About models

Models describe complex relationships in a table’s data. A model consists of one or more outcome variables, one or more explanatory variables, and the hypothesized relationship between them.

Models are useful for testing whether a hypothesized relationship between variables is statistically significant after controlling for the effects of other variables. They are also useful for making predictions. Wizard supports a number of model types:

• Linear models (with optional weights) are used when the outcome is a continuous numeric variable or a date variable
• A probit, logit, or linear model is used when the outcome is an indicator variable, or is a category variables that only takes two values
• An ordered probit or multinomial logit model is used when the outcome is a category variable with more than two values
• A Poisson, geometric, or negative binomial model can be used when the outcome is a numeric variable consisting of count data
• A Cox proportional hazards model can be used when the outcome is a date variable, or a numeric variable consisting of failure-time data

Wizard will report the estimated coefficients of each model, along with standard errors, significance tests, and diagnostic information. After a model has been created, it can then be used to make concrete predictions.

See also:

• Creating a model
• About predictions
• Exporting a model as an interactive spreadsheet
• Copying model residuals to the host table
• Copying a model’s predicted values to the host table
• Exporting a model as R code
• Exporting a model as a Stata, SPSS, or SAS command (Pro only)

---

Back to Wizard Help