ANOVA with SPSS

This exercise is designed to show you how complete a two-way ANOVA by computer.   You may either use SPSS or Excel.  If you wish to conduct this analysis by use of Excel, go to the Web page, ANOVA with Excel

As a preliminary step, you must secure the proper file for data analysis.  These files are available from the Data Sets Web page.  In particular, if you have not already done so, you need to download the file "ATTITUD1.SAV".  This file has been used in previous assignments (Chapters 8 and 13) and may already be in your possession.   

Some background on this assignment may help.  It involves the study of the impact of fallacious arguments on attitudes.  You already probably verified the reliability of the attitude measure by completing an exercise for Chapter 8 and, if you have completed the computer assignments for each chapter, you have computed a t test with this dependent variable as part of an exercise in Chapter 13. The dependent variable is attitude toward the topic (as created in chapter 8).  The study now has two independent variables.  The first independent variable is a true experimental variable: the use or non-use of the non sequitur fallacy in a persuasive messageThe second independent variable is a moderator variable: sex of the respondent.  This second independent variable is a moderator variable and, of course, it is not experimentally manipulated. Thus, each independent variable has two levels.  The experimental variable dealing with the use of the non sequitur fallacy is identified as "falexper" (1= with the fallacy presented to subjects; 2=without the fallacy presented to subjects).  The remaining (moderator) independent variable is "sex" and is identified with 1 equal to males and 2 equal to females.  Because the researcher is interested in examining any main an interactions a 2 x 2 factorial analysis of variance is the chosen variable of interest.  Simple defaults of alpha risk of .05 are applied in this exercise.

bulletStart SPSS and load ATTITUD1.SAV (if you have not already done so).
bulletFrom the "Analyze" menu, click on "General Linear Model" and on a subsequent menu that appears, "Univariate."  In the dialog box titled "Univariate" place "Attitude" in the "Dependent Variable" list.  Place "sex" and "falexper" in the "Fixed Factor(s)" list.  This box should appear as follows:

bulletClick on the "Options" button.  On the "Univariate: Options" dialog box check the "Descriptive statistics" and "Homogeneity tests" boxes and click on the "Continue" button.  The box should look as follows:

Click on the "OK" button. 

NOTE: For purposes of the exercise, only the default methods are employed. In the above case, the result was produced by using the Type III model for calculating the sums of squares.  This model, however, is most suitable for designs with equal sample sizes and no missing cells.  Since unequal sample sizes were present in this study,   you might wish to specify the Type II model for the factorial analysis of variance.   To do so, select the "Analyze" menu, and on subsequent menus that appear, click on "General Linear Model" and "Univariate."  In the dialog box titled "Univariate," click on the "Model" button.  On the "Univariate Model" dialog box select "Full factorial" and enter "Type II" in the "Sum of squares:" box.









bulletSave your results, print them out, and answer these questions:

1.  What are the critical values of F at alpha risk of .05?______________

2.  Are there any significant main effects (if so, for which variables)?


3.  Is there a significant interaction effect (if so, what type of interaction

     is it)?________________________________________________________