The Statistics Advanced module improves the accuracy and reliability of analyses through statistics designed for complex relationships. In particular, it offers the following functionalities:

​

• Bayesian statistics (available from V25)

• General linear models (GLM) and mixed model procedures

• Generalized linear models (GENLIN) which include widely used statistical models such as linear regression for normally distributed responses, logistic models for binary data and loglinear models for counting data.

• LMM (Linear Mixed Model), also known as HLM (Hierarchical Linear Model), which extends the GLM (General Linear Model) used in GLM to allow analysis of data with correlations and non-constant variability.

• GLMM (Generalized Linear Mixed Model) for use with hierarchical data and a wide range of results, including ordinal values.

• Survival analysis procedures to examine permanent and expiring data.

​

In addition, from V25 the following Bayesian statistics are available within the module:

• One Sample and Pair Sample T-test

• Binomial Proportion test

• Poisson Distribution Analysis

• Related Samples

• Independent Samples T-test

• Correlation Pairwise (Pearson)

• Linear Regression

• One-way ANOVA

• Linear-Log regression

 

GLM (General Linear Model)

 

• Describing relationships between a dependent variable and a set of independent variables. Models include: ANOVA (analysis of variance) with fixed effects, ANCOVA (analysis of covariance), MANOVA (multivariate analysis of variance) and MANCOVA (multivariate analysis of covariance).

• Using flexible contrast and design options to evaluate methods and variances and to test and predict methods.

• Associating and comparing categorical and continuous predictors to develop different models.

• Using more accurate LMMs (Linear Mixed Models) for the prediction of non-linear outcomes; e.g. which product the customer is expected to buy, considering hierarchical data structures.

• Formulating dozens of models, including split-plot design, multilevel models with fixed-effects covariance and random full-block design.

 

GENLIN (Generalized Linear Model)

 

• Providing a unified framework including CLM (Classical Linear Model) with normally distributed dependent variables, logistic and probit models, loglinear models for counting data and other non-standard regression models.

• Applying several useful general statistical models, including ordinal regression, Tweedie regression, Poisson regression, Gamma regression and negative binomial regression

 

Linear Mixed Model/Hierarchical Linear Model (HLM)

 

• Modelling methods, variances and covariances in data with correlations and non-constant variability, such as students in the classroom or consumers in the household.

• Formulating dozens of models, including split-plot design, multilevel models with fixed-effects covariance and random full-block design.

• Choosing from 11 types of non-spatial covariances, including independent first-degree, heterogeneous and autoregressive first-degree covariances.

• Obtaining more accurate results when using repeated measurement data, including situations where there are different numbers of repeated measurements, different intervals for different cases or both.

 

Procedure GEE (Generalized Estimating Equations)

 

• Extending GLM (Generalized Linear Model) procedures to handle correlated longitudinal and cluster data.

• Modelling subject correlations.

 

GLMM (Generalized Linear Mixed Model)

 

• Accessing, managing and analysing any type of dataset, including survival data, corporate databases or data downloaded from the Web

• Managing the GLMM procedure with ordinal values to develop more accurate models and predict non-linear results, such as the level of customer satisfaction.

 

Survival analysis procedures

 

• Choosing from a flexible and comprehensive set of techniques for understanding terminal events, such as survival rates, termination or failure.

• Using Kaplan-Meier estimates to measure the duration of an event.

• Selecting Cox regression to perform proportional hazards regression with dependent variables such as response time or duration response.

​

 

​