T-test and Analysis of Variance (ANOVA): Difference between revisions

From psychmethods
(Created page with "=Theoretical introduction= The lecture begins with a basic introduction (what are typical research questions, what...")
 
No edit summary
Line 1: Line 1:
=Theoretical introduction=
=Theoretical introduction=
The [[Media:Correlation_and_regression_analysis_-_Lecture.pdf|lecture]] begins with a basic introduction (what are typical research questions, what kind of variables are used, what assumptions and requirements need to be met to apply the method), continues with an overview over fundamental equations (incl. some do-it-yourself in Excel and Octave), introduces then major regression types (standard, sequential/hierarchical, statistical/stepwise) and ends with some important issues (limitations, aspects to be attentive of, etc.).
 
The lecture includes to practical examples for calculation: An [[Media:Regression_analysis_-_Step-by-step.ods|LibreOffice/Excel-file]] with a demonstration how a correlation (i.e., a very simple regression with one variable; example from Field (2018), Ch. 8) is calculated by hand, as well as a [[Media:Regression_analysis_-_Step-by-step.txt|syntax file]] to run a regression analysis in MATLAB / Octave (from Tabachnik & Fidell (2013), Ch. 5.4; the data can also be found in a [[Media:Regression_analysis_-_Step-by-step.zip|ZIP-file]] to replicate the analyses in SPSS).
The [[Media:Analysis_of_Variance_-_Lecture.pdf|lecture]] briefly introduces how different variable types determine which statistical methods can be used, what assumptions determine when parametric statistical tests can be used (and which procedures can be used to test these assumptions). The the Analysis of Variance (ANOVA) is introduced starting with some background information and how to do a simple ANOVA manually (in Excel), followd by more extensive background information (such as typical designs, contrasts). Finally, more complex ANOVA models are introduced, starting with Analysis of Covariance (ANCOVA) over the multivariate forms (MANOVA and MANCOVA) to an application of these multivariate forms in profile analysis (e.g., for tracing development over time or distinguishing between subtests within a larger test battery).
 
The lecture includes to practical examples for calculation: An [[Media:Analysis_of_Variance_-_Step-by-step.ods|LibreOffice/Excel-file]] with a demonstration how a very simple ANOVA (with one three-step factor; example from Field (2018), Ch. 12) is calculated by hand, as well as a [[Media:ExamplesMATLAB.zip|ZIP file]] to run a several ANOVA models (ANCOVA, MANOVA and a profile analysis) in MATLAB / Octave (from Tabachnik & Fidell (2013), Ch. 6, 7, 8).


=Practical exercises using SPSS=
=Practical exercises using SPSS=
The [[Media:Regression_analysis_-_PC-exercise.pdf|PC exercise]] deals with the practical aspects of carrying out regression analyses in SPSS. There are two major parts dealing with linear regression and logistic regression. The part on Linear regression analysis begins with an assignment (on how to check requirements for calculating a linear regression), then demonstrates the equivalency of correlation to Linear regression (if there is only one predictor), how multiple predictors can be included in the regression model (incl. different methods for adding predictors) and end with how to assess the quality of your model. This is followed by an assignment to test the acquired knowledge practically. The logistic (binary) regression part consists of a basic introduction on the method (focussing on how a logistic function can be used to convert from a continuous to a binary outcome), followed by an assignment to use the method practically.
The [[Media:Analysis_of_Variance_-_PC-exercise.pdf|PC-exercise]] deals with the practical aspects of carrying out Analyses of Variance in SPSS. The exercise begins with an brief introduction on the preparation of ones data (e.g., file organization, hot to treat missing values, etc.), then demonstrates the equivalence of a t-test with an univariate ANOVA (if there is only one predictor with two steps), how multiple predictors / factors can be included in the ANOVA (incl. continuous predictors in an ANCOVA) and how multivariate dependent variables can modelled (MANOVA) to the use of the MANOVA within repeated-measures ANOVA and Profile analysis. This is followed by an assignment to test the acquired knowledge practically.


In addition, there is a file with an [[Media:Regression_analyses_-_Assignment.pdf|additional assignment]] and the [[Media:Regression_analyses_-_Assignment_-_Solution.pdf|solutions]] to it.
In addition, there is a file with [[Media:Analysis_of_Variance_-_Assignment.pdf|additional assignments]] and the [[Media:Analysis_of_Variance_-_Assignment_-_Solution.pdf|solutions]] to it.


Finally, there are two ZIP-files: [[Media:DataFiles.zip|One]] with the data files required in the exercise and the assignment, another [[Media:Regression_OutputSyntax.zip|another]] containing SPSS syntax (with comments) and SPSS output files for the analysis described in the main slides as well as for the additional assignment.
Finally, there are two ZIP-files: [[Media:DataFiles.zip|One]] with the data files required in the exercise and the assignment, another [[Media:Analysis_of_Variance_-_SyntaxOutput.zip|another]] containing SPSS syntax (with comments) and SPSS output files for the analyses described in the lecture slides as well as for the additional assignments.

Revision as of 17:04, 4 June 2019

Theoretical introduction

The lecture briefly introduces how different variable types determine which statistical methods can be used, what assumptions determine when parametric statistical tests can be used (and which procedures can be used to test these assumptions). The the Analysis of Variance (ANOVA) is introduced starting with some background information and how to do a simple ANOVA manually (in Excel), followd by more extensive background information (such as typical designs, contrasts). Finally, more complex ANOVA models are introduced, starting with Analysis of Covariance (ANCOVA) over the multivariate forms (MANOVA and MANCOVA) to an application of these multivariate forms in profile analysis (e.g., for tracing development over time or distinguishing between subtests within a larger test battery).

The lecture includes to practical examples for calculation: An LibreOffice/Excel-file with a demonstration how a very simple ANOVA (with one three-step factor; example from Field (2018), Ch. 12) is calculated by hand, as well as a ZIP file to run a several ANOVA models (ANCOVA, MANOVA and a profile analysis) in MATLAB / Octave (from Tabachnik & Fidell (2013), Ch. 6, 7, 8).

Practical exercises using SPSS

The PC-exercise deals with the practical aspects of carrying out Analyses of Variance in SPSS. The exercise begins with an brief introduction on the preparation of ones data (e.g., file organization, hot to treat missing values, etc.), then demonstrates the equivalence of a t-test with an univariate ANOVA (if there is only one predictor with two steps), how multiple predictors / factors can be included in the ANOVA (incl. continuous predictors in an ANCOVA) and how multivariate dependent variables can modelled (MANOVA) to the use of the MANOVA within repeated-measures ANOVA and Profile analysis. This is followed by an assignment to test the acquired knowledge practically.

In addition, there is a file with additional assignments and the solutions to it.

Finally, there are two ZIP-files: One with the data files required in the exercise and the assignment, another another containing SPSS syntax (with comments) and SPSS output files for the analyses described in the lecture slides as well as for the additional assignments.