Simple t-Tests

Data from experiment with two conditions where the level of measurement of the dependent variable is at the interval or ratio level can be analyzed using one of two variations of the student's t statistic, one for between subject's designs and another for within or repeated subject's designs.

t-Tests for Between Subjects Designs

Burke and Greenglass (1989) have concluded that, "It may be lonely at the top but it's less stressful." These authors found a significant difference between teachers and principles on a measure of burnout (burnout is an extreme form of stress), with teachers exhibiting higher levels of burnout than principals did. Below are some hypothetical data for a further test of this hypothesis.

Burnout Scores:

Teachers:  42, 38, 44, 33, 49, 42, 40
Principals:  28, 35, 40, 38, 30, 24

                - first variable:
                        - Variable Name: job
                        - Type: Numeric1.0
                        - use the Define Label feature to indicate that 1 = teachers and 2 = principals

                - second variable:
                    - Variable Name: burnout
                    - Type: Numeric8.2 (default value)
                    - Variable Label: Burnout Scores

Now enter the data in the SPSS Data Editor. The first column will contain a 1 or a 2 that indicates whether the score is for a teacher (enter 1) or a principal (enter 2). The second column contains the burnout score. For example, the first row of our data would have a 1 in job and a 40 in burnout. Repeat this procedure for all 13 scores.

Once all of the data is entered, use the main menu Analyze/ Compare Means/ Independent-Sample T Test to do the analysis. The screen you will see looks like this. The two variables in the data are noted in the left panel. The variable burnout is the variable we want to test. Highlight this variable and click the arrowhead to move burnout to the Test Variable(s) section.

Now highlight job and click the arrow to move this variable to Grouping Variable window. Your panel should look like this:

The final step is to define the groups that we want compared in this analysis. Press the Define Groups key to produce this panel. We have defined our groups using 1 for teachers and 2 for principals. Just type 1 in the Group 1 box and 2 the Group 2 box. Finally, press Continue to return to the T Test window. Just press OK to do the analysis.

When you press OK the analysis is done and the SPSS Output Viewer is displayed.

The left panel is referred to as the Output Outline box and contains an outline of the type of SPSS procedures that have been performed. The left window contains the output file. (You may need to use the up/down or left/right scroll bars to see all of the output.)

In the case of a t-test, in the top box we are given the N, means, standard deviations, and standard error of the mean for each group (Group Statistics box).

The bottom box provides two sets of results, one when the variances are assumed equal and one when they are not equal. To determine which set to use, look at the results marked Levene’s Test for Equality of Variances. If the value in the Sig column is greater than .05, we conclude that the variances are equal so we use the first set of results. If this value is less than .05, we conclude that the variances are not equal and use the second set. In our case, the value .337 is greater than .05 so we use the Equal variances assumed results..

Ruth, Mosatche, and Kramer (1989) tested the hypothesis that people would state a preference for purchasing a liquor product if the product were advertised with sexual symbolism. Six participants were shown advertisements with and without sexual symbolism. In each condition, the participants were asked to indicate their likelihood of purchasing the product. Sexual symbolism was defined psychoanalytically. For example, one ad presented several skyscrapers with steeples (sexual) while another was a country scene with grazing cows (non-sexual). The data from the study are given below. Higher scores indicate a greater willingness to purchase the product.

Sexual Symbolism               6, 5, 4, 5, 4, 6

No Sexual Symblism           3, 5, 2, 3, 1, 3

Because the same participants were used in both conditions, this is a within-subjects design. That means that the data must be entered in a different manner. Rather than one column indicating the group and one indicating containing the data points, we will have two columns, one for each condition. That is, you should create two variables with the following characteristics:

Once that is done, use the main menu to get the proper statistic (Statistics/Compare Means/Paired-Samples T Test). The first window you will see will look like this.

We need to indicate which two variables we want included in this analysis.

Highlight the Sexual Symbolism variable and then No Sexual Symbolism. Note that the names are now shown in the Current Selections panel at the bottom left of the panel. Press the large arrow in the middle of the screen to indicate that these are the variables you want in the analysis. Your screen should look like this.

Press OK to do the analysis. Your output will look like this:

The top right box gives the summary statistics for the individual variables while the second provides the correlation between the two variables. The bottom box gives us the t value. Use this value and the associated df to determine if the difference is significant.

Back to SPSS Contents