TABFIT <table>,<name of model>,L estimates the parameters of a log-linear model for frequency data given as a multiway table (F table) or estimates the parameters of an ANOVA model for data given as a multiway table (X table). Both types of TABFIT models are special cases of GLM (generalized linear models) and treated here by the same program. The type of the multiway table (F or X) is given as the last parameter of the TABLE definition for <table>. It specifies whether a log-linear model (with the log link) or a classical ANOVA model (with the identity link) is estimated. TABFIT <table>,<name of model>,L Parameter L is optional and gives the first line for the results in the edit field. The model to be fitted has to be defined by a MODEL specification typed on any two consecutive lines in the edit field as follows: MODEL <name of model> <list of terms to be fitted> <table> has to be written in the edit field and it has to be defined by a TABLE specification. The structure of <table> is that of tables produced by the TAB operation (See TAB?). Thus, multiway tables computed by TAB can be used as input for TABFIT without any changes. Each term in the model may consist either of one (main effect) or several (interaction between factors) factors. The factors are notated by the initials of the classifier names in the table. The model is given on two consecutive edit lines in the form: MODEL <name of the model> <list of the terms to be fitted> The terms to be fitted are separated by a '+' operator. If the term is an interaction term of two or several classifiers, the factors must be separated by '.'.For example, A.B represents an interaction of A and B. Also a '*' operator can be used: For example, A*B=A+B+A.B A*B*C=A+B+C+A.B+A.C+B.C+A.B.C The user can remove any existing term from the model by operator '-': For example, A*B*C-A.B.C=A+B+C+A.B+A.C+B.C The constant term is given as '1'. Thus a typical model definition is MODEL M1 1+S+B*K where S,B and K are initials of classifiers of the multiway table. The extent of results given by TABFIT is controlled by the RESULTS specification. If RESULTS=0, only minimal information (i.e. deviance, degrees of freedom and structure of the model) is printed. Otherwise also the estimates of the parameters and their standard errors are given. Furthermore, if a value >70 (like RESULTS=100) is given, the fitted values are displayed as a new multiway table having the same structure as the original table. In all cases, the design matrix generated by TABFIT is saved as a matrix file XTAB.M and the covariance estimates of parameters as a matrix file PCOV.M on the current data disk. T = More information on table management S = More information on statistical analysis

More information on Survo from www.survo.fi

Copyright © Survo Systems 2001-2012.

webmaster'at'survo.fi

Copyright © Survo Systems 2001-2012.

webmaster'at'survo.fi