation, using generalized linear models and ignoring the correlations may cause problems.[3]

Overview and model[edit]


In a hierarchical model, observations are grouped into clusters, and the distribution of an observation is determined not only by common structure among all clusters but also by the specific structure of the cluster where this observation belongs. So a random effect component, different for different clusters, is introduced into the model. Let be the response, be the random effect, be the link function, , and is some strictly monotone function of . In a hierarchical generalized linear model, the assumption on and need to be made:[2] and

The linear predictor is in the form:

where is the link function, , , and is a monotone function of . In this hierarchical generalized linear model, the fixed effect is described by , which is the same for all observations. The random component is unobserved and varies among clusters randomly. So takes the same value for observations in the same cluster and different values for observations in different clusters.[3]


Identifiability is a concept in statistics. In order to perform parameter inference, it is necessary to make sure that the identifiability property holds.[4] In the model stated above, the location of v is not identifiable, since

for constant .[2] In order to make the model identifiable, we need to impose constraints on parameters. The constraint is usually imposed on random effects, such as .[2]

Models with different distributions and link functions[edit]

By assuming different distributions of and , and using different functions of and ', we will be able to obtain different models. Moreover, the generalized linear mixed model (GLMM) is a special case of the hierarchical generalized linear model. In hierarchical generalized linear models, the distributions of random effect do not necessarily follow normal distribution. If the distribution of is normal and the link function of is the identity function, then hierarchical generalized linear model is the same as GLMM.[2]

Distributions of and can also be chosen to be conjugate, since nice properties hold and it is easier for computation and interpretation.[2] For example, if the distribution of is Poisson with certain mean, the distribution of is Gamma, and canonical log link is used, then we call the model Poisson conjugate HGLM. If follows binomial distribution with certain mean, has the conjugate beta distribution, and canonical logit link is used, then we call the model Beta conjugate model. Moreover, the mixed linear model is in fact the normal conjugate HGLM.[2]

A summary of commonly used models are:[5]

Commonly used models
Model name distribution of y Link function between y and u distribution of u Link function between u and v
Normal conjugate Normal Identity Normal Identity
Binomial conjugate Binomial Logit Beta Logit
Poisson conjugate Poisson Log Gamma Log
Gamma conjugate Gamma Reciprocal Inv-gamma Reciprocal
Binomial GLMM Binomial Logit Normal Identity
Poisson GLMM Poisson Log Normal Identity
Gamma GLMM Gamma Log Normal Identity

Fitting the hierarchical generalized linear models[edit]

Hierarchical generalized linear models are used when observations come from different clusters. There are two types of estimators: fixed effect estimators and random effect estimators, corresponding to parameters in : and in , respectively. There are different ways to obtain parameter estimates for a hierarchical generalized linear model. If only fixed effect estimators are of interests, the population-averaged model can be used. If inference is focused on individuals, random effects will have to be estimated.[3] There are different techniques to fit a hierarchical generalized linear model.

Examples and applications[edit]

Hierarchical generalized linear model have been used to solve different real-life problems.


For example, this method was used to analyze semiconductor manufacturing, because interrelated processes form a complex hierarchy.[6] Semiconductor fabrication is a complex process which requires different interrelated processes.[7] Hierarchical generalized linear model, requiring clustered data,is able to deal with complicated process. Engineers can use this model to find out and analyze important subprocesses, and at the same time, evaluate the influences of these subprocesses on final performance.[6]


Market research problems can also be analyzed by using hierarchical generalized linear models. Researchers applied the model to consumers within countries in order to solve problems in nested data structure in international marketing research.[8]


  1. ^ Generalized Linear Models. Chapman and Hall/CRC. 1989. ISBN 0-412-31760-5. 
  2. ^ a b c d e f g Y. Lee; J. A. Nelder (1996). "Hierarchical Generalized Linear Models". Journal of the Royal Statistical Society, Series B. 58 (4): 619–678. JSTOR 2346105. 
  3. ^ a b c Agresti, Alan (2002). Categorical Data Analysis. Hoboken, New Jersey: John Wiley & Sons, Inc. ISBN 0-471-36093-7. 
  4. ^ Allman, Elizabeth S.; Matias, Catherine; Rhodes, John A. (2009). "Identifiability of Parameters in Latent Structure Models with Many Observed Variables". The Annals of Statistics. 37, No. 6A (6A): 3099–3132. Bibcode:2008arXiv0809.5032A. arXiv:0809.5032Freely accessible. doi:10.1214/09-AOS689. 
  5. ^ Lars Rönnegård; Xia Shen; Moudud Alam (Dec 2010). "hglm: A Package for Fitting Hierarchical Generalized Linear Models". The R Journal. 2/2. 
  6. ^ a b Naveen Kumar; Christina Mastrangelo; Doug Montgomery (2011). "Hierarchical Modeling Using Generalized Linear Models". Quality and Reliability Engineering International. 
  7. ^ Chung Kwan Shin; Sang Chan Park (2000). "A machine learning approach to yield management in semiconductor manufacturing". International Journal of Production Research. 38 (17): 4261–4271. doi:10.1080/00207540050205073. 
  8. ^ Burcu Tasoluk; Cornelia Dröge; Roger J. Calantone (2011). "Interpreting interrelations across multiple levels in HGLM models: An application in international marketing research". International Marketing Review. 28 (1): 34–56. doi:10.1108/02651331111107099.