how to exclude outliers in spss

How to Exclude Outliers in SPSS

By Kevin Anderson
AD

Outliers in statistical data can significantly skew results and lead to incorrect inferences. Excluding them is a common practice in SPSS, and this guide will walk you through the process with step-by-step instructions, ensuring your data analysis remains robust and reliable.

Before we dive into the methods of excluding outliers, it's crucial to understand what outliers are and why they pose a problem. An outlier is a data point that significantly differs from other observations in the dataset. They can occur due to various reasons, such as data entry errors, natural variability, or they may represent completely new phenomena. Regardless of the cause, their presence can distort statistical analysis, leading to misleading results.

Detecting Outliers in SPSS

Before excluding outliers, you need to identify them. SPSS offers several methods to detect outliers, including Z-score, modified Z-score, and box plots. Let's explore two common methods:

AD
Testing Linear Regression Assumptions in SPSS
Testing Linear Regression Assumptions in SPSS
8 methods to find outliers in R (with examples)
8 methods to find outliers in R (with examples)
5 Awesome tips and tricks to write an SPSS Assignment
5 Awesome tips and tricks to write an SPSS Assignment

1. **Z-Score**: The Z-score measures how many standard deviations an observation is from the mean. A Z-score greater than 3 or less than -3 generally indicates an outlier.

Calculating Z-Score in SPSS

SPSS doesn't have a built-in Z-score function, but you can calculate it manually using the following steps:

SPSS Tutorial (for Beginners): Intro to SPSS - Statistics How To
SPSS Tutorial (for Beginners): Intro to SPSS - Statistics How To
An evaluation of estimative uncertainty in large language models - npj Complexity
An evaluation of estimative uncertainty in large language models - npj Complexity
  1. Compute the mean (average) of your variable: Analyze → Measure → desserves → Means.
  2. Compute the standard deviation (SD) of your variable: Analyze → Measure → desserves → Statistics → Standard Deviation.
  3. Compute the Z-score: Trans → Compute Variable → Z-score = (Your variable - Mean) / SD.

2. **Box Plots**: Box plots are a useful visual tool that can quickly identify outliers. They are defined by the minimum, lower quartile, median, upper quartile, and maximum values. Any point beyond the whiskers (commonly set at 1.5 x IQR) is considered an outlier.

Creating a Box Plot in SPSS

To create a box plot, follow these steps: Graphs → Legacy Dialogs → Box Plot. Select your variable and click "OK". The box plot will help you visualise and detect outliers.

AD

Excluding Outliers in SPSS

Once you've identified the outliers, it's time to exclude them from your dataset. Here's how you can do it using both the Z-score and Box Plot methods:

**Using Z-Score**: After calculating the Z-score as described earlier, you can use a filter to exclude observations with a Z-score greater than 3 or less than -3. Here's how:

  1. Click Data → Select Cases.
  2. In the dialog box, click "If condition", then "If", and enter: Z-score > 3 or Z-score < -3.
  3. Click "Continue", then "OK". Only the observations that do not meet the condition (i.e., are not outliers) will be selected.

**Using Box Plot**: After creating a box plot, note down the values of your outliers. Then, follow these steps to exclude them:

AD
  1. Click Data → Select Cases.
  2. In the dialog box, click "If condition", then "If", and enter: Your variable < outlier minimum value or Your variable > outlier maximum value.
  3. Click "Continue", then "OK". Only the observations that do not meet the condition (i.e., are not outliers) will be selected.

Finally, always remember that excluding outliers should be done thoughtfully, as they can sometimes carry valuable information. It is essential to understand why you're excluding them and document your decisions for future reference. By learning how to exclude outliers in SPSS, you're equipping yourself with a crucial skill for data analysis.

AD