Error z value Pr ( >| z | ) ( Intercept ) -14.0110 3.09013 -4.534 5.785e-06 H 0.1153 0.04052 2.845 4.438e-03 ( Dispersion parameter for Gamma family taken to be 2 ) Null deviance : 82.943 on 30 degrees of freedom Residual deviance : 72.609 on 29 degrees of freedom -2 * logLik ( max ), constant var. = -69.62317 -2 * logLik ( max ), model = -74.79994 LRT = 5.176767 on 2 df, p - value = 0.075141 Iteration 1 : deviance = -69.62317 Iteration 2 : deviance = -74.73305 Iteration 3 : deviance = -74.73319 Iteration 4 : deviance = -74.73319 Iteration 5 : deviance = -74.73319 Gaussian dispersion model - Call : lm.disp ( formula = y ~ H + D, var.formula = ~ H, data = minitab ) Model for the mean : y ~ H + D Estimate Std. (1987), Modelling variance heterogeneity in normal regression models using GLIM, Applied Statistics, 36, 332вЂ“339. The value of the deviance at the end of the iterative procedure.Д«ased on a similar procedure available in Arc (Cook and Weisberg, ) ReferencesДЄitkin, M. The value of the deviance at the beginning of the iterative procedure, i.e. The summary method can be used to obtain and print a summary of the results.ДЄn object of class 'dispmod' is a list containing the following components:ДЄn object of class 'glm' giving the fitted model for the mean function see glmДЄn object of class 'glm' giving the fitted model for the variance function see glm. Lm.dispmod() returns an object of class 'dispmod'. Where z_i may contain some or all the variables in x_i and other variables not included in x_i z_i is however assumed to contain a constant term.ДЄnd it is fitted by maximum likelihood following the algorithm described in Aitkin (1987). Suppose a response y is modelled as a function of a set of p predictors x through the linear model Gaussian dispersion models allow to model variance heterogeneity in Gaussian regression analysis using a log-linear model for the variance. An offset term can be included in the formula instead or as well, and if both are specified their sum is used. This can be used to specify an a priori known component to be included in the linear predictor during fitting. By default is set to na.omit, but other possibilities are available see na.omit.ДЄn optional list as described in the contrasts.arg argument of. See glm.control.ДЄn optional vector specifying a subset of observations to be used in the fitting process.ДЄ function which indicates what should happen when the data contain NA's. Tolerance value for checking convergence. Integer giving the maximal number of iterations for the model fitting procedure. By default the variables are taken from environment(formula), typically the environment from which the function is called. For the details of model formula specification see lm and formula.ДЄn optional data frame containing the variables in the model. This must be a one-sided formula if omitted the same terms used for the mean function are used. For the details of model formula specification see lm and formula.ДЄ symbolic description of the variance function of the model to be fit. Lm.disp ( formula, var.formula, data = list (), maxit = 30, epsilon = glm.control () $ epsilon, subset, na.action = na.omit, contrasts = NULL, offset = NULL )ДЄ symbolic description of the mean function of the model to be fit.

Minitab 18 gaussian