 Is there a significant difference in the average number of shoes owned by men and women? You can do an independent samples T test to find out. When you configure the test, make sure you have the correct table and then average the number of shoes grouped by gender. When you show the results, you get a P value of 0.105. That's not a significant difference. Take a look at the statistics. You'll see there's one woman with 200 pairs of shoes and one man with 75 pairs of shoes. You can see those as the outliers in the graph. Is this an error on the part of the respondents or is it really good data? There's no way to tell, so you can't just delete the data as you did with the person who was 6 foot 72 inches tall. You knew that was bad data. What would the results look like without those people? You can find out by setting up a filter to specify which values you want to analyze. Select filter and say that you only want data where the number of shoes is less than 70. And then apply. Notice that SOFA lets you know that your results table is being filtered. Now show the results again. This time the P value is less than 0.001, which is a very significant difference. And if you look at the distributions, you will see that those large values are no longer there. An important point here. If you were writing a research paper, you would not include just the significant results. Instead, you'd present the analysis of all the data, including the large values, then point out that they're there, and then present the second set of results. To remove a filter, click filter again, and then remove. And you'll be working with all the data in your table again.