SPSS: Tests

Navigate to Analyze > Nonparametric Tests > Legacy Dialogs > Binomial. Click on the variable in the left-hand column that you would like to test, and use the arrow in the middle to move it into the Test Variable List. Next to Test Proportion, enter the expected proportion for the variable encountered first--that is, the variable in the first row of data.  If your variable is not dichotomous, you can divide it into two groups using the Cut point field under Define Dichotomy.

The p value is found in the circled area. If the test proportion is not .5, it will give you 1-tailed significance, but it will display in the same place in the output.

Navigate to Analyze > Nonparametric Tests > Legacy Dialogs > Chi Square. Click on the variable in the left-hand column that you would like to test, and use the arrow in the middle to move it into the Test Variable List. If there are infrequent values you want to remove from your test, you can set a range under Expected Range. Under Expected Values, you can indicate the expected proportions you are testing. If you are testing whether each value is equally represented, choose "All categories are equal." If you expect different frequencies for each value, you can enter these counts one at a time using the Values field and the Add button, in order from the lowest value to highest.

The p value is found in the circled area.

• Navigate to Analyze > Descriptive Statistics > Crosstabs. Use the arrow in the middle to move one of your categorical variables into Row(s) and one into Column(s).
• Click on the Statistics... button and select Chi-square in the upper left, then Continue and then OK.

The p value is in the circled area.

Navigate to Analyze > Descriptive Statistics > Crosstabs. This test is performed exactly like the Chi square contingency test above, but will only be in the results if the selected variables each have only two values (dichotomous).

The p value is in the circled area.

Navigate to Analyze > Compare Means > One-Sample T Test. Move the variable you want to test to the Test Variable(s) list, and enter the expected mean in the field next to Test Value.

The p value is in the circled area.

• The test requires a hypothesized median, so the first thing to do is remove those cases in which the variable is equal to this hypothesized median. We can do this by going to Data > Select Cases, selecting the variable we will test, checking the button next to "If condition is satisfied," then clicking on the button "If..."
• In this menu, we create an expression for the cases that we want to include. For example, if the hypothesized median for the propblue variable was 0.52, we would enter the expression for propblue is not equal to 0.52, or in SPSS-land, propblue ~= 0.52. You can either type this in the center box, or use the buttons to enter, then click on Continue.
• Now navigate to Analyze > Nonparametric Tests > Legacy Dialogs > Binomial. The rest is the same as the binomial test, with the hypothesized median entered as the cut point, and with a test proportion of 0.5.

The p value is in the circled area.

• Navigate to Analyze > Compare Means > Independent-Samples T Test. Use the arrows to select a Test Variable (dependent variable), and Grouping Variable, which indicates which of two samples an observation belongs in.
• You need to tell SPSS which group is which. Click on the Grouping Variable so it is highlighted in yellow, then click on the Define Groups... button. If you don't know the values of your grouping variable, see instructions for creating a frequency table.
• Enter in a value for Group 1 and Group 2, under "Use specified values." (If you don't know what the values of the variable are, you can make a frequency table using Analyze > Descriptive Statistics > Frequencies). Then click Continue, then OK.

Notice that the grouping variable will have values within the parentheses, rather than question marks.

Move one variable under Variable1, and the other under Variable2, on one line of the Paired Variables sections.

The p value is in the circled area.

• Move the continuous variable into the Dependent List, and the grouping variable into the Factor field, using the arrows in the center of the menu.
• For the Tukey-Kramer test, click on Post Hoc... and check the box next to Tukey. Then click Continue, then OK.

The p value is in the circled area.

• Navigate to Analyze > Compare Means > One-Way ANOVA. Move the continuous variable into the Dependent List, and the grouping variable into the Factor field, using the arrows in the center of the menu. Then click on Options...
• Check the box next to Welch. Then click Continue, then OK.

The p value is in the circled area.

• Navigate to Analyze > General Linear Models > Univariate. Move your dependent (numerical) variable in the Dependent Variable list, and your two independent (categorical variables) in the Fixed Factors list.
• Click on the Options... button, then move everything under Factors and Factor Interactions to Display Means For. Check the box under Display for Descriptive Statistics and Homogeneity Tests. (This will give you Levene's homogeneity of variance test.) Click continue, then OK.

The circled areass, under "Sig." indicate whether either of the independent variables, or their interaction, has a statistically significant effect on the dependent variables.

• Navigate to Analyze > Nonparametric > Legacy Dialogs > 2 Independent Samples. (Mann-Whitney U is the default Test Type.) Use the arrows to select a Test Variable (independent variable), and Group Variable, which indicates which of two samples an observation belongs in.
  
  You also need to tell SPSS which group is which. Click on the Grouping Variable so it is highlighted in yellow, then click on the Define Groups... button. 
• Enter in a value for Group 1 and Group 2, under "Use specified values." Then click Continue, then OK. If you don't know the values of your grouping variable, first see instructions for creating a frequency table.

Notice that the grouping variable will have values within the parentheses, rather than question marks. Make sure that Mann-Whitney U is checked under Test Type.

The p value is in the circled area.

• Navigate to Analyze > Nonparametric > Legacy Dialogs > K Independent Samples. (Kruskal-Wallis should be the default test.) Use the arrows to select a Test Variable (independent variable), and Grouping Variable, which indicates which sample an observation belongs in.
• You need to tell SPSS which group is which. Click on the Grouping Variable so it is highlighted in yellow, then click on the Define Range... button. 
• Enter in values for Minimum and Maximum, under "Range for Grouping Variable." These are the lowest and highest values of the grouping variable that you want to include. If you don't know the possible values, you can create a frequency table of your grouping variable using Analyze > Descriptive Statistics > Frequencies. Then click Continue, then OK.

Notice that the grouping variable will have values within the parentheses, rather than question marks. Make sure that Kruskal-Wallis H is checked under Test Type.

The p value is in the circled area. The test statistic is the number next to "chi square."

• Navigate to Analyze > Regression > Linear. Use the arrow buttons to move your dependent and independent variables into their respective fields, then click OK.
• To check for heteroscedasticity, before running the linear regression click on the Plots... button in the upper right. You can make a scatterplot of the residuals and the predicted values of your regression by moving *ZRESID (standardized residuals) into the Y field, and *ZPRED (standardized predicted values) into the X field, then click Continue and OK. The scatterplot will be created along with the statistical output.
• To add a horizontal line at 0 after you've made this chart, double-click on the chart in the Output window to open up the Chart Editor. In the Chart Editor, click on Options > Y Axis Reference Line. This opens the Properties window; from here, under the "Reference Line" tab, enter a 0 in the Position field. Then click Apply and the line will appear in your chart. Close the Chart Editor and you will see that the chart in the Output window now has the horizontal line. (This works for other charts as well.)

Navigate to Analyze > Correlate > Bivariate. Move your two variables into the Variables box, using the arrow button. Make sure that the box next to "Pearson" is checked under Correlation Coefficients, and click on the box next to "Flag significant correlations."

The output is contained in four boxes. The top right and bottom left boxes should be identical, and contain the useful information. The p value is in the circled area, and the test statistic, r in this case, is shown next to "Pearson Correlation."

Navigate to Analyze > Correlate > Bivariate. Move your two variables into the Variables box, using the arrow button. Make sure that the box next to Spearman is checked under Correlation Coefficients, and click on the box next to "Flag significant correlations."

The output is similar to the output shown above for linear correlation.

• Navigate to Analyze > General Linear Model > Univariate. Use the arrow button to move the dependent variable into the Dependent Variable field. Then move categorical (scale) variables into the Fixed Factor(s) field, and continuous numeric independent variables into the Covariate(s) field.
  The default settings will create interaction terms between all combinations of the fixed factors, but none between the covariates. To specify the interaction terms that you want, click on the Model... button.
• In order to specify your own interaction terms, in the Model menu under "Specify Model," click on the button next to Custom.
  To add a variable as a main effect (not an interaction between variables), click on the variable in the Factors & Covariates list, select "Main effects" under Build Term(s) Type, and then click on the arrow to move it into the Model box. You can add multiple variables as the same time by holding down the Ctrl key while you click on the variables in the Factors & Covariates list.
  To add interaction terms, you must select both variables while holding down the Ctrl key, then select "Interaction" under Build Term(s) Type, then click on the arrow to move the variables over to the Model box. You can tell the variables are entered as interaction terms because they will look like Height*Weight.
• Click Continue and then OK.