go to the School of Psychology home page Go to the UNE home page
Chapter 7 - Analysing the Data Part IV - Analysis of Variance Chapter 1 - Behavioural Science and research Chapter 2 - Research Design Chapter 3 - Collecting the Data Chapter 4 - Analysing the Data Part I - Descriptive Statistics Chapter 5 - Analysing the Data Part II - Inferential Statistics Chapter 6 - Analysing the Data Part III - Common Statistical Tests

 

Chapter 7: Analysing the Data
Part IV : Analysis of Variance

 

Scenario and Data Set #5
SPSS Output 7.2 General Linear Model - General Factorial

Univariate Analysis of Variance





Profile Plots

Figure 7.14 The default chart from selecting the plot options in Figure 7.13


Figure 7.15 A slightly improved version of the default.

Notice a more informative title and axis labels.


Comments on SPSS output

Between Subjects Factors

Here the variables being analysed are identified and the basic design (i.e., a 2 X 2 factorial design).

Descriptive Statistics

Here are all the means, sds, and Ns that we want.

Levene's Test of Equality of Error Variance

Here is the homogeneity test on the four groups of data (notice df = 3). The thing to focus on is the "Sig." value. Here .970 is clearly not significant, so we have no reason to doubt the assumption of homogeneity of variance.

Tests of Between Subjects Effects

Here is the main summary table for the analysis. There is more in this table than we really want. In particular, ignore "Intercept". This is like the constant in normal regression. First look at the "Sig." column and notice that the two main effects are not significant but that the interaction is highly significant. The plot of the means clearly displays this result as well.

Notice the Sums of Squares column. The "model" is the overall, total sums of squares (855.0) in the numcorr variable that is explained by the two main effects and interaction considered together. The SS for Lecture room and Testing room are both = 5.0, whereas the SS for the interaction is 845.0. Obviously these data have been set up to show a highly significant interaction while having two main effects that are not significant just to illustrate their independence.

The error Sums of Squares is 146.0. Together with the explained SS, a total of 1001.0 was the total variability in the numcorr variable. Eta-square for the interaction effect is a very high .884 or 88.4%.

 

 

 

© Copyright 2000 University of New England, Armidale, NSW, 2351. All rights reserved

UNE homepage Maintained by Dr Ian Price
Email: iprice@turing.une.edu.au