Introductory remarks about Chi-Square.
(Explanation)
1. Independence of Observations. Each observed case is independent of each other; in other words, it is from a different person. A chi-square would be inappropriate if a person could have responses in more than one category or could have more than one contribution to a single category. You would not use a chi-square for a repeated-measures design.
2. Size of Expected Frequencies. Many researchers believe that a chi-square test should not be performed when the expected frequency of any cell is less than 5; this is a controversial topic. You can see the expected frequency for each cell under the "Expected N" column in your output.
In this example (from the Social Psych survey), we asked Reedies whether they wanted to go to pursue academia after Reed or not; they could answer Yes, No, or Not Sure. We then asked them what percentage of Reed students would answer Yes, No, and Not Sure (with numbers adding to 100%). Here, we compare whether the distribution of people's responses about themselves pursuing academia is different than the distribution of what people think the distribution is: on average, Reedies think that 40% would say Yes, 36% No, and 24% Not Sure.
1) Go to "Nonparametric tests" under the "Analyze" menu.
2) Select "Chi-Square..." from the list.
3) Enter your categorical variable into the "Test Variable List" box, as shown in Figure 6.1.
Figure 6.1 Chi-Square Tests for Goodness of Fit
4) You may choose to test whether all categories have the same number or percentage of responses in them; in this case, leave the Expected Values as "All Categories Equal." If you have specific expected values, enter them as shown in Figure 6.2; make sure that the order is correct.
Figure 6.2 Setting the Expected Values for Chi-Square Tests
5) To get descriptive statistics for the frequencies of the levels of your variable, you can go to the "Analyze" menu, then to "Descriptive Statistics,"and select "Frequencies."
Explanation (Figure 6.3)
Figure 6.3 Caption
Explanation (Figure 6.4)
Figure 6.4 Caption
Figure 6.5 shows the table you would get by running descriptive statistics for frequencies (step 5).
;
Figure 6.5 Descriptive Statistics for Frequencies
Be sure to include: the number of participants in the study, the percentages of those participants per group, and whether the groups are different. At the end, include the chi-square value, the df, the number of participants, and your p-value.
(explanation)
1. Independence of Observations. Each observed case is independent of each other; in other words, it is from a different person. A chi-square would be inappropriate if a person could have responses in more than one category or could have more than one contribution to a single category. You would not use a chi-square for a repeated-measures design.
2. Size of Expected Frequencies. Many researchers believe that a chi-square test should not be performed when the expected frequency of any cell is less than 5; this is a controversial topic. You can see the expected frequency for each cell under the "Expected N" column in your output.
1) Go to "Descriptive Statistics" under the "Analyze" menu.
2) Select "Crosstabs" from the list.
3) Assign one variable to the rows and one variable to the columns. In Figure 6.6, we are testing whether or not gender and sleep schedule are related.
Figure 6.6 Crosstabs screen
4) Click on the "Statistics" button.
5) Check the box marked "Chi-square" (Figure 6.7) and then click "Continue."
Figure 6.7 Crosstabs Statistics Window
6) You can also choose to have SPSS report the expected values for each cell by clicking "Cells..." in the "Crosstabs" window and then checking the box marked "Observed" (Figure 6.8).
Figure 6.8 Crosstabs Cell Display Window
7) Once you have set the "Cells" and "Statistics" to your specifications, click "OK."
Be sure to include: the pecentages of those participants per group, and whether or not the groups are different. At the end, indicate the chi-square value, the df, the number of participants, and your p-value.