Distribution of Data

When Count data is specified as the Type of Data, and the Distribution of Data field is left empty, columns in the data set are assumed to follow a Poisson distribution and counts analyzed using PROC GLIMMIX. The table below shows the link function that is used in PROC GLIMMIX according to what distribution is selected or used by default.

When Continuous is specified as the Type of Data, the Distribution of Data field is inactive. Columns in the data set are assumed to follow a Gaussian (Normal) distribution and are analyzed using PROC MIXED.

The following link functions are used according to the Distribution of Data selected:

Distribution	Link Function
Beta	Logit
Binary	Logit
Binomial	Logit
Exponential	Log
Gamma	Log
Gaussian/Normal	Identity
Generalized Poisson	Log
Geometric	Log
Inverse Gaussian	Power with Exponent - 2
Lognormal	Identity
Multinomial	Cumulative Logit (Ordinal Data) / Generalized Logit (Nominal Data)
Negative Binomial	Log
Poisson	Log
t	Identity

To Specify a Distribution of Data:

Left-click the drop-down list box.

Left-click a distribution.