Chapter 10
Module 2For self study: Running Tests in SPSS
Assignment 10.6Assignment 6
We take a look again at the data set Moviezone.sav. Open the questionnaire that the students used, and look at Question 10. We can distinguish two types of responses: those indicating a negative attitude (first and third option) and those indicating a positive attitude toward movie houses (the second and fourth), irrespective of whether respondents were ever in one. For this assignment we first have to conduct a RECODE procedure. The first step you need to take is reduce the number of scores on Question 10, and thus reduce the number of cells in your Crosstab. Go to RECODE in the TRANSFORM menu. Select RECODE INTO DIFFERENT VARIABLE. Select Question 10 (“filmh”) and recode values 1 into 1, 3 into 1, 2 into 2, and 4 into 2 for a new variable you may label “filmh1” (check the relevant paragraph in Chapter 7 for details on this procedure). Check the options in Question 10 to see why you do this. To make the SPSS output easier to interpret you may consider entering the value labels “negative” for 1, and “positive” for 2 in the VARIABLE VIEW.
Now conduct a Chi-Square test to see whether male and female respondents are distributed over the two categories (“negative” and “positive”) as one would expect if there is no relation between gender and attitude toward movie houses. (a) Explain why you need to run a Chi-square and not some other test. (b) Interpret the results.
We ran the recode procedure making a new variable for attitude toward movie houses (Filmhatt in the table below). For your reports never trouble your readers with silly variable names, but mention them full out in text and tables. Also, you can sometimes include the extensive tables SPSS produces in an appendix, but you should not use them in the main text of your report. For now, in these assignments you should produce parts of the output, so that you can see whether your results correspond with ours. Let us see whether they do for this assignment.
| Filmhatt * gender Crosstabulation | ||||
|---|---|---|---|---|
| Count | ||||
| gender | ||||
| male | female | Total | ||
| Filmhatt | 1.00 | 103 | 89 | 192 |
| 2.00 | 207 | 268 | 475 | |
| Total | 310 | 357 | 667 | |
Above you see the so-called cross tab, to be found under ANALYZE, DESCRIPTIVE STATISTICS, CROSSTABS. In it you see how the women and men of this sample are divided into two groups: those with a positive and those with a negative attitude toward movie house films. We also have the percentages within columns and rows. It may take you a while to figure out which is which. Take your time for this, in order to avoid an incorrect interpretation. Look at the column for male participants first. What you see is that within this “level” of the variable gender 33.2% (103 out of 310) has a negative and 66.8% (207 out of 310) has a positive attitude toward movie houses. Now look at the other level, the females. Here you see that 24.9% has a negative and 75.1% a positive attitude. Actually, these two patterns are very similar. As you can see, within the group with a negative attitude we see a slightly higher proportion of male (53.6%) than female participants (46.4%). The reverse holds for the participants with a positive attitude (43.6% male versus 56.4% female). These differences are small, but maybe they are significant.
We ran a Chi-Square test because the variables involved are both on a nominal level (see the decision chart in Figure 10.9). You obtain the chi-square statistic if you tick the box indicating it after you clicked on STATISTICS in the CROSSTAB dialog box. Let us see what the results are.
| Chi-Square Tests | |||||
|---|---|---|---|---|---|
| Value | df | Asymp. Sig. (2-sided) | Exact Sig. (2-sided) | Exact Sig. (1-sided) | |
| Pearson Chi-Square | 5.570 a | 1 | .018 | ||
| Continuity Correction b | 5.173 | 1 | .023 | ||
| Likelihood Ratio | 5.563 | 1 | .018 | ||
| Fisher’s Exact Test | 0.21 | 0.12 | |||
| Linear-by-Linear Association | 5.562 | 1 | .018 | ||
| N of Valid Cases | 667 | ||||
0 cells (.0%) have expected count less than 5. The minimum expected count is 89.24.
b.Computed only for a 2x2 table.
As you can see, the results are significant. You can conclude that the participants are not equally distributed over the 4 cells. You may want to use these numbers (in the first table) for a graph. Presenting these results, you may consider focusing on the division of males versus females in the two groups of participants with either a positive or a negative attitude, because this will make it clear what the significant result of the Chi-Square test mean. Look at the two figures below and decide for yourself.
In the choices you make representing the data, try to accentuate what the contrasts are rather than the similarities. The Chi-Square test showed that there is such a contrast in the data. The figures you use to illustrate the findings should too.
Two final notes: we did not use the absolute number of males and females mentioned in the four cells because the totals are not equal (310 male and 357 female students). Using percentages makes it possible to compare the cells. Second, note that the y-axis runs from 0–60% (and 0–80 in the graph to the left). This makes the effect look more dramatic than it actually is. You may consider adjusting this for your own report.