Signal detection theory is a theory in mathematics, statistics and psychology primarily concerned with the descriptive and normative theories of discriminating between signals and noise. In this field, the receiver operating characteristic (ROC) is an important concept, as it allows researchers to plot correct detections versus false positives. SPSS, a powerful piece of statistical software, is capable of plotting such a curve for a researcher's data.
Check and arrange your data. To use ROC, your data must be in the proper form. You will need at least the following variables: "detection type" (a list of the tests or devices used in detection), "detected signals" (1 represents detection and 0 represents lack of detection), and "count" (the number of data points for each test/detection combination). Arrange the data for these three variables in columns, not rows.
Enter the data into SPSS. Open SPSS and select "File" from the menu above. Choose "open" and select the file containing your data set.
Weight cases by "count." SPSS cannot differentiate between whether the data for "count" is representative of a single data point or an accumulation of data points. Thus, you must explicitly tell SPSS that "count" represents more than one data point. Choose "data" at the top menu. Select "weight cases" and a new menu will appear. Click the button to the left of "weight cases by." Highlight "count" and click the arrow below "weight cases by." "Count" will appear under "frequency variable." Click "ok."
Employ the ROC curve. Choose "analyze" from the top menu. Select the "ROC curve" option. Highlight "detection type" and click the arrow beside the box under "test variable" to place "detection type" into this box. Highlight "detected signals" and click the arrow next to the box under "state variable" to place "detected signals" into this box. Type "1" in the box next to "value of state variable." Click "ok" and the ROC curve will appear.