# How can I tell if outliers are influencing my data?

### From PsychWiki - A Collaborative Psychology Wiki

**How can I tell if outliers are influencing my data?**- A quick way to see if outliers are influencing the findings from your data analysis is to look at the "5% Trimmed Mean." The "5% Trimmed Mean" is the same calculation as for the "Mean" except the top and bottom 5% of the data are first removed. For some statistical software, like SPSS, descriptive statistics will provide the "Mean" and "5% Trimmed Mean" for your variables. The difference between the two tells you how strongly, if at all, extreme scores or outliers are distorting your data. - See the following descriptive statistics for variable1 and variable2. Notice how the frequency distributions show variable1 to be relatively normal but variable2 to be positively skewed with a possible outlier. The descriptive statistics parallel this information by showing how the "5% Trimmed Mean" for variable2 is smaller, and more representative of the data, then the "mean". The larger the difference between the "mean" and "5% Trimmed Mean", the stronger the influence of possible outliers.
</td>

- A more accurate way to see if outliers are influencing your data is to analyze your data with the outliers and without the outliers. Are the findings different? How much different? Although this is more accurate, it is also more time-consuming because you have to conduct twice the number of analyses.

- A quick way to see if outliers are influencing the findings from your data analysis is to look at the "5% Trimmed Mean." The "5% Trimmed Mean" is the same calculation as for the "Mean" except the top and bottom 5% of the data are first removed. For some statistical software, like SPSS, descriptive statistics will provide the "Mean" and "5% Trimmed Mean" for your variables. The difference between the two tells you how strongly, if at all, extreme scores or outliers are distorting your data. - See the following descriptive statistics for variable1 and variable2. Notice how the frequency distributions show variable1 to be relatively normal but variable2 to be positively skewed with a possible outlier. The descriptive statistics parallel this information by showing how the "5% Trimmed Mean" for variable2 is smaller, and more representative of the data, then the "mean". The larger the difference between the "mean" and "5% Trimmed Mean", the stronger the influence of possible outliers.

◄ Back to Analyzing Data page