How to Adjust for Confounding Variables Using SPSS
Social scientists use the SPSS (Statistical Package for the Social Sciences) computer program to analyze data. These scientists have an independent variable, for example a man or a woman as a defendant in a trial. They ask participants to respond to a question, such as how guilty is the defendant (the dependent variable). Researchers try to eliminate confounding variables; anything except the independent variable that might influence the dependent variable. For example, the gender of respondents could influence their responses. Researchers control for confounding variables by using an ANCOVA (Analysis of Covariance).
Things You'll Need
Go to "Datasheet" in SPSS and double click on "var0001." In the dialog box, enter the name of your first variable, for example the sex (of the defendant) and hit "OK." Enter the data under that variable.
Click on "var0002" and enter the name of your second variable, for example level of guilt, and click on "OK." Enter the data under that variable.
Click on "var0003" and enter the name of your confounding variable, for example the sex (of participant) and click on "OK." Enter the data under that variable. (See Reference 2)
Go to "File" to save the data. When the menu drops down, select "SaveAS" and enter a file name in the box.
Analyze the Data
Click on "Analyze" at the top of the SPSS screen. Select "General Linear Model" from the drop-down menu. Then select "Univariate."
In the dialog box, highlight your independent variable (sex of defendant) and click the arrow pointing right to put it in the "Fixed Factor" box. Then, highlight your dependent variable (level of guilt) and click on the arrow pointing right to put it in the "Dependent Variable" box.
Highlight the confounding variable (sex of participant) and click the arrow to put it in the box "Covariate." Click "Continue."
Read the Ouput
Review the output that will show a table labeled, "Tests of Between Subjects Effects." Observe that the dependent variable is shown at the top of the table.
Go down the list of sources in the far-left column until you see your independent variable.
Follow the row for your independent variable to the right to find "df" (degrees of freedom). Degrees of freedom are calculated by subtracting 1 from the number of levels of the independent variable. In the example, since you have two levels (male and female), the df is 1.
Continue moving to the right on the independent variable row to find the "F" value (the test statistic) and the "Sig" (Significance). Significance is the most important number in the output. This tells you whether the results are due to the independent variable or to chance. A significance level of less than 5 percent is generally accepted. This would mean that there are five possibilities out of 100 that the results are due to chance when the confounding variable (sex of participant) is controlled.