GLZ Syntax - Keyword Response

  { Variable [(Value List)] };

Example 1

RESPONSE = SURVIVAL ("NO" "YES");

Example 2

RESPONSE = STRESS_R;

Required keyword; specify a dependent (response) variable for the current analysis; if the Distribution selected from the GLZ Startup Panel - Advanced tab is either Binomial, Multinomial, or Ordinal multinomial (i.e., if the response variable is categorical in nature; see Distribution and Link function on the GLZ Startup Panel - Advanced tab), then you can also specify the codes that are used in the response variable to identify the group (category) membership of each observation. For a discussion of the different link functions and distributions, see the Introductory Overview.

Reference level for categorical dependent (response) variables

The last category (level) that is specified for a categorical dependent (response) variable will be the reference category for the comparisons with the other categories. So, for example, if a multinomial dependent (response) variable with k = 3 levels is analyzed, the k-1 = 2 parameters for each predictor (effect column) pertain to the comparison of (1) the first level with the last level, and (2) the second level with the last level of the dependent (response) variable.

Note: response codes for binomial response variables. When selecting (or specifying via syntax) response codes for binomial response variables (e.g., for logit models), the two selected values (numeric codes or text values) are recoded for the analysis, so that the first code that  is specified will be recoded to 1, and the second value will be recoded to 0. For example, if you specified two text values, No and Yes, then the first value (No) will be recoded for the analysis to 1, and the parameter estimates can be interpreted with respect to the probability of responding with No; so, if in such an analysis a positive parameter estimate is found for a particular predictor variable x¸ then it means that the more of x the greater is the likelihood of obtaining the response No. Note that the default coding for binomial response variables in the Nonlinear Estimation module may not be the same as that performed in GLZ. One way to verify the particular coding that was performed is to review the spreadsheet of the predicted responses on the Resid 1 tab of the results dialog; that spreadsheet will report in the first column the recoded values (0 or 1) of the response variable.

Applies to

GLZ