Statistics & Methods Centre - Regression-type models

Regression-type models, presented here, belong to the class of dependence models, in which there is a single response variable (criterion variable, dependent variable) and one or more predictor variables (independent variables). These models are special cases of the general linear model. If the response variable is numerical the models are called multiple regression models, when the response variable is the logarithm of a categorical variable the models are called logistic regression models, and when the response variable is categorical with two or more categories, the models are called discriminant analysis models. When there are only categorical predictors the model is called analysis of variance (see under multivariate statistics).

Multiple regression
Modelling a single numerical response variable by several (mostly numerical, but possibly binary) predictor variables.
Moore & McCabe, Chapter 11: Multiple regression.
Field, Chapter 5: Regression.
Hair et al., Chapter 4: Multiple regression.
Meyer et al., Chapter 5A: Multiple regression, Chapter 5B: Multiple regression using SPSS.
Later sections of the chapters, including analyses of residuals, influence measures.
Moderation and mediation (by David Kenny)
Software
SPSS => Analyze => Regression => Linear
Annotated output multiple regressionn - UCLA-ATS
Reporting regression analysis in publications
Reporting - examples in journal articles
Power: Hair et al. Table 4-7 (p. 195, 6th edition)
Effect size: Field, pp. 172-174 (2nd edition).

Logistic regression
Moore & McCabe, Chapter 16: Logistic regression.
Field, Chapter 6: Logistic regression; Effect size: section 6.5.5
Hair et al., Chapter 5: Multiple discriminant analysis and logistic regression, pp. 335-377.
Meyer et al., Chapter 6A: Logistic regression, Chapter 6B: Logistic regression using SPSS.
Hosmer, D.W. & Lemeshow, S. Applied Logistic Regression(2nd Edition). New York: Wiley.
Software
SPSS => Analyze => Regression => Binary logistic
Effect size:
Annotated output logistic regression - UCLA-ATS
Example with annotated output - UCLA-ATS
Reporting logistic regression in publications
Reporting - examples in journal articles: Pallant (2007). Chapter 14 (p. 178).; Field (2005), Section 6.6.
Power:
Effect size:Field (2005), Section 6.3.2.

Discriminant analysis
Linear discriminant analyis has one categorical response variable and several (most numerical) predictors. Linear combinations of the predictors are sought (the discriminat functions), such that the group means differ as much as possible on the discriminant functions together.
Field, Chapter 14, sections 14.7 & 14.8
Hair et al., Chapter 5: Multiple discriminant analysis and logistic regression, pp. 269-354.
Meyer et al., Chapter 7A: Discriminant function analysis, Chapter 7B: Two-group discriminant function analysis using SPSS.
Huberty, C.J. (1994). Applied discriminant analysis. New York: Wiley
Software
SPSS => Analyze => Classify => Discriminant
Annotated output - UCLA-ATS
Example with annotated output - UCLA-ATS
Example with annotated output - Garson site
Reporting discriminant analysis in publications
Reporting - examples in journal articles
Power:
Effect size: