Home  /  Products  /  Features  /  Other statistical methods

## Other statistical methods

Cronbach’s alpha

• Interitem correlations or covariances
• Generate summative scale
• Automatically reverse sense of variables

Kappa measure of interrater agreement

• Two unique raters
• Weights for weighting disagreements
• Nonunique raters, variables record ratings for each rater
• Nonunique raters, variables record frequency of ratings
• For one-way random-effects models
• Individual and average measurements
• Absolute agreement
• For two-way random-effects models
• Individual and average measurements
• Absolute agreement
• Consistency of agreement
• For two-way mixed-effects models
• Individual and average measurements
• Absolute agreement
• Consistency of agreement

Stepwise regression

• Linear
• Beta
• Competing risks
• Complementary log-log
• Cox
• GLM
• Interval
• Interval-censored parametric survival
• Logistic
• Conditional logistic
• Negative binomial
• Ordered logit
• Ordered probit
• Poisson
• Probit
• Quantile
• Skewed logistic
• Tobit
• Exponential, Weibull, Gompertz, lognormal, loglogistic, generalized gamma parametric survival

Nested model statistics

• Wald or likelihood-ratio tests
• Use with survey data

Kernel-density estimation

• Eight different kernels
• Control band width
• Overlay normal density or Student's t density

Box–Cox transform

• Can be applied to the left-hand side, right-hand side, or both
• Parameters can be the same or different
• Maximum likelihood
• Zero-skewness log

Power transforms

• Search for power transform that converts a variable into a normally distributed variable
• Graphical display of a power-transformed variable

Orthogonal polynomials

• Orthogonalize variables using modified Gram–Schmidt procedures
• Compute orthogonal polynomial for a variable

Tests of normality

• Shapiro–Wilk
• Shapiro–Francia
• Skewness and kurtosis test (D’Agostino, with and without Royston correction)
• Doornik–Hansen
• Henze–Zirkler
• Two by Mardia

Drawing samples from multivariate normal distribution

• Default is orthogonal data
• May specify desired means and covariance or correlation matrix
• Singular covariance matrix is permitted
• Set random-number seed to ensure reproducibility

Creating datasets with specified correlation structure

• Add variables to existing dataset or create new dataset
• Singular covariance or correlation structures are permitted
• Set random-number seed to ensure reproducibility

Collecting statistics into a dataset

• Collection from any command
• Collection of results for each group or subgroup of observations
• Collection from community-contributed or “official” commands