Chapter 11
Module 2For self-study: Running Tests in SPSS
Assignment 1.ANOVA: Univariate
Open the file Moviezone.sav. In this first assignment we will compare participants’ familiarity with art house movies. Scores on this measure indicate the number of selected movie titles participants checked in Question 11 (see the questionnaire). Some of the titles are non-existing movies (e.g., “Talking into the air”). These were included to identify those participants who gave socially desirable answers (see Chapter 5 on Checklists). The variable that was labelled “tottitb” is the sum of (correct) checks per participant. Now, we want to see whether this score differs significantly for three schools.
To do this, go to SELECT CASES in the DATA menu. Select schools 1, 2 and 3 (check Chapter 7, section 7.3, p. 20 for the exact instructions). Now run a Univariate Analysis, examining the effects of “school” on “totitb”, with “age” and “gender” as control variables. Click on OPTIONS and have SPSS provide you with descriptive statistics. Also, make SPSS display means for “school,” compare main effects, and finally, select the Bonferonni post hoc test under CONFIDENCE INTERVAL ADJUSTMENT. Interpret the results of the ANOVA with the help of the descriptive and the Bonferonni results. Which schools differ significantly here?
If you followed the instruction of this assignment correctly, this is what you had on your screen for the univariate test:
And the following for the Options requirement:
First, look at the descriptive statistics. What we see is an increase on the variable “school”. This may already give you an idea about the outcome. The Standard Deviations vary a little but not too much. You see the number of participants in the three groups. Ideally, these should be similar. Here the differences do not seem to be anything to worry about.
| Descriptive Statistics | |||
|---|---|---|---|
| Dependent Variable:tottitb | |||
| school | Mean | Std. Deviation | N |
| 1.00 | 2.2333 | 1.16977 | 60 |
| 2.00 | 2.4630 | 1.00401 | 54 |
| 3.00 | 2.8043 | 1.27575 | 46 |
| Total | 2.4750 | 1.16527 | 160 |
Now look further down your output and you will see the following table. We already suspected that the variable school had an effect on familiarity with art house movies. The results of the Univariate ANOVA confirm this. In your report, include the following information: F(2) = 3.1, p < . 048, that is, the degrees of freedom for your variable (school), the degrees of freedom from your error row (155), the value of the F-statistic (3.095, or rather 3.10; two digits is enough) and the value of p (< .048). You also see that age and gender had no effect on the variable.
| Tests of Between-Subjects Effects | |||||
|---|---|---|---|---|---|
| Dependent Variable:tottitb | |||||
| Source | Type III Sum of Squares | df | Mean Square | F | Sig. |
| Corrected Model | 9.340 a | 4 | 2.335 | 1.752 | .141 |
| Intercept | 2.782 | 1 | 2.782 | 2.087 | .151 |
| age | .054 | 1 | .054 | .041 | .840 |
| gender | .811 | 1 | .811 | .608 | .437 |
| school | 8.249 | 2 | 4.125 | 3.095 | .048 |
| Error | 206.560 | 155 | 1.333 | ||
| Total | 1196.000 | 160 | |||
| Corrected Total | 215.900 | 159 | |||
R Squared = .043 (Adjusted R Squared = .019).
Let us now look at the further results:
| Pairwise Comparisons | ||||||
|---|---|---|---|---|---|---|
| Dependent Variable:tottitb | ||||||
| (I) school | (J) school | Mean Difference (I-J) | Sth. Error | Sig. a | 95% Confidence Interval for Difference a | |
| Lower Bound | Upper Bound | |||||
| 1.00 | 2.00 | −.224 | .220 | .933 | −.757 | .309 |
| 3.00 | −.579 * | .233 | .042 | −1.143 | −.016 | |
| 2.00 | 1.00 | .224 | .220 | .933 | −.309 | .757 |
| 3.00 | −.356 | .250 | .472 | −.962 | .250 | |
| 3.00 | 1.00 | .579 * | .233 | .042 | .016 | 1.143 |
| 2.00 | .356 | .250 | .472 | −.250 | .962 | |
Based on estimated marginal means.
a.Adjustment for multiple comparisons: Bonferroni.
*The mean difference is significant at the .05 level.
| Univariate Tests | |||||
|---|---|---|---|---|---|
| Dependent Variable:tottitb | |||||
| Sum of Squares | df | Mean Square | F | Sig. | |
| Contrast | 8.249 | 2 | 4.125 | 3.095 | .048 |
| Error | 206.560 | 155 | 1.333 | ||
The F tests the effect of school. This test is based on the linearly independent pairwise comparisons among the estimated marginal means.
The last table again indicates that there is a significant effect (p = .048) of school. However, the Bonferroni test shows that it is only the difference between School 1 and 3 that is significant at the .05 level.