Outliers in SPSS: Overview
SPSS uses Tukey’s method to identify outliers, which are visible on boxplots. Finding outliers using that method can be thought of as a rule of thumb, not a statistical rule set in stone. While there are more advanced methods for finding outliers, like Grubb’s Test, SPSS doesn’t offer them. So you’ll want to take the results from Tukey and compare them with a histogram . Then you can make a conclusion based on what is best for your set of data.
Watch the video for the procedure:
Can’t see the video? Click here.
Outliers SPSS: Steps
Step 1: Click Analyze.
Step 2: Choose Descriptive Statistics.
Step 3: Click Explore.
Step 4: Move the variable you want to analyze for outliers into the Dependent list box.
Step 5: Click OK
Step 6: Scroll down the list of results to view the boxplot. SPSS will mark any outliers with a circle. Far outliers, which are more likely to be true outliers, are marked with a star. Next to these icons will be a number; The number corresponds to the numbered item in your Variable View list. Exit the results sheet, return to the Variable view screen, and identify the questionable data point.
Comparison with a Histogram
Next, you’ll want to look at a histogram to confirm the result.
- Click Analyze.
- Choose Descriptive Statistics, then Frequencies.
- Move the variable of interest into the Variables box.
- Click charts, and make sure the histogram option is selected.
- Click Continue, then click OK.
- Scroll down the list of results to view your histogram.
“True” outliers will be obvious on a histogram (either to the far right or the far left), as well as being starred (or circled) on a boxplot. As each dataset is different, striking off outliers is usually a judgment call, based on you—the researcher—who usually knows their data the best.